Document LRdORd05ag8B9M2XDQo685aq

International Journal of Epidemiology 0International Epidemiological Association 1986 Vol. 15, No. 3 Printed in Great Britain Multiple Myeloma: Propoxyphene and ther Drugs, Radiation and Occupation GARY D FRIEDMAN Friedman G D (Department of Medical Methods Research, Kaiser Pvrmanente Medical Care Proqram, 3451 Piedmont Avenue, Oakland, CA 94611, USA). Multiple myeloma: relation t c propoxyphene and other drugs, radiation and occupation. Jnternational Journal of Epidemiology 1986, 15: 423-425. A case-control study involving 327 cases of multiple myeloma (MM) did not confirm that propoxyphene use is a predisposing factor. The evidence that any drugs are involved in thc aetiology of MM is very weak. There was some suggestive evidence that x-ray therapy and certain occupations werr predictors of MM. ref -. <2 2 14 6 8 IO I2 14 1- 16 18 ' 20 -2 2 In screening medicinal drugs for possible carcinogenic re\ iew). h,leclical charts wei e niasked after a relerciicc Fk effects, we noted an association between propoxyphene daie, six ninnilis before the case's diagnosis of Mhl, w: use and subsequent multiple myeloma (MM).] Among aticl wcre abstracted by researchers unaware o f the case- I 01 17927 people who received the drug between 1969 and coiitrol status. Test-based confidence intervals were t I1 1973, 17 developed MM in follow-up through 1978 and c;i Iculatecl using the Kothman-Boice program^.^ if? 9.4 were expected (standardized morbidity ratio [SMRI = 1.8, p<0.05). This hypothesis seemed well K I :S U LTS A N 1) D I SCUSS ION worth pursuing further since little is known about the S I Ihjecls aetiology of MM and drugs are among the few AI diagnosis the cases were aged 32 to 88 years witli a suspected causes.* Also, the incidence of MM has nicdian of 64 years; for coiitrols, the age range \ v a ~33 increased over the past few decadesZand propoxyphene to 87 yearn witli a median of 63 years. Men made rip has been commonly prescribed during this p e r i ~ dW. ~e 58.7% of both the cases and the controls. Most of the therefore conducted a case-control study drawing upon siihjects were white-74.3?'n of cases and 72.5% of a much larger group of cases, in an attempt to confirm controls. The duration of KFHP membership prior IO and further characterize the association. the refcrence date (ie. follow-back time i n our recodt) raiiged up to 35 years, witli a rnediari of 16 years for METHODS borli cases and controls. In a 1969-1982 computer file of cancer cases among Kaiser Foundation Health Plan (KFHP) subscribers i n Pi-opnyjydiene Northern California we identified 327 people with MM, (herall, propoxyphene usc was predictive o f MM; 135 11: subsequently confirmed as follows: 3 16 (96.6%) had (41.3%) of tlic cases and 1 IO (33.6%) of the controls '1 diagnostic evidence based on pathological examination of bone niarrow or other tissue and 1 1 (3.4%) had other strong evidence such as typical abnormal serum received propoxyphcne at least once before tlic a1 rel'ercncc date (x2= 4.08, 0.05>p>0.02). The niatclictl bc estimate of relative risk ( R R ) based on a ratio of 82 , di proteins. For each case a control subject was selectcd case-only discordant pairs to 57 control-only from the KFHP membership file of the same year as the discordant pairs was I .44 (95% confidence limits [C'l.l, caw's year of first hospitalization for MM. Control I .03. 2.01). However, the largest case-control dif- subjects were matched to the cases by sex, year of birth, fei.ence was i n the two-year period jurt before 1lic date of joining the KFHP, area of residence (postal zip rel'erence date (Table I ) . This suggested that sonic c a w code), and, wlien feasible, race (race matched by chart had been given the drug for h~1Msymptonis (eg, back pain) before MM was diagnosed. When attentioil iva\ i i _________ Deparitneiil of Medical Meilindc Ke,carLIi, Kaiser I'cttiiaiienie hlcdical Care Program, 3451 I'iedniont Avenue, Ociklatd, C A Y4hl I, tectricted to the period two years or inore befotc tlic the R R wa' (95*'n c'L ' 0'843 USA ~ 1 <l'# I+ .< ratio of diccordant pair s 74/63). Siinilarly, the (higlil!. w a L I 2' * , '* 3.3 Great Britain t MULTIPLE MYELOMA 425 I T M l E I Nunibers of rnultiple myeloma cases and conlrols who cytotoxic drugs8 have been suggested as predisposing 1 rrtervrd propoxyphene by two-year intervals before reference date. SU~JJCCIJ were counled in any interval rrr which they received propoxyphene. factors for MM.'.Y We have screened tlie first two of these; no association with MM was found. Through - 1980, among 954 phenytoin users there were 1 observed No. who received and 0.6 expected cases of MM; among 849 colchicine Y r ~ i sbuloir IClslCllLr dpte --__ propoxyphrne - Cases Controls Case:coni rol ratio users there were 1 observed and 1.1 expected cases. With a t wo-year lag between dispensing and diagnosis there were no cases of MM observed among users of the ! \2 38 18 2. I cytotoxic drugs, cyclophosphamide (291 users), vin- 2 -<4 4 -C 6 6-Cbl 1 u (10 'snt IO (12 34 25 1.4 cristine (62 users), procarbazine (55 users), chloram- 38 23 29 25 21 27 1.7 I .2 0.8 bucil (142 users), or fluorouracil (355 users). In our drug screening through 1978I.' a few drugs showed 18 22 0.8 nominally statistically significant associations with iild I2 <I4 14 <I6 s a 16 <I8 me in <XI 1 20 -<22 22424 23 14 1.6 subsequent MM. These, with numbers of users, 17 9 I .9 observed cases, and expected cases, respectively, were: 9 12 4I 0.8 0.6 carisoprodol, 837, 5, 0.3; dimenhydrinate, 435, 2, 0.2; 2 4 0.5 methocarbamol, 5453, 7, 2.6; and secobarbital, 2884, 1 0 - 6, 1.9. Since carisoprodol and methocarbamol are 1 I referencc ' sl\ewetl) distribution of number of times propoxyphene 5 o f MM, { was prescribed differed significantly ( p =0.05 by the iI1 the casc- v a h were owtailed Komolgorov-Smirnov (KS) test), but when die two years just before the reference date were lll5.J II igiiorcd, the p value was 0.60. Table 1 shows that more cases than controls received commonly prescribed for mulsculoskeletal pain, their association could be artifactual due to treatment of early symptoms of MM. These drug-MM associations must be regarded as possible chance associations or hypotheses derived from the thousands of drug-cancer combinations screened. pr-opuxyphene between zero and eiglit years and X-rudiution bc.tivcc.ri12 and 16 years before the reference date. The Radiation has been reported as a causal factor in MM.' ais with a reverse was true between eight and 12 and between 16 Cases had not received significantly more diagnostic ige was 33 aiid 22 years before the reference date. Between 22 and x-ray examinations before the reference date than : 24 years before the reference date, only one case arid no controls (median 14.1 versus 12.8 per person, p = 0 . 3 0 controls received propoxyphene. Although one could by KS test). For x-ray at least two years before the Iiypothesize post hoc that propoxyphene acts as a reference date the case-control difference was smaller pi'oiiiotrr ot' MM during the period of up to about eiglit (respective medians, 10.8 and 10.3). years before diagnosis, it seems likely that the observed Eighteen cases and ten controls had received some relation to duration is explained by both chance form of x-ray therapy before the reference date (1719 variation and the receipt of propoxyphene for pain due discordant pairs: RR = 1.89, 95% CL =0.85,4.18). For IO M M in the few years before the diagnosis was made. the nine cases and seven controls with number of rads Altogether 114 cases and 86 controls were believed to recorded the respective means were 2451 and 2060 rads have estimable total durations of propoxyphene use. per person. Although not statistically significant our 'I'he mean total duration of use was 90.2 days in cases data are consistent, within wide confidence limits, with and 129.1 days in controls. The median duration i n MM associated with x-ray therapy, but not with the both groups was eight days. The distributions did not amount of radiation ordinarily received from diag- dil't'er significantly (KS test, p=0.60). nostic x-rays. Aniong the many indications for propoxyphene use iioted, tliose for which cases clearly exceeded controls Occupution weIe largely a variety of types of musculoskeletal pain. All occupations noted in the medical records were This is consistent with the clinical course of MM and considered. Although cases and controls differed little ,me cases with the likelihood that many of the cases were having with respect to broad occupational groupings, there (eg, bach syniptoms well before diagnosis. were some notable differences for the 150 individual ition was occupations looked at (Table 2). Although not statis- cforc I l k -84, 1.64; \:(Iiiglily- Orker drugs Based on a few case reports phenytoin (diphenylIiydantoin),sa6 colchicine and sulfinpyrazone,' and tically significant here, some occupations with excesses among tlie cases have been linked to MM previously, ie, carpenters (wood-related industries1-12), and , TABLE 2 Occupalions with a1 leas1four more cases lhan conlrols. Occupation No o f cases No o f controls Student+ Housekeeper Longshoreman Janitor Warehouseman Hospital attendant Guard*+ Other crafts Carpenter Painter Other clerk 7I 51 15 7 14 9 11 7 83 60 11 6 IO 5 62 84 ~~ ~~~ * Verified from manual records because o f our surprise at having so many students in this age group. *+ flO.05 by two-tailed Fisher's exact test. p a i n t e r ~ . 'T~he other sizable excesses, particularly in longshoremen, janitors, warehousemen, hospital attendants, and guards, may be clues to some MM-inducing exposure. If exposures to hair sprays or cosmetics become suspect, it should be noted that fewer cases than controls were hairdressers (0 versus 4) or barbers (2 versus 6). CONCLUSIONS A case-control study involving 327 cases of MM failed to confirm that propoxyphene use is a predisposing factor. An association between its use and the development of MM up to about eight years later could conceivably represent a causal link, but is most likely due to the use of propoxyphene to treat pain due to MM before i t was diagnosed. Follow-up of users of certain other drugs, implicated in a few case reports, did not confirm an association with MM, and associations with still other drugs among the 215 screened were very few. Thus, the evidence that medicinal drugs are involved in the aetiology of MM is very weak. X-radiation therapy but not diagnostic x-rays had an appreciable but non-significant association with subsequent MM. Although numbers were small and therefore of limited reliability, there were some intriguing associations of MM with occupation, two of which support previously reported links to wood and paint exposure. ACKNOWLEDGEMENTS Supported by Public Health Service Grant R01 CA 19939 from the National Cancer Institute. The technical assistance of Bill Frank, Merril Jackson, Belly Jue, .loan Thomas and Donna Well$ is gratefully acknowledged. lnternationa 0 lnternatk Let1 Does I REFERENCES i Friedmati G I),Ury CI K. Screening Cor possible drug carcitto. genicity: second report of findings. J No// Cancer lnsr I9RJ; 71: 1165-75. Blattner W A. Multiple myeloma and niacroglobirliiiemia. In. Sclio[tenfeld D. Fraumeni J F Jr. eds. Cancer t-pideniidop and p r e i ~ n / i o r i .Philadelphia, W 13 Saunderc. 1982, pp 79!813. Friedman G 1). Collcn M I-,tiarrir 1. E , Van Brunt E E , Iheir 1.S Experience in monitoring drug reactions in outpaiients: thc Kaiser-Pernianente Drug Reaction Monitoring System J A M A 1971; 217: 567-72. Rothnian K J . Boice J D Jr. Epidemiologic analysis with a pro. grammable calculator. Washingion, DC. US Department or Healih, Education and Welrare, 1979, N l t i publication no 79- 1649. blatzner Y , Polliack A. 1.yrnphoproliferat ive disorder$ in fotir patientr receiving chronic diplienylliytlatttoin tlirrnlry: ctio lopic correlation or chance aswciat ion? I ? / .I A!rd 5c.i 1971; 14: 865-9. Aymard J P, Lederlin 1'. Witz F, e / a/. Multiple myeloma aftcr phenytoin therapy. ScandJ Haerfiaiol 1981: 26: 330- 2. Witwer M W, Schmid 1: R, Tesar J T. Acute myelonionocytic leukaemia and multiple myeloma after sulphinpyrazone and colchicine treatment of gout. Br MedJ 1976; 2: 89. Aksoy hl. Erdem S. Dincol G. Kutlar A, Bakioglu I , HepytrkselT. Clinical observations showing the role o f ronie factors in thc etiology o f muliiple myeloma. A study in seven patientc. Acto Ffaerfrar1984;. 71: 116-20. Friedman G 1). l l r y H K. Initial screening for carcinogenicity or coninionly used drugs. J Na/l Cancer Intt 1980; 65: 723- 31. lo Milhani S Jr. Occupational mortality in Warhington Stale, 19501971. WaFhington. DC, US Government Printing Ollkc. 1976, D H E W Pubication No (NIOSH) 76-175. I t Brinton 1. A, Stone B J. Illot W .I, e / a/. Nasal cancer in US lurniturc industry counties. Lancer 1976; 2: 628. l 2 Decoufle P. Stanislawczk K. fiouten L . e/ a/. A retrocpecti\,e siiricp o f cancer in relation to occupation. 1977, 1)IiEU' (NIOSIiI Publication N o 77- 178. Adelstein A M. Occupational mortality: cancer. Ann O w / p /!,y 1972; 15: 53--7. (Revised version received Deceniber 1985) From N I Sir-Acco question o From tb for child'' older siblii home, ai' pregnancy significant old childr from 5o(u and Scot 1: In an ea traps in Tirst suspi regression leading re eg the so( homogenc Rut on conclud ec blatant, 11 In Tahl home sm 0,0012. T value by n = 5075, - O.(X)34 cigarette$ nieaciiren for age, sumably, The ge the inves analyses 1. Child' 2. Paren 3. Numt -- Ikparlmen American 1 USA.