1394 Gut 1993; 34:1394-1396

Increased risk of and pancreatic

in familial adenomatous polyposis Gut: first published as 10.1136/gut.34.10.1394 on 1 October 1993. Downloaded from

F M Giardiello, G J A Offerhaus, D H Lee, A J Krush, A C Tersmette, S V Booker, N C Kelley, S R Hamilton

Abstract gland,'67 "p"X ,' 67 1`21 biliary Familial adenomatous polyposis has been tree,'6722 and .6 A formal risk assess- associated with several extraintestinal , ment, however, ofthese in FAP has but the relative and absolute risks of these not been reported. malignancies have not been determined. Extra- The purposes of this study were to define intestinal cancers reportedly associated with further the cancers in the FAP phenotype, to adenomatous polyposis (thyroid gland, adrenal determine the magnitude ofrisk ofthese extrain- gland, pancreas, and ) were identi- testinal cancers, and to evaluate the need for fied in polyposis patients and their at risk surveillance ofFAP patients for these . relatives in The Johns Hopkins Registry. The The risks of of the thyroid gland, adrenal rates ofidentified tumours were then gland, pancreas, and biliary tree identified in compared with the general population through subjects affected with and at risk for FAP in The person year analysis with adjustment for Johns Hopkins Polyposis Registry were com- population. For comparison, the incidence pared with the general population of the United rates of the two most common cancers not States through person year analysis. For com- associated with polyposis ( in parison, the risks of the two most common women and cancer) were also calculated. cancers that have not been associated with FAP There was an increased of thyroid (lung and breast cancer) were also calculated. cancer (relative risk 7-6; 95% confidence limits (CL) 2-5-17.7) and pancreatic (relative risk 4*46; 95% CL 1.2-11.4) in poly- Methods posis patients and at risk relatives. The abso- Data were collected from The Johns Hopkins lute risk was 26-8 and 21-4 cases/100 000 person Polyposis Registry that contains 197 pedigrees years, respectively. No cases of adrenal or with FAP.9 Patient information was obtained on biliary cancer were found in this cohort. There all registry subjects through chart review and was http://gut.bmj.com/ was no increased relative risk of subsequently computerised using dBase III Plus (95% CL 0.04-1.4) or breast cancer (95% CL software (Ashton-Tate, Torrance, California). 0.04-1.4) over the general population. The For risk assessment, FAP patients were con- relative risks of thyroid and sidered at risk for extraintestinal cancer from are increased in familial adenomatous poly- birth until death. First degree relatives of FAP posis, but the absolute lifetime risk is low. patients were considered at risk from birth up to Screening for pancreatic cancer may not be 55 years, by which time the phenotype is on September 23, 2021 by guest. Protected copyright. worthwhile with currently available methods, expressed in virtually all affected subjects.' Both but careful physical examination ofthe thyroid FAP patients and at risk relatives contributed gland is warranted along with consideration for person years accordingly. Study time was until Departments of ultrasonography. date oflast contact (n= 285), death (n= 360), date Medicine, Pathology and (Gut 1993; 34: 1394-1396) of diagnosis of extraintestinal cancer (n= 13), or Center, The closing date ofthe study (n=733). The period for Johns Hopkins University School of analysis started on 1 January 1969, and ended 31 Medicine and Hospital, Familial adenomatous polyposis (FAP) is an December 1987. These dates were chosen Baltimore, USA autosomal dominant disease characterised by the because population rates of cancer for com- F M Giardiello D H Lee development ofhundreds ofcolorectal parison were available from Surveillance, A J Krush in young adults.' Ifprophylactic colectomy is not , and End Results (SEER)23 only S V Booker performed, virtually all affected subjects will for this period. Person years at risk were calcu- N C Kelley develop by the fifth decade of lated according to sex, race, and age specific S R Hamilton life. Recently, investigators have discovered that categories up to 84 years of age during five year and the Department of this disorder is caused by germline mutation of calendar time periods of study using a computer Pathology, Academic the adenomatous The ratio Medical Center, polyposis coli located on program for cohort analysis.24 of Comprehensive Cancer chromosome 5q21.2- observed tumours to the expected number was Center West, Amsterdam, FAP can be associated with various extra- computed with a test of significance and 95% The Netherlands colonic . Benign extraintestinal manifesta- confidence limits assuming a Poisson distribu- C J A Offerhaus A C Tersmette tions include osteomas, exostoses, epidermoid tion.2425 This observed/expected ratio indicated Correspondence to: , dental abnormalities, pigmented ocular the relative risk and compared the specific extra- Dr F M Giardiello, fundic lesions, and small bowel adenomas." intestinal cancer risk in the study population Division, Blalock 935, The Johns FAP patients are also at increased risk for with that in the general population. The expected Hopkins Hospital, 600 N intestinal cancers especially duodenal and peri- numbers for specific extraintestinal cancers were Wolfe Street, Baltimore, Maryland 21205, USA. ampullary .9 Extraintestinal cancers calculated by multiplying the person years with Accepted for publication have been reported in association with FAP as corresponding age, race, sex, and calendar time 2 February 1993 well. These include of the thyroid specific incidence rates for specific observed Increased risk ofthyroid and pancreatic carcinoma infamilial adenomatous polyposis 1395

TABLE I Patients with familial adenomatous polyposisfrom carcinoma. No patient in this cohort developed The3johns Hopkins Registry who developed extraintestinal cancer adrenal or biliary cancer. Table II shows the results of person year Age at diagnosis Location of analysis of thyroid, pancreas, lung, and breast Gut: first published as 10.1136/gut.34.10.1394 on 1 October 1993. Downloaded from ofcancer (y) Race Sex cancer cancer. Statistically significantly raised relative 29 W M Thyroid risks ofthyroid and pancreatic cancer were seen. 27 W F Thyroid 27 W F Thyroid The absolute risk was 26-8 and 21-4 29* W F Thyroid cases/100 000 person years respectively. Based 18 B F Thyroid 52 W F Pancreas on an 80 year life span, the lifetime, absolute risk 47 W M Pancreas for was 2*1/100 persons and for 32 W M Pancreas 78 W M Pancreas pancreatic cancer was 1-7/100 persons. The 60 W F Lung relative risks for lung and breast cancer were not 45 W M Lung 57 W F Breast statistically significantly different from the gen- 50 W F Breast eral population of the United States. *Paient at risk for FAP. All others were affected with FAP. Discussion In this study, statistically significantly increased extraintestinal cancers obtained from SEER relative risks of thyroid and pancreatic cancer data.23 Thus, the relative risk was adjusted for were found in FAP families. Of note, all but one race, age, sex, and also calendar time (to account ofthe thyroid and pancreatic cancers occurred in for possible secular trends). The absolute risk known affected patients. These relative risks are was defined as the number of observed car- presumably conservative for two reasons. cinomas per total person years. Firstly, the analysis included first degree rela- Risk assessment was performed for selected tives of FAP patients from birth to withdrawal at cancers reported to be associated with FAP the relatively late age of 55 years. Fifty five years including thyroid carcinoma (ICD 193-0, 9th old was chosen because the FAP phenotype can revision) and pancreatic cancer (ICD 157-0-0-9, first be expressed in middle age and therefore 9th revision). No adrenal or biliary cancers affected and unaffected persons could not be occurred in the study population. The relative distinguished with certainty until this age. Con- risks of FAP associated malignancies were com- sequently, some persons included in the study pared with the two most commonly occurring group do not have germline mutation ofthe APC cancers in the United States, lung cancer (ICD gene, and patients with the mutated gene are 162-00-99, 9th revision) and breast cancer (ICD probably at even higher relative risk. When 174-0-0-9, 9th revision), which are not thought genotyping for the APC gene mutation becomes to be associated with FAP. Only women were readily available, analysis of only affected http://gut.bmj.com/ included in the breast cancer analysis. persons can be done. Secondly, we used overall risks without distinction to specific morphology, further diluting the risk estimates because the Results SEER data contain all histological types of The person year analysis included 1391 affected malignancy for a particular organ site. This or at risk subjects from FAP families and they approach was used because underreporting ofthe contributed 18 682 6 person years of follow up. morphology of specific types of cancer in the on September 23, 2021 by guest. Protected copyright. For breast cancer risk analysis, 711 women SEER data would lead to overestimation of risk. contributed 9698 13 person years. These sub- Importantly, no biliary or adrenal cancers were jects included 604 white men, 646 white women, identified in this cohort indicating no signifi- 76 black men, and 65 black women. cantly increased risk. Table I lists the FAP patients with extraintes- A concern raised by the comparison of a tinal cancers analysed in this study. These registry based population to the general United cancers all occurred within different pedigrees. States population is detection bias - that is, All patients except one (asterisk in Table I) were surveillance ofthe population in the registry may affected by FAP. Of the five patients with lead to a higher diagnostic yield of certain thyroid cancer, three had papillary carcinoma diseases compared with the SEER data. For this and two patients had mixed papillary/follicular reason, the relative risk in this cohort of the two most common cancers (lung and breast) in the United States population were also analysed. These malignancies have not been reported in TABLE II Risk analysis ofextrainterestinal cancers in patients withfamilial adenomatous association with FAP, and the risk of both these polyposis (TheJohns Hopkins Polyposis Registry) as compared with the generalpopulation of cancers was not significantly different from the the United States (SEER data) general population. This result strengthens the 95% Rate per specificity ofour findings and makes it less likely Site, ICD Patient number Number of Relative confidence 100000 carcinomas risk limits that variables account for the 9th revision (person years) (OIE) (personyears) association ofthyroid and pancreatic cancer with Thyroid 1391 5 7-6 2 5 to 17-7 26-8 FAP. factors could be a if 193-0 (18682-6) Confounding problem Pancreas 1391 4 4 5 1-2 to 11-4 21-4 certain lifestyle characteristics, for example, 1570-009 (18682-6) were more common in our Lung 1391 2 0 4 0-4to 1-4 10-7 smoking, etc, 162-0-09 (18682-6) study group than in the reference population. In Breast 711 2 0 4 0-04 to 1-3 20-6 the event, however, an overall increase in the risk 1740-009 (9698-1) ofmalignancy would be expected instead of only O/E = Observed/expected. the specific sites previously associated with FAP. 1396 Giardiello, Offerhaus, Lee, Krush, Tersmette, Booker, Kelley, Hamilton

In our series of FAP patients, thyroid cancer 5 Joslyn G, Carlson M, Thliveris A, et al. Identification of deletion mutations and three new at the familial occurred in the third decade of life. This is polyposis locus. 1991; 66: 589-600. consistent with previous reports.'67 ""i Of 39 6 Boland CR, Itzkowitz SH, Kim YS. Colonic polyps and the gastrointestinal polyposis syndromes. In: Sleisenger MH, Gut: first published as 10.1136/gut.34.10.1394 on 1 October 1993. Downloaded from other reported patients with FAP and thyroid Fordtran JS, eds. : pathophysiology, malignancy, 34 patients had thyroid malignancy diagnosis, and management. 4th ed. Philadelphia: W B Saunders, 1989: 1500-7. diagnosed between 15 and 35 years of age. 7 Haggitt RC, Reid BJ. Hereditary gastrointestinal polyposis Although the relative risk of both thyroid and syndromes. AmJ SurgPathol 1986; 10: 871-87. 8 Giardiello FM, Offerhaus GJA, Traboulsi El, Graybeal JC, pancreatic cancer was much higher than the Maumenee IH, Krush AJ, et al. The value of combined general population, the absolute lifetime risk of phenotypic markers in identifying inheritance of familial adenomatous polyposis. Gut 1991; 32: 1170-4. each of these malignancies in FAP patients was 9 Offerhaus GJA, Giardiello FM, Krush AJ, Booker SV, about 2%. With present diagnostic techniques, Tersmette AC, Kelley CN, et al. The risk of upper gastro- intestinal cancer in familial adenomatous polyposis. Gastro- screening for pancreatic cancer in this population enterology 1992; 102: 1980-2. seems unwarranted, but may become worthwhile 10 Ono C, Iwama T, Mishima Y. A case of familial adenomatous polyposis complicated by thyroid carcinoma, carcinoma of with advances in molecular biological markers, the ampulla ofvater and adrenalcortical .J7pnJ Surg endoscopic ultrasonography or other techniques. 1991; 21: 234-40. 11 Reed MW, Harris SC, Quayle AR, Talbot CH. The association By contrast with the pancreas, the thyroid gland between thyroid neoplasia and intestinal polyps. Ann R Coll is easily accessible to physical examination, and SurgEngl 1990; 72: 357-9. 12 Herrera L, Carrel A, Rao U, Castillo N, Petrelli N. Familial careful in FAP patients is certainly adenomatous polyposis in association with thyroditis. warranted at each examination. In addition, Report of two cases. Dis Colon Rectum 1989; 32: 893-6. 13 Delmarre J, Capron JP, Armand A, Dupas JL, Deschepper B, examination in the third decade oflife Davion T. Familial thyroid cancer and rectocolic adeno- may be justified, but a formal cost benefit study matous polyposis. Gastroenterol Clin Biol 1989; 13: 229. 14 Bulow S, Holm NV, Mellemgaard A. Papillary thyroid ofthis strategy is needed. carcinoma in Danish patients with familial adenomatous Finally, the present study calls renewed atten- polyposis.J7 Colorectal Dis 1988; 31: 29-31. 15 Plail RO, Bussey HJ, Glazer G, Thompson JP. Adenomatous tion to the importance of genetic changes of the polyposis: an association with carcinoma of the thyroid. APC gene on chromosome 5q2 1. When mutated BrJ Surg 1987; 75: 377-80. 16 Perkins JT, Blackstone MO, Riddell RH. Adenomatous in the germline, this gene may be instrumental in polyposis coli and multiple endocrine neoplasia type 2b. A the formation ofextraintestinal cancers as well as pathogenetic relationship. Cancer 1985; 55: 375-81. 17 Arvanitis ML, Jagelman DG, Fazio VW, Lavery IC, the intestinal adenomas and carcinomas that are McGannon E. Mortality in patients with familial adeno- hallmarks of FAP. What part somatic mutation matous polyposis. Dis Colon Rectum 1990; 33: 639-42. 18 Camiel MR, Mule JE, Alexander LL, Benninghoff DL. of this tumour suppressor gene plays in thyroid Association of thyroid carcinoma with Gardner's syndrome and pancreatic cancer unassociated with FAP has in siblings. N Englj Med 1%8; 278: 1056-8. 19 Marshall WH, Martin FIR, MacKay IR. Gardner's syndrome yet to be determined. with adrenal carcinoma. Australian Annals ofMedicine 1967; 16:242-4. 20 Naylor EW, Gardner EJ. Adrenal adenomas in a patient with Supported in part by The Clayton Fund and The McAshan Fund. Gardner's syndrome. Clin Genet 1981; 20: 1. We thank Dr Paul Ladenson for his advice and Ms Linda Welch 21 Painter TA, Jagelman DG. Adrenal adenomas and adrenal

for secretarial support. carcinomas in association with hereditary adenomatosis of http://gut.bmj.com/ the colon and rectum. Cancer 1985; 55: 200-1. 22 Walsh N, Qizilbash A, Banerjee R, Waugh GA. Biliary 1 Bussey HJR. Familial polyposis coli. Family studies, histo- neoplasia in Gardner's syndrome. Arch Pathol Lab Med patholog, , and results of treatment. 1987; 111: 76-7. Baltimore, Maryland: Johns Hopkins University Press, 23 Surveillance, Epidemiology, and End Results (SEER) Program 1975. public use tape. Washington, DC: National Cancer Institute, 2 Nishisho I, Nakamura Y, Miyoshi Y, et al. Mutations of DCPC, Surveillance Program. Cancer Statistics Branch, chromosome 5q21 genes in FAP and colorectal cancer 1990. patients. Science 1991; 253: 665-9. 24 Coleman M, Doublas A, Hermon C, Peto L. Cohort study 3 Kinzler KW, Nilbert MC, Su LK, et al. Identification of FAP analysis with a Fortran computer program. IntJ Epidemiol locus genes from chromosome 5q21. Science 1991; 253: 1986; 15: 698-703. on September 23, 2021 by guest. Protected copyright. 661-5. 25 Breslow NE, Day NE. Statistical methods in cancer research. Vol 4 Groden J, Thliveris A, Samowitz W, et al. Identification and II. The design and analysis ofcohort studies (IARC Scientific characterization of the familial adenomatous polyposis coli Publications No 82). Lyon: International Agency for gene. Cell 1991; 66: 589-600. Research on Cancer, 1987.