Acromegaly and Gastrointestinal Cancer

Elaine Ron, PhD,* Gloria Gridley, MS,* Zdenek Hrubec, ScD,* William Page, PhD,t Shobhit Arora, and Joseph F. Fraumeni Jr, MD*
A cohort of 1041 men who were discharged from the hospital with a diagnosis of acromegaly were examined for subsequent cancer. With a mean follow-up time of 8.3 years, an increased rate of cancers of the digestive organs was observed (27 cases; standard incidence ratio [SIR], 2.0; 95% confidence interval [CI], 1.3 to 2.9). Rates were elevated for cancers of the esophagus (7 cases; SIR, 3.1), stomach (4 cases; SIR, 2.51, and colon (13 cases; SIR, 3.1). The increased risk of colon cancer in acromegaly is consistent with previous clinical reports and suggests opportunities for etiologic research and early cancer detection. It would seem prudent to also evaluate this risk in current research on the use of growth hormone in older individuals to increase muscle mass and reduce body fat. Cancer 68:1673-1677,1991.

distinct clinical disease of somatic growth and metabolic derangements caused by increased secretion of growth hormone. It usually results from a benign pituitary somatotrophic tumo? that is often present for 10 to 20 years before diagnosis. In recent years, several clinical and case reports6- have suggested an increased incidence of colon cancer among subjects with acromegaly. In addition, there have been case reports of acromegaly associated with multiple myeloma, bone cancer, and malignant pituitary tumor. A follow-up study of 166 patients with acromegaly reported an increased number of cancer deaths overall, but no instances of colon cancer were found.12 To clarify the cancer experience associated with acromegaly, we evaluated the risks arising in a sizable cohort of patients with acromegaly who were discharged from all Veterans Administration (VA) hospitals between 1969 and 1985.
Methods

CROMEGALY is a rare but

A cohort of 12 I2 patients with acromegaly (diagnosis code 253.0 in the eighth and ninth revisions of the International Classification of Diseases) was identified from a computer-based file: of 4,364,184 persons discharged beFrom the *Epidemiology and Biostatistics Program, National Cancer Institute. Bethesda, Maryland; and the ?Medical Follow-up Agency, National Academy of Sciences, Washington, DC. Address for reprints: Gloria Gridley. MS, National Cancer Institute, EPN 443. Bethesda, M D 20892. Accepted for publication June 24, 1991.

tween July 1, 1969 and September 30, 1985 from all VA hospitals in the United States. Due to the small number of female patients, the study population was restricted to 1 190 male veterans. An additional 149 patients with acromegaly were excluded for the following reasons: they died during their first admission for acromegaly (43 patients); they were older than 100 years of age at first admission (1 patient); they had a diagnosis of cancer along with or before their first diagnosis of acromegaly in the VA system (87 prevalent cancers); or they were not white or black (1 8 patients). The final acromegaly cohort consisted of 1041 male patients (859 were white and 182 were black). Statistical analyses used standardized incidence ratios (SIR) (ratios of observed to expected events) in which the expected numbers of cases of cancers were determined by applying internally generated incidence rates. These rates were obtained by following through the VA hospital system all 3.7 million first admissions among white and black male veterans between the ages of 18 and 100 years (excluding, as mentioned above, patients who died or who had a diagnosis of cancer on their first visit, but not excluding the patient with acromegaly). The expected number of cases of cancer was calculated by applying 5-year age and 5-year calendar-time and race-specific cancer incidence rates from the 3.7 million patients of the VA hospitals to the person years at risk of the acromegaly cohort. Person years at risk were computed for each patient from the date of first hospital discharge (for the acromegaly cohort, the reported date of first discharge with a diagnosis of acromegaly was used) to the date of first admission for

1673

1674

CANCER October 15 1991

Results

Vol. 68

TABLE Selected Characteristics of Men Diagnosed With 1. Agromegaly and a 1% Sample of All Male Veterans Discharged From Veterans Administration Hospitals, 1969-1985 Characterstic No. of men identified from admissions records Person years of follow-up Mean years of follow-up Mean age at study entry (yr) Mean age at end of follow-up (yr) Mean year of cancer diagnosis Acromegaly
1041 8619 8.3 52.7 61.0 1979.3 1% Sample 37036 314141 8.5 50.1 58.6 1979.7

cancer, the date of death, age of 100 years, or the end of follow-up (September 30,1989, whichever occurred first. Patients were assumed to be alive and free of cancer unless death or cancer were documented in the VA hospital record. Exact 95% confidence intervals (CI) for the SIR were Because the findings for white patients and black patients were essentially the same, the results were combined for this report. A 1% systematic sample of the white male patients and the black male patients of the VA hospitals (37,000 patients) was used as a comparison group for demographic characteristics, multiple cancer patterns, and prevalence of malignant and benign disorders. Statistical significance was assessed by Fishers exact test. Hospital records of patients with digestive, pituitary, and brain tumors, as well as a small, randomly selected sample of patients with acromegaly without cancer, were reviewed to validate diagnoses of cancer and acromegaly. Because the confirmation rate for acromegaly was similar for patients with and without cancer, we did not adjust the SIR presented.

The 1041 patients with acromegaly had a total of 8,6 19 person years of follow-up (mean follow-up time, 8.3 years) (Table 1). The mean ages at study entry and at the end of follow-up were 52.7 and 6 1.O years, respectively, approximately 2 years older than a 1% sample of patients admitted to VA hospitals during this time. For the patients with acromegaly, the mean calendar years of entry and diagnosis of cancer were 1976.0 and 1979.3, respectively. Altogether, 1 16 patients had a diagnosis of cancer subsequent to acromegaly, compared with the expected number of 72.3 (SIR, 1.6; 95% CI, 1.3 to 1.9) (Table 2). The increased incidence of cancer was attributable to significantly increased numbers of cases of cancer of the digestive organs (SIR, 2.0), pituitary gland (SIR, 135.0), and brain and other nervous system components (SIR, 3.6). As shown in Table 3, the increased incidence of cancer of the digestive organs was mainly caused by an increase in the number of cases of cancer of the colon (SIR, 3. l), esophagus (SIR, 3. l), and stomach (SIR, 2.5). Black patients and white patients had a similar increased incidence of these tumors, although the risk of esophageal cancer was slightly higher among black patients and the risk of colon cancer was greater among white patients. Only one cancer of the rectum was observed, but 2.24 cases were expected. Exclusion of patients whose diagnosis of acromegaly was ever coded as uncertain ( 192 patients) did not reduce the SIR for colon or stomach cancers, but somewhat decreased the SIR for esophageal cancer (SIR, 2.3; CI, 0.6 to 5.9). Table 4 shows that the risk of gastrointestinal cancers decreased with time after the first admission for acro-

TABLE Standardized Incidence Ratios of Cancers After the Diagnosis of Acromegaly Compared With Those for All Male Veterans 2. Discharged From Veterans Administration Hospitals, 1969-1985 Cancer site
A11 cancers

Observed cancers
1 16*.t

Expected cancers
72.28 6.94 13.50 25.41 0.45 0.77 0.90 14.50 1.38 0.23 0.20 2.10 1.95 0.82 I .70

SIR
1.o

95% CI 1.3-1.9 0.4-2.0 1.3-2.9 0.6-1.3 0.0-6.6 0.4-8.6 0.7-9.7 0.6-1.7 1.3-8.0 0.2-2 1.4 91-193 0.0- 1.42 0.6-4.9 0.1-6.0 0.2-3.9

1.6

Buccal cavity, pharynx (140-9) Digestive organs, peritoneum ( 1 50-9) Respiratory system ( 160-5) Bone ( 170) Connective tissue ( I7 1 ) Melanoma (172) Male genital and urinary ( 1 85-9) Brain & other nervous system ( I9 1, 192) Thyroid (193) Pituitary gland (194.3) Ill defined, unspecified (195, 199) Lymphoma (200-202) Multiple myeloma (203) Leukemia (204-208)

7 27 22 0 2 3 15 5* I 27t 0 4
1

2.0 0.9 0.0 2.6 3.3

1.o

3.6 4.3 135.0 0.0 2.0 1.2 1.2

SIR: standardized incidence ratio; CI: confidence interval. * Two of the three hospital charts of brain cancer were reviewed and both tumors were found to be benign pituitary tumors.

t Eighteen charts of pituitary cancer were reviewed and all tumors were found to be benign.

No. 8

ACROMEGALY N D GASTROINTESTINAL. CANCER A

Ron e a/. f

1675

TABLE Standardized Incidence Ratios of Gastrointestinal Cancer 3. Subsequent to Acromegaly Diagnosis Compared With Those for All Male Veterans Discharged From Veteran Administration Hospitals. 1969-1985 Cancer site Esophagus ( I 50) Stomach ( I 5 I ) Small intestine ( I 52) Colon ( 1 53) Rectum ( 154) Liver ( I 55) Gallbladder ( I 56) Pancreas ( 157) Other ( I 59) Observed cancers Expected cancers
2.29 I .62 0.16 4.22 2.24 0.83 0.24 I .76 0.17

SIR
3.1 2.5 6.2 3.1 0.4 0.0 0.0 0.3 0.0

95% CI 1.3-6.0 0.8-6.0 0.3-30.8 1.7-5.1 0.0-2.2 0.0-3.6 0.0-17.1 0.0-2.8 0.0-17.6

7
4

I
13 1

hospitals. No unusual combination of malignant neoplasms was observed in the acromegaly cohort. There was a significant increase in the number of patients with lipoma in the acromegaly cohort, along with an insignificant increase in the number of colon polyps, benign prostatic hypertrophy, and nonmelanoma skin cancer. However, none of these conditions was found to cluster in patients with acromegaly with cancer overall or with digestive or colon cancer.

0 0 I 0

Discussion
The results of this study showed a significantly increased risk of cancers of the gastrointestinal tract in patients with acromegaly. The increase in the incidence of colon cancer is consistent with several clinical report^.^-^ Our findings, however, indicate that risk may extend to other parts of the digestive tract, including the esophagus and possibly the stomach. Because the epidemiologic profiles for colon, esophageal, and stomach cancer are different, it is unlikely that these associations reflect known risk factors. Because smoking and drinking are major causes of esophageal ~ a n c e r ,it ~ interesting that the patients with acromegaly ' is had a smaller proportion of smoking and alcohol-related diagnoses than the entire VA cohort. No malignant or benign disorders were identified as clustering with the gastrointestinal cancers in the patients with acromegaly, but an unexpected increase in the incidence of lipomas was observed for the larger cohort of patients with acromegaly. Whether the lipomas represent one of the cutaneous manifestation of acromegaly (e.g., fibromas or sebaceous cysts) or a chance finding is unclear. Benign colon polyps and skin tags have been previously
TARI.I: Standardized Incidence Ratios of Selected Cancers by 4. Years Since First Admission for Acromegaly Compared With Those for All Male Veterans Discharged From Veterans Administration Hospitals, 1969-1985
Years since admission for acromegaly Cancer site Gastrointestinal ( I 50-9) Observed Expected SIR 95B CI Colon ( 153) Observed Expected SIR 95% CI Esophagus ( 150) Observed Expected SIR 95% CI <I 5
1.51 3.3 1.2-7.3

SIR: standardized incidence ratio: CI: confidence interval.

megaly, but remained elevated throughout the study period. For colon cancer, the largest increase in risk occurred in the 5 years after the diagnosis of acromegaly, but a twofold (although not significant) increase in risk persisted until the end of follow-up. A number of the patients with acromegaly had concomitant pituitary disorders (20%) or polyglandular syndrome (170). After exclusion of these patients from the analysis, the SIR for esophageal cancer remained constant (SIR, 3.4; CI, 1.2 to 7.3) but the SIR for total gastrointestinal cancers (SIR, 1.8; CI, 1. I to 2.8) and colon cancer (SIR, 2.7; CI, 1.2 to 5.2) were reduced slightly. The increase in stomach cancer incidence disappeared totally (SIR, 0.8; CI, 0.0 to 4.4). Significant increases in the number of cases of colon cancer still occurred in the 5 years after the diagnosis of acromegaly (< 1 year SIR, 5.7: 1 to 4 year SIR, 4.1), but only two cases of colon cancer appeared after 5 years (SIR, 1.2). A review of the medical charts was attempted for all patients with cancers of the gastrointestinal tract, pituitary gland, and brain and other central nervous system components, as well as a sample of patients without cancer. Of the 89 (59 patients with cancer and 30 without cancer) charts requested, 66 (74%) were obtained. Of the 2 1 charts ofgastrointestinal cancer that were reviewed, all cases were confirmed, although one stomach cancer was diagnosed before the acromegaly. A review of the charts of 20 pituitary and brain cancers showed that all tumors were benign. One tumor coded to the central nervous system turned out to be a malignant astrocytoma, whereas another was a meningioma. Of the 66 charts received, the diagnosis of acromegaly was definite in 50 (76%) cases, probable in 9, and uncertain in 7. To explore whether gastrointestinal cancer and acromegaly might reflect an underlying syndrome, we examined the frequency of a number of conditions reported in the acromegaly cohort compared with a I % sample of white male patients and black male patients of the VA

1-4

5-9 8 4.4 1 I .8 0.8-3.4 3

10+ 4 2.45 I .6 0.5-3.9 2 0.79 2.5 0.4-8.4

10 5.08 2.0 I .O-3.5

3 0.45 6.7 1.7-1 8. I

5 I .55 3.2 I .2-7. I

3 0.88 3.4 I 0.9-93

I .43 2. I 0.5-5.7
3 0.77 3.90 1.0-10.6

0
0.25

I
0.3 I 3.23 0.2-15.9

0
0- 16.4

SIR: standardiicd incidence ratio: CI: confidence interval.

1676

CANCER October 15 199 1

Vol. 68

reported in association with a~romegaly,~ the increase but in the number of colon polyps that we observed was not statistically significant. Despite case reports in the literature linking acromegaly with pituitary cancer,'' bone cancer," and multiple myeloma,' we found no increased incidence of these malignant conditions. Furthermore, no increase in the incidence of leukemia was seen despite recent reports of childhood leukemia after the use of synthetic growth hormone to treat growth retardation. l6 In a review of the charts of digestive cancers, we found no evidence of neuroendocrine or carcinoid tumors, which cause acromegaly in approximately I % of patients. ''.I8 We also investigated the timing of acromegaly and digestive cancer. Of the 2 I charts of cancer reviewed, 20 were incident cases that followed the diagnosis of acromegaly. However, the proportion of gastrointestinal cancers among the 87 patients with acromegaly with prevalent cancer who were excluded from the main analysis was 3 times higher (18% versus 6%)than that found among the 1% sample of prevalent cancers. Previous clinical reports have also noted an increased incidence of digestive cancers before the diagnosis of a~rornegaly.'.~,~ Because acromegaly usually cannot be diagnosed for 10 to 20 years,' it is possible that excess production of growth hormone exerts an effect on the gastrointestinal tract even before the diagnosis of acromegaly, although one might speculate that both conditions are manifestations of an underlying syndrome that needs to be fully defined. For approximately 80% of patients with acromegaly, excess secretion of growth hormone is accompanied by an increase in other compounds such as prolactin or glycoprotein.2 However, the exclusion of 20% of our cohort with a complex pituitary dysfunction did not substantially affect the increased incidence of gastrointestinal cancers, except for stomach cancer. This computer-based data set of more than 4 million hospitalized veterans offers the opportunity to study comparatively rare diseases such as acromegaly and their sequelae. The resource provides a relatively easy method for generating and testing hypotheses related to etiology, natural history, and therapy. In this analysis the increased incidence of esophageal and stomach cancers was not previously reported with acromegaly and should prompt further study. This computer resource has its limitations. Hospitalized male veterans are not representative of United States males in general, and the assumption that all ofthese men were alive and free of cancer until 1985 was not verified. These problems were handled by using internally generated cancer rates. These internal rates have been compared with United States population-based cancer incidence rates of the Surveillance, Epidemiology, and End Results program19 and were found to be generally lower (as expected because of underascertainment of cancers that are

diagnosed outside the VA hospital system). The increased incidences that we found for stomach and colon cancers could be partially due to the lower comparison rates, if no patients with acromegaly had stomach or colon cancers diagnosed outside the VA system. However, the internal VA rates are 40% higher than the Surveillance, Epidemiology, and End Results program rates for buccal, laryngeal, and esophageal cancers, so the increased incidence of esophageal cancer in our study would have been even greater if we had used Surveillance, Epidemiology, and End Results program rates. A review of the medical records disclosed other problems. Although we tried to obtain 89 medical charts, only 66 (74%) could be reviewed. Although the charts confirmed the increased incidence of gastrointestinal cancers, we found that the reported increased incidences of pituitary and brain cancers actually resulted from miscoding of benign pituitary tumors. The threefold increase in the incidence rate of colon cancer, which persisted over time, provides epidemiologic evidence that patients with acromegaly are part of a highrisk group that should undergo periodic screening, similar to patients with a personal or family history of colorectal cancer or polyps. Our findings are of special interest in view of recent reports indicating that growth hormone may be useful in increasing muscle mass and reducing Before such treatment body fat in older is accepted as standard, it would seem prudent to consider the potential risks pointed out by Hennine and Condamine.22 Although the mechanism linking acromegaly to an increased incidence of colon and perhaps other tumors is uncertain, further studies are needed to clarify the role of somatomedins (insulin-likegrowth factors) that mediate the effects of growth hormone and appear to have mitogenic or proliferative effects on colonic e p i t h e l i ~ m . * ~ - ~ ~
REFERENCES
I . Karpf DB, Braunstein GD. Current concepts in acromegaly: Etiology. diagnosis, and treatment. Compr Ther 1986: 12:22-30. 2. Laws ER Jr, Scheithauer BW, Carpenter S. Randall RV. Abboud CF. The pathogenesis of acromegaly. J Neurosurg 1985; 63:35-38. 3. Klein I, Parveen G , Gavaler JS. Vanthiel DH. Colonic polyps in patients with acromegaly. Ann Inlern Med 1982; 97:27-30. 4. Pines A, Rozen P, Ron E, Gilat T. Gastrointestinal tumors in acromegalic patients. A m J Gastroenlerol 1985: 80:266-269. 5. Brunner JE, Johnson CC, Zafar S, Peterson EL, Brunner JF. Mellinger RC. Colon cancer and polyps in acromegaly: Increased risk associated with family history of colon cancer. Clin Endocrinol 1990: 32: 65-7 I . 6. ltuarte EA, Petrini J, Hershman JM. Acromegaly and colon cancer. Ann Iniern Med 1984; 101:627-628. 7. Ritter MM, Richter WO, Schwandt P. Acromegaly and colon canccr. Ann Iniern Med 1987; 106:636-637. 8. Ziel FH, Peters AL. Acromegaly and gastrointestinal adenocarcinomas. Ann Inlrrn Med 1988; 108:5 14-5 15. 9. H a g E, Asplund K. Holm J. Acromegaly and multiple myeloma. Ann Intern Med 1988; 108:437438. 10. Ratner RE. Hare JW. Association of acromegaly and chondrosarcoma. Soufh MedJ 1983; 76:1181-1182.

No. 8

ACROMEGALY AND GASTROINTESTINAL CANCER

Ron et d.

1677

I I . Gordon DA, Hill FM, Ezrin C. Acromegaly: A review of 100 cases. Can Med Assoc J 1962; 87:1106-1109. 12. Bengtsson B, Eden S, Ernest I, Oden A, Sjogren B. Epidemiology and long-term survival in acromegaly. Acta Med Scand 1988: 223:327335. 13. Liddell FDK. Simple exact analysis of the standardized mortality ratio. J Epidemiol Community Health 1984; 38:85-88. 14. Miettinen 0s. Standardization of risk ratios. Am J Epidemiol 1972: 961383-388. 15. Day NE, Munoz N. Esophagus. In: Schottenfeld D, Fraumeni JF

19. Horm JW, Asire AJ, Young JL Jr, eds. SEER Program. Cancer Incidence and Mortality in the United States, 1973-81. Bethesda, MD: National Institutes of Health, 1984 [DHHS publication no. (NIH)851837). 20. Rudman D, Feller AG, Nagraj HS ef al. Effects of human growth hormone in men over 60 years old. N Engl J Med 1990; 323: 1-6. 2 I . Vance ML. Growth hormone for the elderly? N Engl JMed 1990 323152-54. 22. Hermine 0, Condamine L. Effects of human growth hormone in men over 60 years old (Letter). N Engl JMed 1990 323:1561-1562. 23. Tricoli JV, Rall LB, Karakousis CP et a/. Enhanced levels of in-

Jr, eds. Cancer Epidemiologyand Prevention. Philadelphia: W B Saunders.

1982; 596-623. 16. Watanabe S, Yamaguchi N, Tsunematsu Y, Komiyama A. Risk factors for leukemia occurrence among growth hormone users. Jpn J Cancer Res 1989; 80:822-825. 17. Daughaday WH. A new treatment for an old disease. N Engl J Med 1985: 3 I 3: 1604- 1605. 18. Scheithauer BW, Kovacs K, Randall RV, Horvath E, Laws ER

Jr. Pathology of excessive production of growth hormone. Clin Endocrinol Mefah 1986; 15:655-681.

sulin-like growth factor messenger RNA in human colon carcinomas and liposarcomas. Cancer Res 1986; 46:6169-6173. 24. Pollak MN, Perdue JF, Margolese RG, Baer K, Richard M. Presence of somatomedin receptors on primary human breast and colon carcinomas. Cancer Lett 1987; 38:223-230. 25. Conteas CN, Desai TK, Arlow FA. Relationship of hormones and growth factors to colon cancer. Gastroenterol Clin North Am 1988;
17:761-772.