Text

theses

Pancreatic cancer (PC) is one of the most lethal cancers in the Unites States (US). It ranks number four among the leading causes of cancer death, and number ten in cancer incidence. In the absence of effective screening and methods of early diagnosis, most pancreatic cancers are diagnosed at a late stage. It has an overall 5 year survival of only 6.7%, so improvements in identifying a risk profile for the cancer and new etiological knowledge are urgently needed. We used the Surveillance, Epidemiology, and End Results (SEER) and the SEER-Medicare database, in which the SEER database is matched to the Medicare database by the National Cancer Institute and CMS (Centers for Medicare & Medicaid Services), to study the association between pancreatic cancer and other chronic diseases: prior cancers, prior non-cancer chronic diseases, and new-onset and long-term diabetes. In chapter one, we performed a retrospective cohort study (1,755,816 Whites, 205,751 Blacks, 176,855 Asians, and 196,192 Hispanics for all sites of cancers) for analysis of standardized incidence ratio (SIR) dividing the observed number of new cases of pancreatic cancer in cohorts of survivors of other cancers by the expected number of cases of pancreatic cancer based on age, gender, and race reference rates derived from SEER for the US general population. We found an interesting pattern of associations with other cancers such as ascending and hepatic flexure colon cancer (White: SIR=1.63, 95% CI 1.35-1.94, Black: SIR=1.89, 95% CI 1.15-2.91, Asian: SIR=1.39, 95% CI 0.63-2.63, and Hispanic: SIR=1.42, 95% CI 0.61-2.80) and stomach cancer (White: SIR=2.03, 95% CI 1.50-2.68, Black: SIR=1.37, 95%CI 0.55-2.82, Asian: SIR=1.49, 95% CI 0.79-2.55, Hispanic: SIR=2.05, 95% CI 0.98-3.77), but not colorectal cancer of sigmoid, junction, and rectum, and not lung, breast, or prostate cancers. SIR’s for the association of other cancers with PC decline for cancers further down the gastrointestinal tract: bile duct 3.75 (95% CI: 2.22-5.93), gall bladder 3.26 (95% CI: 1.49-6.19), cecum 1.08 (95% CI: 0.86-1.34), ascending colon 1.54 (95% CI: 1.23-1.90), hepatic flexure 1.90 (95% CI: 1.32-2.66), transverse colon 1.01 (95% CI: 0/67=1.56), sigmoid colon 0.93 (95% CI: 0.76-1.13), rectum 0.86 (95% CI: 0.67-1.08). Our findings are consistent with the idea that Western diets rich in fats and proteins may stimulate more pancreatic exocrine secretion that could contribute to inflammation in the pancreas as well as other digestive organs. The strength of this hypothesis is that it might be testable through comparisons of pancreatic secretions in persons on high protein/fat and low protein/fat diets in rich and poor resource settings. In chapter two, we did a case-control study with frequency matching on age and length of enrollment in traditional Medicare plans. 28,375 new histologically confirmed PC cases with at least two years of prior enrollment in Medicare Parts A and B were identified. Potential controls were drawn from a 5% random sample of Medicare enrollees who were living in the same SEER-defined geographic areas and met the same enrollment criteria. Controls were frequency matched to cases on age and length of enrollment in a 1:5 ratio (28,375 cases, 141,875 controls). Our findings confirmed male gender and Black race as important risk factors for pancreatic cancer (crude odd ratio of female vs. male: 0.72, 95% CI 0.71-0.74, Black vs. White: 1.24, 95% CI 1.19-1.30). Other minority groups, whose lifestyles may differ from the dominant White pattern, had lower risk (crude odd ratio: 0.74, 95% CI 0.68-0.81). After adjustment for age, gender, and race, the strongest association of PC was with pancreatitis (OR=5.37, 95% CI: 4.98-5.78), implying role of inflammation induced by pancreatic digestive enzymes, in carcinogenesis. These enzymes are secreted into the small intestine and presumably are progressively degraded as they move distally. Odds ratios for the association of PC with inflammatory disease decline as one moves down the GI tract: pancreatitis 5.37, liver diseases 2.07 (95% CI: 1.94-2.20), appendicitis 1.60 (95% CI: 1.30-1.95), diverticulitis 2.30 (95% CI: 2.21-2.39). Most of the chronic diseases that we tabulated are believed to be associated with Western lifestyles or related risk factors and several were associated with pancreatic cancer (adjusted odd ratio of coronary heart disease 1.44, 95% CI 1.38-1.50, hyperlipdemia 2.27, 95% CI 2.20-2.33, hypertension 2.08, 95% CI 2.02-2.13, COPD 1.52, 95% CI 1.45-1.56, diabetes 1.89, 95% CI 1.83-1.95, asthma 1.42, 95% CI 1.34-1.50; adjusted by age, gender, race/ethnicity, and length of enrollment in Medicare). Atherosclerosis of the pancreatic and coronary blood supplies is likely correlated and could explain the association. These associations are only of moderate strength, but the multiplicity of associations suggests several aspects of Western lifestyles might contribute to pancreatic cancer risk. The associations with the atopic diseases would also support this view. The protective effect of dementia, which has been reported before, seems likely to be factitious but deserves further study. In chapter three, we used SEER-Medicare database as in chapter two to perform a case-control study with 1:5 frequency matching on age and length of enrollment (cases: 24,004, controls: 120,020), but categorized diabetes into new-onset diabetes that occurred within two years of diagnosis of pancreatic cancer and long term diabetes known to have been present for more than two years. In this very large population-based case-control study of older Americans, we found highly significant, adjusted odds ratios of 2.07 (95% CI 2.00-2.15) and 1.46 (95% CI 1.35-1.49) for diabetes of >2 years and <=2 years duration, respectively. The robust association between diabetes and pancreatic cancer is not clearly understood and remains an important target for further investigation. It is notable that in the SEER Medicare population the prevalence of diabetes of greater than two years duration in both cases and controls exceeded the prevalence of diabetes of shorter duration in each race and ethnic group examined. These findings suggest that if future screening for pancreatic cancer is to target persons with diabetes, that all such persons, not just those with diabetes of recent onset, should be considered. We suggest that Western high fat/protein diet promotes increased secretion of pancreatic enzymes contributing to tissue inflammation in GI tract sites as people age. This could be tested in small bowel aspirates in people eating Western and non-Western diets. Atherosclerosis in the coeliac and other relevant arteries could reduce blood flow and weaken tissue resistance to exocrine secretions within the pancreas. Arterial plaques could be assessed by review of imaging studies done for PC and for other conditions. Our studies are based on the SEER cancer registries and the Medicare claims database, which both lack exposure records for environmental and dietary risk factors as well as for genetic traits. So the associations found could be confounded by differences in these other characteristics, which is an important limitation here. Although we selected population from 1992 to 2012 in the study of relationship between pancreatic cancer and other prior cancers, some individual prior cancer were too few for analysis after stratification by race and ethnicity, gender, age groups, and subtypes of cancer, which is another limitation in our study.

ETD_8660

Ph.D.

Includes bibliographical references

by Zhuliang Tao

doi:10.7282/T3PG1VZD

ETD doctoral

The author owns the copyright to this work.