Document 20zMMk8ypVokzQ35Gk4XaYBa

EPA 560/6-81-002 9 EPIDEMIOLOGY STUDIES SCREENING FOR THE EARLY DETECTION OF DISEASE IN INDIVIDUALS EXPOSED TO VINYL CHLORIDE JANUARY 1981 FINAL REPORT U.S. ENVIRONMENTAL PROTECTION AGENCY OFFICE OF PESTICIDES AND TOXIC SUBSTANCES WASHINGTON, D.C. CMA. 052872 EPA 560/6-81-002 January 1981 SCREENING FOR THE EARLY DETECTION OF DISEASE IN INDIVIDUALS EXPOSED TO VINYL CHLORIDE Carlos H. Tamburro^ Charles Kupchella^ Kenneth Taylor^ Emanuel Landau4 by Richard Greenberg^ 2 Hildegarde Maricq Joseph Whelan, JrJ 5 Joseph Seifter 1. University of Louisville Louisville, Kentucky 2. University of South Carolina Charleston, South Carolina 3. Yale University New Haven, Connecticut 4. American Public Health Association Washington, D. C. 5. U.S. Environmental Protection Agency Washington , D. C. Project Officer Jane Keller Office of Pesticides and Toxic Substances Washington, D.C. U.S. ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, D.C. 20460 CM* 052873 DISCLAIMER This project has been funded with Federal Funds from the Environmental Protection Agency under contract number 68-01-3859. The content of this publication does not necessarilyref1ect the views or policies of the U.S. Environmental Protection Agency, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. CMA. 052874 ABSTRACT A prospective collaborative study was conducted to compare the effectiveness of four clinical techniques in the detection of liver damage due to vinyl chloride monomer exposure. A chemically exposed and medically monitored worker population was identified by histopathologlcal and biochemical documentation. Three techniques were non-invasive: a) grey scale ultrasonography of the liver, b) microvascular skin capillary assessment, and c) urinary analysis of glycosaminoglycan excretion. The fourth technique was the standard 99mTc sulfur colloid radionucleotide liver spleen scan. The screening studies were performed on a randomly selected single cohort of chemical workers; some of whom were known to have disease. All four techniques were analyzed for their sensitivity and specificity as compared to results of the liver biopsy and biochemical blood test classification. Although all four screening techniques had a sensitivity and specificity sum greater than one, none were significantly better than could be explained by chance or the use of a biased coin. Reclassification of the population into those with more severe biochemical abnormalities improved the sensitivity of all screening tests, but only the sensitivity and specificity sun. for the GAG test were statistically significant at the 0.05 level. There was no significant correlation between any pair of screening tests. None of the four screening tests agreed with the biopsy results better than might be obtained by biased coin or chance. These screening studies as presently constituted, do not provide sufficient sensitivity and specificity to warrant their use in conmunity screening for subclinical asymptomatic hepatic injury due to chemical exposure. CMA. 052875 INTRODUCTION The initial reports of primary liver cancers (angiosarcoma) in rats exposed to vinyl chloride by Maltoni, et al (1) and the discovery of similar liver tumors / in vinyl chloride polymerization workers by Johnson and Creech (2) has lead to considerable environmental concern regarding communities surrounding chemical industries which utilize potentially carcinogenic agents such as vinyl chloride. Several investigators have reported on various screening techniques as effective indicators of vinyl chloride chemical injury. Four such techniques - ultrasono graphy (3), radionucleotide scanning (4), nailbed capillary visualization (5), and glycosamincglycan (GAG) (6) excretion - were reported to have seme possible usefulness in detecting early chemical injury to the liver. In order to determine the useability of these techniques for conmunity screening, the American Public Health Association (APHA) and Environmental Protection Agency (PA) funded a multi-center collaborative study designed to determine the comparative sensitivity and specificity of these various techniques in detecting and identifying chemical related hepatic Injury in asymptomatic individuals. Materials and Methods Population Selection The chemical worker population consisted of 1,178 active (Group B, Figure 1) employees as of September 1, 1977, and 70 employees who had had liver biopsies regardless of current employment status (Group A, Figure 1); they were undergoing annual medical screening for the idertification of work-related disorders. The medical screening consisted of an annual or semi-annual (for those employees with 10 or more years of employment) comprehensive history and physical examinations, laboratory screening studies consisting of 35 biochemical tests, chest and abdomi nal X-rays, and radionucleotide liver-spleen scan. CMA. 052876 2 This population was selected for the collaborative study to determine the comparative effectiveness of four screening techniques in detecting liver damage as indicated by 1) past histopathological documentation of liver injury of 2) current hepatic dysfunction identified biochemically. Three of the four techniques were included as potential non-invasive pro cedures suitable for determining the effects of vinyl chloride in a community population. These included 1) grey scale ultrasonography of the liver as devel oped by Taylor and colleagues (7, 13, 14), 2) a nailbed skin capillary evaluation of the middle and distal phalanges of the fingers as developed by Maricq and associates (8), and 3) urinary analysis of glycosaminoglycan excretions (GAG) as published by Kupchella and associates (9). The fourth method included for comparison purposes was the standard radio* QQ nucleotide liver-spleen scan utilizying mTc colloid and interpreted by Whelan and associates (10). The non-invasive screening studies were performed during a single week on a group randomly selected from all chemical company employees. The workers were selected on the basis of complete medical and work data for 1976-1977, and all those employees who had investigative liver biopsies performed during the screening program (1974-1977). The selection process for these workers is illustrated in Figure 1. The biochemical data and radioisotopic scans were part of the routine medical surveillance system for the employees. The pathological data was based on the last or most recent liver biopsy (s) which were performed for medical reasons, both related and not related to their work. Positive and negative results were determined as defined in Table 1 which list the technique, the evaluation or evaluators, and the criteria used. The employees targeted for examination were selected by simple random sampling from all available employees. CMA. 052877 Ons hundred and twenty cne cf the targeted 170 employees (71 Dercent) participated. Twenty six ceciMea zo participate cr could rot ce scheduled; 13 did not keep their scheduled appointment. After participation, nine employees were discovered to have been misclassified as biochemically abnormal. They had some biochemical abnormalities but not all (classified intermediate) and have been eliminated from the final analysis. Four biochemically abnormal individuals had abnormally low test values and they were included in the analysis. Liver Biopsies Liver biopsies were performed by the transjugular technique (11) and pro vided two to five biopsies from various areas of the liver. In addition, some individuals had second biopsies performed by the percutaneous or wedge biopsy via mini-laparotomy procedures. Pathological data was recorded in a computerized format identifying all histological abnormalities in a semi-quantitative fashion. Biopsies were read without knowledge of the individual's medical history or chemical exposure by two pathologists and a hepatologist with extensive experience in hepatic chemical injury. All biopsies were classified as 1) normal, 2) abnormal, a) chemical injury, and b) non-chemical injury. The non-biopsy groups were drawn from those currently employed, and based on biochemical liver "function tests" individuals were sorted into positive, negative and indeterminate for hepatic disease. Only the positive and negative are included in this study. The ultrasonic evaluation and its relative ability to identify hepatic damage due to vinyl chloride has been published elsewhere (3). Microvascular techniques and the method of evaluation by Dr. Maricq and co-workers are also published in part (8). The experimental work on GAG excretion in vinyl chloride workers and the techniques for differentiating the electrophoretic patterns in patients with CMA 052878 4 angiosarcoma and connective tissue damage of the liver has been published else where (9). The effectiveness of radioisotopic (radionucleotide) scanning as a technique for identifying anatomical lesions in vinyl chloride workers is in preparation (12). Method of Analysis The biopsied group and the non-biopsied (biochemical) group were analyzed independently. For the biopsied group sensitivity and specificity were estimated for each screening test by assuming that the biopsy was correct. For the bio chemical group the biochemical classification was assumed to be correct. For each analysis the data consisted of a simple cross classification. In a perfect screening test, the sum of sensitivity and specificity would be two. In a screening test which provided results no better than could be obtained by using a biased coin, the sum of sensitivity and specificity would be equal to one. We, therefore, estimated 95% confidence limits for the sum of sensitivity and spec ificity and observed whether or not one is included within these limits. As a test of statistical significance this is equivalent to the usual X test for independent proportions. Finally, in Table 4 we looked at the association (as measured by the phi coefficient) between each pair of screening tests. For these comparisons we used all employees who received both screening tests regardless of their biopsy status. In this case, we assumed both tests were subject to error and estimated the phi coefficient (r$) between them. Finally, for completeness, we give the biochemical classification for the 51 employees included in the biopsy group. Results Table 2 compares the histological and biochemical results for the 51 biopsied employees included in the study. Twenty-two of these employees had biochemical abnormalities as defined in Table 1. There was no significant CMA. 052879 rtyT*resT a ^biochemical 3rd bVODSV C^ aSS 1 71 CSt"I OTt TCT* >,.. 5 c.^ e"*. i o'-'ees v/ith positive or negat^'/e biochemical classifications *, i = 0.21: at 0.20). The biochemical studies used in this analysis were those determined at the time of this and not at the time the biopsy was.performed. In all instances of disagreement, the biopsy was positive and the biochemical results negative (P< 0.001). Figure 2 provides the sensitivity and specificity for each of the screening tests when compared to biopsy results. In no case is the sum significantly greater than one, indicating that the results are not statistically significantly better than could be obtained using the biased coin. With the exception of the GAG studies, similar results are obtained when comparing the sum of sensitivity and specificity in the biochemical group (Figure 3). The sum of sensitivity and specificity for the GAG studies are just statistically significant with 95 percent confidence limits of 1.06 to 1.52. Table 3 gives the frequency distribution of the results of the GAG test for employees with normal and abnormal biochemical results. The distributions differ In their spread (variance) and not in their location (means). Finally, the correlation matrix for the four tests are given in Table 4. There is no significant correlation, as measured by r$, between any pair of the screening tests. This is also true when they are sorted by biopsy status. After reviewing the results, a reclassification of the 51 employees who had biopsies was assessed in regard to whether there was chemically induced liver damage. Ultrasonographic evaluation was reclassified by Dr. Taylor and the liver biopsies by Drs. Tamburro and Popper. Table 5 gives the results of this additional analysis which demonstrates that there was no agreement that could not be explained by chance (P * 0.60). CMA 052880 6 Discussion The increasing industrialization in highly developed Western countries, such as the United States, continues to provide concern, not only for the health and safety of the industrial workers, but also for the surrounding comuunities in the areas of these industries. It is highly desirable to identify and validate the reliability of screening and diagnostic techniques which will Identify the early development of injury due to the exposure of a variety of chemicals such as vinyl chloride. The assessment of newly developing techniques on a high-risk, exposed worker population, who have been carefully screened and prospectively followed, provide the most reliable method for determing both the sensitivity and specificity of these technical procedures in the asymptomatic subclinlcal highrisk exposed community population. The failure of such techniques to provide sufficient sensitivity in the presence of a required specificity is of critical clinical importance. This is especially so where the incidence of disease is relatively low and the population exposed large. Tests which provide a high sensitivity but of low specificity can and do medically stigmatize the population under surveillance leading to unnecessary anxiety and socioeconomic disturbances which can far outweigh the benefit of early detection of even serious disease in a smaller population. Far too often screening techniques which have been developed in a highly diseased, clinically overt, hospitalized population are applied to an asymptomatic, clinically well-working populations without adequate determination of the sensi tivity and specificity at this earlier stage of disease development. In this study, all four techniques had, in the highly diseased hospitalized population, demonstrated either a sensitivity or specificity suggestive for the Identification of underlying chemically related liver disease. Some techniques (Maricq and Kupchella) appeared ideal for community studies since they were noninvasive, relatively inexpensive, and provided a means of screening which would be highly accepted by a comnunity. 052881 7 ~ -s ^r;:;ctive1y designed study has allowed us to estimate the ability of jltrescnography, nailbed capillary assessments, radioisotopic scanning, and glycosaminoglycan excretions to correctly predict the presence and absence of hepatic disease as documented by an exposed population. These studies clearly show that none of the four screening techniques sufficiently agree with either the biopsy, the biochemical results, or each other in a well defined population; they do not provide sufficient sensitivity or specificity to be useful as early indicators of chemical exposure injury. The inclusion of nine individuals with intermediate biochemical results, who were originally misclassified as abnormal, would decrease the sum of sensi tivity and specificity in all three screening tests. With their exclusion only the GAG tests provided results better than might be expected by chance (P< 0.05). Even the GAGs, from a pragmatic point of view, provide too high a false positive rate to be useful in its present stage. Possibly, with increased refinement and further study, this might provide a simple non-invasive technique for the identification of increased collagen changes related to chemical injury. More inmediately, it should be duplicated to rule out chance. Had we eliminated the four employees with an abnormally low biochemical test values, the sum of sensitivity and specificity for the GAGs and nailbed skin capillary screening tests would have been slightly reduced and that for the scan slightly increased. The GAG would still be of bordering significant (X^ = 3.73), and the scan not significant (X^ = 1.46). Finally, the tests not only disagree with each other, but within the biopsy group there was no agreement between the biopsies and biochemical results. It should be noted, however, that the biochemical studies used in analysis were those chronologically closest to September, 1977, and not to the date of biopsy. CMA 052882 8 The test determinations and the biopsies may have been as long as three years apart. An analysis of the biochemical tests value done at the time of the biopsy would have more accurrately reflected the liver status, as shown by histology (15). It has been shown, in previously published studies, that biochemical and histological findings each correlate with chemical Injury and chemical exposure (16, 17). CMA 052883 a**sr s 1. Maltoni, C. and Lefemine, G. Carcinogenicity to bioassays of vinyl chloride I. Research plan and early results- Environmental Perpectives, *.387-405, 1974. 2. Creech, J. L. and Johnson, M. N. Angiosarcoma of the liver in the manufac ture of polyvinyl chloride. Journal of Occupational Medicine, 16:150-151, 1974. 3. Taylor, K. J. W., Williams, 0. M. J., Smith, P. M. and Dach, B. W. Grey scale ultrasonography for monitoring industrial exposure to hepatotoxic agents. Lancet, U1222-1224, 1975. 4. Whelan, J. G. Jr., Creech, J. L. and Tamburro, C. H. Angiographic and iso topic characteristics of hepatic angiosarcoma found in vinyl chloride workers. Radiology, 118:549-557, March 1976. 5. Maricq, H. R., Johnson, M. N., Whitstone, C. L. and LeRoy, E. C. Capillary abnormalities in polyvinyl chloride production workers. JAMA, 236, 1368- 1371 , 1976. ----- 6. Kupchella, C. E. and Tamburro, C. H. Urinary and tissue glycosaminoglycan patterns in angiosarcoma and other vinyl chloride exposure - associated liver injury. In Prevention and Detection of Cancer, Part I, 1:915-926, ed. Niebergs, H., Marcel Dekker, Inc., 1977. 7. Taylor, K. J. W., Carpenter, D. A., Hill, C. R. and McCready, V. R. Grey scale ultrasound imaging the anatomy and pathology of the liver. Radiology, 119: 415-423, 1976. 8. Maricq, H. R. and LeRoy, E. C. Patterns of finger capillary abnormalities in connective tissue disease by wide field microscopy arthritis pheum, 16: 619-629, 1973. -- 9. Curran, K. L., Kupchella, C. E. and Tanfcurro, C. H. Urinary glycosaminoglycan patterns in angiosarcoma of the liver. Cancer, 40:3050-3053, 1977. 10. Whelan, J. G. Jr., Greenberg, R. and Tanfcurro, C. H. The effectiveness of radioisotopic scans and grey scale ultrasonography in. the detection of liver damage. Gastroenterology, 79:1129, 1980. 11. Rosch, J., Antonovlc, R. and Dotter, C.T. Transjugular approach to the liver, biliary system and portal circulation. American Journal of Roentgenology, 125:602-608, 1975. 12. Whelan, J. G. Jr., Creech, J. L. and Tantiurro, C. H. Primary liver cancer detection in vinyl chloride workers by radioisotopic scanning (In Preparation). 13. Taylor, K. J. W, Glees,J. P., Smith, T. A. and Carpenter, 0. A. Ultrasonic examination of the liver. In Ultrasound in Medicine. Vol, , pp. 173-174 (Eds.) White, D. N. and Barnes, R., Plenum Press, New York, 1976. CMA 052884 10 14. Taylor, K. J. W. and Carpenter, 0. A. Comparison of radioisotopic and ultra sound examination in the investigation of hepatobiliary disease. In U1trasound in Medicine. Vol. 1, pp. 159-167 (Ed.) White, D. N., Plenum Press, New York, 1976. 15. Clarmont, R.J. and Chalmers, T-C. The transarfrinase tests in liver disease. Medicine, 46:197-207, 1967. 16. Tamburro, C.H. and Greenberg, R. Effectiveness of federally-required medical laboratory screening in the detection of chemical liver Injury. Environmental Perspectives, 1980. In Press. 17. Tamburro, C.H. and Greenberg, R.A. Identification of human toxicity and carcinogenicity by ethylene derivatives in mechanisms of toxicology and hazard evaluation (eds. Holmstedt, B., et al) Elsevier/North-Holland Biomedical Press, New York, New York, pp. 319-334, 1980. 18. Tamburro, C.H., Makk, U and Popper, H. Early hepatic histological alterations among chemical (vinyl monomer) workers. Gastroenterology, 77:A43, 1979. CMA 052885 Table 1 CMA. 0 5 2 8 8 6 CLASSIFICATION Biopsy Biochemical Ultrasound Microvascular DETERMINED BY CRITERIA a) Positive by consensus agreement 1) Dr. Popper, Pathologist if medically significant pathology 2) Dr. Makk, Pathologist is present. Negative otherwise. 3) Dr. Tamburro, Hepatologist b) Pathology is of chemical or non chemical origin. (18) Liver "Function" Tests Group 1 Group 2 SGPT GGTP ICG BILIRUBIN SGOT ALK. PHOSPHATASE Positive if two or more Group tests were abnormal or if one Group 1 and both Group 2 tests were abnormal. Negative if all six tests were normal. Intermediate otherwise. Dr. Taylor Defined as negative if the over all impression was normal. Posi tive otherwise. Dr. Maricq Defined as negative if there were no microvascular abnormalities. Positive otherwise. Glycosaminoglycans Uronic _ UG Urnlc Acid nc*d MG Creatinine Defined as positive for values less than 2.0 or greater than 4.8. Negative otherwise. Liver Scan Dr. Whelan Defined as positive if any patho logical defect was detected. Negative Otherwise. TABLE 2 COMPARISON OF BIOPSY AND BIOCHEMICAL DETERMINATIONS OF THE PRESENCE OF LIVER DISEASE 12 BIOPSY BIOCHEMICAL POSITIVE ' NEGATIVE BIOCHEMICALLY TOTAL SUM INDETERMINATE POSITIVE NEGATIVE SUM 3 0 3 18 21 15 36 88 7 15 26 29 22 51 r 0.20966 = 0.200 N.S. 2 Matched xj - (18 " 1} /l8 - 16.1 P < 0.001 CMA 052887 TABLE 3 FREQUENCY (F) AND RELATIVE FREQUENCY (R.F.) OF GAGS FOR NORMAL AND ABNORMAL EICCBEI-IICAaj RESUi-iT^ GAG <2 2<3 3<4 4<5 5+ Sum Mean Variance Normal_____________ Abnormal R.F. f R.F. 4 0.111 16 0.444 13 0.361 1 0.028 2 0.056 6 0.250 7 0.292 6 0.250 1 0.042 4 0.167 36 1.000 2.886 0.746 24 1.000 3.160 1.776 t = 0.968 N.S. T Test Independent Means F = 2.381 P < 0.05 F Test Independent *3'3S Variances ( Two Tailed) CMA 052888 14 TABLE 4 CORRELATION MATRIX (Ro) BETWEEN FOUR SCREENING TESTS USED TO PREDICT TOE PRESENCE OR ABSENCE OF LIVER DISEASE GAG CAPILLARY ULTRASOUND RADIOISOTOPIC ANALYSIS ASSESSMENT STUDY SCAN GAG ANALYSIS 0.08 -0.03 0.005 CAPILLARY ASSESSMENT (113) 0.03 0.07 ULTRASOUND STUDY (87) (84) 0.06 RADIOISOTOPIC SCAN (120) (114) (88) None of the correlations are statistically significant (a = 0.05). The correlations are given above the diagonal. The sanple size is given in parenthesis below the diagonal. CMA. 052889 15 TABLE 5 mrrAggTPTOVTTgj CF BICPSUD EMPLOYEES FOR CHEMICALLY Tranrrm ABNORMALITIES Biopsy Ultra Sound Chemical Abnormality Other Chemical Abnormality Other 7 26 5 13 SUM 33 18 SUM 12 39 51 Specificity 13/39 " 0.333 Sensitivity = 7/12 = 0.583 Sum = 0.916 CMA 052890 GROUP A FIGURE 1 GF3XP B 16 *Not seen by Doctor Taylor CMA 052891 FIGURE 2 S e n s itiv ity and S p e c ific ity CMA. 0 5 2 8 9 2 \ 1 Nimbfr Fl.ployees Positive Biopsies Negative Biopsies SENSITIVITY AND SPECIFICITY. FOR FOUR SCPEENING TESTS FOR THE PREDICTION OF LIVER ABNORMALITIES (AS DETERMINED BY BIOPSY) 02 2's 1.13 Specificity T ^ Sun: Sensitivity fPlus g5% Limits 1.04 .53 .75 80 .60 40 .64 [Sensitivity [Specificity 40 GAG (51) 36 15 Microvascular Ultrasound (49) 34 15 (51) 36 15 Scan (51) 36 15 CMA 0 5 2 8 9 3 CO >. JP tOH) <H OH d) a CO 2C3 co '>rt P o d) w Figure 3 SENSITIVITY AND SPECIFICITY FOR FOUR SCREENING TESTS FOR THE PREDICTION OF LIVER ABNORMALITIES (AS DETERMINED BY REVIEWED BIOCHEMICAL TESTS) Ntmber Ehployees Positive Biochemical Negative Biochemical (60) 24 36 (57) 22 35 (37) (61) 24 37