Document ymbE2bbnYMjZXw7MyRZvgeZ5r

1 UNITED STATES DISTRICT COURT FOR THE DISTRICT OF MASSACHUSETTS 2 3 ) 4 BRIAN K. MILWARD, et al, ) ) 5 Plaintiffs, ) ) Civil Action 6 v. ) No. 07-11944-GAO ) 7 ACUITY SPECIALTY PRODUCTS ) GROUP, INC., et al, ) 8) Defendants. ) 9) 10 11 BEFORE THE HONORABLE GEORGE A. O'TOOLE, JR. UNITED STATES DISTRICT JUDGE 12 13 DAUBERT HEARING - DAY 3 14 15 John J. Moakley United States Courthouse 16 Courtroom No. 9 One Courthouse Way 17 Boston, Massachusetts 02210 Thursday, April 23, 2009 18 9 a.m. 19 20 Marcia G. Patrisso, RMR, CRR Official Court Reporter 21 John J. Moakley U.S. Courthouse One Courthouse Way, Room 3510 22 Boston, Massachusetts 02210 (617) 737-8728 23 Mechanical Steno - Computer-Aided Transcript 24 1 1 APPEARANCES: 2 ALLEN STEWART, P.C. By: Allen Stewart, Esq. 3 Steve Baughman Jensen, Esq. 325 N. St. Paul Street - Suite 2750 4 Dallas, Texas 75201 - and 5 KREINDLER & KREINDLER, LLP By: James D. Gotz, Esq. 6 277 Dartmouth Street Boston, Massachusetts 02116 7 On Behalf of the Plaintiffs 8 NIXON PEABODY LLP By: Joseph J. Leghorn, Esq. 9 Nancy C. Antonellis, Esq. 100 Summer Street 10 Boston, Massachusetts 02110 On Behalf of the Defendant CRC Industries, Inc. 11 WEATHERS, RILEY & SHEPPEARD, LLP 12 By: Wesley A. Weathers, Esq. 4848 SW 21st Street - Suite 202 13 Topeka, Kansas 66604 On Behalf of the Defendant NCH Corporation 14 FORMAN, PERRY, WATKINS, KRUTZ & TARDY, LLP 15 By: Tim Gray, Esq. 200 South Lamar Street - Suite 100 16 Jackson, Mississippi 39201-4099 - and 17 MELICK, PORTER & SHEA, LLP By: Raymond H. Tomlinson, Jr., Esq. 18 28 State Street - 22nd Floor Boston, Massachusetts 02109-1775 19 On Behalf of the Defendant Aristech Chemical Corporation 20 CETRULO & CAPONE, LLP By: Lawrence G. Cetrulo, Esq. 21 Kevin R. Powers, Esq. Jeffrey D. Adams, Esq. 22 Two Seaport Lane - 10th Floor Boston, Massachusetts 02210 23 On Behalf of the Defendants Henkel Corporation; Sunoco, Inc.; and Sherwin-Williams Company 24 25 2 1 DICKIE, McCAMEY & CHILCOTE, P.C. By: James R. Miller, Esq. 2 Two PPG Place - Suite 400 Pittsburgh, Pennsylvania 15222-5402 3 On Behalf of the Defendant Sherwin-Williams Company 4 CAMPBELL, CAMPBELL, EDWARDS & CONROY, P.C. By: Brian P. Voke, Esq. 5 One Constitution Plaza Boston, Massachusetts 02129 6 On Behalf of the Defendant Berryman Products, Inc. 7 CAMPBELL, CAMPBELL, EDWARDS & CONROY, P.C. By: Charles K. Mone, Esq. 8 One Constitution Plaza Boston, Massachusetts 02129 9 On Behalf of the Defendant The Clorox Company 10 ADLER, POLLOCK & SHEEHAN, P.C. By: Brian R. Birke, Esq. 11 175 Federal Street Boston, Massachusetts 02110 12 On Behalf of the Defendant Radiator Specialty Company 13 GORDON & REES, LLP By: Leslie A. Sheehan, Esq. 14 633 West Fifth Street - Suite 4900 Los Angeles, California 90071 15 On Behalf of the Defendant Sunnyside Corp. 16 McDONOUGH, HACKING & LAVOIE, LLC By: William R. Corino, Esq. 17 One Washington Mall Boston, Massachusetts 02108 18 On Behalf of the Defendant Boyle-Midway, Inc. 19 RYAN, COUGHLIN & BETKE, LLP By: Elizabeth A. Doubleday, Esq. 20 175 Federal Street Boston, Massachusetts 02111 21 On Behalf of the Defendant WD-40 Company 22 TUCKER, HEIFETZ & SALTZMAN, LLP By: Syd A. Saloman, Esq. 23 Three School Street Boston, Massachusetts 02108 24 On Behalf of the Defendant AGS Company 25 3 1 INDEX 2 DIRECT CROSS REDIRECT RECROSS WITNESSES FOR THE 3 DEFENDANTS: 4 DAVID HAY GARABRANT 5 By Mr. Leghorn (Cont'd) 5 80 By Mr. Stewart 32 6 7 JOHN M. BENNETT 8 By Mr. Weathers 84 136 By Mr. Stewart 114 137 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 4 1 PROCEEDINGS 2 THE CLERK: All rise. 3 Continuation of the Milward Daubert hearing. 4 Please be seated. 5 MR. LEGHORN: Good morning, your Honor. 6 THE COURT: Good morning. 7 MR. LEGHORN: If I may ask Dr. Garabrant to resume the 8 stand? 9 THE COURT: Please. 10 DAVID HAY GARABRANT, resumed 11 MR. LEGHORN: Your Honor, I think I'm on the same 12 input. I don't know if you're getting it or not. 13 THE COURT: I have it. Do you want everyone to have 14 it? 15 MR. LEGHORN: Yes, your Honor. Please. 16 CONTINUED DIRECT EXAMINATION 17 BY MR. LEGHORN: 18 Q. Good morning, Dr. Garabrant. 19 A. Good morning. 20 Q. We ended up yesterday after the break in the court talking 21 about the Golomb study. Other studies that Dr. Smith mentioned 22 were studies done by Mele, M-E-L-E. Did you look at the Mele 23 studies? 24 A. Yes. 25 Q. And what were your observations about the Mele studies 5 1 from looking at them? 2 A. I think I prepared a slide on that. Okay. 3 Q. Who is Mele, by the way? 4 A. Well, Mele is part of a group of investigators in Italy. 5 The -- I believe he's with the Italian Leukemia Study Group, 6 and they've published a number of papers, epidemiology studies, 7 looking at leukemia. 8 Q. And did you look at those studies? 9 A. I did. 10 Q. And your observations about those studies, Doctor? 11 A. Well, the first point that's worth putting -- making is 12 that one of the Mele publications, this one published in the 13 American Journal of Epidemiology in 1994, provides some of the 14 history of benzene use in Italy. And I've just taken a clip 15 from that. It basically says in Italy the use of glues with a 16 benzene content greater than 2 percent was prohibited by law in 17 1963. And then he goes on to say, "In our study, only two 18 cases out of every 14 leukemia or RAEB cases who worked as a 19 shoemaker began such work after 1963. Thus, significant 20 exposure to benzene may have occurred, for most cases, before 21 1963, providing further evidence of a role for exposure to 22 organic solvents." 23 Okay. So that's important history when we come to the 24 study at issue. 25 Q. Was this a study that Dr. Smith referenced? 6 1 A. Well, he didn't reference this article; he referenced the 2 next article by Mele. 3 Okay. So this is the one that Dr. Smith referenced. And 4 this was published in 1995, but it's the same group of 5 investigators. And what this article showed was that among APL 6 cases, the center column, 5.7 percent of them were shoemakers 7 compared to controls, the left-hand column, in which 1.5 8 percent were shoemakers. And then Mele calculates an odds 9 ratio of 6.3, which is statistically significant. 10 Now, I have two points to make regarding that: This is 11 actually a very small number of observations. Of the 36 APL 12 cases, 5.7 percent represents two cases. So the observation's 13 based on two cases. And it's important, if you go back and 14 look at both of the Mele papers, to recognize that the cases 15 were identified in 1987 to 1990. That actually matters, okay? 16 And then if you would click. 17 Okay. So if we recall from the other paper that benzene 18 in glues was restricted in 1963. What's really going on in 19 this paper is that substantial exposure was limited to a period 20 at least 24 years prior to diagnosis, all right? Because the 21 cases were diagnosed in '87 to '90, benzene use was restricted 22 in glues in the shoemaking industry 24 years before that. And 23 so this is inconsistent with a topoisomerase-induced mechanism, 24 which Dr. Eastmond's paper says would act within two to 15 25 years of exposure. 7 1 Q. Doctor, a paper that -- do you have any other observations 2 on the paper by Mele? 3 A. No. 4 Q. Doctor, one of the other papers that Dr. Smith talked 5 about was the paper by -- Lois Travis published, that she 6 published concerning the study of the Chinese workers; is that 7 correct? 8 A. Yes. 9 Q. Did you have an opportunity to review that? 10 A. I did. 11 Q. Do you have any observations about that paper and Dr. 12 Smith's analysis? 13 A. Yes, I do. 14 Q. And what are they? 15 A. If you could go to the next slide. 16 The Travis paper reported four cases of ANLL M3, as Dr. 17 Smith's slide here points out, and they are -- if you look at 18 the far right -- far left-hand column -- they are patients 19 Number 5, 6, 7 and 8, okay? So in that paper they're recorded 20 as M3, or APL. If you would click. 21 Okay. I went through the tables and the text in the 22 Travis study to try to figure out how they made these diagnoses 23 of M3. And what you see here is this little table I've 24 prepared that summarizes the data from the Travis paper. Look 25 at -- for Cases 5, 6, 7 and 8, looking at the peripheral blood 8 1 smear, the percent blasts, the peripheral blood smear 2 morphology, the bone marrow percent blasts, the comments 3 regarding what the bone marrow looked like, the cytogenics, 4 immunophenotype, and the white blood cell count in the 5 peripheral blood. 6 And when you -- it's a little hard to get all of this out 7 of the tables, but when you put it together what you realize is 8 that for Case 5 they didn't have any bone marrow examination; 9 it appears that the diagnosis was based entirely on the 10 peripheral blood smear morphology, which showed hypogranular 11 promyelocytes. So it's my impression that that's not adequate 12 to establish a diagnosis of APL. So that one, I think, is in 13 serious question, and there's little reason to think that 14 that's an APL. 15 Cases 6, 7 and 8 they did have bone marrow, and it's my 16 understanding that the bone marrow was consistent with APL, but 17 they did not have cytogenics or immunophenotyping on any of the 18 these cases. And so this is not -- to the extent I'm aware, 19 this is not adequate to really establish that these are APL 20 cases, although Cases 6, 7 and 8, that's a reason -- I believe 21 that's a reasonable decision, but that's certainly not a 22 rigorous way of determining the diagnosis. 23 So -24 Q. Okay. 25 A. -- one of these cases is unlikely to be an APL and the 9 1 other possibly or probably are. 2 Q. Doctor, in your answer you mentioned "phenotyping." What 3 does phenotyping -- what information does that give you? 4 A. Well, immunophenotyping tells you about cell surface 5 markers that helps you to identify the cell types. 6 Q. And what typing gives you the genetic karyotype? 7 A. Well, for APL, the -- a very high proportion of the cases 8 have the translocation on Chromosomes 15 and 17 -- the 9 t(15;17). 10 Q. Is that the cytogenetic testing that would give you that 11 information? 12 A. Yes. 13 Q. Did you have any further observations with regard to the 14 Travis paper and Dr. Smith's analysis? 15 A. Yes. 16 If you would go to the next slide. 17 Okay. So now the question is Dr. Smith's handling of the 18 data in the Travis paper. Recall in the Golomb study Dr. Smith 19 excluded one of the APL cases because the karyotype was normal. 20 Remember, one of them didn't have the t(15;17), didn't have any 21 abnormalities to 17. Dr. Smith excluded that case. Now in the 22 Travis study he includes all four of these APL cases, and none 23 of them had any karyotype done. In addition, as we're going to 24 see on the next few slides, he actually creates five more APL 25 cases that were not known to exist; they were not reported in 10 1 that study. And, of course, since they weren't in the study 2 and didn't exist, as far as we know, they didn't have any 3 karyotype, and they certainly didn't have t(15;17). 4 So my point is that Dr. Smith's arbitrary creating the 5 data and choosing data, excluding a case in Golomb but now 6 including cases that don't fit the rule he used in Golomb, is 7 not a reliable scientific method and it is not a generally 8 accepted methodology in the scientific community, neither in 9 medicine or in epidemiology; he is simply creating data that 10 fits his beliefs. 11 Q. Did you go through -- you mentioned that he created five 12 cases. What did you mean by that? 13 A. I'm going to show you. I want to walk you through his 14 calculations. 15 Okay. Here's his presentation of his calculations. This 16 is from his slide. And, you know, instead of making what are 17 relatively standard calculations of rate ratios, he does some 18 fairly unorthodox things that are based on assumptions that I 19 think are not reliable. 20 Okay. So these bulleted points, I'm going to walk through 21 and show what these actually represent. If you would go to the 22 next slide, please. 23 Okay. So the first thing is to abstract the data from 24 Travis's text in Tables 1 and 2. If you read Travis carefully, 25 what you see is in Table 1 there were a total of 82 cases of 11 1 lymphohemapoietic malignancies among which there were 32 cases 2 of acute leukemia. Travis and her colleagues reviewed the 3 pathology reports and medical records for 51 of the 82 total 4 cases and 17 of the 32 acute leukemias. 5 Now, so they reviewed the diagnostic material from China, 6 all right? In addition to that, they reviewed the 7 histopathologic materials; in other words, they had the bone 8 marrow biopsies, they had the pathologic materials. They 9 reviewed those for 31 of the 82 cases including 14 of the 32 10 acute leukemias. 11 Now, there's a missing case of acute leukemia which can't 12 reconcile with the Travis paper. You see the total of 32 in 13 that first row. Seventeen of them they reviewed the pathology 14 reports and medical records; 14 of them actually had the bone 15 marrow -- the biopsy materials. There's one case unaccounted 16 for. 17 Okay. If you look at the non-exposed workers -- I should 18 have said we were talking about the benzene-exposed workers on 19 the left. 20 If you look now at the right side of the paper with the 21 non-exposed workers, there were a total of 13 total cases of 22 lymphohematopoietic malignancies, six of which were acute 23 leukemias. They reviewed histopathologic materials for three, 24 and they saw one M2. 25 Okay. Now, if you would go to the next slide. 12 1 Q. Doctor, I've got a question here. On the benzene-exposed 2 workers, under the 14 of the acute leukemias where Dr. Travis 3 and her colleagues reviewed the pathology materials, you have 4 another series of numbers: a nine, a four, a four and a one. 5 Tell us what those represent. 6 A. Okay. Those 14 acute leukemias were broken down into the 7 following subcategories: Nine of them were ANLLs, acute 8 nonlymphocytic leukemia; five were acute leukemia not otherwise 9 specified, AL NOS. Within the nine ANLLs, Travis and her 10 colleagues felt that four of them were M2s, four of them were 11 M3s, and one of them was an M4. 12 Okay. Now if you'd go to the next slide. 13 Okay. So here's what Dr. Smith does with that. He 14 assumes that the proportion of all acute leukemias that were M3 15 can be obtained from the cases that had histopathological 16 materials. So in other words, he says let's look at the 14 17 acute leukemias that had histopathology that Travis and 18 colleagues could look at. There were four M3s. So four out of 19 the 14 is 28.6 percent. 20 Okay. Now, I take issue with that assumption. And what 21 he's going to do is he's going to apply that back to the 32, 22 okay? What we have here is 32 acute leukemias. 14 had bone 23 marrow biopsies; the others did not. People get bone marrow 24 biopsies for non-random reasons, such as having access to 25 medical care or better medical care, having an unclear 13 1 diagnosis that needs further evaluation. So it's not a 2 reasonable assumption to think that the people who don't have 3 bone marrow biopsies would have the same distribution of 4 diagnoses as those who do. Furthermore, the diagnoses that 5 were made by the treating physicians based on the pathology and 6 medical records and whatever else they did to make the 7 diagnoses and the review of those diagnoses by Travis I think 8 would take precedence over Dr. Smith's assumption that he can 9 now infer what the diagnoses were. The treating physicians did 10 not reach -- or, say, there's no record that they reached any 11 other diagnoses. 12 So if you would go to the next slide. 13 So what he does is he takes the 28.6 percent. He 14 multiplies it times the 32 acute leukemias and says there must 15 have been a total of 9.2 M3s in the total cohort of 16 benzene-exposed workers. So in essence, where there were only 17 four M3s diagnosed, he's creating 5.2 additional M3s that were 18 not diagnosed and which Dr. Travis and Dr. Lee, who is a 19 hematopathologist at the Mayo Clinic, did not feel -- or I 20 should say made no decision existed [sic]. 21 Okay. So if you would go to the next slide. 22 All right. Now he does the same calculation for the 23 non-exposed workers. He says the proportion of acute leukemias 24 that are M3 must be 28.6 percent. I'll multiply that times the 25 six acute leukemias. That means there must have been 1.7 M3s 14 1 in the non-exposed workers. And again, I take issue with that 2 assumption. I don't think that has any basis. 3 Q. Doctor, did Dr. Travis and her colleagues make an 4 observation about the occurrence of M3 in both the exposed and 5 the unexposed population? 6 A. Well, she doesn't comment on the unexposed, but, yes, she 7 does comment on the four cases of M3 among the benzene-exposed 8 workers. 9 Q. And what was that comment, do you recall? 10 A. I would have to go back and pull the paper to see it. I 11 don't recall. 12 Q. Well, Doctor, do you recall whether she says that the 13 incidence of M3 in the exposed population was similar to the de 14 novo rate in the Chinese population? 15 A. Yeah. Thank you for refreshing my memory. She says on 16 page 99, "Although AML-M3 occurred in at least four patients in 17 this series, its general representation among the subtypes of 18 ANLL was similar to its distribution in de novo ANLL in China," 19 and she references reference 27. And if you go to her 20 bibliography, 27 is a paper by Yang and Zhang in 1991 called 21 "Incident Survey of Leukemia in China." Okay. And we'll come 22 to that paper. 23 So she says, you know, the proportion of M3s among ANLLs 24 in our study was similar to what was seen in China in the Yang 25 paper. Okay. 15 1 Q. Did you do any further analysis of Dr. Smith's? 2 A. Yes. If you would go to the next slide. 3 Okay. Then Dr. Smith does the following calculation: He 4 says, okay. Among the benzene exposed, there were 9.2 M3s out 5 of a total of 74,828 people. And so let's divide that by the 6 1.7 M3s among 35,805 people. And that division yields a ratio 7 of 2.59. So that's Dr. Smith's calculation. And he calls that 8 an odds ratio, which it is not. 9 Q. And why is it not? 10 A. Well, it has nothing to do with odds. It's proportions; 11 it's a different type of calculation. 12 Q. What's wrong with that? 13 A. Okay. So the entire calculation is based on the premise 14 that Dr. Smith can diagnosis APL by imagining that it must have 15 been present among people who were diagnosed and treated by 16 physicians who did not reach this diagnosis and among people 17 who, after careful review of medical records and pathology 18 reports, were not diagnosed with APL by Dr. Travis and her 19 colleagues at the NCI. Travis and her colleagues were 20 responsible for the diagnostic validity in this major 21 epidemiological study. Dr. Smith's creation of these cases is 22 pure speculation; it has no basis in the paper. 23 Q. And does his approach that he's taken here -- is it 24 consistent with methods and techniques used in the field of 25 epidemiology? 16 1 A. No. I have not seen this done. 2 Now, there's an alternative that could have been done that 3 I think is consistent with epidemiologic methods in common use, 4 and that would have been to calculate the number of APL cases 5 that would have been expected in the Travis cohort of benzene 6 workers using the background rates of APL in the general 7 Chinese population, and then it would have been reasonable to 8 compare the observed and expected numbers. And this sort of 9 calculation is routinely done in epidemiology to come up with 10 either standardized incidence ratios or standardized mortality 11 ratios. 12 So the idea is essentially you're saying, well, look, if 13 the benzene cohort had the rates that are seen in the 14 non-exposed population, how many cases would they have gotten? 15 And then you say, okay. Here's how many we saw, compare it to 16 how many we expected to see -- or how many we observed versus 17 how many we expected to see -- and that's essentially a rate 18 ratio. That's routine. And that's essentially what I showed 19 when I taught the epidemiology methods yesterday. 20 Q. Doctor, you mentioned a study by Yang. And I'm going 21 to -22 MR. LEGHORN: Your Honor, this has been tendered to 23 the Court as an exhibit to the brief. I don't know whether you 24 would prefer that we mark it separately here, but I don't know 25 if you have it. 17 1 THE COURT: If it's already in the record, it's not 2 necessary. 3 MR. LEGHORN: Would you like a copy to follow along? 4 THE COURT: I could use it. 5 BY MR. LEGHORN: 6 Q. And, Doctor, you mentioned that you found -- or there was 7 a source of the incidence rates in the general Chinese 8 population; is that correct? 9 A. Yes. 10 Q. And you mentioned a particular paper, correct? 11 A. Yes. 12 Q. And what did you find in that paper? 13 A. Okay. Well, this is the Yang paper that Travis referenced 14 as her Reference 27. And if you go to -- and it's an 15 "Incidence Survey of Leukemia in China" published in 1991. If 16 you go to Table 4 you will see that for M3, which is the fourth 17 row in the table, the incidence rate is .35 per 10 to the 5th. 18 So it's .35 times 10 to the -5, or 3.5 times 10 to the -6, 19 which is 3.5 per million people per year. Okay. And that's in 20 the range that Dr. Smith talked about of three to eight cases 21 per million people per year. So this is an estimate of the 22 rate in China. It's, to my knowledge, the only estimate 23 available. 24 In addition to providing that rate, Yang also reports the 25 percent of ANLL that is M3. And that's 18.7 percent. Now, we 18 1 saw in Dr. Smith's calculations that he assumed it was 28.6 2 percent, and I think that that's simply wrong. 3 Q. Now, Doctor, when Dr. Smith was on the stand, he was 4 critical of using information drawn from the general population 5 because it didn't account for the healthy-worker effect. Does 6 Yang report on the population from which he gathered -- or that 7 they gathered the data? 8 A. Yes. I didn't make a slide of this, but in Table 1 of the 9 Yang paper, Yang lists the different populations that were 10 surveyed in order to come up with this estimate of the 11 incidence of the various types of ANLL. And there were 46 12 different areas in China where they do the survey. They 13 calculate total person years of 60,557,127. So this is a lot 14 of data. 15 And if you look down, what these -- at the list of what 16 these populations are, about -- I don't know -- somewhere 17 between -- I'd say roughly a third are working populations. So 18 they come from the Daqing Oil Field, the Baoutou Steel Company, 19 Yanji Forestry Farm, and so on. So some of these are general 20 populations; some of these are working populations. So I think 21 that's a reasonable way to collect the data. 22 Q. Doctor, can you explain how the healthy-worker effect may 23 influence the results of epidemiological studies? 24 A. Yeah. Okay. The healthy-worker effect is a term that is 25 used to describe the fairly common observation that working 19 1 populations have lower mortality than the general population, 2 at least in the United States and Western Europe. And there's 3 been a fair amount of investigation and thought about why 4 working populations are healthier than the general population. 5 And it varies from population to -- working population to 6 working population. It's largely due to the selection of 7 healthy workers at the time of hire. 8 So in the United States and Western Europe, it was a 9 general practice to hire healthy people and to not hire people 10 who had preexisting diseases, people with low-back problems, 11 people with orthopedic problems, people with heart disease. 12 You know, that was common, okay? Now, that's actually gone 13 away, in part, because of the Americans with Disabilities Act. 14 You can't do that anymore in the United States. But it was 15 done, all right? 16 The real question is, does the healthy -- you know, if 17 we're going to talk about leukemia, and particularly M3, can 18 you select people at the time of hire with any idea who's going 19 to not get APL or acute leukemia? And the answer is: You 20 can't. You can't select people at the time of hire because we 21 don't know what -- we don't know who's at increased risk for 22 leukemia. 23 So the idea there ought to be some adjustment for the 24 healthy-worker effect for leukemia really has no basis. If you 25 were talking about a study of orthopedic problems, then I would 20 1 say, yeah, you really would have to pay attention to the 2 healthy-worker effect; if you were talking about a study of 3 cardiovascular disease, where you can select people who are at 4 lower risk of cardiovascular disease -- and essentially you 5 would be selecting against smokers and people with high blood 6 pressure and people with uncontrolled diabetes and people with 7 hypercholesterolemia -- yes, you can select against people who 8 had increased risk of future cardiovascular disease. But you 9 can't do it for leukemia. So I don't think there is any 10 healthy-worker effect for this diagnosis. 11 Q. Doctor, using the Yang data from this article, did you do 12 any calculations? 13 A. I did. Okay. So I calculated the expected number of APL 14 cases, or actually, I should say the number of APL cases you 15 would expect to occur in the benzene cohort if they had the 16 background rates seen in the general population of China from 17 the Yang paper. So what you do is you take the number of 18 person years in the -- actually, I made another typo -- I 19 apologize -- the number of person years in the benzene cohort, 20 and that's reported as 782,368, and that's in another one of 21 the papers by Travis and her colleagues; you take the incidence 22 rate in the -- from the Yang paper, which is 3.5 times 10 to 23 the -6; and then you multiply those two numbers together to get 24 the expected number of APL cases in the benzene cohort. And 25 you would expect about 2.74 if they had the same rates as the 21 1 general population. 2 Then I calculated rate ratios. And I did it two ways. I 3 said, first off, let's assume that all four of the APLs that 4 Travis diagnosed were APLs. So we observed four, we expected 5 to see 2.74 based on the background rates, the ratio of 4 over 6 2.74 is 1.46. I calculated a p-value for that, it's .29, and I 7 calculated the 95 percent confidence interval, which is .4 to 8 3.74. So there is a -- let's say a moderate association that 9 is not statistically significant and the confidence interval 10 includes one. So, okay, there's a small association. 11 I did the calculations again based on three observed 12 cases; in other words, eliminating one of them that I think is 13 unlikely to have been an APL. This time you get a rate ratio 14 of 1.09, so it's almost 1, p-value of .52, and the confidence 15 interval goes from .23 to 3.20. 16 If you would click again. 17 Okay. So these results indicate there is no significant 18 association between benzene exposure and APL in the Travis 19 paper. And it doesn't matter which calculation you use, 20 whether you use the four or the three observed, you still come 21 out with a result that is nowhere close to statistical 22 significance. 23 Q. Doctor, did you do a calculation taking into account or 24 making some attribution for a healthy-worker effect? 25 A. Yes, I did. I said, okay, if there were a healthy-worker 22 1 effect -- and here you have to have some background knowledge 2 of the magnitude of the healthy-worker effect. In the U.S. and 3 in Europe we commonly see healthy-worker effect that range from 4 none to about a 30 percent reduction in mortality, okay? So 5 working populations may have only 70 percent of the mortality 6 of the general population -- in some settings. 7 So I said, okay. Let's adjust for that. Let's bump up 8 the observed by 25 to 33 percent -- so add one case to the 9 four, that would be a 25 percent adjustment for the 10 healthy-worker effect, or add one case to the three, that would 11 be a 33 percent adjustment. I did that -- it increased the 12 rate ratios in both of those calculations; the p-value was 13 still nowhere close to significant. 14 So even if you make an adjustment for the healthy-worker 15 effect, which I don't think is appropriate to do, you do not 16 have a statistically significant association in this data set. 17 Q. Did you do any further calculations with regard to any of 18 Dr. Smith's calculations? 19 A. No, I think that was everything. 20 Q. And, Doctor, at this point can you summarize your 21 observations with regard to Dr. Smith's methodology and their 22 consistency with epidemiological methods? 23 A. Yeah. It's my view that Dr. Smith's methods for choosing 24 data and excluding data is arbitrary and unreliable. His 25 methods for analyzing data are arbitrary, they're not reliable, 23 1 and they're not accepted in the scientific community. These 2 are not the sorts of calculations we see. 3 Number 2, none of the authors of the studies he cites -4 Rinsky, Glass, Travis, Golomb -- made any of the calculations 5 Dr. Smith made, nor is there any information in those papers 6 that supports any of Dr. Smith's methods for choosing the data 7 as he did or for analyzing the data as he did. 8 Number 3, Dr. Smith made a number of errors and 9 misrepresentations. Number 4, the reliability of the methods 10 he uses is not known and to my knowledge has never been tested. 11 Number 5, much of the evidence he relies upon has nothing to do 12 with benzene and APL; it either addresses other diseases or it 13 addresses other exposures. So it is not relevant to the issue 14 of whether benzene causes APL. 15 Number 6, Dr. Smith has failed to show any reliable 16 evidence that benzene is associated with APL. And Number 7, 17 Dr. Smith ignores the evidence that shows no association 18 between benzene and APL. He simply doesn't look at it. 19 Q. Now, Dr. Garabrant, in your work, you've become familiar 20 with the Hill criteria? 21 A. Yes. 22 Q. And in your work, have you been involved in studies 23 exploring the issues of cause and effect of exposure to an 24 agent and the potential development of a medical condition? 25 A. Well, much of the work I've done over my career has been 24 1 focused on that issue. 2 Q. And did you, looking at what Dr. Smith did, consider that 3 in connection with the Hill criteria? 4 A. I did. 5 Q. And using the Hill criteria, do you have an opinion as to 6 whether there is adequate evidence to determine whether benzene 7 is capable of causing APL? 8 A. I do. 9 Q. And what is that? 10 A. Okay. If you'd go to the next slide. Well, actually, 11 we'd have to go to the next one. 12 I wanted to point out -- I think I already covered this 13 earlier -- the claim was made that John Snow didn't need 14 statistics to take the handle off the Broad Street pump. And 15 this is right out of Hill's paper. John Snow, in fact, did use 16 statistics to calculate the death rates, which we would call 17 mortality rates, in different regions of London. And he 18 figured out which regions were supplied by which water supply. 19 And it was quite clear -- and this was using statistics as they 20 were available in 1855. It was quite clear that the death rate 21 in some areas of London, which were supplied by water from the 22 Thames downstream of the city, was dramatically higher than the 23 regions of London that were supplied with water from the Thames 24 from upstream of London. 25 So he did use statistics. They were rudimentary compared 25 1 to what we do today, but he did. 2 Okay. Next slide. 3 Q. Doctor, looking at a 14-times increase given what Dr. Snow 4 observed, would it be fair to come to a cause-and-effect 5 relationship with regard to the wells without doing the 6 techniques that we do today? 7 A. Well, I think you've asked two questions, and you've got 8 to separate them out. It would certainly be fair to come to a 9 conclusion that there was a very strong association between the 10 source of the water supply and the risk of death from cholera. 11 That is quite clear. And you don't really need a p-value; it's 12 obvious. 13 Okay. As to whether you would make a causal 14 determination, I think you would want to step through the 15 considerations that Hill puts forth. Okay. So it's more than 16 simply saying, "Gee, I've got numbers." You have to think 17 about what the numbers mean. 18 Q. Now, going through the Hill considerations with regard to 19 Dr. Smith's observations, what did you conclude? 20 A. All right. So these are Hill's considerations. The first 21 is "Strength of Association." You have to have an association. 22 And Hill says stronger associations are less likely to be due 23 to subject errors like bias and confounding. So the question 24 is how large is the association between benzene exposure and 25 APL -- 26 1 If you would click. 2 -- and the answer is none. We don't have an association, 3 period. Okay. 4 Hill next says, well, has this association been observed 5 consistently by different persons; in other words, different 6 investigators in different places in different circumstances, 7 different times? And the answer is no, we don't have any 8 consistent association. 9 Third, he asks, is the association limited to specific 10 workers and to particular sites and types of diseases? And the 11 answer is no, we don't have an association. 12 Number 4, temporal relationship: Which is the cart and 13 which is the horse? Well, hold on. You've gone ahead of me 14 there. 15 The answer is, yeah, we have temporality in all of these 16 studies; we know whether the benzene preceded the diagnoses. 17 Okay. So we have temporality. 18 Okay. Five: Biological gradient, which we call "dose 19 response." No, there is no dose response. There is no study 20 that shows the risk of APL increases with increasing benzene 21 exposure. None. 22 Okay. Number 6: Plausibility. Okay. Now, Hill's 23 cautious about plausibility. He says, "Is the causation we 24 suspect biologically plausible? This is a feature we cannot 25 demand. What is biologically plausible depends on the 27 1 biological knowledge of the day." All right. I think this is 2 questionable. 3 So if you would click on that. 4 Q. Doctor, Dr. Hill also, with plausibility, says you 5 shouldn't reject an association even if you can't understand 6 what's bringing it about. 7 A. Absolutely. So if you see a strong association -- if you 8 see dose response and you can't explain it based on your 9 biological understanding of the disease -- you should not 10 reject it, okay? But he doesn't say that you should continue 11 to consider that an association is plausible when you can't 12 find an association, okay? And I think this is -- plausibility 13 is not satisfied. You know, what we heard about yesterday from 14 Dr. Cranor is the way the NTP and IARC assess data. And they 15 rely heavily on human epidemiology studies but they're also 16 willing to consider mechanistic evidence; they're willing to 17 consider animal evidence. 18 So if I were to look at plausibility here for benzene and 19 APL, I would consider the following: First off, we don't have 20 epidemiologic evidence that supports it; in fact, we have 21 epidemiologic evidence that says there's no association. 22 Secondly, we do not have any evidence that benzene causes 23 translocations of Chromosomes 15 and 17, which is essential. 24 Don't have it. 25 Thirdly, I'm not aware of any animal model that 28 1 demonstrates benzene causes APL. And, fourth, I'm not aware of 2 any mechanistic evidence in any cell culture, in any animal, in 3 any setting that gives us a mechanism by which benzene can lead 4 to APL. We don't have the steps. 5 So I would say, you know, thinking in the way Dr. Cranor 6 did yesterday, well, could we rely on other forms of evidence? 7 The answer is we don't have any other forms of evidence, and 8 the epidemiology says there's no association. So I think it's 9 implausible. 10 Q. Coherence? 11 A. Coherence. All right, Hill says, "The cause and effect 12 interpretation of our data should not conflict with generally 13 known facts of the natural history and biology of the disease." 14 And my answer is that there's no conflict here; there's no 15 evidence that benzene causes APL. 16 Okay. "Experiment: Do we have experimental evidence that 17 preventative action does, in fact, prevent the disease?" No, 18 we don't. And then "Would it be fair to judge by analogy?" I 19 don't think so. 20 Okay. So I don't think that this set of information 21 satisfies Hill's criteria in any way with the exception of we 22 know in these various studies that benzene preceded the various 23 diagnoses of lymphohematopoietic leukemia. 24 Q. Or at least the way Dr. Smith interpreted the study. He 25 didn't know whether it was actual benzene exposure. 29 1 A. Well, in the studies that look at benzene, we do -- and 2 remember, we have Yang; we have Rinsky -- I mean, the two 3 studies that really are pure benzene-exposed, they don't 4 support it; they don't show it. 5 Q. And, Doctor, finally, Dr. Smith mentioned a study by a 6 Joli Weiss concerning acetaminophen use and risk of adult 7 leukemia; is that correct? 8 A. Yes. 9 Q. Did you have an opportunity to review that? 10 A. I did. 11 Q. And do you have any observations on that? 12 A. This study was reported in Leukemia Research in 2006. It 13 came from Roswell Park Cancer Institute in Buffalo, which is a 14 very fine institution. And what it says is that at Roswell 15 Park they ask every patient to fill out a 16-page questionnaire 16 regarding a long list of things. And they've analyzed that. 17 And what they found was that there was a positive association 18 between ever having used acetaminophen, so Tylenol and other 19 drugs like that -- or I should say "and other products like 20 that" -- the odds ratio was 1.53. The confidence interval went 21 from 1.03 to 2.26. And they found a weak inverse association 22 with aspirin use, which I think was not statistically 23 significant. 24 Okay. When they looked further at the acetaminophen data, 25 they did not find any evidence of increasing risk with the 30 1 frequency of acetaminophen use with the degradation of 2 acetaminophen use or with the cumulative acetaminophen use. So 3 what we have is an association. We don't have dose response; 4 we don't have a body of results that presents a consistent 5 picture. 6 I would say this is a suggestion. It's -- if we went back 7 to the scientific method, I would say this is a basis for a new 8 hypothesis that needs to be tested. And I think the author's 9 basically said that in their discussion. "Gee, this is 10 brand-new. It certainly needs to be confirmed by other 11 studies." It's an awfully important issue. If we think that 12 acetaminophen causes adult acute leukemia, that's a big issue. 13 And so I would say this is a hypothesis that really 14 deserves a rigorous evaluation and further testing before we 15 should make any further conclusions about it. 16 Q. And did they conclude there's a cause-and-effect 17 relationship between acetaminophen and leukemia? 18 A. No, I think they're appropriately cautious and do not make 19 that conclusion. 20 Q. Thank you, Doctor. 21 MR. LEGHORN: No further questions, your Honor. 22 THE COURT: All right. 23 Mr. Stewart, are you going to be using any of the 24 equipment? 25 MR. STEWART: I will. I will be using the Elmo. 31 1 THE COURT: Okay. 2 CROSS-EXAMINATION 3 BY MR. STEWART: 4 Q. Good morning, Dr. Garabrant. 5 A. Good morning. 6 Q. I'm not going to attempt to cover everything you covered 7 or we'll never finish this week, and I want to finish this 8 week, so I'm going to try to focus on some of the things that 9 you covered. But first, before I do that, I would like to 10 spend some time with a document that you ended up producing 11 about the testimony that you've given over the last five years, 12 all right? And what I would like to do is I would point out -13 and this makes sense because you filled this out November 19, 14 2008 -- that this chart stops November 18, 2008. So it was 15 very, very current up to that time period. Thank you for doing 16 that. I want to go through this list in a summary fashion and 17 then try and have you update it so we have a better 18 understanding of how often you do this, all right? 19 So with respect to this list, I counted some 62 times from 20 November 2004 to November 18, 2008, where you gave deposition 21 testimony or trial testimony in some form or fashion in a legal 22 context, fair? 23 A. Yes. 24 Q. Okay. Now, what is your current occupation? 25 A. I am emeritus professor of Occupational Medicine and 32 1 Epidemiology at the University of Michigan. 2 Q. Okay. And does that mean you're still on the payroll? 3 A. Yes. 4 Q. Okay. What are your duties in connection with that? 5 A. I have an active research agenda. We're running one very 6 large study, and I have three other studies we're still writing 7 papers and analyzing data and reporting at. 8 Q. Okay. Now, in the year 2008, how much money did you make 9 doing this kind of work, being an expert in litigation? 10 A. I don't know exactly. I would estimate it was probably in 11 the range of 400-, maybe $450,000. 12 Q. And of that $450,000 that you made in 2008, how much of 13 that as a percentage came from giving testimony or doing work 14 on behalf of companies in a toxic tort context? 15 A. I'm not sure I could break it out in a toxic tort context. 16 It -- I believe it was all defense work. 17 Q. Okay. So in 2008 450-some thousand dollars for this type 18 of work. How about in 2007? How much money did you make in 19 2007 doing this kind of work for companies in the defendant 20 role? 21 A. Again, I don't know exactly. I believe it was less in 22 2007. I would estimate 300- to $400,000. 23 Q. Okay. So now let's talk about -24 MR. STEWART: Your Honor, would you mind if I go to 25 the chart, please? Thank you. 33 1 BY MR. STEWART: 2 Q. So in 2008 we were at $450,000. I believe in 2007 you 3 gave us a range -- and I don't want to get it wrong. What was 4 the range you gave us? 5 A. I think I said 300- or $400,000. 6 Q. And in 2006 what would be your estimate of the amount of 7 money that you made doing this kind of work on behalf of 8 defendant companies? 9 A. I don't know. I believe it would have been less. I would 10 have to guess at this point. I don't actually know. 11 Q. I don't want you to wildly guess, but if you could give us 12 a reasonable estimate, that would be -13 A. It was probably in the range of $300,000. 14 Q. Okay. And then in 2005 -- and we'll stop there for the 15 sake of time -- give us a reasonable estimate of the amount of 16 money that you made doing this kind of work for defendant 17 companies. 18 A. To be honest, I don't know. I mean, it was appreciable. 19 I don't know whether it was 200- or 300- or 350-. I just don't 20 recall. 21 Q. Okay. So -22 A. I don't know. 23 Q. -- let's do this: Conservatively $200,000? 24 A. Yeah, that would -- it was above that. 25 Q. Okay. More than $200,000. 34 1 So over the past four years it's safe to say that you have 2 made more than $1.2 million doing this kind of work; fair? 3 A. Yes. 4 Q. And it could even be more than $1.6 million doing this 5 kind of work for companies, correct? 6 A. I don't know. I mean, those numbers don't total to that. 7 I'm not sure. 8 Q. $1.3 million? 9 A. Yes. 10 Q. All right. Now, in connection with these sums, is it fair 11 to say that over the last -- from 2005 to 2008 that this is the 12 manner in which you make most of your money? 13 A. It's varied. In some years, yes; in some years, no. 14 Q. In 2008 was it the manner in which you made most of your 15 money? 16 A. Yes. 17 Q. How about 2007? Was this the manner in which you made 18 most of your money? 19 A. No. 20 Q. Okay. How about 2006? 21 A. I'm not sure. To be honest, I don't know. 22 Q. Okay. Now, I told you that I realize this was as 23 up-to-date as you could possibly be, and to be fair there was 24 one thing that wasn't on here which was a trial you and I were 25 involved in in California, true? Turner versus Chevron? 35 1 A. Yes. 2 Q. But to be fair, Turner versus Chevron -- your deposition 3 is on this list -- it looks like you just inadvertently left 4 off the trial portion of that. No big deal. I want to ask 5 this question, though: If we were to update this list from 6 November 18, 2008, to the present -- we have 62 things up to 7 November 2008 -- how many more items would go on this list from 8 November 18, 2008, to when you came into this courtroom and 9 began testifying yesterday? 10 A. I would estimate -- so that was a five-month period, a 11 little less than half a year -- if you took 2008, and took 12 about half of the number for 2008, that would be a reasonable 13 estimate for 2009. 14 Q. Does that mean that you're doing more of this on a monthly 15 basis in 2009 than you were doing in 2008? 16 A. No, I think about the same. You use about the same rate. 17 Q. Okay. So in connection with the year 2009, if we put 2009 18 on the board as well, how much money have you made or billed in 19 2009 in relation to doing this kind of work for defendant 20 companies? 21 A. I would estimate somewhere between 150- and $200,000. 22 Q. Now, it's true that you haven't published any studies of 23 benzene-exposed workers and leukemia risks, true? 24 A. I'd have to look at my CV. I think my colleagues and I, 25 when I was at the University of Southern California, did a 36 1 number of studies where the answer would be yes. We did a 2 study of workers who had a high exposure to electromagnetic 3 fields in -- which we assessed their benzene exposures and we 4 reported out their leukemia risks. The goal of that study, 5 however, was to look at leukemia in relation to electromagnetic 6 fields, but benzene was a potential confounder. 7 We also published a study of childhood leukemia in which 8 we looked at parents' occupations and parental occupational 9 exposure to solvents including benzene. And I've done other 10 studies that are not yet published that have also looked at 11 that issue. 12 Q. When you say "not yet published," they're in the queue to 13 be published or they just weren't published? 14 A. In the queue. I haven't gotten around to writing the 15 manuscripts yet. 16 Q. Okay. So in connection with these at issue, have you 17 published any article, textbook, chapter, any publication in 18 which you've expressly discussed the issue of whether benzene 19 exposure is capable of causing some subtypes of AML but not 20 others? 21 A. No. 22 Q. Now, I want to turn to something that you said yesterday. 23 I want to get to the heart of some of the things that you have 24 testified about. And the first thing that I want to do in that 25 category is I want to talk about your discussion of Rinsky and 37 1 what Rinsky said. So I'd like to do it this way. You were 2 asked this question by Mr. Leghorn yesterday and, Dr. 3 Garabrant, I am reading from the transcript which everybody's 4 getting on a daily basis here. You were asked this question -5 MR. STEWART: Your Honor, is that light enough here? 6 THE COURT: Yes. 7 MR. LEGHORN: Thank you. 8 BY MR. STEWART: 9 Q. You were asked this question: "And, Doctor, do you have 10 an opinion whether Rinsky supports the conclusion or that 11 Rinsky -- the Pliofilm cohort supports the data there" -12 "supports the conclusion that benzene causes APL?" And you 13 said, "Well, it does not. There is no case of APL in that 14 cohort." 15 That was your testimony, correct? 16 A. Yes. 17 Q. And in addition, what you did was you had a slide that you 18 showed -- and I've got a black-and-white of this slide. And 19 you showed the slide and said in 1981, the Rinsky study, a 20 single case of leukemia coded ICD 205 was acute myeloblastic 21 leukemia, not APL. Then you brought out this slide and said in 22 the 1994 updated data, two cases of leukemia were coded to 23 205.0; neither was diagnosed as APL. "Dr. Smith claims the 24 AMLs could have been APL, but he doesn't know." 25 That's what you put, correct? 38 1 A. Yes. 2 Q. Now, in connection with the Rinsky case and in follow-up 3 to the Paxton, here's my question: How did you know that the 4 AMLs were not APLs? 5 A. I don't. The point is they were not written down as APLs; 6 they were written down as acute myelocytic leukemia. It would 7 be speculation to infer that they were something also. 8 Q. Well, it would be actually speculation to say that there 9 is no case of APL in that cohort, wouldn't it? 10 A. There is no case of APL identified in that cohort. 11 Q. That doesn't mean there's not an APL in that cohort, does 12 it? 13 A. The Paxton paper reports what the diagnostic information 14 was. That's all there is to go on. So there could be an APL 15 there, but there's no information that says there is an APL 16 there. 17 Q. Well -- and this is because you'd have to write down on 18 the death certificate what the subtype of AML was, correct? 19 A. In order for the Rinsky paper or the Paxton paper to know 20 the diagnosis, it has to be written on the death certificate. 21 Q. So what I said was correct, right? 22 A. Yes. 23 Q. Uh-huh. And when we look at the Rinsky paper -- this is 24 page 240 of the Rinsky paper -- it says -- Case 3 was in a 25 60-year-old male who died in 1958; the cause of death, recorded 39 1 on the death certificate, was acute myelocytic leukemia, 2 correct? 3 A. Yes. 4 Q. Now, that's not a subtype, right? 5 A. Well, that could be all that the diagnosis provided. 6 There are many patients where that's what the hemopathologist 7 said it is. That's it. 8 Q. Just answer my question, sir. That is not subtyped, 9 correct? 10 A. It does not give one of the FAB subtypes. And sometimes 11 you can't reach a further diagnosis. So it is what it is. 12 Q. Well, so that part that you just put there with the comma, 13 sometimes you can't reach that, you don't know if that was the 14 case here or not, correct? 15 A. That's correct. 16 Q. So this is an example of something that could have been an 17 APL, might not have been an APL; we just don't know, correct? 18 A. I certainly don't know. All you have is what's written on 19 the death certificate. 20 Q. Right. So Rinsky doesn't know, true? 21 A. I assume Rinsky doesn't know. 22 Q. Dr. Paxton doesn't know? 23 A. All any of us have, to my knowledge, is what's on the 24 death certificate. 25 Q. Yes. 40 1 A. That's it. 2 Q. So when you came yesterday and said that there were no 3 APLs in that cohort, that was an overstatement of what the 4 scientific evidence was showing, correct? 5 A. I think that was a fair statement. 6 Q. Well, it's only a fair statement if you can say that 7 people looked for APLs and identified them when they found them 8 in subtyping, correct? 9 A. Well, it would have been fair to have said there were no 10 APLs recorded on the death certificates. 11 Q. Right. And then it would have been fair to say, "So we 12 cannot say whether or not there were APLs," correct? 13 A. I cannot say whether there were or were not APLs. All I 14 can say is that there were none recorded. 15 Q. Uh-huh. So I'm back to this slide. When you say Dr. 16 Smith claims the AMLs could have been APL but he doesn't know, 17 he was accurately describing the Rinsky article, wasn't he? 18 A. Well, I think that that's going a little beyond what 19 Rinsky said. I mean, you know, you've got written records that 20 come from death certificates. They say what they say. If you 21 want to say, you know, "I really think that these were 22 something else," I think that's going beyond what the death 23 certificate says. That's it. So, I mean, I don't think Dr. 24 Smith knows that they were APLs, nor do I. I don't know they 25 weren't APLs. All I know is what was recorded and presented in 41 1 the Rinsky paper; that's it. 2 Q. Yes, I agree with you, Dr. Garabrant. But you're making 3 the insinuation that Dr. Smith came and said that the Rinsky 4 paper said there were APLs. Did you hear him say that? 5 A. I'm not sure -- I don't understand the question. I said 6 the Rinsky papers said there were APLs? 7 Q. I'll rephrase it. This statement right here by Dr. Smith, 8 you say Dr. Smith says he doesn't know whether there were APLs 9 or not, correct? 10 A. Okay. We're going to parse the words. I think what I'm 11 saying there is Dr. Smith claims the AMLs could have been APL. 12 I am saying he doesn't know. I didn't mean to say that Dr. 13 Smith said he didn't know; I'm saying that I don't think he 14 knows. Maybe I should have written that a little differently. 15 Q. And were you trying to infer from this that he said he did 16 know? 17 A. No. To the best of my understanding of what he said, he 18 said they could have been APLs. And my comment is he doesn't 19 know. 20 Q. Well, isn't it true that you agree they could have been 21 APLs? 22 A. It's possible that the ones that were coded to ICD 205.0 23 using the ICD-9 could have been APL because APL is coded to 24 205.0 in the ICD-9. So it could have been. 25 Q. Well, there are some things that are coded to 204 that 42 1 could have been APLs, too, aren't there? 2 A. Which ICD is that? 3 Q. It's 204. 4 A. No. No. Which version of the ICD is that? 5 Q. This was 7. 6 A. Then you would have to go back to the ICD-7. I'm not sure 7 I recall it from memory, but I think the coding changed. I 8 think under ICD-7, a 204 was acute myelogenous leukemia. But 9 we would have to pull up the ICD-7. 10 Q. Well, let's just look at what Rinsky actually says. What 11 Rinsky says -- and let's go back to Case 3, since we were 12 looking at that before -- Case 3 was in a 60-year-old male who 13 died in 1958. The cause of death recorded on the death 14 certificate as acute myelocytic leukemia was coded as ICDA 204, 15 right? 16 A. That's what it says. 17 Q. Yes. And that could have been APL, true? 18 A. Conceivably it could have been. 19 Q. So when you said maybe the 205s, in fact, some of the 204s 20 could have been APLs as well, right? 21 A. Okay. Could you go back to Paxton's table? 22 Q. I can. 23 A. Well, I mean, what you're talking about now is the 24 reclassification of the ICD-7 to the ICD-9. They changed the 25 codes. This is from memory; I don't memorize the ICD versions. 43 1 I think in the ICD-7 AML was coded to 204. I think in the 2 ICD-9 it was coded to 205. 3 Now, the question is, when Travis presented those ICD 4 results, did she update them all to 209 or did -- or to the 5 ICD-9, or did she leave them with the original codes and 6 indicate that these were the original codes from the ICD-7, the 7 ICD-8, the ICD-9? I can't -8 Q. So you don't know? 9 A. Well, not from memory, I don't know. Could you put the 10 chart up? 11 Q. I can. 12 A. Okay. And what's her footnote say? Okay. So her 13 Footnote A says, "International Classification of Disease Code 14 currently on the 1987 update of the NIOSH tape." And if we 15 look back -- if we look at the top half where we see some of 16 the 204s are written as acute myelocytic leukemia/acute 17 myelogenous leukemia, that would suggest to me those are the 18 ICD-7 codes in which AMLs were coded to 204. 19 Q. Yes. 20 A. Okay. 21 Q. So what I said earlier about the 204s, things that were 22 coded as 204 could also have been APLs, correct? 23 A. Well, they are what they were written as, and conceivably 24 some of them may have been APLs. 25 Q. Well, let's look at how Rinsky recorded them so there's no 44 1 misunderstanding. Case No. -- we already talked about Case No. 2 3. Let's look at Case No. 4. It says Case 4 was in a 3 65-year-old man who died in 1960. The cause of death, recorded 4 on the death certificate as, quote, "acute myelogenous 5 leukemia," was coded as ICDA 204, correct? 6 A. That's correct. 7 Q. Case 4 is another example of someone in the Rinsky study 8 who could have been an APL, correct? 9 A. Could have been. 10 Q. Let's look at Case No. 6. Case No. 6 was a 57-year-old 11 man who died in 1961. The cause of death recorded on the death 12 certificate as acute granulocytic leukemia was coded as ICDA 13 204. That is another example of someone who could have been an 14 APL, correct? 15 A. You know, as far as I know, it could have been. I think 16 before it is fair to do this, we should get the ICD-7 in front 17 of us and see where APL is coded, okay? Because I don't know 18 where APL's coded in the ICD-7. 19 Q. Well, now, wait a second, Doctor. In order to be fair, 20 before you came and gave testimony to this Court it was only 21 205 that APL would have been in, shouldn't you have already 22 done that? 23 A. My point was that in not one instance in the Rinsky paper, 24 or the Paxton paper, was APL diagnosed. So it's not critically 25 important to know which ICD code they're coded to. The point 45 1 is there is nothing recorded there as APL. If you're going to 2 take me through now which codes in which ICD revisions would 3 have been AML and could they have been APL, then I have to go 4 back and look at the ICD-7 and say, well, where would APL have 5 been coded? In other words, there's an inference here that the 6 coding was right and that when you say 204.1, that that not 7 only included acute myelogenous and acute myeloblastic and 8 acute granulocytic, but that that code also included acute 9 promyelocytic leukemia. And I'm saying my memory isn't that 10 good. I don't know whether they would have all been coded 11 204.1 in the ICD-7. 12 Q. You just said that no one in the Rinsky-Paxton study was 13 diagnosed with APL. That's what you just said, correct? 14 A. I said what I said. Nobody was recorded as having been 15 diagnosed with APL. 16 Q. Okay. But that's different than being diagnosed with APL, 17 isn't it? Because this is a study where you have -- the 18 recordation is something is written on the death certificate. 19 This is all we have: what's written on the death certificate 20 and the ICD code, correct? 21 A. That's correct. 22 Q. Okay. And so if there's not an ICD code for APL in ICD 23 Code No. 7, there's nowhere to put "APL," is there, in its own 24 distinct category? 25 A. I'm not sure I understand. "In its own distinct 46 1 category." What do you mean? 2 Q. Well, isn't it true, Doctor, that it wasn't until the 3 ICD-10 codes that APL was recognized with its own distinct ICD 4 number? 5 A. It is correct that under the ICD-9 APL was coded to 205.0. 6 It's also correct that the ICDO system, which was created by 7 the World Health Organization and public in 1986 to allow more 8 precise coding of cancers, had clearly identified a unique code 9 for APL. So the ICDO, it was there. And I've been a cancer 10 epidemiologist since the late '70s. The ICDO made it quite 11 clear you could code APL as a distinct diagnostic category as 12 early as 1976, and then for those who were doing this, well 13 before that. 14 Q. Hmm. So you're saying that you believe there was an ICD 15 code for APL that could have been used in the Rinsky study 16 specifically. That's what you're saying? 17 A. I said ICDO. 18 Q. Uh-huh. ICDO is not ICDA, correct? 19 A. That is correct. 20 Q. Uh-huh. 21 A. The ICDO is a classification system for cancers put out by 22 the World Health Organization in addition to the ICD system. 23 Q. Yes. But the ICDA is what is recorded in Rinsky, correct? 24 A. Well, what's recorded in Rinsky is the cause of death. 25 That's what's important. 47 1 Q. No, Doctor, what's recorded in Rinsky is both the cause of 2 death and a code, ICDA 204, right? 3 A. Correct. 4 Q. Okay. So when you said "ICDO codes," those aren't the 5 codes that are being used in the Rinsky paper, are they? 6 A. That's correct. Your question was whether there was a 7 unique code for acute promyelocytic leukemia prior to the 8 ICD-10, and my answer was, yes, there was. 9 Q. But the codes that are being talked about in Rinsky and 10 Paxton, these papers you've looked at, are not that code that 11 you've just mentioned, are they? 12 A. No. They're the ICD-7 or ICD-8 or ICD-9. 13 Q. Yes. Thank you. So now going back to other cases, Case 14 10 was in a 38-year-old male who died in 1955. The cause of 15 death recorded on the death certificate as acute myelocytic 16 leukemia was coded as ICDA 204. That's another example of 17 someone who could have been an APL, correct? 18 A. Could have been. 19 Q. And 12, Case 12, 48-year-old male died in 1954, diagnosis 20 was acute myelocytic leukemia. That's another example, 21 correct? 22 A. Could have been. 23 Q. Someone who could have been an APL? 24 A. Could have been. 25 Q. Okay. Now, I want to go to a different topic. And the 48 1 topic I want to go to now is the Travis paper. And the Travis 2 paper -- what you said in the Travis paper was there were four 3 APLs that were listed in the Travis paper. And I'm going to 4 put that chart up. It's Table 4. You said it was 5, 6, 7 and 5 8, correct? 6 A. Yes. 7 Q. And let's just make sure we're clear here on the Travis 8 paper. These folks -- and we're looking at, I think, 12 9 authors on the Travis paper, correct? 10 A. Yes. 11 Q. All right. And these 12 authors who are studying this 12 group of people have reached the conclusion from the evidence 13 that they are looking at, 12 people looking at this evidence, 14 that they believe four of these people are best classified as 15 APLs, correct? 16 A. That's correct. 17 Q. And you disagree with them, correct? 18 A. I think that Case 5 has very scant information upon which 19 that diagnosis is based; that's all. 20 Q. I understood that. But there's 12 people who came 21 together, looked at the evidence, wrote a paper, true? 22 A. Yes. 23 Q. And those 12 people had that paper peer-reviewed by other 24 people, true? 25 A. As far as I know. 49 1 Q. Who looked at the same kind of information that you looked 2 at, right? 3 A. Well, they had the manuscript; yes. 4 Q. Exactly. And those 12 people and their peer-reviewers -5 and in those peer-reviewers there was an editor of that journal 6 in addition to the peer-reviewers, true? 7 A. Yes. 8 Q. All right. And so with 12 authors, peer-reviewers and 9 editors who came to the conclusion that this information was a 10 fair and accurate representation of what they saw and what they 11 found, you've basically come to say you disagree with it, 12 correct? 13 A. I think, as you saw in my presentation, that I did my 14 calculations assuming that there were four APLs, and I did them 15 again assuming there were three, because I think the diagnosis 16 of Case 5 is uncertain -- is based on uncertain data; 17 nonetheless, I will defer to a qualified hematopathologist to 18 make that decision. 19 Q. Well, Dr. Garabrant, so I'm clear, you are unwilling to 20 defer to 12 experts who looked at all the data, collected it, 21 analyzed it, presumably talked about it, came to conclusions 22 about it, put it in front of peer-reviewers, put it in front of 23 an editor, put it out there for the world to see -- you are 24 unwilling to defer to them to say that their conclusion is a 25 reasonable one, true? 50 1 A. No. 2 Q. That's not true? 3 A. As you saw in my presentation, I considered that they were 4 right. I included all four cases. I also considered that one 5 of them, based on reasonable grounds, might not be APL. I did 6 it both ways. The point of my presentation was to show that 7 you can get a significant association either way. 8 Q. But you do get an association, correct? 9 A. Uh-huh. 10 Q. And these Bradford Hill considerations that you were 11 talking about, the Bradford Hill considerations don't talk 12 about a significant association; they talk about the strength 13 of an association, true? 14 A. No. 15 Q. That is not true? 16 A. That is not true. 17 Q. So let's look at what Bradford Hill says. It says, 18 "Strength. First upon my list, I would put strength of 19 association." That's what it says, right? 20 A. It's one of the things it says, yes. 21 Q. Yes. It doesn't say "significant"; it says "strength," 22 right? 23 A. Well, it is talking about associations, and the strength 24 of the association matters. 25 Q. I don't disagree with you. "Under this same heading of 51 1 'Strength of Association,' Hill said we must nevertheless look 2 at the obverse of the coin. We must not be too ready to 3 dismiss a cause-and-effect hypothesis merely on the grounds 4 that the observed [sic] observation appears to be slight. 5 There are many occasions in medicine when this is in truth so." 6 That's what Hill said, right? 7 A. Well, you're missing one of Hill's important points which 8 is that you must assess the play of chance, okay? That's woven 9 throughout the paper. He's talking about once you have done an 10 appropriate assessment of the play of chance, then you look at 11 strength of association and dose response and consistency, et 12 cetera. 13 Q. Well, that's funny, Doctor, because he doesn't get to 14 chance until later in the paper. He first talks about strength 15 of association and then he talks about the -- this concept of 16 don't be too ready to dismiss a cause-and-effect hypothesis 17 merely on the grounds that the observed association appears to 18 be slight. That's what he said, right? 19 A. Well, yes. And you certainly wouldn't accept a slight 20 association that was entirely compatible with chance. If you 21 read his article carefully, you'll see that quite clearly. You 22 don't accept weak associations that are chance or are entirely 23 compatible with chance. He has a whole section on tests of 24 significance after he finishes the nine considerations. 25 Q. Yes, there are. I agree with you. 52 1 A. Right. 2 Q. It says, "Tests of significance. No formal tests of 3 significance can answer those questions." 4 That's what he says, right? 5 A. Yeah. Then if you want to interpret that, you have to go 6 back to what the questions are. 7 Q. Uh-huh. Well, he says here -8 A. If you'd read the paragraph just above that. 9 Q. I'm happy to. Just above it? "Here then are nine 10 different viewpoints, all of which we should study" -- "all of 11 which we should study association before we cry causation. 12 What I do not believe, and this has been suggested, is that we 13 can usefully lay down some hard and fast rules of evidence that 14 must be obeyed before we accept cause and effect. None of my 15 nine viewpoints can bring indisputable evidence for or against 16 the cause-and-effect hypothesis, and none can be required as a 17 sine qua non. What they can do with greater or less strength 18 is to help us to make up our minds on the fundamental question: 19 Is there any other way of explaining the set of facts before 20 us, is there any other answer equally or more likely than cause 21 and effect," correct? 22 A. And then he talks about tests of significance, which means 23 you have to evaluate chance as a possible explanation that is 24 more likely than cause and effect. 25 Q. Well, actually, he says there's no formal test of chance. 53 1 That's what he says, right? 2 A. There's no formal test of significance that can answer 3 those questions, but the point is you have to address it. 4 Q. But, Doctor, you have been using a formal test of chance 5 the entire time you have been in this courtroom, haven't you? 6 A. Hum? 7 Q. You've been using a formal test of significance, and your 8 formal test of significance is the 95th percent confidence 9 interval or the P at .05. That's your formal test, correct? 10 A. No. No. I've been doing exactly what Hill says you must 11 do, okay? "No formal tests of significance can answer those 12 questions. Such tests can and should remind us of the effects 13 that the play of chance can create, and they will instruct us 14 in the likely magnitude of those effects." In other words, 15 what he's saying is before you accept that there is a causal 16 association, you've got to assess the role of chance, you have 17 to -- and it will instruct you in the magnitude of the effect 18 of chance. 19 Now, as to whether you choose a p-value of .05 or a 20 p-value of .04 or a p-value of .01 as your nominal threshold 21 for statistical significance I think is arguable. By 22 convention we use .05. My calculations show repeatedly that 23 you're nowhere close to .05; you've got p-values of .2 and .3 24 and .5, okay? These results are entirely compatible with 25 chance findings; they are not indicative of cause and effect. 54 1 This is chance stuff. That's what Hill says you need to do. 2 Q. Okay. So you didn't share with the Court all the 3 calculations you did on the Travis study, did you? 4 A. Well, I spent quite a bit of time trying to figure out 5 what Dr. Smith was doing, and I tried to make calculations 6 trying to follow his reasoning in addition to the ones I 7 presented. I'm not sure what you're referring to. 8 Q. I'm referring to this slide right here. That starts, "It 9 is possible to make other more plausible assumptions about the 10 cases of APL in the Travis cohort. These would yield different 11 estimates of the association between benzene and APL." 12 This is your slide, right? 13 A. Yes. 14 Q. Didn't show this one to the Court, correct? 15 A. I did not. 16 Q. Uh-huh. And in this slide what you did is you made an 17 assumption that 18.7 percent of the AMLs in the Travis study 18 were APLs, right? 19 A. Well, I was following Dr. Smith's reasoning, okay? 20 Q. Yes. 21 A. And what Dr. Smith reasoned was that the proportion of 22 ANLL that was APL was 28.6. And he referenced the statement 23 made by Travis where she said the proportion of ANLLs or APLs 24 was similar to what we saw in the general population of China, 25 and referenced the Yang paper. So I went back to the Yang 55 1 paper to see if, in fact, that was true and, in fact, we see -2 or as we saw, the Yang paper said it was 18.7 percent. So I 3 said, you know, that's a better -- that's a more reliable 4 estimate than 28.6 as Dr. Smith used. So let's do Dr. Smith's 5 calculations using the 18.7 percent, not the 28.6. So I did 6 those calculations. 7 Q. Yes. And once you did those calculations that we're 8 presently seeing for the first time, what you found -- let's 9 just back up to say you described this as a more plausible 10 assumption to do with respect to the Travis paper than what Dr. 11 Smith did, correct? 12 A. Well -13 Q. That's how you start this line. 14 A. The proportion of ANLL that was APL is -- the number 18.7 15 percent is more plausible than Dr. Smith's assumption of 28.6 16 based on a very small amount of data. 17 Q. So your answer is yes, correct? 18 A. Yes. 19 Q. And in connection, once you did this calculation to, 20 quote/unquote, correct what Dr. Smith did, what you found was 21 -- when you counted the APLs that the Travis 12 authors 22 determined were APLs, when you counted those four, you had an 23 observed of 5.98, an expected of 2.3, and a rate ratio of 2.6, 24 correct? 25 A. Right. Now, let's walk through the calculations. All 56 1 right. So what this is about is saying, look, if we assume 2 that 18.7 of the ANLLs were APL, and we follow Dr. Smith's 3 method, which I don't think is reliable, we would estimate 5.98 4 cases of APL in the Travis cohort. I then said, now, let's try 5 and figure out what the expected is, and I'll use Dr. Smith's 6 estimates that range from three per million to eight per 7 million -- eight per million per year -- so 380. That's a 8 range he gave. That would generate expecteds in the range of 9 2.3 to 6.3. So these are boundary calculations. 10 So if you use the 5.98 observed -- or I should say there 11 were four observed plus two created -- versus 2.3 expected, you 12 get a rate ratio of 2.6 and a p-value that is not quite 13 statistically significant. If you use Dr. Smith's other 14 assumptions that the background rate is eight per million 15 person years, you get 6.3 expected. Compare that to the 5.98 16 that you have calculated, now you have a rate ratio of 0.95; in 17 other words, no association. 18 The point of these calculations is to indicate that, 19 depending on what you assume, you might or might not get any 20 association at all. Both of these are guesses. These are just 21 guesses. And depending on what you guess at, you can find an 22 association or no association. 23 Q. Well, now, let's back up because I want to get back to my 24 question. You've just given an answer, but that wasn't my 25 question. My question was: When you -- when there were four 57 1 APLs in the Travis paper, the amount of expecteds -- the amount 2 of expecteds, the rate ratio would be 2.6, right? Am I right? 3 A. No, you're wrong. 4 Q. I'm wrong? Okay. The top number that we have here in 5 this line, that rate ratio of 2.6, would you agree with me that 6 if that assumption is a correct one -- and I understand you 7 don't believe it's a correct one, but if that assumption is a 8 correct one -- that that is a rate ratio that appears to be 9 showing an association? 10 A. Which assumption? 11 Q. Line 1. 12 A. But which assumption is a correct one? There's a bunch of 13 assumptions in there. 14 Q. Okay. Well, I'm not trying to play a word game with you. 15 Line 1. Does line 1 where you have observed of 5.98, expected 16 of 2.3, rate ratio of 2.6, that rate ratio -- would that rate 17 ratio be showing an association? 18 A. By definition, a rate ratio of 2.6 is a positive 19 association. It is based on three assumptions that are 20 questionable. 21 Q. Well, when you worked this math, this was what you laid 22 out in a slide as more plausible assumptions, correct? 23 A. No. This is what I laid out in a slide to show that 24 depending on what you assume, you can get answers that range 25 from no association to a 2.6-fold association. That's the 58 1 point of the slide. 2 Q. I understand. In fact, the slide starts, "It is possible 3 to make other more plausible assumptions about the number of 4 cases of APL in the Travis cohort," correct? That's what you 5 wrote? 6 A. That's what I wrote. And I meant "more plausible than 7 what Dr. Smith presented." 8 Q. Uh-huh. And when you made those more plausible 9 assumptions -- when you put together the more plausible 10 assumptions based on your understanding, and you got to do the 11 math -- in the first category you found a rate ratio that would 12 show a positive association, correct? 13 A. Yes. More plausible than what I regard as an implausible 14 assumption made by Dr. Smith; however, the bottom line on this 15 slide is still the correct one. 16 Q. Okay. Well -17 A. If you make a bunch of guesses, you can come up with 18 widely varying results. That's it. 19 Q. Well, Doctor, that might be what you want to say that it's 20 about, this slide, but I'm entitled to talk about what I want 21 to talk about on this slide, and I want to talk about the 22 p-value with respect to that rate ratio. Because on that 23 p-value with respect to that rate ratio, what that p-value is 24 telling us is that number, that rate ratio, 2.6, is a p-value 25 that shows 92 percent confidence, correct? 59 1 A. That's not what it means. 2 Q. Oh, doesn't it mean that there's an 8 percent chance that 3 that -- that there's an 8 percent likelihood that that number's 4 due to chance? 5 A. Okay. Let's go back to what p-values mean. 6 Q. Is that a yes or no? And then you can explain your 7 answer. 8 A. That's a no. 9 Q. Okay. 10 A. Okay? What the p-value is is the probability that you 11 could have gotten the data you got by chance alone when there 12 is no association whatsoever in the data. Now, what the 13 p-value doesn't do -- and of course this really goes back to 14 Austin Bradford Hill -- is it doesn't tell you anything about 15 the reliability of the data. So where we've created data that 16 really don't have any reliability and then make calculations of 17 p-values, the p-value doesn't in any way fix the unreliability 18 of the data; it just says that if there were really no 19 association -- and you assume the data was reliable -- this is 20 the probability that you could have gotten these data by chance 21 alone when there was no association to the data. 22 Q. Okay. So let's get back to the p-value. And I understand 23 that you don't like the data points that you put in here, 24 correct? 25 MR. LEGHORN: Objection to the form, your Honor. 60 1 Argumentative. 2 THE COURT: No, go ahead. 3 BY MR. STEWART: 4 Q. Am I correct? 5 A. As I've said now repeatedly, the assumptions that created 6 these data I think are not reliable. 7 Q. Uh-huh. I understand. 8 A. Okay? So what that means is that the data have no 9 particular reliability. 10 Q. Okay. And so my question was: You don't like the data 11 points that you put in. Yes? 12 A. Well, I think I've answered it. 13 Q. Okay. Well, just for the sake of time, I'm really trying 14 to ask you yes-or-no questions. So -15 A. And, sir, I'm doing my best to answer them correctly. 16 Q. Okay. 17 A. They don't reduce to yes and no so simply. 18 Q. I'll do my best to keep asking them as most of the time we 19 can get yes and noes if we try. 20 Now, let's go to the next issue, which is the p-value 21 component that we have here on Line No. 1. What you're saying 22 is that that p-value of 0.08 doesn't have any meaning because 23 you don't like the inputs into the system, true? 24 A. This is an example of garbage in, garbage out, okay? If 25 you put garbage numbers in, you can create p-values that are 61 1 very small. Okay? That's what it is. The p-value's just a 2 calculation; it doesn't comment on the validity of the data. 3 It's just a calculation. And, yes, it's 0.08. 4 Q. Okay. And you've made the decision that it's garbage 5 you're putting in so that you can now come testify that you 6 just think this calculation is garbage coming out, right? 7 A. The point of the slide is that if you make assumptions 8 about what the data ought to be and change the data, you can 9 get different answers. That's it. 10 Q. I wasn't asking about the slide, sir. 11 A. Okay. 12 Q. My question didn't have anything to do with the slide. My 13 question really was really simply: You have come to say "I've 14 made the decision that this is garbage in," so that you can 15 come into the courtroom and say, "and now this is garbage out," 16 right? 17 A. No. I've tried to use accepted methods that are widely 18 applied in epidemiology to try to make reasonable inferences 19 from the Travis data. I've also tried to follow Dr. Smith's 20 reasoning and to show that, depending on what Dr. Smith 21 assumes, you can find or not find an association. And, yes, 22 you can calculate p-values and confidence intervals. And there 23 they are. 24 Q. Thank you. Because one of your criticisms of Dr. Smith 25 was that he didn't calculate p-values, right? 62 1 A. Yes. Dr. Smith made no attempt to assess the role of 2 chance in his results. 3 Q. Yes. And so what you have done is you've crunched some 4 numbers that were Dr. Smith's numbers, yes? 5 A. Well, they are based on Dr. Smith's assumptions, yes. 6 Q. That's fair. And once you crunched those numbers based on 7 Dr. Smith's assumptions, you then -- after doing that, you 8 crunched some more numbers with a more plausible assumption 9 made by you which would reduce the numbers, correct? 10 A. Which are you referring to -11 Q. I'm referring to your 18.7 percent of ANLL was APL as 12 observed in Yang. 13 A. Well, that was actually the number that Travis referred 14 to. 15 Q. I'm not quarrelling with you. What I'm saying is you 16 quarrelled with Dr. Smith using a number different than 18.7 17 percent, correct, in his calculation, all right? 18 A. Yes. I said that had no basis. 19 Q. Yes. So what you did was you threw out the number that 20 Dr. Smith used and you inserted the number 18.7 percent, 21 correct? 22 A. Based on a published paper by Yang, yes. 23 Q. Okay. And once you did that, now what you're doing is you 24 were working on correcting, with your logic, Dr. Smith's 25 calculation, correct? 63 1 A. Based on Dr. Smith's logic and still allowing some of Dr. 2 Smith's other assumptions. So, again, what I'm doing is 3 looking at the effect of his assumptions on the results. And 4 so you can do that for a number of different assumptions and 5 work through. It's sort of a sensitivity analysis. If you 6 question this assumption, you put in something different, what 7 happens? Question this assumption, do something different, 8 what happens? Yes, I did that. 9 Q. Yes. And so these numbers that we are seeing here that 10 are calculated on the screen are you putting in corrections to 11 Dr. Smith's assumptions, right? 12 A. No. I'm doing what I would say is a sensitivity analysis 13 and showing how sensitive his results are to changing his 14 assumptions one at a time. So if you change his assumption 15 about the proportion of ANLL that was M3, here's what happens. 16 If you change his assumption about the background rate, here's 17 what happens. 18 Q. Yes. And I'm trying to make this easy. You made 19 corrections to what Dr. Smith's assumptions were and generated 20 these numbers, right? 21 A. Well, I think I've answered that. I have -- I wouldn't 22 say I've made corrections. I've showed that the effect of 23 changing the assumptions has dramatic effects on the results; 24 that's all. 25 Q. Yes. Now, I'm back to line 1. When we insert the 64 1 correction to -- your correction to Dr. Smith's assumption of 2 28-some-odd percent of AMLs had APL, and when we insert what 3 you think is a more plausible assumption -- let's use Yang. 4 When you do that, when you put in the 18.7 percent number, you 5 still get a rate ratio of 2.6, right? 6 A. Yes. 7 Q. You still get a positive association, right? 8 A. Okay. But here's the point: I don't believe any of this 9 has validity. This is a bunch of guesses. 10 Q. Okay. But wait a second. 11 A. And the guesses -- and the results are very sensitive to 12 which combination of guesses you use. And you can find a set 13 of guesses that will give you a rate ratio of 2.6 that has 14 borderline significance. You can find a set of guesses that 15 will do that, okay? 16 Q. Okay. Dr. Garabrant, respectfully, you've now told me 17 that theory three or four times. I'm asking very discrete 18 questions about what this shows, not how you interpret it. 19 Just that 2.6 is a positive association, right? 20 A. Yes, by definition. 21 Q. Yes. Now I want to get to the p-value on that 2.6. And 22 once you've made your correction to Dr. Smith's assumptions and 23 you generated this p-value, that p-value of 0.08 that you're 24 showing there, that is not an insignificant p-value, is it? 25 A. Well, there is a convention that P less than .05 is 65 1 statistically significant. If you want to change that, okay. 2 Q. Well, I'm not saying I do want to change that. I'm 3 pointing out you've said earlier that while that's a 4 convention, other people can disagree with it, right? 5 A. I suppose there are people who disagree with that, but it 6 is a convention that's widely followed. 7 Q. Yes. And in that convention -- so a .05 p-value in that 8 convention, what you're saying is there is a 95 percent 9 probability that the result is not due to chance, correct? 10 A. You know, we use the words very carefully, and I have to 11 think whether what you've said is correct mathematically. Let 12 me say the correct conclusion. When you have a p-value of .05, 13 it says there is a 5 percent probability that you could have 14 gotten these data, all right? It's not about the rate ratio; 15 it's about the data. You could have gotten these data by 16 chance alone when the truth is that there's no association at 17 all. And then the calculation of the rate ratio follows from 18 the data. That's what the p-value means. 19 Now, what you've said is the -- what -- the inverse of 20 that, there's a 95 percent chance that it's not due to chance. 21 And I don't know whether that's right or not. But let's just 22 leave it the way we say it. When you have a p-value of .05, 23 there's a 5 percent chance that you could have gotten data such 24 as this by chance alone when there's really no association. 25 But, again, you have to note that there's nothing in the 66 1 calculation of p-values that comments on whether the data are 2 valid; in fact, you shouldn't calculate p-values from invalid 3 data. You shouldn't do it when you've got a biased set of 4 data. It's wrong to do so. 5 Q. Well, you're the one who did it, right? 6 A. For the purpose of showing that Dr. Smith's various 7 assumptions and alternatives to those assumptions give wildly 8 different answers; that's all. 9 Q. Let's go back to the p-value and let's use your 10 definition. This means that your calculated p-value there, 11 there's an 8 percent -- based on the p-value there's an 8 12 percent chance that you could come to that data by chance 13 alone, right? 14 A. When there is, in fact, no association at all. 15 Q. Uh-huh. 16 A. You have to put that part of the definition in as well. 17 Q. That's fair. And in that first line it's showing a 18 positive association, right? 19 A. I've answered that, yes. 20 Q. Okay. Now, you said, I believe, that on your chart for 21 the Bradford Hill considerations, that there was absolutely no 22 association. "Strength of association." And you said, "How 23 large is the effect?" And you said none, right? 24 A. When we look across the studies, okay, when we looked at 25 Yang, there was no association. After Yang did his 67 1 multi-varied analysis and controlled further factors in the 2 model, we didn't see an association. When we looked at Rinsky, 3 we saw not a single case of APL recorded. Now, I'll agree with 4 you it's possible that what was recorded didn't capture 5 something that was known, but there's no evidence it was known. 6 You have to go with what you've got. So we've got studies that 7 have looked -- and these were benzene-exposed studies, okay? 8 Q. I'm sorry. I just want to interrupt. Why are you saying 9 that the Rinsky study looked for APL when you've just admitted 10 that the Rinsky study didn't look for APL? Nobody knows how 11 many APLs were in the Rinsky study. 12 A. No. Excuse me. That's not true. If there had been a 13 case of APL recorded on the death certificate, the Rinsky study 14 would have shown it. And as you saw from those death 15 certificates, there were very specific diagnoses listed for 16 many of those cases. You know, you see DiGuglielmo's acute 17 myelocytic leukemia. That is a specific FAB subtype. So for 18 some of them, it was quite clear that there were specific types 19 of AML recorded. There was no APL recorded. Now, I agree with 20 you it's possible there was an APL that was recorded as an AML, 21 okay? The point is there's no data in Rinsky that says there's 22 an APL. 23 Q. Uh-huh. Okay. 24 A. Okay. So and -25 Q. I want to go back to your statement of none -- that 68 1 there's none. No data. But this right here is some data, 2 isn't it? 3 A. No. These are guesses. These are just guesses. The data 4 in Travis does not show any association between benzene 5 exposure and APL. 6 Q. Wait a second. You said these are guesses and yet you 7 told me that what these calculations are, are your corrections 8 of Dr. Smith's assumptions by using data, right? 9 A. No. I said these are -- this is a sensitivity analysis 10 where if you changed Dr. Smith's assumptions, you can show that 11 the results change dramatically. 12 Q. Okay. But you changed his assumptions to your liking, 13 correct? 14 A. I changed -- I changed the assumption about the proportion 15 of ANLL that was APL based on the Yang paper which was, in 16 fact, the source of Dr. Travis's claim that the proportion was 17 similar in the general population which was, in fact, the 18 source of Dr. Smith's idea that you could use the proportions 19 somehow. 20 Q. Yes. And when you put in these assumptions, these 21 corrections to the assumptions based on data, these are the 22 numbers you got, right? 23 A. I would say it slightly differently. When I made 24 different assumptions, I got different results. 25 Q. Yes. And these were -- what you did was you made more 69 1 conservative assumptions than what you think Dr. Smith made, 2 right? 3 A. I made assumptions that had a little better foundation in 4 data than Dr. Smith's did. 5 Q. And you would agree, I think, that by picking these 6 numbers, which you say are grounded in data, that these numbers 7 end up creating a scenario where the rate ratio is lower than 8 what Dr. Smith would have gotten, true? 9 A. Well, all right. Yes. Let's back up. These rate ratios 10 are based on at least three assumptions by Dr. Smith which I 11 think cumulatively lead to a wrong answer. 12 Q. I understand that, Dr. Garabrant. You've made it clear 13 you think Dr. Smith is wrong, okay? And what I'm trying to 14 point out is, by then putting in corrections to make Dr. Smith 15 more right, you are still getting a rate ratio that's 2.6 in 16 one instance, correct? 17 A. You've misrepresented what I've said. 18 Q. I haven't. I haven't tried to, anyway. That's what your 19 first line -- your first line has "Corrections to Dr. Smith's 20 assumptions" using, in your words, "data." 21 A. The last sentence in bold. The point is that Dr. Smith's 22 guesses, his assumptions, are no better than other guesses. 23 The results obtained for guessing can be anything you want them 24 to be. You can make a set of assumptions that yield 25 associations, but the association is just based on the 70 1 assumptions you make. If the assumptions are wrong, the 2 association has no meaning. It's not reliable. That's the 3 point. 4 Q. So, Dr. Garabrant, what you've really come to do here is 5 to say that you don't like the assumptions that Dr. Smith has 6 made, correct? 7 A. Well, having 30 years of experience as an occupational 8 epidemiologist and designing these studies and collecting the 9 data and analyzing the data and publishing the papers, I have 10 some experience in what are reasonable assumptions and 11 reasonable methods for calculating things. It is my opinion 12 that what he's done is not reliable. 13 Q. Yes. And as a matter of fact, this is what you routinely 14 do over the last four years: You come into courtrooms and you 15 sit up on the stand, or in depositions, and you say about other 16 scientists, what they have done is unreliable. That's what you 17 do, right? 18 A. Only when it's unreliable. 19 Q. Well, you do it to the tune of more than a million dollars 20 over the last four years, right? 21 A. Only when it's unreliable. 22 Q. Hmm. That's a lot of unreliability, isn't it? 23 A. There are things that are done that I think don't 24 withstand scrutiny. 25 Q. Well, sir, when you get involved in cases there are very 71 1 few things that withstand your scrutiny; isn't that true? 2 A. No, sir. 3 Q. So can you point to me on this list of these cases, 62 4 cases, where you said that you thought that the expert for the 5 plaintiff had been reasonable and reliable in their opinion? 6 A. Well, I'd have to see the list. I can't do it from 7 memory, and to be honest, I'm not sure I recall everything that 8 was said in every one of these cases. Sure. 9 Karen Brown versus Christus Spohn Health System. This is 10 a lady who got a barium enema and they perforated her colon. I 11 felt there was no question they had perforated her colon. I 12 agreed with the plaintiff. The question was whether she had 13 barium poisoning. I felt she did not have barium poisoning. 14 She got peritonitis from the barium preparation, the contrast 15 agent. So I agreed they perforated her colon, I agreed they 16 pumped a whole bunch of barium material into her peritoneum, 17 and I agreed she got peritonitis and she was disabled as a 18 result of that. 19 The one issue that I disagreed was she didn't have barium 20 toxicity; she did not have any specific toxicity from barium. 21 But other than that, yeah, I agreed with the plaintiff. 22 Q. Okay. So you disagreed with the plaintiff on the central 23 point in the case, which was barium poisoning, right? 24 A. No, I don't think it was the central point. I think the 25 central point was that a mistake had occurred and she had been 72 1 damaged by it. And it was a peripheral point as to whether she 2 had barium poisoning. She did not have barium poisoning. 3 Q. The only reason I'm asking for this back, it's the only 4 one I've got. 5 THE COURT: Are you going to turn to something new? 6 We'll take the morning recess. 7 MR. STEWART: That will be great. 8 THE CLERK: All rise. 9 The Court will take the morning recess. 10 (There is a recess in the proceedings from 11:07 a.m. 11 to 11:30 a.m.) 12 THE CLERK: All rise. 13 Resuming the Milward Daubert hearing. 14 Please be seated. 15 MR. STEWART: Thank you, your Honor. 16 BY MR. STEWART: 17 Q. So, Dr. Garabrant, I want to talk about two more issues. 18 I had more, but I cut them down at the break. We need to get 19 you off the stand and get things done. 20 The next issue I want to talk about is I want to talk 21 about the Mele study called "Epidemiology of Acute 22 Promyelocytic Leukemia" which you talked about a little bit on 23 direct. And you would agree that in this study in 1995 there 24 was a strong association between shoemakers and APL in Italy, 25 true? 73 1 A. Yes. 2 Q. Yes. And you would also agree that these authors, in 3 their discussion section in connection with that, said the 4 following: "The significant relationship between shoemaking 5 and APL is possibly related to benzene exposure, a well-known 6 leukemogenic agent present in the glues," correct? 7 A. Yes. 8 Q. Okay. Now, these are the same authors that you cited from 9 an earlier paper that showed that the benzene content that was 10 supposed to be in the glues after 1963 in Italy was not 11 supposed to be any more than 2 percent, correct? 12 A. They said whatever I showed they said, that -- I think it 13 was that the content -- it said in Italy the use of glues with 14 the benzene content greater than 2 percent was prohibited by 15 law in 1963. 16 Q. Yes. And what that statement means is that it would have 17 been illegal to use glues that had more than 2 percent in 18 Italy, correct? 19 A. Yes. 20 Q. All right. But the statement also means that glues in 21 Italy after 1963 could still have benzene in them, right? 22 A. As far as I know, the law restricted it to 2 percent. 23 Q. Which means the law allowed up to 2 percent benzene in the 24 products, right? 25 A. As far as I know. 74 1 Q. And, sir, that's something these authors in 1995 -- these 2 Mele authors knew when they wrote the sentence that "The 3 significant relationship between shoemaking and APL is possibly 4 related to benzene exposure, a well-known leukemogenic agent 5 present in the glues," right? 6 A. I assume they knew it. They had written the other paper 7 the year before. 8 Q. Yes. And, sir, this is a reasonable statement by original 9 researchers about their topic, what they had researched, and 10 what their possible conclusions could be from that research, 11 true? 12 A. Well, it's as they said, it's possible, or possibly 13 related. It's important to recognize the Mele study had no 14 measurements of benzene exposure. They had no documentation of 15 benzene exposure. They reported an association between APL and 16 working as a shoemaker. That's it. 17 Q. Well, that's not all they said. They then reached the 18 conclusion that this could possibly be related to benzene 19 because it's a logical connection that benzene is used in 20 shoemaking work, true? 21 A. Well, they said it's possibly related. Okay. 22 Q. That's a reasonable conclusion from their research, right? 23 A. I think it's exactly what they said. It's a possibility. 24 That's it. 25 Q. Do you consider that a reasonable statement? 75 1 A. That it's a possibility? I think based on what we know at 2 the present time, which is now, what, 14 years after this was 3 published, that possibility has gotten a little smaller. 4 Q. So what I would like to do is I want to put your Hill 5 considerations list back up on the screen here. And this is 6 going to be the last thing that I do. You went through these 7 factors, the Hill considerations. And this is what I want to 8 ask you: With respect to these factors, Number 1, there is a 9 significant strength of association for benzene causing AMLs 10 generally, true? 11 A. I believe -- well, I don't know what you mean "there's a 12 significant strength of association." I believe it is well 13 established that benzene can cause AML under circumstances of 14 high exposure. And I can give you the full definition of my 15 view on that. But benzene can cause AML. 16 Q. And you would agree that, with respect to Number 2, 17 consistency of association, there is ample evidence of a 18 consistency of association of benzene causing AMLs generally, 19 true? 20 A. I believe there is an adequate body of knowledge to 21 adequately support that conclusion. It's not perfectly 22 consistent; there are some studies that show variation. But I 23 would say, yeah, there's enough consistent evidence that 24 supports it. 25 Q. Yes. Number 3: Is there enough evidence to support 76 1 specificity in Number 3 for the proposition that benzene causes 2 AMLs similarly? 3 A. I think that there is adequate evidence -- you know, 4 "specificity" has two elements. Is the exposure specific? Is 5 the outcome specific? What we know, in my opinion, clearly is 6 that benzene specifically is linked to AML whereas related 7 materials such as toluene, xylene, petroleum distillates, 8 gasoline have not been related to AML. So when you get to the 9 specificity, you get to the benzene, yes, on that side. 10 On the AML side, when you group all AMLs together, there's 11 clearly evidence of an association. But it is not clear that 12 each and every type of AML. So that's, of course, the issue 13 today. It is not clear that APL, specifically, is related to 14 benzene. In fact, we're looking at the evidence today. That's 15 what we're doing. 16 Q. Okay. So, Dr. Garabrant, for the sake of time, please, 17 I'm keeping my question specifically to AMLs generally and 18 benzene, all right? That's my question. 19 And with respect to Number 4 -- you adequately answered 20 Number 3. Number 4, temporal relationship, you'll agree that 21 there are studies out there that show benzene exposure first, 22 AMLs afterwards, right? 23 A. Yes. 24 Q. All right. Now, Number 5, biological gradient. You will 25 agree that there's ample evidence of a dose-response curve for 77 1 benzene exposure and AMLs generally, true? 2 A. Yes. 3 Q. Number 6: Plausibility. You would agree that there's 4 ample evidence of the biological plausibility of benzene 5 causing AMLs generally, true? 6 A. You know, that's actually a harder issue. To the extent I 7 understand the biology, it took many, many years to come up 8 with an animal model that would support that -- the association 9 seen in humans. And the mechanism by which benzene causes AML 10 still has some elements that are not clear. So that one's not 11 a yes/no answer. But largely I would say yes, for benzene and 12 AML as a group of diagnoses, the plausibility is reasonable. 13 Q. Okay. Coherence: You would agree that the coherence with 14 benzene and AMLs generally, that that is satisfied, those 15 considerations? 16 A. Yeah. It doesn't conflict with any generally known facts 17 about the natural history and biology of AML. In general, 18 again, not looking at APL specifically. 19 Q. Now, with respect to experiments, you would agree there is 20 ample experimental evidence that preventative action does, in 21 fact, prevent AMLs generally? 22 A. No, I don't think there's any experimental evidence. What 23 Hill is talking about is instituting some control. And he was 24 actually talking about the cotton industry where, when they 25 controlled exposure, they were actually able to reduce the 78 1 incidence of byssinosis. 2 No, I don't think there's any experimental evidence where 3 we can show that reducing benzene exposures has prevented the 4 disease. I don't think it's there. 5 Q. Oh, okay. And I understand your distinction. I was 6 actually thinking about studies of industries that had large 7 benzene exposures, and when those benzene exposures were 8 eliminated, the rates of AMLs generally went down. 9 A. I'm not aware of any such study that has shown that. 10 Q. If you don't know, that's fine. 11 Number 9: Analogy. Do you believe that the analogy 12 consideration is satisfied in benzene causing AMLs generally? 13 A. No, there's no analogy. There's no other compound like 14 benzene that causes AML. 15 Q. Okay. So what I hear you saying is even though Number 8, 16 you're not aware of; Number 9, by analogy you're not aware of; 17 Number 6, there's not really a yes or no on plausibility, 18 that's still with respect to benzene causing AML generally, 19 that the Hill considerations that it does satisfy are such that 20 you believe benzene causes AMLs in humans -21 A. Yes. 22 Q. -- with sufficient -23 A. Yes. With sufficient exposure, with the right timing, 24 yes, I do. 25 Q. All right. I think that's all the questions that I have, 79 1 Doctor, at this time. 2 MR. LEGHORN: Your Honor, just a brief redirect? 3 REDIRECT EXAMINATION 4 BY MR. LEGHORN: 5 Q. Dr. Garabrant, while I get set up, you talked about 6 sensitivity testing. What do you mean by that? 7 A. Well, what I was doing with the various calculations is 8 assessing how sensitive the association in the Travis study was 9 to different assumptions made by Dr. Smith. And he made a 10 bunch of assumptions in doing his calculations. And so I was 11 testing different assumptions and seeing what the effect was. 12 MR. LEGHORN: Your Honor, I'm going to use this to 13 make it work, not in the full-screen mode -14 THE COURT: All right. 15 MR. LEGHORN: -- because I may be flipping back and 16 forth. I just want to warn the Court about that. 17 BY MR. LEGHORN: 18 Q. This was the slide -- and I'll try to make it as big as I 19 can here -20 A. Put it on the slide-show version. 21 Q. I don't want to because when we flip back and forth... 22 A. Okay. 23 Q. And one of the things, Doctor, here is you said there were 24 a number of assumptions there, correct? 25 A. Yes. 80 1 Q. And here the issue -- can you tell us, in the "observed," 2 what were the assumptions that you were making to come up with 3 that number? 4 A. Okay. Well, the first assumption is that it's even 5 meaningful to infer that there were cases of APL where Travis 6 and colleagues didn't find them. And that's a very strong 7 assumption -- I should say that's a very questionable 8 assumption, that you can do that. I don't think you can; I 9 don't think you should. But that was Dr. Smith's method. 10 So I said, all right. We'll allow the idea that you can 11 create cases where none exist, and so let's use Yang's evidence 12 of the proportion rather than Smith's assumption of a 13 proportion. 14 Q. And when you're talking about creating cases, that's what 15 was displayed on this slide here; is that correct? 16 A. Yes. 17 Q. That we now see the one where you have the red arrows and 18 where under the total cases 9.2 apply, right? 19 A. Yes. Travis reported four cases of APL. Dr. Smith's 20 method creates five additional cases that were not diagnosed. 21 Q. And so when you did the sensitivity testing you, in fact, 22 in the -- created an additional two cases? 23 A. So I said if you change the proportions from Smith's 24 assumption of .286 to Yang's reported .1 -- I can't remember 25 it. I apologize -- to Yang's reported 18.7 percent, this is 81 1 the effect of changing that assumption -- you get 5.98 2 observed. So you've invented two cases instead of 5.2. 3 Q. And to do this you used Dr. Smith's estimate of three to 4 eight cases per million; is that correct? 5 A. Okay. Here I am using, in order to calculate the expected 6 number of cases, Dr. Smith's testimony that the background rate 7 is somewhere between three and eight cases per million, which 8 yields expected numbers of 2.3 to 6.3. 9 Q. And that's three times .187 and eight times -- you 10 adjust -11 A. No. No. No. No. No. No. That's using Smith's 12 estimate of background rate, okay, three per million person 13 years or eight per million person years multiplied by the 14 roughly 750,000 person years at risk in the benzene cohort. 15 Q. And then when you did the calculation that you showed in 16 your direct testimony using information from the Zhang study 17 here, you were using the four actual cases reported as APL by 18 the 12 doctors in the Travis study, correct? 19 A. Okay. This slide shows, yes, the four observed by Travis 20 that they felt existed. Right, that's the four observed. 21 Q. And in the other slide, just to make sure that we're 22 clear, you're doing the Smith technique of generating extra 23 cases? 24 A. Yes. Generating an extra 1.98 cases using Dr. Smith's 25 method. 82 1 Q. And in the field of epidemiology, can you tell us whether 2 generating additional cases where the diagnosis is unknown is 3 an accepted method? 4 A. I've never seen that done. Not to my memory, no. 5 MR. LEGHORN: No further questions, your Honor. 6 MR. STEWART: I don't want to do any more math. 7 THE COURT: All right, sir. Thank you. You may step 8 down. 9 (The witness is excused.) 10 MR. WEATHERS: Your Honor, the defense calls Dr. John 11 Bennett. And as Dr. Bennett is making his way to the witness 12 box, the time issue: Is there a chance of us extending a bit 13 today, if we need to, beyond one? 14 THE COURT: It depends on what you mean by "a bit." A 15 bit, yes; a bunch, no. 16 MR. WEATHERS: So we don't have an hour more; we have 17 minutes more? 18 THE COURT: No. 19 MR. WEATHERS: All right. We will do our very best, 20 your Honor. 21 And in that regard, there are two papers that I 22 believe have not been previously identified in this proceeding. 23 I would like to hand those up to the Court. I'm providing a 24 copy to Dr. Bennett and a copy to counsel. 25 JOHN M. BENNETT, sworn 83 1 THE CLERK: State your name, spell your last name for 2 the reporter, keep your voice up, and speak into the mic. 3 THE WITNESS: I'm John M. Bennett, B-E-N-N-E-T-T, 601 4 Elmwood Ave., Rochester, New York 14642. 5 DIRECT EXAMINATION 6 BY MR. WEATHERS: 7 Q. What's your current position, Dr. Bennett? 8 A. I am professor emeritus in Laboratory Medicine and 9 Pathology at the University of Rochester, and currently in 10 active practice of hematopathology. 11 Q. Doctor, the Court -- Judge O'Toole has your report, will 12 have your CV. And in the interest of time, if you will allow 13 me, sir, I'm going to summarize a bit of your considerable 14 background. Is that fine? 15 A. That's fine. 16 Q. Doctor, you graduated from Harvard College, cum laude; you 17 graduated from Boston Medical College, from Boston University 18 College of Medicine, cum laude; took an internal medicine 19 rotation at Boston University Medical Center and a hematology 20 fellowship at Boston City Hospital, correct? 21 A. Correct. 22 Q. And you then had some experience with the U.S. Public 23 Health Service, correct? 24 A. Yes. Before that I was a staff hematologist at the Beth 25 Israel Hospital in Boston and in charge of the Hematology 84 1 Diagnostic Laboratories from 1963 until 1966. 2 Q. And then later you had some involvement with Tufts Medical 3 School where you taught, and with an outpatient hematology lab 4 at Boston City, correct? 5 A. That's correct. 6 Q. Tell us just briefly, Doctor, have you had involvement on 7 advisory commissions or editorial boards prior to the time that 8 you became an emeritus professor? 9 A. Oh, yes. Extensively. 10 Q. Give us examples. 11 A. I've served as chairman of an NIH advisory board on 12 chemotherapy in a variety of different cancers sponsored by the 13 National Cancer Institute; I serve currently as chairman of the 14 Scientific Advisory Committee of BioReference Laboratories; 15 chairman of the Myelodysplastic Syndrome Foundation; and also 16 editor-in-chief of the International Journal of Leukemia 17 Research, which is one of the articles you heard about this 18 morning. 19 Q. What did you do professionally, Doctor, prior to the time 20 that you became -- that you took emeritus status? 21 A. I served as clinical director of the cancer center at the 22 University of Rochester and saw many patients, primarily with 23 hematologic diseases, and also served as a pathologist of 24 reference for the Eastern Cooperative Oncology Group looking at 25 anywhere from six to seven hundred cases a year of acute and 85 1 chronic leukemias. 2 Q. Doctor, during that time period, so we're all clear, did 3 you actually diagnose and treat leukemia patients? 4 A. Oh, yes. 5 Q. Was that a major portion of your professional life? 6 A. It was a majority of what I did. 7 Q. Post-emeritus -- after you took emeritus status -- tell us 8 about your professional practice in summary, please. 9 A. I serve as a consultant to the medical staff of the cancer 10 center, and then one week a month actually sign out all the 11 bone marrows and peripheral blood smears of patients who had 12 marrows performed looking for a diagnosis. So it's a formal 13 sign-out. 14 Q. And, Doctor, is that only cases referred from within your 15 institution? 16 A. About 85 percent from my own institution, 15 percent from 17 other institutions. 18 Q. Doctor, in testimony -- first of all, you were not here 19 with us Tuesday, correct? You were not present in the 20 courtroom? 21 A. That's correct. 22 Q. But you have subsequently had an opportunity to read Dr. 23 Smith's testimony; is that correct? 24 A. That's correct. 25 Q. All right. And, Doctor, I think that Dr. Smith's made 86 1 some comment about he wasn't sure whether you had done any 2 research regarding causation of disease. I would ask you that 3 question, sir: Have you? 4 A. Yes. In addition to a series of publications on diagnosis 5 as well as management of leukemias, I have participated in at 6 least a half a dozen epidemiologic studies looking at causation 7 of a variety of different agents including cancer chemotherapy 8 drugs, ionizing radiation that occurred in Japan after the 9 detonation of the two atomic bombs, and at least one study on 10 potential benzene exposure in a prospective study for the Shell 11 Oil Company. 12 Q. All right. Doctor, to this point we've all heard a great 13 deal about the -- what's called the FABs. I'm sorry. I 14 apologize. We apparently don't have that slide. But -15 A. It's disappeared. 16 Q. It seems to have. But nonetheless, what I was going to 17 put up was a list of all of the FABs, correct? 18 A. Right. And I founded the French-American-British working 19 leukemia group in the early 1970s. And over the past 35 years 20 we have produced approximately 15 papers codifying the various 21 types of both acute and chronic leukemias. And they're 22 referred to as the FAB classification. 23 Q. So, Doctor, when we talk about the French-American-British 24 system of hematological identification, would it be fair to say 25 that you were the American portion of that? 87 1 A. I and one other investigator from the National Institute 2 of Health, Dr. Harvey Gralnick. The two of us. There were 3 seven members of the group. 4 Q. And when was that, approximately, Doctor? 5 A. The first meeting occurred in 1972; the first publication 6 in 1976; and there have been a series of publications including 7 the most recent one in January of this past year. 8 Q. Now, Doctor, we've also heard a good deal about the WHO, 9 the World Health Organization, and the ICDs. Are you familiar 10 with those? 11 A. Yes, I am. 12 Q. And have you had any involvement with those directly at -13 A. Yes. I was one of the original members that put together 14 the chapters for both the acute and -- acute leukemias and the 15 myeloplastic syndromes, the first edition that was produced ten 16 years ago and the more recent update that was just released 17 last year. 18 Q. Doctor, the defense asked you to serve as an expert 19 witness in this case and to consider whether there is reliable 20 science to connect or associate the disease, acute 21 promyelocytic leukemia, with exposure to benzene. And have you 22 reached an opinion on that question, Doctor? 23 A. Yes, I have. 24 Q. And what did you do to inform yourself prior to reaching 25 that opinion? 88 1 A. Well, based on my experience as a hematopathologist and 2 diagnosing and managing cases, serving as an inpatient director 3 of the leukemia service for many years, talking with patients 4 and then reviewing the literature, much of which you have heard 5 already over the past two days. 6 Q. All right. And, Doctor, what was your opinion after 7 reviewing those sources and obviously bringing to bear your 8 considerable background? 9 A. That there is no reliable evidence to support a linkage 10 between benzene exposure and APL. 11 Q. Doctor, I'm going to generally characterize that one of 12 the principal predicates of Dr. Smith's testimony was that 13 benzene causes AML, and APL is one of the -- APL is one of the 14 AMLs. Now, Doctor, my question is in that regard: Do you have 15 an opinion as to whether APL, acute promyelocytic leukemia, is 16 a unique and distinct disease? 17 A. Yes, I believe it is. 18 Q. And we have now a slide in front of us. Can you refer to 19 that and give the Court your basis for that conclusion? 20 A. So of all of the leukemias we see, acute promyelocytic 21 leukemia has a very unique morphology. By that we mean what 22 the cells look like under the microscope. It has very unique 23 chromosome abnormalities. And then there are a series of other 24 steps that I'll come back to. 25 So that if we could go to the next slide. 89 1 Q. Is this it? 2 A. Yeah. I can't read... 3 Q. I'm sorry. It's too dark? 4 A. Well, it's not the font that we originally had which is -5 I don't know how it shows up. 6 Q. It shows up fine on my -7 A. Okay. I -- it's showing up as black on my monitor. 8 Q. So you can't read it? 9 A. Well, I can -- yeah, I guess I can read it. All right. 10 So, your Honor, what I've tried to do here is to define 11 the various types that are commonly called APL. And as you can 12 see, over on the right-hand side I've listed the approximate 13 percentages, as though we were dealing with 100 consecutive 14 cases, of all of these different cases of APL. Obviously, the 15 most common type is what we call hypergranular APL that has the 16 15;17 translocation, and the rearrangement between the PML and 17 the RAR-alpha gene. And that constitutes about 83 percent. 18 Then there is what we call the microgranular variant. And 19 we've used that term back in the early '80s to describe about 20 15 percent of patients who, for reasons that we still do not 21 understand, have a different morphology of most of their cells. 22 Occasionally you'll see the classical cell, but you really need 23 to have the chromosome information to be sure you're dealing 24 with APL. It has the same genetic rearrangements; its biologic 25 behavior is the same as far as its responsiveness to 90 1 chemotherapy and all-trans retinoic acid, which has become the 2 hallmark of the treatment success of this disease. 3 Thank you. 4 (Laughter.) 5 THE WITNESS: I thought maybe it was my glasses. 6 So now we're down to two. And the problem in the 7 terminology is that we describe what we call M3v the variant, 8 which was agranular or hypogranular. I'll illustrate that in a 9 moment. But then others have said, yes, but there are some 10 variant translocations. And they're, then, characterized by A, 11 B and C. And these represent a trivial percentage of the cases 12 we see. And we believe on morphologic grounds that we can 13 separate these out pretty well, and we have so published in 14 Leukemia on that. 15 All of these variants involve Chromosome 17. Every 16 one. But they have a different part of the chromosome. One 17 has -- and there are two, interestingly enough, that are both 18 11;17s, but the breakpoints on the long arm of the chromosome 19 is at a different site; and, therefore, you get a different 20 rearrangement of the gene products. But they all have an 21 RAR-alpha, but they are different genes that are defined there 22 for you. The importance of this is that when you recognize 23 this, these patients do not respond as well to combination 24 chemotherapy with all-trans retinoic acid. 25 And then very recently -- and Dr. Smith referred to 91 1 this in his testimony two days ago -- two pediatric cases of an 2 unusual 5;17 -- so that's Chromosome 5 hooked the Chromosome 3 17, so they formed a fusion chromosome -- have been described 4 that has yet another part of the gene -- and this does not 5 respond to all-trans retinoid acid that we call ATRA, either. 6 So that basically defines APL in all of its variants in sort of 7 a word fashion. 8 Now, if we can go back. So we've now covered Points 1 9 and 2. Points 3 and 4 I'll come back to. 10 Characteristic immunologic phenotype was referred to 11 earlier. This means that there are proteins on the cell 12 membranes of all of these cells which we can identify with 13 antibodies. So we purchase these antibodies, we attach them to 14 a chromophobe, which is a fluorinated colored compound, and we 15 flow them in a machine that flows literally millions of cells, 16 sorts them out by the type of protein that is present on the 17 cell membrane. And we have published on this. We can come up 18 with a characteristic phenotype of APL that is present in about 19 92 percent of the cases. It can be seen rarely in some other 20 patients with different types of leukemia. And I've just 21 listed here for you three of the antigens that we see 22 characteristically: CD33; '13; and HLA-DR, which is on the 23 minus side, is not present. 24 Now, '33 characterizes a cell that has gone one step 25 beyond a stem cell. So you have to keep in mind when we talk 92 1 about stem cells, we're talking about at least a cell that has 2 CD34 on its membrane, and the vast majority of patients with 3 APL do not have CD34. So try to remember that. 4 High curability with chemotherapy and ATRA: For 5 classical APL we currently are curing 85 percent of all 6 patients, pediatric as well as adults, which has been a major 7 achievement over the past decade. So it is the most curable of 8 all of the acute leukemias that we see in adults, whether it's 9 acute lymphocytic leukemia or acute myeloid leukemia. 10 Now we can go to the pictures. 11 All right. So this is an example of a classical case 12 of acute myeloid leukemia. And it's acute myelogenous 13 leukemia. And if you look at the lower right, the second cell 14 up in the cytoplasm -- now, think about a fried egg, okay? The 15 yellow part of the egg is the nucleus, and the white part of 16 the egg is the cytoplasm. So in all of these cells the nucleus 17 is purple and the lighter color is the cytoplasm. 18 And in that cell, on the lower right you'll see a 19 little small rod-like structure that is colored pink. That is 20 called an Auer, A-U-E-R, and was defined by Dr. Auer almost 100 21 years ago. It is a classical defining feature of acute myeloid 22 leukemia. We can see cases of AML without it, but when we see 23 that particular cell, you know you're dealing with acute 24 myeloid leukemia. 25 This particular case -- and you have to take my word 93 1 for this -- comes from a patient who has an 8;21 translocation, 2 one of the specific translocations which is highly curable as 3 well in AML, but the cure rate is closer to 60 percent rather 4 than 85 percent. 5 So if we go to the next slide, this is APL, classical 6 hypergranular APL. Why hypergranular? If you look where the 7 white should have been, it's now replaced by innumerable 8 pinkish granules and darker rod-like structures. And if you 9 look at the upper right, below the red cells you'll see a 10 little piece of cytoplasm has split off the cell and has a 11 little rod-like structure in it. 12 Now, this is what happens: These cells self-destruct. 13 They circulate in the bloodstream. They release an enzyme that 14 converts fibrinogen to fibrin. And if you don't treat these 15 patients quickly and promptly, they will bleed out. So this is 16 an example of a case of hypogranular promyelocytic leukemia, 17 and these granules contain proteins that trigger the 18 coagulation system to clot. And it will clot, and you will 19 lose all your clotting factors, and patients bleed out. They 20 have brain hemorrhages or pulmonary hemorrhages. So very -21 treatment is required instantaneously. 22 Thank you. That's perfect. Next? 23 All right. Now, this is an example of an agranular, 24 or microgranular, variant. And it looks totally different. 25 There are very few cells with granules, and instead you see 94 1 this doubling of the nucleus. A very characteristic feature. 2 Now, occasionally this can be confused with another variety of 3 acute leukemia that we call acute myelocytic leukemia. And, 4 again, experts in this field of hematopathology, obviously 5 experts, are able to pretty well look at this and say, "I 6 strongly suspect this is APL, but let's wait until we get back 7 the flow and wait until we get the cytogenetics." With a case 8 like this we would institute therapy on a morphology without 9 waiting for anything else to come back. 10 Now, in the Travis paper, Dr. Garabrant was concerned 11 about one case that was called microgranular because it didn't 12 fulfill the usual criteria that we expect to see in a patient 13 with APL, which is a very high white count and a lot of 14 circulating blasts, well above the 20 percent that that 15 particular case had. 16 I think we can go back. Okay, we can go forward. We 17 can go back one. 18 BY MR. WEATHERS: 19 Q. You said you were going to come back to bullet points -20 A. Okay. So we're going to discuss a couple of papers that 21 address Bullet Points 3 and 4. We've now -- and I don't think 22 that that -- the other translocation slide helps very much. 23 All right. So about a decade ago a paper appeared in the 24 journal of Blood. Number one, what does that mean? Well, 25 Blood is believed to be -- and I think everyone would probably 95 1 agree -- the number one hematology journal in the world. They 2 accept no more than one out of ten papers that they receive. 3 So there's a very high rejection rate, which means that they 4 are selecting only the very best papers. They also have 5 reviewers who are the best reviewers in the country. So I'm 6 editor-in-chief of Leukemia Research. We reject 60 percent of 7 our papers. So we're not ranked quite as high. 8 So ranking, what we call "impact factor" which is 9 calculated by an international agency that looks at how often a 10 paper is quoted by someone else, turns out to be an important 11 mechanism of where people send papers. You don't want to send 12 a paper to a junk journal even though they might publish it. 13 So you try to send your papers to the very best journals. 14 Leukemia Research ranks in the top 25. Leukemia ranks about 15 20th; we rank about 25th; Blood ranks number one. So that's 16 just sort of background. 17 MR. WEATHERS: To clarify one other thing, your Honor, 18 that's the Turhan paper that we just handed up. 19 THE WITNESS: So this paper has four, five, six, 20 seven -- nine authors. All of these authors are from the same 21 institution in Paris. The hematopathologist in that hospital 22 is one -- when that was written; he's subsequently retired -23 was one of the FAB authors. So there is no question about the 24 diagnosis of the three cases of APL that are described in this 25 paper. He was an authority; he would not make a mistake. 96 1 And the bottom line in this paper was for the first 2 time they demonstrated by fairly sophisticated methodologies, 3 which have now been supplanted by others, that the 15;17 4 translocation and its molecular byproduct, the PML RAR-alpha 5 gene fusion, could only be identified in a cell that was later 6 than a stem cell, and by their definition a stem with CD34, 7 CD38, negative. 8 So that basically what they're saying is that based on 9 their study they thought that the cell that was capable of 10 immortality was a cell that was a step beyond the common stem 11 cell that exists in all of us that allows us to repopulate our 12 bone marrow on a regular basis as long as we live. 13 THE COURT: Let me just ask before you go any further: 14 You said that their finding was that the t(15;17) PML RAR-alpha 15 could only be identified in a cell that was later than -16 THE WITNESS: A little later than the blast. 17 THE COURT: Is this an observational finding? 18 THE WITNESS: This was done by flow cytometry, by 19 morphology, by cytogenics and molecular genetics. 20 THE COURT: So when you say "could only be 21 identified," meaning could only be observed? 22 THE WITNESS: Observed. Correct. Thank you, your 23 Honor. 24 Now, I like this paper because the implication is 25 that, you know, if the defect by a carcinogen is so early that 97 1 it involves the primordial stem cell, one that is resting and 2 only reproduces enough cells to populate us to keep us healthy 3 and then goes to sleep until it has to respond again, and this 4 is going on on a continual basis, why is it -- a rhetorical 5 question -- why is it that we cure anybody? In other words, 6 you have to have somewhere in your bone marrow a population of 7 stem cells that survive the assault of our drug therapy and 8 that are normal. They must be sitting there in a niche 9 somewhere. Otherwise, we would give combination chemotherapy 10 to these patients, we would wipe our their bone marrows, they 11 would become an empty bone marrow like aplastic anemia, which 12 is a very uncommon disease that we see, and there would be no 13 repopulation. So that just, as an aside, makes me think that 14 this -15 So this paper was published and it just sort of sat 16 there for about ten years. No other papers that I could find 17 attempted to reproduce this until we come to a very recent 18 paper that is so recent that it's still in press but is 19 available through the wonderful system of the web, and the fact 20 that now we're required to put all our papers up as soon as 21 they've been accepted and not necessarily in hard copy in a 22 journal. So this paper, which is in Leukemia won't be in hard 23 copy in Leukemia for several months, but it is nevertheless 24 available. 25 Now, this is totally different. What these 98 1 investigators did, basically, was to use a mouse model that is 2 susceptible to leukemia, and then they gave us a product that 3 would cause acute promyelocytic leukemia, which is the PML 4 RAR-alpha fusion. So this is not the first time this has been 5 done. In chronic myeloid leukemia, which we call CML, the 6 DCR-able product, which is a result of a fusion of two 7 chromosomes, is able to induce in experimental rabbits a 8 disease that is exactly analogous to chronic myeloid leukemia. 9 So we know that this -- these products that we can 10 identify in at least 15 to 20 percent of leukemias that we see 11 play a major role in the development of that leukemia. Now, 12 that doesn't tell us what the causative agent was that forced 13 this chromosome abnormality to occur; it just says that once it 14 occurs, this is the triggering product that arrests the cells, 15 does not allow them to grow, crowds out the bone marrow and 16 produces acute leukemia. 17 Okay. So now let's go to the next illustration. And 18 this is going to blow you away. I'm sorry. And I'll apologize 19 at the outset. The first thing you should see is there are a 20 lot of colors. If you're color-blind, you cannot be a 21 hematopathologist. Impossible. We are totally dependent upon 22 recognition of colors. This is a six-color flow cytometric 23 analysis; in other words, there are six different chromogens, 24 or chromophobes, that have been attached to a variety of 25 proteins, and then the cells are extracted from the mouse and 99 1 flowed automatically, and we get the ability to produce these 2 graphs. 3 And all I want you to appreciate is on the left-hand 4 side there are three boxes. Contained within these boxes are a 5 series of patterns or colors. Each little dot represents a 6 cell. So if you look at the upper left, the big blot of red 7 are promyelocytes and some later myeloblasts. The plot above 8 it are even earlier cells, but they still are in the myeloid 9 family. To the right, the blue color, are normal progenitor 10 cells, the earlier cell to date that any investigator has been 11 able to identify using different antigens and proteins than the 12 paper from ten years ago. Okay. That's the control, all 13 right? No leukemia. 14 The next panel down is the same animal strains in 15 which the investigators had injected a vehicle. But it's a 16 dummy vehicle; it doesn't contain the gene product that we've 17 been talking about, the ATRA PML. And that's exactly the same. 18 So the vehicle is not capable of causing leukemia. There is no 19 difference. And on the right where it says "progenitors," 20 those are all the developing cells: the megakaryocytes, the 21 erythrocytes, the monocytes. And you can see there's no 22 difference. 23 The third panel down is the leukemic mouse. So what's 24 happened? The green has disappeared; the blue are only a few 25 little dots left. So what's happened is you get a crowding 100 1 effect of the normal stem cells, which are in the upper right, 2 and it's replaced by a huge number of these malignant abnormal 3 promyelocytes that contain the PML RAR-alpha and the 15;17 4 translocation. And then if you look at the right, everything's 5 gone. Everything has disappeared except the granulocyte 6 macrophage colony, GMC, which you would expect, because those 7 are the progeny of the promyelocytes. And in every case of 8 promyelocytic leukemia, we do see a small percentage of some 9 more mature cells. 10 The right is just a bar graph giving you a 11 quantitative picture of the qualitative pictures that I showed 12 before. 13 BY MR. WEATHERS: 14 Q. Doctor, in the interest of time, let me ask you this 15 question: Do these two papers, together with your considerable 16 experience in this field, inform your opinion as to where the 17 mutation that leads to APL occurs in terms of cell maturation? 18 A. Both papers strongly suggest -- we can go back to the 19 previous one. No, one more back. 20 Q. Okay. 21 A. -- strongly suggests that the leukemic cell is a step 22 beyond the common stem cell. And I've illustrated in the first 23 bullet point that in addition to this study, there are other 24 studies that have shown similar findings with other, rarer 25 types of acute leukemia, one of which is acute myelocytic 101 1 leukemia, or in the FAB classification, what we call M5. 2 Then we can go forward. 3 So where does this put us? Anybody can do a schemer. And 4 you can do a vertical; you can do a horizontal. It's all in 5 the way you want to try to illustrate what you think is going 6 on in the bone marrow. But I don't know what's going on in my 7 own bone marrow. These cells are all mixed together. They 8 don't have instructions like this, but they do respond to 9 signals that tell them what to do, where to go, when to divide, 10 and when to mature. 11 So all of these hematopoetic hierarchies, or schemers, 12 basically, are drawn up by individuals who are knowledgeable in 13 the field, may have done research in the field, and are trying 14 to set up a pathway for people to look at and say, you know, it 15 begins with a very long-term cell that is immortalized. 16 Now, I hope you all appreciate the fact that 17 "immortalized" means you never die. We're mortals. I don't 18 have to say anything further, right? Okay. So we're going to 19 die. That's built into our system. But as long as you live 20 there are stem cells in the bone marrow, in the 21 gastrointestinal tract, in the testicle, that will continue to 22 be produced in the skin as long as you live. So that does not 23 dictate why you die. You die because, unfortunately, we don't 24 have those same kind of stem cells in the heart or the liver or 25 the brain or the central nervous system or the kidneys. So 102 1 sooner or later these organs fail or we get cancer and we die. 2 So be it. 3 Q. Doctor -4 A. So if you go down this pathway, based on those two papers 5 I would place the leukemic cell of APL at at least the CMP 6 line, or the GMP line. And it's really just a guess on my part 7 based on those two -- thank you -- observations. But it's a 8 long way from the HSC. 9 Q. And, Dr. Bennett, just to put this in the context of a 10 slide that we moved past in the interest of time, Dr. Smith's 11 opinion, as stated in his original declaration and elsewhere, 12 is that the mutation occurs which leads to APL and all other 13 AMLs at what he called the pluripotential stem cell, correct? 14 A. That's correct. 15 Q. Is that the HSC? 16 A. Yes. Well, yes, or it might be one step -17 Q. We're giving -18 A. It's not rocket science, unfortunately. 19 Now, I would agree with Professor Smith that there are 20 clusters of different types of leukemia that appear to involve 21 more than just a single line. So, for example, in 22 myelodysplasia and some of the secondary leukemias which were 23 mentioned by Dr. Smith where there are chromosome aberrations 24 including, like, Chromosome 5, Chromosome 7, we tend to see a 25 picture in which all of the bone marrow cells, except the 103 1 lymphocytes which, your Honor, over on that the right side of 2 that tree share morphologic abnormalities which strongly 3 suggests that you have to go at least above where you have 4 drawn your circle. How far above, no one knows. But as I've 5 indicated, there are several types of acute leukemia that 6 involve just the monocyte pathway or the granulocytic pathway 7 which appear to be what we call downstream rather than upstream 8 of the multipotential progenitor. 9 Q. And this is, in your opinion, Doctor, unique for APL; that 10 is, the -11 A. What I have been describing is unique for APL. 12 Q. Thank you, sir. 13 Now, I'm going to pass by this slide. I think we -14 A. This just shows us, your Honor, that you can do it 15 vertically or you can do it horizontally. It doesn't make any 16 difference at all. It depends on how you like to look at 17 things. 18 Q. And just for clarification, the same area would be -19 A. That's good right there. 20 Q. That's where -- on this slide that Dr. Smith used on 21 Tuesday, that's where you believe -22 A. Yes. And the authors of the Wojiski paper state that it 23 could be there or it could be one step earlier. 24 Q. Okay. Thank you. 25 Conclusion on the cell of origin, Doctor. 104 1 A. We've already discussed this so -2 Q. All right. Then in the interest of time and being fair to 3 counsel to allow them cross-examination, we'll move on. 4 Doctor, my question is -- with regards to this next slide, 5 my question is this: First of all, I think at one point Dr. 6 Smith said so far as he knows researchers aren't even looking 7 at the uniqueness of APL as regards benzene exposure and any 8 secondary causes of APL. The paper that we now have up, is 9 that of Dr. Douer? 10 A. The reason for showing this, I'm a clinical hematologist. 11 I've done some bench research years ago. I'm not a molecular 12 biologist; I'm not a stem cell researcher. Basically, I take 13 care of patients and I make diagnoses, and I've published a 14 fair amount in this area. 15 Dr. Douer, who is at the University of Southern California 16 and a renowned clinical hematologist, one of the experts in 17 APL, has written a chapter. And that is his statement that I 18 don't need to read to you. So basically, he and I are in 19 agreement. Do we talk at meetings? Yes. Do we write papers 20 together? We are co-authors on some publications. But he is 21 speaking independently from his own personal perspective on 22 APL. 23 Q. And I've moved to the next slide, and I'll just read the 24 pulled quote from it. "APL is an example of a truly unique and 25 distinct entity within AML because of its distinct clinical, 105 1 therapeutic, biological and molecular differences." 2 Do you agree with that? 3 A. I agree with that statement. 4 Q. And Dr. Douer also reached some conclusions as regards 5 whether there are any known causes, either environmental or 6 occupational risk factors, for APL, correct? And in the pulled 7 quote that we have there he concludes by saying, "So far no 8 environmental and/or occupational risk factors have been found 9 for APL." 10 A. That's correct. I mean, one could modify that a bit by 11 saying that we know that we can cause APL in patients in whom 12 we administer chemotherapy. And that's what we refer to as 13 secondary AML. And there are well-described multiple case 14 reports and series of chemotherapy compounds that interfere -15 and you heard this discussion from Dr. Smith two days ago -16 with topoisomerase II that are capable of causing APL. 17 Now, you have to survive to get it, so, you know, don't be 18 too critical of us, because our attempts are to cure or prolong 19 life of these patients with diseases that would destroy them if 20 we didn't offer them some treatment. But overall, 21 approximately 5 percent of patients that we cure will come down 22 with a secondary AML. And I give lectures on this to staff 23 physicians and doctors at various conferences. And we have a 24 couple of slides that we could illustrate this for you as to 25 how we think about occupational- or therapy-related, or what I 106 1 prefer to call event-related, a more generic term. So there 2 are a lot of events that can cause leukemia to occur, and a lot 3 of these have been identified. 4 Q. I'm going to move on, Doctor -5 A. Oh, my. 6 Q. -- in the interest of time. 7 A. We're seeing this again, right? 8 Q. We are. And this needs little introduction. We've talked 9 at length about the Travis study. I simply want to essentially 10 re-create for you a slide that Dr. Garabrant has also already 11 discussed. And I'll represent to you that these are the four 12 alleged APLs from the Travis study, and this is a summary of 13 all of the histological/morphological data contained in the 14 paper. 15 And I want to first -- with that representation I'm going 16 to first refer you to Case 5 and ask you, sir, to take a look 17 at the data reported and tell me as a hematopathologist, and 18 one with your experience and background, is there enough data 19 about Case 5 to allow you to reach a diagnosis that the 20 individual had APL? 21 A. Obviously I would have loved to have had the opportunity 22 to review this material myself, but Dr. Lee was chosen by the 23 Mayo Clinic I think partly because -- Number 1, he's an expert; 24 Number 2, he speaks the language, and that obviously helps 25 considerably. My problem with the Travis paper is a little 107 1 deeper than that; that is, that Dr. Travis states in various 2 parts of the paper that they were unable to review all of the 3 pertinent material. They had to accept what was told to them 4 by doctors who took care of these patients that they did have 5 what they were supposed to have. 6 And you have to remember that China went through a 7 horrendous period of ten to 15 years in which their top 8 physicians, their top professors, were out farming and raising 9 rice in the rice patties rather than working at the medical 10 school. So there was a period of time in China, and it still 11 exists today, where the reliability of their database has to be 12 questioned. 13 What they tried to do, and I give them tremendous credit 14 for this, is to assemble all of this data. And you cannot 15 determine from reading this paper how many slides, microscope 16 slides, they actually looked at themselves. But they did the 17 best they could, they analyzed it as well as they could, and 18 this is what they came up with. 19 I am perfectly willing to accept Cases 6, 7, 8 as highly 20 likely to be hypogranular promyelocytic leukemia. I make that 21 diagnosis without waiting for the cytogenics to come back, 22 without waiting for the flow cytometry. I feel comfortable -23 and I'm sure if Dr. Lee saw those cases -- that's what they 24 were. 25 Case 5 is the questionable case because, as you remember, 108 1 that's the one that we had problems with. When you don't see 2 the granules, is it APL or is it another type of FAB like a 3 monocytic leukemia that can resemble APL? And I can assure you 4 that many of the cases that we see are misclassified because of 5 that observation. Moreover, at the final column, the white 6 count of 1.6 is distinctly uncommon in the variant form of APL. 7 The usual white count is 25- to 30,000 for that particular type 8 of leukemia. And moreover, this is much less common in adults 9 than it is in children. So when you look at that and you say, 10 "Well, how confident are you that Case 5 -- are you as 11 confident as Case 6, 7 and 8," I just don't have that same 12 confidence. 13 Q. All right, Doctor. You mentioned a moment ago that there 14 are some slides that you use in teaching having to do with 15 therapy-related AMLs and MDSs? 16 A. Correct. 17 Q. Is the slide in front of us now one of those? 18 A. This is one. We can either use this one or use -- there 19 was another -- okay. Let's start with this because I think 20 it's -21 Q. And, again, in the interest of time, summarize for us -22 A. I'll summarize. 23 Q. I want to be fair to counsel and give them time to 24 cross-examine. 25 A. I certainly want counsel to question me, no question. 109 1 Okay. So basically what I've done here is to summarize a 2 huge amount of literature which looks at event-related MDS -3 or myelodysplastic syndrome -- which is sort of a preleukemia. 4 A big percent of patients with MDS will develop acute leukemia. 5 But they all have a malignant leukemic blast in their bone 6 marrow; it's just a question of how many you count. We have a 7 convention that says until you get to 20, you don't call it 8 preleukemia. 20 percent. 9 So on my left is the typical alkylating/radiation, so 10 ionizing radiation. Chemotherapy drugs that we used to use for 11 the treatment of Hodgkin's disease, for ovarian cancer, we 12 continue to use for the treatment of patients with lymphomas of 13 all types, like Cytoxan. Everyone agrees that if you've been 14 exposed to these agents, with enough exposure there is a long 15 latency period -- "latency" means from the time of first 16 exposure to the end when the disease is diagnosed. And that 17 exposure is usually somewhere between three and 12 years -- in 18 virtually every paper, and we've written several papers on 19 that. 20 The verbiage to the right of the arrow says that when you 21 have this kind of leukemia, the vast majority of patients have 22 certain chromosome changes that are seen characteristically. 23 And that was found in going back to the Golomb paper that you 24 heard about yesterday and the day before, Dr. Reilly's work, 25 Mitelman's work. There are innumerable papers that show that. 110 1 So there is a characteristic chromosome pattern when you 2 see these secondary leukemias. A primary leukemia in which we 3 cannot identify any cognitive agent -- we call that idiopathic, 4 and you've heard that term used -- you would expect to see 5 abnormalities of Chromosome 5 and 7 no more often that 15 to 20 6 percent. 7 So this defines, then, the kinds of changes we see. And 8 it produces a kind of leukemia that is very difficult to treat, 9 unfortunately. So it's bad news twice. You've cured the 10 patient, they come down with this kind of leukemia, and there's 11 not much we can do for them. If they're young, we can offer 12 them sometimes a bone marrow transplant. But basically this is 13 bad news. This is terrible. 14 Q. Doctor, before you -15 A. This accounts for about 85 percent of the events that we 16 see. 17 Q. And the exposures that you're dealing with here are 18 alkylating agents and radiation? 19 A. They're well-defined exposures. 20 Now, it has been thought for a long time, including Dr. 21 Smith until his more-recent studies, that benzene sort of fell 22 into the upper area. If you look at the time, the latency 23 periods where it's been examined, if you look at a few studies 24 that have shown chromosome changes in occupationally exposed 25 individuals, you see a trend that suggests that benzene belongs 111 1 up there, at least traditionally. 2 The second group is the short-term latency. So I had a 3 patient who happened to be married to the head of our 4 veterinarian medicine program -- it always happens that way -5 with acute myelocytic leukemia that came on four months after 6 she completed her therapy for Hodgkin's disease. Was in 7 remission with the Hodgkin's disease and had an explosive onset 8 of acute myelocytic leukemia and died within a month. Exposed 9 to a topoisomerase II agent. There are a whole family of 10 these. Some of them appear to be more potent in causing a 11 second event than others. And this was also mentioned by 12 Professor Smith. Mitoxantrone is an anthracenedione not 13 related to anthracycline but functioning in a similar way, and 14 it has the highest incidence of secondary leukemias. 15 What we find fascinating, and I'll conclude on this, is 16 that if you look at some of the translocations that are 17 described, you'll see some of the same translocations that we 18 see in de novo, or idiopathic, AML. And, indeed, if you are 19 unfortunate to have leukemia but fortunate enough to have this 20 type having been exposed, they're highly treatable. And indeed 21 some of these patients can be cured. 22 The last point is that none of these patients have a 23 prelude of preleukemia, or myodysplasia. So you can really 24 draw a line between the two columns, and they don't -- they 25 don't cross over. I mean, nothing is pure in medicine. And we 112 1 have some patients who have gotten a mixture of both agents, 2 and you'd expect them to see a mixture of chromosome 3 abnormalities and a latency period that is somewhere halfway 4 between short-term and long-term, but that's only in 1 and 2 5 percent. 6 The last slide -- the slide before that one, which really 7 just puts this up so that clinicians can understand what we're 8 dealing with. So Class I was the top that I showed you before, 9 the alkylating agents, the radiation. The Class II are the 10 topo II inhibitors. The age tends to be older with the 11 alkylating agents, younger with the topo II. 12 "Unbalanced translocations"? What that means is that you 13 don't see a reciprocal translocation where the chromosomes 14 maintain all of their material; the chromosome can be lost, or 15 portions can be lost. There is a myelodysplastic phase; 16 there's a preleukemic phase on the first column. The latency 17 period is long, whereas it's short on the other one. 18 And I've tried to summarize as best I can my own 19 experience, my own publications, on what we have seen on the 20 morphology. And we have published extensively on alkylating 21 agents in ovarian cancer and in bowel cancer. And we have 22 found that a lot of these patients tend to have a type of 23 leukemia that is more recognizable in that category. M6 is 24 erythroleukemia, a combination of bad blood cell precursors and 25 myeloblasts. 113 1 And then on the right-hand side we see the other FAB 2 types. And believe it or not, one can even see therapy-related 3 acute lymphocytic leukemia as well as myeloleukemia. And, 4 finally, "Poor response to treatment," clinical response, very 5 likely, with the shorter exposure. 6 Q. Dr. Bennett, I have one last question for you. Having 7 seen Dr. Smith's original declaration, his supplemental report 8 and now his testimony unrobed here in these proceedings, has 9 anything changed your opinion that you have provided this Court 10 about APL and benzene? 11 A. No. 12 Q. Thank you, sir. That's all I have. 13 CROSS-EXAMINATION 14 BY MR. STEWART: 15 Q. Doctor, we've never met before, right? 16 A. Correct. 17 Q. Nice to meet you. I'm Al Stewart. 18 A. Same here. 19 Q. You saw me earlier go through this exercise with Dr. 20 Bennett, and I would like to do it with you since you sort of 21 understand where I'm heading. What I would like to do is go 22 through basically how much money you have made over the course 23 of 2008, '7, '6 and '5, doing this kind of thing, which I will 24 define "this kind of thing" as providing expert testimony or 25 assistance to defendants in litigation. Can we do that 114 1 quickly? 2 A. Well, number one, I'm not prepared to respond because I 3 did not understand, nor was I told, that I should bring my tax 4 return forms to the courtroom. So what I'm giving you is off 5 the top of my head. And basically, what I can say is my 6 university salary -- number one, I'm 50 percent full time with 7 the university. So that means 20 hours a week. Doctors work 8 80 hours a week, so you can -- so it's 40 hours. But it's 20 9 hours a week according to the law. And I get $110,000 for 10 that, okay? 11 I receive from BioReference Laboratories and Leukemia 12 Research about $50,000, serving as chair of those two 13 organizations. And I give lectures, do -- write chapters for 14 books, royalties from that, and that totals somewhere in the 15 neighborhood of $50,000 a year from the various talks that I 16 give. 17 I also have a -- provide a service to about eight 18 pharmaceutical companies who are doing studies of leukemia and 19 related disorders, and I sign a contract with them to review 20 those slides for quality control, and they pay me to do that. 21 I give 25 percent of that to the University of Rochester. And 22 that amounts to another 50- or $60,000 a year. I have a 23 microscope at home as well as at the office. So it's a lot of 24 work. 25 Approximately $70,000 a year to $80,000 a year over the 115 1 past four, five years comes from toxic-tort type of work where 2 I give declarations, review pathology slides, confirm a 3 diagnosis, I do a deposition, and rarely appear in court. This 4 is my first time at federal court. And I think I've been in 5 three other court appearances and two workmen's compensation. 6 That's it. 7 Q. Okay. So let's get back to my question. So that's what I 8 want to get to. My question was: How much money in 2008, 9 approximately, did you make doing this kind of work for 10 defendants in toxic court work? 11 A. $70,000. 12 Q. And is that answer going to be the same for '08, '07, '06 13 and '05, approximately? 14 A. Plus or minus 10 percent, one way or the other. 15 Q. Fair enough. So do you want me to do 70- to 80- or make 16 it 70- to 75-? 17 A. Whatever you want. 18 Q. I don't care. I want the best approximation that you 19 have. 20 A. And the reason that it's fixed is because I limit myself. 21 Because of my responsibilities as a journal editor, et cetera, 22 there's just so much time that I can devote to this activity. 23 So it's not surprising that the amount would be the same. 24 Q. No problem. I'm just looking for that range. 70- to 25 what? What do you want the top number to be? 116 1 A. 75-, 80-? Whatever. That's fine. 2 Q. Great. All right. Now I'm going to go to the next topic. 3 The next topic, Dr. Bennett, is these two papers that you 4 mentioned. You mentioned two papers of Turhan which you 5 mentioned in Blood, true? 6 A. True. 7 Q. And then the other paper you mentioned was Wojiski, which 8 isn't technically out, but because of the Internet it's 9 accessible, right? 10 A. Correct. 11 Q. So it could change between now and the time it goes to 12 press? 13 A. No. 14 Q. It can't? 15 A. No. 16 Q. Okay. Fair enough. 17 Now, with these two articles, the Wojiski and the Turhan, 18 neither of these articles were on your list of reliance 19 materials when you gave a report in this case, true? 20 A. Correct. 21 Q. Right. And so on November the 20th, 2008, when you were 22 asked to write a report for us and this Court about what the 23 basis for your opinions were, and you listed on your reference 24 list all the articles for your opinions, neither of those will 25 be found, correct? You listed 14 things -- 117 1 A. That's correct. 2 Q. Neither of those. 3 Okay. Now, what I think you ended up telling this Court 4 was that it is upon the basis of these two articles that you 5 have reached your opinion about what is going on in the stem 6 cell that you've testified about today, true? 7 A. True. 8 Q. Okay. So what you've just said is, number one, one of 9 those articles you couldn't have had in your possession because 10 it just came out, so it wasn't the basis of your opinion, 11 correct? 12 A. Which opinion? 13 Q. The opinion about stem cells and where -- and where the 14 effect of stem cells happens. At the time of your report. 15 A. At the time of my report -- yes, that's correct. I mean, 16 this -- these papers support -- give further weight -- to my 17 opinion that I offered previously. 18 Q. I understand that. But I guess my point is when you 19 rendered your opinion in your report, you didn't have any 20 papers that you cited for that support, correct? 21 A. That's correct. 22 Q. And yet it was your opinion, even without the papers, 23 right? 24 A. Yes. 25 Q. You didn't feel like you needed these papers to have that 118 1 opinion, right? 2 A. At the time I wrote the letter, that's correct. 3 Q. Right. And then when you wrote -- after you wrote the 4 report you gave a deposition, correct? 5 A. Correct. 6 Q. And at the time of that deposition, you didn't have either 7 of these papers, correct? 8 A. Correct. 9 Q. Uh-huh. And even though you didn't have those papers, you 10 formulated the opinion that you have in this case that you've 11 expressed to this Court, and you were able to formulate that 12 opinion without any scientific paper that you had in your 13 possession supporting that opinion, correct? 14 A. That's correct. 15 Q. So, sir, it's safe to say, isn't it, that after your 16 report and after your deposition, that you went out and found 17 two papers to support your opinion, right? 18 A. Yes. I was concerned about the brevity of some of my 19 remarks at the deposition and some of the problems that we had 20 with communication, and dropping the line and so forth. And 21 when I read over the deposition, I realized that my statement 22 was overly broad and I modified it. And that was appended to 23 the pages that I subsequently submitted to the deposition. 24 Q. Yes. And so, Dr. Bennett, I guess my point is this, and 25 tell me if you agree with this: that you were able to reach 119 1 your opinion on the subject of the initiation of leukemia in 2 stem cells without scientific papers to support it, right? 3 A. Without these two papers; that's correct. 4 Q. Yes. 5 A. And that was based on my judgment, as I previously 6 indicated, a review of the literature that I had available to 7 me at the time, and my sense that this type of leukemia was 8 extremely unique. 9 Q. Okay. But, Doctor, there are no other papers that you 10 cited in support of this proposition besides these two, 11 correct? 12 A. Correct. 13 Q. All right. Now, with respect to this opinion that you 14 have, there's not universal agreement that this is actually the 15 way things happen with leukemia in the stem cells, true? 16 A. I don't know. 17 Q. Well, there weren't so many papers out there that you were 18 able to list them in your report, November 20, 2008, true? 19 A. When you asked whether there was universal agreement, I'm 20 a sole person of a very large universe, and there's no way that 21 I can have discussions with the entire universe, as you define 22 it. 23 Q. That's fair. I'll change the question. 24 A. Thank you. 25 Q. You would agree with me that there are reasonable 120 1 scientists in your field that disagree with your opinion on how 2 leukemia occurs via stem cells. You would agree with that, 3 right? 4 A. I'm sure there are. 5 Q. And those people are not unreliable people, true? 6 A. I can't attest to that; I can just agree with your first 7 statement. 8 Q. Fair enough. 9 Now, I want to go to -- well, let's just back up. Dr. 10 Smith -- you believe that Dr. Smith disagrees with your view of 11 what's happening with leukemia in the stem cells, correct? 12 A. Well, my recollection is that Dr. Smith was impressed by 13 this Wojiski paper and felt that more research needed to be 14 done. And, indeed, if you read the paper carefully -- and I've 15 read it probably 50 times because it's not easy -- the authors 16 do not rule out the possibility that the leukemic event 17 couldn't have occurred at an earlier point, but that there's no 18 way that they can detect that. 19 Q. Understood. And I want to put this up on the chart that 20 we have all seen just so we're comparing things. What I think 21 I hear you saying -22 A. My Xs came back. 23 Q. Yeah, well, this was in your original slide. 24 What I think that you're saying here is even the Wojiski 25 authors agree that it is consistent with their paper for these 121 1 events to be occurring with respect to APL prior to here, 2 correct? 3 A. Yes. What they say is -- you can move the pen back at 4 least one level and possibly another level. 5 Q. Correct. 6 A. Right there. 7 Q. And, Doctor, isn't it true that at this level with this 8 cell, this cell can become all these cells that it has lines 9 to, correct? 10 A. Correct. 11 Q. All right. So by an effect on the stem cell at this 12 level, if that effect occurs, it could become erythrocytes, 13 platelets, neutrophils, monocytes, eosinophils, basophils, 14 true? 15 A. The reason I said that I can support this cell being a 16 cell of interest, if you want to use the legal sense, because 17 we know that there are cases like chronic myeloid leukemia in 18 which the Philadelphia chromosome, which identifies the defect 19 in CML, can be found in all of those cells that you listed 20 until you get down to the lymphocytes line. So clearly the 21 CFU-GEMM is a target cell for chronic myeloid leukemia. Now, 22 having said that, the only known cause of chronic myeloid 23 leukemia that has been associated with exposure has been 24 ionizing radiation. 25 If you're going to suppose that that cell can be hit by, 122 1 say, benzene, and only produce neutrophils or promyelocytes and 2 their progeny, I know of no biologic mechanism that has ever 3 been considered that would say that there is that type of 4 permissive behavior. It is much more likely that if that cell 5 became malignant because it has built in the potential to do 6 anything, it would do anything. And that is what usually 7 happens with malignancy. With most cancers, the cancer 8 recapitulates as best as it can the normal environment. And 9 the normal environment for the CFU-GEMM is to move to all of 10 these various compartments. 11 Q. Well, let me just try and say it simply: This Wojiski 12 paper that you have recently seen and you are now relying on, 13 the authors themselves say it's consistent that the stem cell 14 event is happening before X and right here, correct? 15 A. No. What they say is that they can't rule out the 16 possibility that there is a latent cell that their techniques 17 do not allow them to identify. 18 Q. Yes. And you basically told us, I think in very simple 19 terms, that rules in this cell here -- or this stage here and 20 this stage here, true? 21 A. As I indicated previously, this isn't happening. This is 22 only happening in the minds of people who decide they want to 23 show a flow diagram. So we don't know what's going on in the 24 bone marrow. So this is as good as we could get. 25 Q. All right. Let's move -- 123 1 MR. STEWART: What's your pleasure, your Honor? It 2 will be a little -3 THE COURT: Go ahead. Go ahead. 4 MR. STEWART: Great. 5 BY MR. STEWART: 6 Q. Let's move on to another issue, and that issue is the fact 7 that you don't view yourself as an expert on the relationship 8 between benzene and AML, true? 9 A. True. 10 Q. Now -11 A. Correct. 12 Q. In contrast to that, Dr. Bennett, you do view Dr. Smith to 13 be an expert on the relationship between benzene and AML, true? 14 A. True. 15 Q. And as a matter of fact, you consider him to be such an 16 expert that you invited him to be one of the principal authors 17 for the chapter in your book entitled "Causative Agents in the 18 Etiology of Myelodysplastic Syndromes and the Acute Myeloid 19 Leukemias," correct? 20 A. Absolutely correct. 21 Q. Yes. Now, Doctor, I know this is apparent, but I want it 22 to be in the record. You did not invite Dr. Garabrant to write 23 this portion of your book, did you? 24 A. No, I did not. 25 Q. You did not invite Dr. Pyatt to write this portion of the 124 1 book? 2 A. No, I did not. 3 Q. And you did not invite yourself to write this portion of 4 the book? 5 A. I would never think of doing that. 6 Q. All right. Fair enough. 7 Now, you did edit this book, though, correct? 8 A. Yes. We have a second edition out that just came out a 9 few months ago, but we have no section on epidemiology since it 10 really hasn't changed very much. So that's why no one else was 11 invited. 12 Q. Okay. Fair enough. 13 In this book -- in this chapter, your book says, "Although 14 benzene has been most strongly associated with AML and aplastic 15 anemia, there is evidence to suggest that other subtypes of 16 leukemia, MDS, non-Hodgkin's lymphoma, and possible other 17 hematopoetic and lymphoproliferative malignancies and related 18 disorders are linked with this exposure to this chemical," 19 correct? 20 A. Yes. And he referenced some nine articles to support that 21 statement. 22 Q. Yes, he does. 23 Now, you would agree, wouldn't you, Doctor, that there 24 have been innumerable studies that demonstrate that benzene 25 actually works at multiple levels to create damage to the DNA 125 1 structure of the hematopoietic system, true? 2 A. That's an overly broad statement, but by and large, most 3 studies show that benzene -- it's not obviously benzene because 4 benzene does not cause the damage; the hydroquinone has to be 5 metabolized to that active compound. But let's say that the 6 studies are consistent that show that a variety of different 7 cells in animals and patients, primarily lymphocytes, can be 8 damaged by benzene, which strongly implicates benzene as 9 causing damage to DNA. 10 Q. Okay. Let's do it this way: You gave testimony in, let's 11 see, the case of Hooper versus Bridgestone Tire, and I believe 12 Chevron was in that case as well, August 8, 2005. Yes? 13 A. Yes. 14 Q. And in that case you were asked this question: "And there 15 have been enumerable studies that demonstrate that benzene 16 actually works at multiple levels to create damage to the DNA 17 structure of this hematopoietic stem cell; isn't that correct?" 18 And you answered "yes," correct? 19 A. Yes. 20 Q. And then you were asked, "And so this hematopoietic stem 21 cell is sort of the mother cell. All of these cells that we 22 see on the chart are flowing from the stem cell; isn't that 23 correct?" 24 And you answered "correct," right? 25 A. Correct. That's the flowchart we looked at earlier. 126 1 Q. Now, you're familiar with the studies done by the National 2 Cancer Institute through Dr. Smith that showed low-dose 3 exposures, doses of less than one part per million, showing 4 chromosomal damage to progenitor stem cells, true? 5 A. You're talking about in vitro studies now? 6 Q. I'm talking about the Qu-Smith article; yes. 7 A. Dr. Smith and his group have written several articles on 8 damage of cells and using different techniques and so forth; 9 yes. 10 Q. Well, let's look at how you answered this question in 11 2005. You were asked this question: "Are you familiar with 12 the recent studies done by the National Cancer Institute of 13 low-dose exposures, doses less than one part per million, 14 showing chromosomal damage to these progenitor stem cells?" 15 You said, "I'm not sure which article you are referring 16 to." 17 The lawyer said, "The Qu article." 18 And you said, "Is that Qu, Smith, et al?" 19 The answer was "yes," and you say "yes." 20 Here's the question: "And they have written in research, 21 in fact, Dr. Martyn Smith -- in fact, most of that group is 22 associated with the Berkeley -- University of California at 23 Berkeley; isn't that correct?" And you answered "correct." 24 And the question was, "And they have recently published 25 under a grant, 'Cellular Effects Generated by Benzene'; isn't 127 1 that correct?" And you answered, "That is correct." 2 And then the question was, "And they are of -- they have 3 shown multiple levels of chromosomal damage done at exposures 4 below one part per million; isn't that correct?" And you 5 answered, "I believe that is true," right? 6 A. You're reading it. 7 Q. Well, that's what you testified to. 8 A. I can't argue. Right. 9 Q. Okay. 10 A. Okay. 11 Q. Now, you know that Dr. Pyatt has been deposed in this 12 case, right? 13 A. Yes. 14 Q. You read his testimony? 15 A. Deposition? 16 Q. Yes. 17 A. Yes. 18 Q. Okay. So with respect to Dr. Pyatt, I want to know if you 19 agree with Dr. Pyatt's statement -- and I will show it to 20 you -- his statement that says, "The progenitor cell, 21 hematopoetic progenitor cell population is a target issue for 22 benzene toxicity." Do you agree with that statement? 23 A. No. I think it's a matter of -- when I showed the 24 illustration of being absolutely crystal clear to be very, very 25 specific about what is the progenitor cell, what is the 128 1 hematopoetic stem cell, what is the cell that leads down the 2 myeloid pathway. And I think there can be, often, confusion in 3 terminology. 4 Q. All right. Well, let's try this -- let me ask you one 5 other question about this deposition and then I'll ask you 6 something further. 7 Dr. Pyatt also told us in his deposition -- and I will 8 read this to you. It says, "With respect to the last question 9 that I just asked you about the different subvariants of AML 10 and the extent to which they arise from the same progenitor 11 cell, is that issue something that you would say reasonable 12 scientists could disagree about?" And Dr. Pyatt says, 13 "Reasonable scientists do disagree." 14 Do you agree with that statement by Dr. Pyatt? 15 A. I'm sorry. You'll have to put that back up again because 16 you're reading from what Mr. Leghorn said. It says, "Mr. 17 Leghorn: Object to form. Reasonable scientists do disagree." 18 Q. That's not Mr. Leghorn -19 A. "A" is from -20 Q. That's Dr. Pyatt. 21 A. Okay. "Reasonable scientists do disagree." Okay. 22 Q. All right? And what I'm asking you: Do you agree with 23 Dr. Pyatt that reasonable scientists can disagree about the 24 difference subvariants of AML and the extent to which they 25 arise from the same progenitor cell? 129 1 A. Yes. 2 Q. All right. So on this issue of your opinions on 3 subvariants of AML and where they arise, you've looked at the 4 literature and come to a conclusion -- made scientific 5 judgments about that and come to a conclusion, correct? 6 A. Correct. 7 Q. Dr. Smith has done the same thing, read literature and 8 come to conclusions, correct? 9 A. Correct. 10 Q. All right. He has not come to unreasonable conclusions, 11 has he? 12 A. From my perspective, he has come to an unreasonable 13 conclusion, yes. 14 Q. I understand you disagree with his conclusion -15 A. I disagree with his conclusion. 16 Q. -- but you would agree that reasonable scientists can and 17 do disagree with you, true? 18 A. Correct. I would hope so. 19 Q. Now, I want to talk about the Douer paper that you talked 20 about. 21 A. It's not a paper; it's a chapter. It's just a review. 22 Q. It's just a review? 23 A. It's a review of a chapter in a textbook, review of AML. 24 Q. All right. Now, you pointed out that this says, "It is 25 unknown whether the risk for APL is associated with any 130 1 environmental or occupational exposures," right? 2 A. That's what he said. 3 Q. Yes. But isn't it true that when you looked -- went and 4 looked at the literature, you said there have been occupational 5 exposures that have been identified, right? 6 A. Oh, you're referring now to the Italian cohort cases that 7 I -- yes. 8 Q. Right now I'm referring to what your report -9 A. Okay. Fine. Right. 10 Q. And you say in your report, "Two Italian cohort 11 case-control studies have identified certain occupational 12 exposures as potentially having a higher risk. These including 13 shoemaking" -- and you have an odds ratio there of 6.3, 14 correct? 15 A. Correct. 16 Q. You would consider that a significant association, 17 wouldn't you? 18 A. Yes. 19 Q. And, in fact, you know that in that paper it's a 20 statistically significant association, true? 21 A. Yes. 22 Q. And this is the Mele paper that we were talking about 23 earlier? 24 A. Right. 25 Q. And you understand that these authors in this Mele paper 131 1 believe that this relationship could be related to benzene, 2 true? 3 A. They suggested that. 4 Q. Yes. That's not an unreasonable suggestion, is it, 5 Doctor? 6 A. Based on their study it's not an unreasonable suggestion. 7 Q. Yes. So when you mentioned the Douer paper, he doesn't 8 mention -- Douer doesn't mention in his discussion any of the 9 things that you found, does he? He doesn't talk about 10 shoemaking or electricians or agricultural and textile workers 11 or machine operators, does he? 12 A. I think, with due respect to Dr. Douer's report, what he 13 was assuming was sort of a global analysis or inference about 14 the scope of the literature, and comes down on the side of 15 there's not much out there. You can always find one or two 16 articles to support a claim of relationship. And, again, it's 17 based on the weight of the evidence and the number of articles 18 that have appeared. 19 Q. Yes. But to suggest -20 A. I can find nothing wrong with either of the two Mele 21 papers as far as what they did, nor -- I know several of their 22 workers -- met them at meetings -- and their capacity to make 23 proper diagnoses. 24 Q. Yes. But to suggest that what Dr. Douer is saying lays 25 out for readers what the studies are, what they say to evaluate 132 1 them, he didn't do that in his study, did he? And it's not a 2 study, it's a review paper. I apologize. 3 A. It's a review paper, and he's limited by amount of time 4 and effort he could put into it. 5 Q. So simply what we have here is Dr. Douer's opinion in a 6 review paper? 7 A. Yes. 8 Q. Okay. Let me ask you this question: You testified that 9 APL may arise at the CMP progenitor level, right? 10 A. Yes, I believe so. 11 Q. Yes. Isn't it true that all of the various types of AML 12 involve abnormalities in cell types that all descend from the 13 CMP progenitor cell? 14 A. No, I think I indicated that it's a granulocyte, not 15 the -- the CMP is earlier -- a little earlier stage. And the 16 authors indicated that it was the granulocyte precursor cell. 17 The CMP is a more multifocal type of cell. But what they're 18 inferring is that they can't rule that out; that's correct. 19 Q. I didn't actually ask about that study. I asked isn't it 20 true that all the various subtypes of AML involve abnormalities 21 in cell types that all descend from the CMP progenitor cell? 22 A. All the myeloid cells descend from that common precursor 23 cell. 24 Q. Yes. 25 MR. STEWART: I'm done with my cross-examination. 133 1 THE COURT: Okay. Let me just ask, Doctor: I think 2 you referred to something as "the malignancy event." Is 3 that -- or a -- that is, when the healthy cell becomes 4 malignant or leukemic. Is that -- am I touching on a concept 5 that I'm accurately recalling or not? 6 THE WITNESS: Well, when a healthy cell is -- take any 7 healthy cell in the body. When it becomes malignant, it then 8 has the capacity to continue to reproduce itself without 9 maturing. Without maturing. Now, it can mature a little bit. 10 So, for example, in colon cancer, we can see 11 well-differentiated colon cancer or we can see poorly 12 differentiated colon cancer. And that's all based on signal 13 coaching that exists within the tumor. 14 THE COURT: I may have gone down the wrong line. What 15 I'm interested in is the so-called leukemic stem cell. Maybe I 16 can just ask you to tell me what that is. 17 THE WITNESS: The leukemic stem cell is a cell that 18 has been identified that is capable of continually reproducing 19 itself and not maturing, not developing into normal developing 20 cells like neutrophils, pus cells, monocytes and so forth. 21 THE COURT: But it's a stem cell in the sense that it 22 both reproduces itself and generates other cells? 23 THE WITNESS: No, not the latter; the former. 24 THE COURT: Isn't that what a stem cell does? 25 THE WITNESS: Well, that's why these terms are so bad, 134 1 and you're correct in having concern about the terminology. So 2 the normal stem cell -- the normal stem cell -- receives 3 responses where it sits in the niche in the bone marrow, that 4 triggers it to release a flood of cells that are capable of 5 maturing and repopulating, keeping the peripheral blood system 6 intact. 7 As soon as it does that, one of two daughter cells 8 stops producing and goes to sleep until it gets another 9 message. Then it wakes up, reproduces itself, goes back to 10 sleep. The leukemic cell never sleeps. It's incapable of 11 sleeping. So it just continues to reproduce itself; it 12 divides, it divides, it divides. 13 Now, the rate of division is very much dependent upon 14 the kind of leukemia or lymphoma that we see. So there are 15 some tumors that divide very rapidly, like small-cell lung 16 cancer, or Burkitt's lymphoma in which the cell literally 17 divides and doubles itself every 24 hours. Horrendous growth 18 rate. There are some, like in MDS, in which the growth rate is 19 so slow that the patient may live for ten years before that 20 leukemic cell divides enough to become acute leukemia and crowd 21 out the bone marrow. So it's highly variable. 22 But by definition the leukemic -- once we say 23 "leukemic stem cell," we are saying a cell that is immortal 24 unless we impact on it with treatment. 25 THE COURT: So that characterization doesn't 135 1 necessarily say anything about the stage of maturation at which 2 that occurs? 3 THE WITNESS: No. 4 THE COURT: Okay. 5 MR. STEWART: Can I just follow up? 6 THE COURT: I was going to give him a chance and then 7 you. 8 REDIRECT EXAMINATION 9 BY MR. WEATHERS: 10 Q. Two quick things in the interest of not pressing the 11 Court's patience any more than we have. 12 A. The Court has been very patient with me, I must say. 13 Q. Doctor, the Wojiski paper: Counsel took you through the 14 fact that the authors of that paper can't rule out that the 15 mutation leading to APL could have occurred at an earlier 16 stage, correct? 17 A. Correct. 18 Q. But they've reached a scientific conclusion from their 19 work, didn't they? 20 A. They did. That it was unlikely based on their work, 21 right. 22 Q. But they reached a conclusion as to where they thought it 23 occurred? 24 A. Right. 25 Q. And what was that? 136 1 A. At a later stage. 2 Q. All right. And one last thing. There was talk about the 3 book in which Dr. Smith kindly wrote a chapter for you. When 4 was that book published? 5 A. Ten years ago. 6 MR. WEATHERS: Thank you. That's all I have. 7 THE COURT: Mr. Stewart? 8 RECROSS-EXAMINATION 9 BY MR. STEWART: 10 Q. Yes. The version of the book I have -11 A. '92? '94? 12 Q. The version of this book that I have was copyrighted in 13 2002. 14 A. So it -- 2002 means that it was put to bed around 1998 or 15 '99, because it takes that long to get the chapters in. 16 Q. Fair enough. Here are my -- I'll move to the microphone. 17 Here are my two questions: Isn't it true a leukemic stem 18 cell can result in different variants of leukemia? 19 A. Yes. 20 Q. And most scientists think this is a multistep process, so 21 that more than one event or attack on the cell line must occur 22 before you see full-blown leukemia? 23 A. So the latter is more in question, in my mind, than the 24 former. 25 MR. STEWART: Fair enough. That's all I have. 137 1 THE COURT: All right. Thank you very much, Doctor. 2 We'll recess until tomorrow morning. 3 MR. WEATHERS: Thank you. 4 THE CLERK: All rise. 5 Court is in recess. 6 (The proceedings adjourned at 1:25 p.m.) 7 8 CERTIFICATE 9 10 I, Marcia G. Patrisso, RMR, CRR, Official Reporter of 11 the United States District Court, do hereby certify that the 12 foregoing transcript constitutes, to the best of my skill and 13 ability, a true and accurate transcription of my stenotype 14 notes taken in the matter of Civil Action No. 07-11944-GAO, 15 Brian K. Milward, et al, v. Acuity Specialty Products Group, 16 Inc., et al. 17 18 /s/ Marcia G. Patrisso MARCIA G. PATRISSO, RMR, CRR 19 Official Court Reporter 20 21 22 23 24 25 138