Background: Pancreatic cancer is a highly lethal malignancy that is often diagnosed at advanced stages where treatment options are limited and prognosis is poor. As the immune system is intricately involved in the detection, elimination, and initiation of cancer, there has been major interest in understanding the role of immune dysregulation in the development of pancreatic cancer. Conditions associated with immune overactivity (e.g. allergies), chronic inflammation (e.g. type 2 diabetes, metabolic syndrome) and the epigenetic modification of immune genes (e.g. cytokines) may perhaps provide insight into the underlying mechanisms of pancreatic carcinogenesis. Past retrospective epidemiological studies have observed inverse associations between allergies and pancreatic cancer risk, but this finding has not been detected in prospective studies. In addition, the relationship between type 2 diabetes onset and pancreatic cancer risk has not been well explored in racially/ethnically diverse populations. Metabolic profiles of diabetes patients and pancreatic cancer risk have similarly not been evaluated in non-white individuals. Finally, there is limited literature on the influence of methylation in cytokine genes and pancreatic cancer progression.
Objectives and Specific Aims: The objective of this dissertation is to further elucidate the influence of atopic allergic conditions, type 2 diabetes, and methylation of cytokine genes on pancreatic cancer risk and survival. The specific aims were: 1) to conduct a prospective analysis of atopic allergic conditions (AACs) and the treatment of these conditions on pancreatic cancer risk in a multiethnic population; 2) to investigate the association between type 2 diabetes onset and pancreatic cancer risk in a racially diverse patient-based cohort, and to establish a metabolic profile of diabetes patients who are at high risk of developing pancreatic cancer; and 3) to evaluate the relationship between the methylation of cytokine genes in pancreatic cancer tumors and overall survival.
Methods: For specific aim 1, we analyzed prospective data from 187,266 participants from the Multiethnic Cohort (MEC). Information on AACs and antihistamine medication use was assessed via a baseline questionnaire when individuals joined the MEC in 1993-1996. Adjusted risk ratios (RRs) and 95% confidence intervals (CIs) for pancreatic cancer incidence by AACs and antihistamines were estimated using Cox regression models, adjusting for age, sex, ethnicity, education, smoking status, family history of pancreatic cancer, body mass index, diabetes, and alcohol intake. We further evaluated associations among subgroups defined by age, sex, ethnicity, follow-up time, and known pancreatic cancer risk factors. For specific aim 2, we conducted a retrospective cohort study of 1,499,627 patients from Kaiser Permanente Southern California (KPSC) from 2006-2016. Diabetes patients were identified using glucose and hemoglobin A1c (HbA1c) measurements based on diagnostic criteria by the American Diabetes Association (ADA). We used Cox regression with time-varying exposures to assess the relationship between diabetes status/duration and pancreatic cancer. For individuals with incident diabetes, we evaluated longitudinal changes in glucose, HbA1c and weight measurements leading up to the diabetes diagnosis between those with and without pancreatic cancer. We further performed Cox proportional hazards regression to examine the associations between the percent differences in glucose, HbA1c and weight and pancreatic cancer risk. All models included gender, race/ethnicity, smoking, BMI, alcohol use, family history of pancreatic cancer, history of pancreatitis, and education as covariates. For specific aim 3, we performed a retrospective cohort study of 162 pancreatic cancer patients from The Cancer Genome Atlas (TCGA). We evaluated methylation on CpG probes in proximity to seven genes: IL10, IL6, IL8, TGFβ1, TGFβ2, TGFβ3, and TNF. Methylation was assessed using the Illumina Infinium HumanMethylation450 (HM450) BeadChip methylation assay. We used Cox proportional hazards regression to evaluate the relationship between site- and region-specific DNA methylation and pancreatic cancer survival, adjusting for age, gender, stage and five independent surrogate variables associated with potential confounders measured with error (smoking, receipt of radiation, sample plate). We assessed the relationship between methylation at each given site/region and the expression of the gene in closest proximity using the Spearman’s correlation coefficient.
Results: For specific aim 1, 1,455 incident cases of pancreatic cancer were identified among the white, African American, Latino, Japanese American and Native Hawaiian participants of the MEC with an average 16-year follow-up. AACs (relative risk/risk ratio [RR] 1.00, 95% CI 0.88-1.12) and antihistamines (RR 0.92, 95% CI 0.78-1.07) were not clearly associated with pancreatic cancer incidence. While these associations were also null for most subgroups, we did observe inverse associations of AACs (RR 0.74, 95% CI 0.56-0.98) and antihistamines (RR 0.66, 95% CI 0.45-0.96) among the oldest participants (70+). For specific aim 2, we identified 2,002 incident cases of pancreatic cancer among the white, African American, Latino and Asian members of KPSC with 7.5 million person-years of follow-up. Compared to those without diabetes, individuals with incident diabetes and shorter diabetes durations (≤1 year: RR 6.91, 95% CI 5.76-8.30; >1 year: RR 1.96, 95% CI 1.62-2.38) had the highest risk of pancreatic cancer. Among incident diabetes patients, those with pancreatic cancer had steeper increases of glucose, HbA1c and greater weight loss during the time prior to diabetes. Racial/ethnic differences in metabolic changes by pancreatic cancer status were observed, with more pronounced differences in Asians and blacks for glucose, in blacks and whites for HbA1c, and in Asians and whites for weight (p-values for heterogeneity ≤0.01). For specific aim 3, we observed poorer survival for increased methylation in several CpG probes in TGFβ1 (cg03313751: hazard ratio [HR] 1.62, 95% CI 1.14-2.31), TGFβ2 (cg16658719: HR 1.58, 95% CI 1.21-2.06), TGFβ3 (cg16292972: HR 1.45, 95% CI 1.01-2.09) and IL6 (cg15703690: HR 1.03, 95% CI 1.00-1.05), specifically those near transcription start sites. Conversely, higher methylation in one CpG locus in IL10 (cg14789529: HR 0.97, 95% CI 0.94-1.00) was associated with increased survival.
Conclusions: Our study provides additional evidence of the intricate relationships between immune dysregulation and pancreatic cancer risk and progression. The findings for atopic allergic conditions support the null association for pancreatic cancer observed in prior prospective cohorts. Moreover, increased pancreatic cancer risk for new-onset diabetes is observed among racial/ethnic minorities, while the metabolic profiles for incident diabetes patients appear to vary by pancreatic cancer status and race/ethnicity. Lastly, epigenetic alterations and regulation of cytokine-related gene expression may be involved in the progression of pancreatic cancer.