Are people with family history of pancreatic cancer enriched with genetic propensity for diabetes mellitus?

Introduction

Pancreatic cancer ranks among the deadliest cancers, with a 5-year survival rate of just 13% (1). The vast majority (about 90%) of cases are pancreatic ductal adenocarcinoma, an especially aggressive and difficult-to-treat form (2). Globally, pancreatic cancer cases and deaths are expected to rise sharply by 2045, according to the Global Cancer Observatory (3). With low survival rates and no screening available, prevention through risk factor management is essential.

A small number of factors are associated with pancreatic cancer risk, including type II diabetes, which is especially important due to its growing global impact (4,5). Diabetes mellitus affects a significant portion of the U.S. population, with approximately 10% of Americans, or over 38 million individuals, living with the condition (6). Most of these cases, ranging from 90% to 95%, are classified as type II diabetes mellitus. While this form of diabetes mellitus has traditionally been associated with middle-aged and older adults, there is a concerning trend of increasing diagnoses among children, adolescents, and young adults (6). This increase in diabetes prevalence in the U.S. and globally is likely due to an increase in the prevalence of modifiable risk factors (7). Yet, there remains strong evidence supporting an underlying genetic predisposition. The genetic basis for developing type II diabetes mellitus is characterized as polygenic, with over 50 genes detected (8), and there is no evidence that a single gene fully explains the risk of type II diabetes mellitus, except in limited populations (9). As a result, polygenic scores (PGS) for diabetes have been developed to better capture this polygenic risk (10).

As for most cancers, persons with a first-degree family history of pancreatic cancer (FH+) have about twice the risk of pancreatic cancer (11). This increased risk is attributed to inherited genetic factors, lifestyle risk factors, and their interactions (5). There is clear evidence linking familial cases of pancreatic ductal adenocarcinoma to mutations in critical genes (12-14), and Mendelian segregation analysis further supports the hereditary aspects of the disease (15). Genome-wide association studies (GWAS) have also identified numerous common single nucleotide polymorphisms (SNPs) that contribute to increased susceptibility (16). Obesity, smoking, and heavy alcohol drinking are risk factors (17), and these may interact with genes to increase pancreatic cancer risk (18).

Whether diabetes causes pancreatic cancer was debated for many years, although now consensus for a causal association is being reached (15,19). Several approaches have been used to attempt to distinguish between diabetes as a cause versus the result of pancreatic cancer, including modeling longer versus shorter time between diabetes onset and pancreatic cancer diagnosis, and using Mendelian randomization (MR) with genetic instruments for diabetes to evaluate the association with pancreatic cancer. For example, diabetes mellitus is a recognized feature of pancreatic cancer (20,21), with new-onset diabetes (diagnosed within the past two years) being particularly common among pancreatic cancer patients (22,23).

Previous MR studies, which use genetic data to assess causality, have produced mixed results: some suggest a causal relationship, while others do not (24,25). The inconsistency may be due to limitations in the genetic instruments used, which can weaken the statistical power of these studies. However, more recent MR studies have demonstrated a more consistent positive association, suggesting a strengthening body of evidence in support of causality (26).

Distinguishing between new-onset diabetes resulting from yet undetected pancreatic cancer and diabetes as a risk factor for pancreatic cancer is essential for public health action. The former may provide information to improve earlier detection of pancreatic cancer (15,16). The importance of implementing a screening program for the early detection of pancreatic cancer is critical due to its deadly nature. However, because pancreatic cancer is rare in the general population, a broad screening initiative is not feasible. Instead, it has proposed that focusing on targeted screening for high-risk individuals would be more effective in reducing the incidence of this cancer (27-29), a strategy that is being tested in pilot study in the United Kingdom’s National Health Service (30,31). The latter may provide information to support pancreatic cancer prevention. Preventing disease by intervening on risk factors is a strategy that avoids morbidity and reduces healthcare costs (32). Thus, being able to determine whether diabetes is also a cause of pancreatic cancer is essential evidence for (I) including in diabetes prevention and treatment guidelines the avoidance of pancreatic cancer as benefit, and possibly for (II) testing whether targeting persons with a FH+ for more extensive screening for diabetes and diabetes management would reduce their subsequent risk of pancreatic cancer.

This study aimed to determine whether genetic risk of diabetes, as measured by a PGS, is associated with a FH+ in persons without a personal history of pancreatic cancer. If diabetes mellitus causes pancreatic cancer, then we hypothesized that individuals with a first-degree FH+, reflecting both inheritance and lifestyle risk factors, are more likely to be enriched with genetic propensity to diabetes than individuals without a first-degree FH+. Our hypothesis assumes that the major causal genes for diabetes and pancreatic cancer are not identical. We present this article in accordance with the STROBE reporting checklist (available at https://ace.amegroups.com/article/view/10.21037/ace-2025-11/rc).

Methods

Study design and population

We conducted a cross-sectional study among 3,911 participants using data from “All of Us”. The “All of Us” Research Program is a National Institutes of Health (NIH) initiative to improve medical precision and population health (33). The goal is to enroll about one million participants with the majority coming from historically underrepresented in biomedical research. “All of Us” incorporate electronic health records (EHRs), surveys, physical measurements, biospecimen (e.g., blood and urine), survey responses, and data from wearable devices. The diverse dataset enables researchers to study the interplay between genetics, environment, and lifestyle on health outcomes (33,34).

“All of Us” is Institutional Review Board (IRB)-approved (35), all participants provided informed consent. For this study, we analyzed de-identified data through the secure “All of Us” Researcher Workbench without direct access to raw participant data. This platform provides a controlled environment for data analysis while maintaining participant privacy and security (35,36).

Figure 1 shows inclusions and exclusions. We included adults aged 30 years or older in the study. The minimum age range helped ensure that we could observe individuals with a FH+, as the mean age of pancreatic cancer diagnosis is 70 years. We excluded individuals with a personal history of pancreatic cancer or a known diagnosis of type 1 diabetes or genetic syndromes that predispose to both pancreatic cancer and diabetes (e.g., hereditary pancreatitis, cystic fibrosis).

Figure 1 Inclusions and exclusions, cross-sectional study of 3,911 participants in “All of Us”.

Assessment of first-degree FH+

FH+ was assessed in “All of Us” through structured EHR reviews and self-reported questionnaires capturing first degree relatives’ cancer histories. Participants completed the personal and family health history survey, indicating whether any first-degree biological relatives (parents, siblings, children) had been diagnosed with pancreatic cancer. Survey responses were mapped using OMOP PPI vocabulary codes (standardized codes used to represent all survey questions and answers). EHR-derived data were assessed using ICD-10 Z80.0 (“family history of malignant neoplasm of digestive organs”). We relied primarily on self-reported data. For example, a participant was classified as ‘yes’ only if they responded ‘yes’ to the self-report question and then were confirmed with corresponding EHR data with the relevant ICD-10 code that explicitly captures family history of digestive cancer.

Calculation of a PGS for type 2 diabetes

The PGS for type II diabetes was calculated using pre-trained weights based on publicly available summary statistics (i.e., discovery GWAS) from the PGS Catalog (37). The discovery GWAS included ~6.9 million SNPs within a sample of persons of European ancestry (26,676 cases, 132,532 controls). It explains ~3% of variance in diabetes risk and shows good predictive performance [area under the receiver operating characteristic (AUROC) ~0.73] in European samples. We limited the ~6 million SNPs from the discovery GWAS to SNPs that were present in the “All of Us” sample. We then used the scoring function in Plink 1.9 to create the Type 2 diabetes PGS in “All of Us”. Notably, this approach does not involve intentional selection of any genetic variant(s), but instead leverages all available variants in our sample, which are weighted based on the effect size from the existing GWAS (38). The resulting PGS was then regressed on the top 10 genetic ancestry principal components (provided by “All of Us”) and standardized prior to analysis.

Statistical analysis

Means and prevalences of baseline demographic and clinical characteristics of the study participants, stratified by FH+, were calculated (33). Logistic regression models were used to estimate the odds ratio (OR) of FH+ associated with a high diabetes PGS, before and after adjusting for factors that are associated with diabetes and/or pancreatic cancer. We viewed the underlying genetic risk of diabetes as the risk factor for pancreatic cancer, and as such, we hypothesized that persons with a higher genetic risk of diabetes should be more likely to have a FH+. Thus, we modeled FH+ as the dependent variable and diabetes PGS as the independent variable. However, statistically, both modeling approaches—PGS predicting FH+ or FH+ predicting PGS examine the same association in a cross-sectional analysis.

Assessing bias

To explore potential sources of bias, we performed sensitivity analyses and simulations. We examined the possibility of selection bias due to our excluding participants with a personal history of pancreatic cancer from the study population. Additionally, we evaluated the potential for survival bias by stratifying the analysis by age (≤50, >50 years old, a typical cut-point for young onset cancer).

Ethical consideration

This study was conducted in accordance with the principles of the Declaration of Helsinki and its subsequent amendments. The research utilized de-identified data from the “All of Us” Research Program Researcher Workbench. The “All of Us” Research Program has obtained approval from a centralized IRB, and all participants provided informed consent at the time of enrollment. Because this study involved secondary analysis of de-identified data, it was considered non-human subjects research and did not require additional institutional ethics approval or individual informed consent for this analysis.

Results

Table 1 shows the demographic and lifestyle factors of the 3,911 participants included in the analysis. Of them, 173 (4.4%) had a first-degree FH+, and 3,738 participants did not. The diabetes PGS is standardized to a mean of 0. Mean PGS was close to 0 in both those with and without a FH+. However, individuals with a FH+ had slightly lower mean diabetes PGS (mean −0.07, standard deviation 0.48; range, −1.54 to 1.14) than those without a family history (mean 0.003, standard deviation 0.50; range, −1.82 to 1.93), and their range was narrower. Of those with a FH+, 18.5% (32 of 173) had a high PGS, compared to 25.3% (946 of 3,738) of those without a family history.

In both unadjusted and adjusted models—first controlling for demographic variables, and then additionally for lifestyle risk factors for pancreatic cancer (multivariable-adjusted model), the quartile analysis showed a non-linear association between PGS and FH+. Participants in the highest diabetes PGS quartile (Q4) had a lower odds of having a FH+ compared to those in the lowest quartile (Q1) [multivariable-adjusted OR =0.66, 95% confidence interval (CI): 0.41–1.04; Table 2]. In contrast, the second (Q2) and third (Q3) quartiles of the diabetes PGS were not associated with FH+ (Table 2). Although the prevalence of daily alcohol drinking was higher in those with FH+ (Table 1), alcohol drinking did not confound the association between diabetes PGS and FH+.

Given that the association appears to be non-linear based on the quartile analysis (Table 2), we next modeled the association of the top quartile (≥75th percentile, “higher genetic risk of diabetes”) versus the bottom three quartiles (<75th percentile, “lower genetic risk of diabetes” (Table 3). In the unadjusted model, higher diabetes PGS was statistically significantly inversely associated with FH+ (P<0.04). Likewise, the OR was similar in the multivariable adjusted model (P<0.04).

Assessing bias

Given that the findings were not consistent with the hypothesis, we performed sensitivity analyses and simulations to assess the possibility of bias.

Bias due to a pre-defined exclusion

In the analytic cohort, we excluded participants who had a personal history of pancreatic cancer (Figure 1) because we were concerned that some participants with a personal history of pancreatic cancer would already be deceased or too ill to participate in “All of Us”. We had assumed that this exclusion would be non-differential with respect to genetic propensity to diabetes. However, this assumption could be incorrect. Persons with a personal history of pancreatic cancer may be more likely to have a FH+ than persons without a personal history of pancreatic cancer. If our hypothesis is true, then those with both a personal and FH+ would also be even more enriched with a genetic propensity to diabetes, than participants without a personal history but with a family history. Thus, it is possible that we induced selection bias in making this exclusion. To assess this possibility, we included participants with a personal history of pancreatic cancer and re-calculated the OR for higher PGS and FH+ (unadjusted given the lack of confounding by covariates). We still observed an inverse association (OR =0.74), albeit attenuated, compared with the main analysis excluding those persons (OR =0.67). We also simulated these ORs for the scenario that diabetes PGS is not associated with FH+. When including those with a personal history of pancreatic cancer, OR =1 (by design) and when excluding them, OR =0.90. These assessments suggest a modest (~10%) selection bias resulting from our exclusion of those with a personal history of pancreatic cancer (assuming no survival bias for entry into “All of Us”).

Bias due to differential survival

We also considered whether a higher risk of death or illness among persons with both a higher genetic risk of diabetes and a FH+ (e.g., because they have both diabetes and a personal history of pancreatic cancer) than only one or neither state may have led to their differential underrepresentation among participants in “All of Us”. First, we stratified by age. In younger participants (≤50 years), among whom we expected a lower likelihood of survival bias (e.g., too young, on average, to have developed both diabetes and pancreatic cancer), the association was less inverse (OR =0.75, 95% CI: 0.47–1.18). In contrast, the OR was more inverse in older participants (>50 years, OR =0.60, 95% CI: 0.38–0.94).

Second, we performed a simulation in which we sequentially increased the proportion of those in the higher PGS/FH+ category (i.e., including those who did not survive/were too ill to participate) relative to those in the higher PGS/no FH+ category. As the percentage increased from the observed 0.94% in that category, the OR became less inverse, null, and then became positive (Table 4). For example, when the category proportion was doubled to 1.88%, the OR was 1.39 when individuals with a personal history of pancreatic cancer were excluded. When those with a personal history of pancreatic cancer were included, the OR increased to 1.54. We next reviewed the probability of being diagnosed with pancreatic cancer or dying of it from birth to various ages prior to entry into “All of Us” using Surveillance, Epidemiology and End Results Program data (Table S1). For birth to age 50 years, the probability of developing or dying of pancreatic cancer is <0.1% (39), suggesting that in the ≤50-year-old stratum of our analysis, the likelihood of survival bias is very small.

Discussion

The relationship between a genetic predisposition to diabetes mellitus and FH+ likely reflects a complex interplay of shared and distinct risk factors. We hypothesized that if diabetes is a cause of pancreatic cancer, then persons with a FH+, reflecting both inheritance and lifestyle risk factors, would be enriched with a genetic propensity for diabetes. Contrary to our hypothesis, persons with a FH+ were not enriched with a genetic risk of diabetes as defined by a higher diabetes PGS. Through sensitivity analyses and simulations, we conclude that the inverse association between diabetes PGS and FH+ that we observed may be, in part, due to survival bias. This work points to complexities in the conduct of etiologic research using the cross-sectional design.

Prior studies reported a positive association between a FH+ and pancreatic cancer risk (OR =2.79 for first-degree relatives) (40), while a family history of diabetes is linked to a 30–50% elevated risk of pancreatic cancer (40). However, genetic overlap between diabetes and pancreatic cancer appears limited, as we assumed for our hypothesis (38). GWASs have identified distinct susceptibility loci for each condition, with minimal shared genetic architecture. For example, NR5A2 and PDX1 are implicated in both diseases, but were not included in standard diabetes PGS (41,42).

These data from “All of Us” suggest that a genetic propensity for diabetes, as measured by PGS, is not positively associated with a FH+. This contrasts with epidemiological positive links between diabetes and pancreatic cancer but possibly aligns with genetic studies showing limited overlap in susceptibility loci.

First, we considered the possibility of non-causal explanations for the findings. In the “Results”, we considered in detail two possible sources of selection bias. With respect to the exclusion of those with a personal history of pancreatic cancer, we estimated that selection bias accounted for about ~10% of the relative difference from the unbiased OR (assuming no survival bias). With respect to survival bias, we determined that differential survival is very unlikely to account fully for the inverse association in participants ≤50 years old, given the rarity of pancreatic cancer in that age group. Additionally, lifestyle factors (e.g., smoking, alcohol use) and comorbidities like obesity may contribute to familial clustering of pancreatic cancer in addition to genetic predisposition. Indeed, the covariate, alcohol drinking, was positively associated with FH+. However, alcohol was not a confounder of the diabetes PGS and FH+ association, and thus, did not explain the observed inverse association.

Second, given the cross-sectional design and study population age distribution, incomplete ascertainment of FH+ is possible leading to misclassification. We restricted the analysis to a minimum age of 30 years to increase the opportunity for the observation of a family history. However, we cannot rule out differences in age of onset of pancreatic cancer in family members based on the genetic propensity to diabetes leading to differential opportunity to ascertain family history of pancreatic at younger ages in the analytic cohort. Pleiotropy could account for the inverse association if the same gene positively influences one condition (e.g., diabetes) and inversely influences the other (e.g., pancreatic cancer). However, the diabetes PGS we used captures common variants (e.g., in TCF7L2, PPARG) and does not include rare, high-penetrance genes (e.g., BRCA2, PALB2) linked to pancreatic cancer (43), and genes influencing pancreatic development (e.g., PDX1) may contribute to both conditions were not included in diabetes PGS (44,45). Moreover, it is possible that genetic propensity to diabetes without the development of diabetes (e.g., do not have lifestyle risk factors that may interact with genetic propensity) is insufficient for the development of pancreatic cancer (in relatives). Also, diabetes can develop in persons with a low genetic propensity to diabetes due to the presence of lifestyle risk factors, which could be sufficient to lead to the development of pancreatic cancer (in relatives). The relative balance of these scenarios could lead to inverse, null, or positive associations. Finally, as with any observational study, we cannot rule out chance as an explanation for the inverse association observed.

The study has other possible limitations. First, FH+ was largely obtained from self-reported questionnaires, which introduces the possibility of misclassification due to recall errors or incomplete awareness of relatives’ diagnoses. Although EHRs were incorporated, the ICD-10 code Z80.0 denotes a family history of malignant neoplasms of digestive organs broadly and does not specifically distinguish pancreatic cancer, which may reduce the specificity of case identification. Second, the relatively low prevalence of FH+ (4.4%) may have limited statistical power to detect subtle associations. Third, the diabetes PGS applied in this analysis was developed primarily in populations of European ancestry. While the analytic sample was predominantly White, reduced predictive performance in individuals from other ancestral backgrounds remains a potential source of bias in a diverse cohort such as “All of Us”. Fourth, the cross-sectional study design precludes assessment of temporality and limits causal inference. Finally, the decision to categorize the PGS into quartiles or binary may have led to a loss of information relative to modeling it as a continuous construct.

Conclusions

Our hypothesis that a genetic propensity to diabetes would be positively associated with a FH+, if diabetes is a cause of pancreatic cancer, was not supported in this cross-sectional study in “All of Us”. We explored non-causal explanations for the observed inverse association and determined that selection bias, in part, explanatory. Our work highlights the inferential limitations of cross-sectional studies, even when embedded in state-of-the-art epidemiologic databases. Future research should address these biases through prospective designs or simulation-based approaches.

Acknowledgments

This research was conducted using data from the “All of Us” Research Program, a program supported by the National Institutes of Health. The “All of Us” Research Program is supported by the NIH Office of the Director and other NIH Institutes and Centers under Award Number OT2OD035580. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the “All of Us” Research Program. All of Us Research Program. All of Us Researcher Workbench. Version [All of Us Controlled Tier Dataset v7]; [access date - 24 January 2025]. https://www.researchallofus.org/

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://ace.amegroups.com/article/view/10.21037/ace-2025-11/rc

Peer Review File: Available at https://ace.amegroups.com/article/view/10.21037/ace-2025-11/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://ace.amegroups.com/article/view/10.21037/ace-2025-11/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the principles of the Declaration of Helsinki and its subsequent amendments. The research utilized de-identified data from the “All of Us” Research Program Researcher Workbench. The “All of Us” Research Program has obtained approval from a centralized Institutional Review Board, and all participants provided informed consent at the time of enrollment. Because this study involved secondary analysis of de-identified data, it was considered non–human subjects research and did not require additional institutional ethics approval or individual informed consent for this analysis.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Kratzer TB, Giaquinto AN, et al. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10-45. [Crossref] [PubMed]
2. Hu JX, Zhao CF, Chen WB, et al. Pancreatic cancer: A review of epidemiology, trend, and risk factors. World J Gastroenterol 2021;27:4298-321. [Crossref] [PubMed]
3. Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
4. Rawla P, Sunkara T, Gaduputi V. Epidemiology of Pancreatic Cancer: Global Trends, Etiology and Risk Factors. World J Oncol 2019;10:10-27. [Crossref] [PubMed]
5. Klein AP. Pancreatic cancer epidemiology: understanding the role of lifestyle and inherited risk factors. Nat Rev Gastroenterol Hepatol 2021;18:493-502. [Crossref] [PubMed]
6. CDC. Diabetes. 2024. Available online: https://www.cdc.gov/diabetes/about/about-type-2-diabetes.html
7. Klonoff DC. The increasing incidence of diabetes in the 21st century. J Diabetes Sci Technol 2009;3:1-2. [Crossref] [PubMed]
8. Herder C, Roden M. Genetics of type 2 diabetes: Pathophysiologic and clinical relevance. Eur J Clin Invest 2011;41:679-92. [Crossref] [PubMed]
9. Fuchsberger C, Flannick J, Teslovich TM, et al. The genetic architecture of type 2 diabetes. Nature 2016;536:41-7. [Crossref] [PubMed]
10. Brīvība M, Atava I, Pečulis R, et al. Evaluating the Efficacy of Type 2 Diabetes Polygenic Risk Scores in an Independent European Population. Int J Mol Sci 2024;25:1151. [Crossref] [PubMed]
11. Schenk M, Schwartz AG, O'Neal E, et al. Familial risk of pancreatic cancer. J Natl Cancer Inst 2001;93:640-4. [Crossref] [PubMed]
12. Salo-Mullen EE, O'Reilly EM, Kelsen DP, et al. Identification of germline genetic mutations in patients with pancreatic cancer. Cancer 2015;121:4382-8. [Crossref] [PubMed]
13. Jones S, Hruban RH, Kamiyama M, et al. Exomic sequencing identifies PALB2 as a pancreatic cancer susceptibility gene. Science 2009;324:217. [Crossref] [PubMed]
14. Roberts NJ, Jiao Y, Yu J, et al. ATM mutations in patients with hereditary pancreatic cancer. Cancer Discov 2012;2:41-6. [Crossref] [PubMed]
15. Andersen DK, Korc M, Petersen GM, et al. Diabetes, Pancreatogenic Diabetes, and Pancreatic Cancer. Diabetes 2017;66:1103-10. [Crossref] [PubMed]
16. American Cancer Society. Cancer facts & figures 2025.atlanta. American Cancer Society. 2025. Available online:
17. Anderson LN, Cotterchio M, Gallinger S. Lifestyle, dietary, and medical history factors associated with pancreatic cancer risk in Ontario, Canada. Cancer Causes Control 2009;20:825-34. [Crossref] [PubMed]
18. Malats N. Gene-environment interactions in pancreatic cancer. 2002;:63-7.
19. Ke TM, Lophatananon A, Muir KR. Risk Factors Associated with Pancreatic Cancer in the UK Biobank Cohort. Cancers (Basel) 2022;14:4991. [Crossref] [PubMed]
20. Huxley R, Ansary-Moghaddam A, Berrington de González A, et al. Type-II diabetes and pancreatic cancer: a meta-analysis of 36 studies. Br J Cancer 2005;92:2076-83. [Crossref] [PubMed]
21. Everhart J, Wright D. Diabetes mellitus as a risk factor for pancreatic cancer. A meta-analysis. JAMA 1995;273:1605-9.
22. Chari ST, Leibson CL, Rabe KG, et al. Pancreatic cancer-associated diabetes mellitus: prevalence and temporal association with diagnosis of cancer. Gastroenterology 2008;134:95-101. [Crossref] [PubMed]
23. Pannala R, Leirness JB, Bamlet WR, et al. Prevalence and clinical profile of pancreatic cancer-associated diabetes mellitus. Gastroenterology 2008;134:981-7. [Crossref] [PubMed]
24. Lu Y, Gentiluomo M, Lorenzo-Bermejo J, et al. Mendelian randomisation study of the effects of known and putative risk factors on pancreatic cancer. J Med Genet 2020;57:820-8. [Crossref] [PubMed]
25. Yuan S, Kar S, Carter P, et al. Is Type 2 Diabetes Causally Associated With Cancer Risk? Evidence From a Two-Sample Mendelian Randomization Study. Diabetes 2020;69:1588-96.
26. Ke TM, Lophatananon A, Muir KR. Strengthening the Evidence for a Causal Link between Type 2 Diabetes Mellitus and Pancreatic Cancer: Insights from Two-Sample and Multivariable Mendelian Randomization. Int J Mol Sci 2024;25:4615. [Crossref] [PubMed]
27. Canto MI, Goggins M, Hruban RH, et al. Screening for early pancreatic neoplasia in high-risk individuals: a prospective controlled study. Clin Gastroenterol Hepatol 2006;4:766-81; quiz 665. [Crossref] [PubMed]
28. Canto MI, Goggins M, Yeo CJ, et al. Screening for pancreatic neoplasia in high-risk individuals: an EUS-based approach. Clin Gastroenterol Hepatol 2004;2:606-21. [Crossref] [PubMed]
29. Ludwig E, Olson SH, Bayuga S, et al. Feasibility and yield of screening in relatives from familial pancreatic cancer families. Am J Gastroenterol 2011;106:946-54. [Crossref] [PubMed]
30. NHS England. NHS launches drive to catch one of the most lethal cancers. 2025. Available online: https://www.england.nhs.uk/2025/06/nhs-launches-drive-to-catch-one-of-the-most-lethal-cancers/
31. Bryony Gooch. GPs to comb patient records to spot early signs of killer disease. 2025. Available online: https://www.the-independent.com/bulletin/news/pancreatic-cancer-nhs-test-b2772587.html
32. AbdulRaheem Y. Unveiling the Significance and Challenges of Integrating Prevention Levels in Healthcare Practice. J Prim Care Community Health 2023;14:21501319231186500. [Crossref] [PubMed]
33. All of Us Research Program. From promise to progress. the future of health research is now. 2025:1. Available online: https://allofus.nih.gov/
34. All of Us Research Program. Data types and organization. 2025:1. Available online: https://support.researchallofus.org/hc/en-us/articles/4619151535508-Data-Types-and-Organization
35. All of Us Research Program. Research projects directory. 2023:1. Available online: https://allofus.nih.gov/protecting-data-and-privacy/research-projects-all-us-data
36. All of Us Research Program. Data access tiers. 2025:1. Available online:
37. PGS Catalog. Polygenic score (PGS) ID: PGS000014. 2019. Available online: https://www.pgscatalog.org/score/PGS000014/
38. PLINK 2.0 alpha. Available online: https://www.cog-genomics.org/plink/2.0/
39. SEER*Explorer. An interactive website for SEER cancer statistics. Surveillance Research Program, National Cancer Institute. 2025. Available online: https://seer.cancer.gov/statistics-network/explorer/
40. Austin MA, Kuo E, Van Den Eeden SK, et al. Family history of diabetes and pancreatic cancer as risk factors for pancreatic cancer: the PACIFIC study. Cancer Epidemiol Biomarkers Prev 2013;22:1913-7. [Crossref] [PubMed]
41. Ueno M, Ohkawa S, Morimoto M, et al. Genome-wide association study-identified SNPs (rs3790844, rs3790843) in the NR5A2 gene and risk of pancreatic cancer in Japanese. Sci Rep 2015;5:17018. [Crossref] [PubMed]
42. Wang X, Sterr M, Burtscher I, et al. Genome-wide analysis of PDX1 target genes in human pancreatic progenitors. Mol Metab 2018;9:57-68. [Crossref] [PubMed]
43. Hofstatter EW, Domchek SM, Miron A, et al. PALB2 mutations in familial breast and pancreatic cancer. Fam Cancer 2011;10:225-31. [Crossref] [PubMed]
44. Zhang Y, Fang X, Wei J, et al. PDX-1: A Promising Therapeutic Target to Reverse Diabetes. Biomolecules 2022;12:1785. [Crossref] [PubMed]
45. Roy N, Takeuchi KK, Ruggeri JM, et al. PDX1 dynamically regulates pancreatic ductal adenocarcinoma initiation and maintenance. Genes Dev 2016;30:2669-83. [Crossref] [PubMed]