Sex-Specific Associations between Adiponectin and Leptin Signaling and Pancreatic Cancer Survival

Pancreatic cancer is currently the third leading cause of cancer-related deaths in the United States, with a 5-year survival of 12% (1). Obesity, defined as a body mass index (BMI) higher than 30 kg/m², has been associated with both increased risk and decreased survival of patients with pancreatic cancer (2, 3).

Several mechanisms have been proposed for the role of obesity in pancreatic cancer, including altered levels of hormones secreted by adipose tissue (adipokines), elevated systemic inflammation, hyperinsulinemia, altered cholecystokinin signaling, and gut dysbiosis, but the pathways leading from obesity to increased risk and higher mortality from pancreatic cancer are not well understood (4, 5). Given that more than 40% of the U.S. population is obese (6) and the prevalence of obesity is expected to further increase, it is important to understand how the obesity-associated pathways drive pancreatic cancer progression.

Obese individuals have significant differences in plasma levels of adiponectin (7) and leptin (8), which are hormones primarily responsible for regulating energy balance and metabolism (9). However, by activating their respective receptors (ADIPOR1/ADIPOR2 and LEPR) that are expressed in pancreatic tumor tissue (10, 11), adiponectin and leptin also interact with several signaling pathways that regulate pancreatic tumor cell proliferation, angiogenesis, and metastasis (5). We and others have previously shown an association of high leptin and low adiponectin levels with increased risk of pancreatic cancer (12–16). However, the association of these hormones in the prediagnostic period with patient survival has not been investigated.

To evaluate exposure to adipokine levels in the prediagnostic period in relation to patient survival, we measured circulating adiponectin and leptin levels in blood samples collected up to 26 years before diagnosis in 472 patients with pancreatic cancer. In a subset of 415 patients, we also evaluated polymorphisms in genes coding for the adiponectin and leptin receptors in relation to patient survival.

This study included patients with pancreatic cancer identified among participants from five prospective cohort studies: Health Professionals Follow-up Study (HPFS), Nurses’ Health Study (NHS), Physicians’ Health Study (PHS), Women's Health Initiative–Observational Study (WHI-OS), and Women's Health Study (WHS). HPFS started in 1986, when 51,529 male U.S.-based health professionals ages 40 to 75 completed a mailed biennial questionnaire (17). NHS is a prospective study of 121,700 female U.S.-based nurses initiated in 1976, when nurses ages 30 to 55 returned a mailed questionnaire (18). In both HPFS and NHS, participants were followed by biennial questionnaires that collected demographic, lifestyle, and medical history information. PHS is a clinical trial initiated in 1982 to study aspirin and β-carotene supplementation among 22,071 male physicians’ ages 40 to 84 years (19). The trial was completed in 1995 and participants were further followed as an observational cohort. WHI is an observational cohort study that enrolled 93,676 women ages 50 to 79 years from 1994 to 1998 (20). Participants were asked to return a completed baseline questionnaire and subsequent annual questionnaires. WHS is a randomized clinical trial of low-dose aspirin and vitamin E that enrolled 39,876 female health professionals ages 45 years or older between 1992 and 1995 (21). Following trial completion in 2004, 33,682 women continued being followed as an observational cohort. This study was conducted in accordance with recognized ethical guidelines and approved by the institutional review boards of Brigham and Women's Hospital (Boston, MA) and the Harvard T.H. Chan School of Public Health, and those of participating registries as required. All the participants provided written informed consent.

Pancreatic cancer cases were identified on self-administered questionnaires (returned annually in PHS, WHI and WHS, and biennially in NHS and HPFS), or through death records reporting pancreatic cancer as a cause of death (ICD code 157). All cases included in this analysis were confirmed by medical records, death certificates or cancer registry data that were reviewed by a physician blinded to exposure status. Deaths were identified from next-of-kin, postal services, and the National Death Index, which captures >98% of deaths (22).

Blood samples were collected among 18,225 HPFS participants (1993–1995), 32,826 NHS participants (1989–1990), 14,916 PHS participants (1982–1984), 93,676 WHI participants (1994–1998), and 28,345 WHS participants (1992–1995). Details of blood collection, plasma processing, and storage have been described previously (19, 20, 23–25).

We identified 492 patients with pancreatic cancer with available plasma samples. Because patients with pancreatic cancer often report weight loss in the months before pancreatic cancer diagnosis, which can itself lead to altered adipokine levels (26), we excluded 20 patients with pancreatic cancer diagnosed within 1 year after blood collection, leading to the final dataset of 472 patients.

Plasma adiponectin and leptin were assayed in the laboratory of Dr. Nader Rifai (Children's Hospital, Boston, MA). Adiponectin was measured using an ELISA from ALPCO Diagnostics. Leptin was measured using ELISA with reagents from R&D Systems. All samples were handled identically in a single batch, and laboratory personnel was blinded to patient outcomes. The mean intra-assay coefficients of variance for quality control samples were <10% for both markers.

A total of 29 SNPs in the ADIPOR1/ADIPOR2 and 39 SNPs in LEPR genes (±20 kb) were selected using Haploview tagger algorithm, with minor allele frequency of ≥5% among Whites from the HapMap project database and using r² ≥ 0.8. From 415 patients (Supplementary Table S1), DNA was extracted from buffy coat using QIAGEN QIAmp, and the amplification of genome was performed using GE Healthcare GenomiPhi. Genotyping was performed using a custom-designed Illumina Golden Gate assay at the Partners HealthCare Center for Personalized Genetic Medicine (Boston, MA). Seven SNPs in LEPR and 3 SNPs in ADIPOR1/ADIPOR2 were not supported by the platform. In total, 2 SNPs in LEPR (rs913199, rs2148683) and 3 SNPs in ADIPOR1/ ADIPOR2 (rs35916161, rs16850799, rs16928759) deviated from Hardy Weinberg equilibrium (P < 0.05) and were thus removed from the analysis. In total, 23 SNPs in ADIPOR1/ADIPOR2 and 30 SNPs in LEPR were analyzed (Supplementary Tables S2 and S3). Mean genotype concordance for replicate samples used for quality control was 98%.

Information on participant characteristics, including age, race/ethnicity, weight, height, physical activity, smoking status, and history of diabetes, was obtained from the baseline questionnaire in PHS, WHI, and WHS, and from the questionnaire preceding the blood draw in HPFS and NHS. Dates of cancer diagnosis and cancer stage at diagnosis were obtained by medical records review as previously described (27).

Adiponectin and leptin levels differ significantly between men and women (28, 29). Furthermore, we and others have reported sex-specific associations between those adipokines and several health outcomes, including diabetes, colorectal cancer, and pancreatic cancer risk (12, 25, 30–32). We therefore investigated the association between adiponectin and leptin levels with pancreatic cancer survival separately for men and women.

Adiponectin and leptin levels were divided into sex-specific quartiles. To evaluate the association between adipokines and mortality in patients with pancreatic cancer, we used multivariate Cox proportional hazards models using age as time scale and calculated hazard ratios (HR) and 95% confidence intervals (CI). Models were adjusted for study cohort, variables associated with adipokine levels [fasting time (<8 hours, ≥8 hours, missing), time between blood collection and diagnosis (continuous)], variables associated with pancreatic cancer survival [cancer stage (localized, locally advanced, metastatic, unknown; ref. 1), year of diagnosis (continuous; ref. 1)], and variables with prediagnostic levels or values associated with both adipokine levels and pancreatic cancer survival [race/ethnicity (White, Black, other, missing; ref. 1), smoking (never, past, current with <25 cigarettes per day, current with ≥25 cigarettes per day, missing; ref. 33)]. Although the association between use of hormone replacement therapy (HRT) and pancreatic cancer survival in women has not been evaluated, we included this variable (categorized as premenopausal, postmenopausal HRT non-users, postmenopausal HRT users, unknown) in the multivariate models due to the impact of hormone therapy on metabolic (34) and immune systems (35) that could potentially be associated with pancreatic cancer survival. In the second multivariate model, we also adjusted for BMI (continuous) and diabetes (yes, no) as those variables were previously associated with adipokine levels and pancreatic cancer survival (3, 36–38). We used the Wald test of the cross-term product between sex and adiponectin or leptin quartiles to evaluate the interaction between sex and adipokine levels. To further examine the linear association with continuous adipokine measurements and to evaluate the possibly non-linear association with pancreatic cancer risk, we used the likelihood ratio test comparing the model including linear and cubic spline terms with the model containing only linear term (39). To reduce the influence of extreme values, we excluded the patients with more than two standard deviations from the mean adiponectin or leptin measurements.

Proportional hazards assumption was verified by creating a time-dependent variable (product of adiponectin or leptin levels and time) and was satisfied for both adiponectin (women, P = 0.16; men, P = 0.14) and leptin (women, P = 0.22; men, P = 0.74).

Because adiponectin and leptin levels are moderately correlated (40, 41), we also considered a model adjusted for both markers. Linear trend across sex-specific quartiles was evaluated by entering quartile medians into models.

We performed stratified analysis by time between blood collection and diagnosis (1–10 and ≥10 years), BMI (<30 and ≥30 kg/m²), cancer stage (localized/locally advanced, metastatic), smoking status (never, ever), and HRT use among postmenopausal women (never, ever). The Wald test of product of stratifying variable and marker levels was used to evaluate statistical significance of interaction.

The association between SNPs with patient survival was evaluated using an additive model, where each genotype was modeled as number of copies of minor allele in a Cox proportional hazards model using age as time scale and adjusting for cohort, race/ethnicity, fasting time, smoking status, cancer stage, year of diagnosis, BMI, and diabetes. P values adjusted for multiple hypotheses were calculated using the Benjamini and Hochberg method (42). Analyses were performed using SAS 9.4 and R statistical software, and all the P values were two-sided.

To explore the functional effect of evaluated SNPs on LEPR and ADIPOR1/ADIPOR2 expression, we used the publicly available blood eQTLGen database (https://eqtlgen.org/cis-eqtls.html). This database was generated in 31,467 blood samples from 37 cohorts of the eQTLGen Consortium to identify regulatory mechanisms for genetic variants identified in genome-wide association studies (43). More specifically, we queried the cis-eQTL database, because SNPs located proximally (<1 megabase) from the gene of interest have a stronger effect on gene expression (44).

Data are accessible through the established data sharing policies described at https://www.nurseshealthstudy.org/researchers (email: [email protected]) for NHS, https://sites.sph.harvard.edu/hpfs/for-collaborators for HPFS, https://dbgap.ncbi.nlm.nih.gov/aa/wga.cgi?page=login for PHS, https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs001964.v1.p1 and http://whs.bwh.harvard.edu/ for WHS, and at https://www.whi.org/page/working-with-whi-data for WHI.

Characteristics of patients with pancreatic cancer included in this analysis are shown in Table 1. Average (standard deviation) age at diagnosis was 71.4 (8.1) years, and the average time between blood collection and cancer diagnosis was 8.2 (5.1) years. Most patients (69%) were women. Most patients were diagnosed with metastatic disease (44%), followed by locally advanced (24%), and localized disease (14%), whereas cancer stage was unknown for 85 (18%) patients. By the end of follow-up, 446 (94%) patients had died, and median overall survival was 6 months among all patients.

Consistent with previous studies (28, 29), women had higher mean levels of both leptin (27.3 vs. 9.6 ng/mL in men) and adiponectin (8.7 vs. 5.1 μg/mL in men). Patients with higher prediagnostic leptin had higher BMI, lower levels of physical activity, and were less likely to be White (Table 2). Conversely, patients with higher baseline adiponectin levels had lower BMI, higher levels of physical activity, and were more likely to be white (Table 3). After adjusting for age, study cohort and fasting status, leptin and adiponectin levels were inversely correlated (women, Spearman correlation coefficient = −0.32, P < 0.001; men, Spearman correlation coefficient = −0.24, P < 0.0001; Supplementary Table S4), similar to previous reports (40, 41).

There was no significant association between leptin levels and survival among women (HR, 1.22; 95% CI, 0.74–2.04, comparing top with bottom quartile; Table 4), and no statistically significant linear association (P_trend = 0.11). Similar results were observed in the restricted cubic spline analysis (P_linear = 0.09; Supplementary Fig. S1A). We observed no significant association between leptin and survival in men (HR, 1.35; 95% CI, 0.54–3.33, comparing top with bottom quartile; Table 4; Supplementary Fig. S1B). We observed no significant association between adiponectin levels with survival in men (HR, 0.89; 95% CI, 0.46–1.70, comparing top with bottom quartile; Supplementary Fig. S2B). In women, we observed higher mortality in adiponectin quartile 2 (HR, 1.49, 95% CI, 1.01–2.19), quartile 3 (HR, 2.01, 95% CI, 1.32–3.06), and quartile 4 (HR, 1.71; 95% CI, 1.15–2.54) compared with bottom quartile, with a statistically significant linear trend across quartiles (P_trend = 0.03; Table 4). Similar results were observed in the restricted cubic spline analysis (P_linear = 0.05, Supplementary Fig. S2A).

To account for correlation between adiponectin and leptin levels (Supplementary Table S4), we mutually adjusted for the two hormones in multivariate models and observed no change in associations. In men, there was no association with either adiponectin (HR, 0.80; 95% CI, 0.41–1.57, comparing top with bottom quartile) or leptin levels (HR, 1.48; 95% CI, 0.59–3.72) with survival. In women, we observed inverse association between adiponectin and mortality (HR, 1.65; 95% CI, 1.11–2.47) and no association with leptin (HR, 1.19; 95% CI, 0.71–2.00).

We next performed stratified analyses in women comparing top with bottom quartile of adiponectin levels. To evaluate whether higher adiponectin levels at the time of diagnosis were due to weight loss caused by the occult disease, we stratified the analysis by time between blood collection and cancer diagnosis. The association between adipokine levels and patient survival among women appeared stronger when blood was collected ≥10 years before diagnosis (HR, 2.33; 95% CI, 0.65–8.34, comparing top with bottom quartile) compared with 1 ≤ 10 years before diagnosis (HR, 1.47; 95% CI, 0.91–2.35; P_interaction = 0.65; Fig. 1). Because smoking was previously reported as a modifier of the association between adiponectin and risk of pancreatic cancer (45), we performed a stratified analysis by smoking status at time of blood collection. We observed a significant association between adiponectin levels and survival among never smokers (HR, 2.70; 95% CI, 1.30–5.63) and no association among ever smokers (HR, 1.16; 95% CI, 0.63–2.13; P_interaction = 0.02; Fig. 1). Because adiponectin levels are inversely correlated with BMI (7), we next examined the association separately by obesity status at the time of blood collection. We observed a significant association between adiponectin and survival in non-obese patients (HR, 2.57; 95% CI, 1.57–4.21) but not in obese patients (HR, 0.47; 95% CI, 0.10–2.21; P_interaction = 0.003; Fig. 1).

Because the association between adiponectin and patient survival was observed only in women, to explore a potential role of sex hormones in adipokine signaling pathway, we performed stratified analyses by HRT use among postmenopausal women. We observed no significant association between adiponectin levels and survival in never HRT users (HR, 1.07; 95% CI, 0.43–2.70), and a significant association in ever HRT users (HR, 1.79; 95% CI; 1.06–3.02; P_interaction = 0.09; Fig. 1), suggesting an interaction between sex hormones and adiponectin levels on patient survival.

We next examined the association between polymorphisms in genes coding for the adiponectin receptors ADIPOR1 and ADIPOR2. In women, two SNPs in ADIPOR1 remained associated with patient survival after multiple hypotheses correction (Fig. 2; Supplementary Table S2). We observed a positive association between rs1418445 and mortality (per minor allele HR, 1.40; 95% CI, 1.15–1.71; P_adjusted = 0.02). We also observed an inverse association between rs10753929 and mortality in women (HR, 0.67; 95% CI, 0.51–0.88; P_adjusted = 0.03). Although the association between these two SNPs and survival did not reach statistical significance in men, the associations were relatively similar for both rs1418445 (HR, 1.16; 95% CI, 0.78–1.74) and rs10753929 (HR, 0.71; 95% CI, 0.34–1.49), with no statistically significant interaction by sex (P = 0.75 and 0.23, respectively). Interestingly, these SNPs have previously been associated with increased (rs1418445) or decreased (rs10753929) expression of ADIPOR1 in the blood eQTL analysis (ref. 43; Supplementary Table S5).

We next examined polymorphisms in LEPR in relation to patient survival. We observed a significant association between rs11585329 and survival in male (Fig. 3; HR, 0.39; 95% CI, 0.23–0.66; P_adjusted = 0.02) but not in female patients (HR, 1.04; 95% CI, 0.78–1.39; P_adjusted = 0.99), with a statistically significant interaction by sex (P = 0.0002). In addition, three SNPs were associated with survival in the combined population: rs12025906 (HR, 1.54; 95% CI, 1.25–1.90; P_adjusted < 0.0001), rs3790431 (HR, 0.72; 95% CI, 0.59–0.88; P_adjusted = 0.02), and rs17127601 (HR, 0.70; 95% CI, 0.56–0.89; P_adjusted = 0.03), with similar estimates between men and women (Fig. 3; Supplementary Table S3; all P_interaction > 0.2). Out of four SNPs in LEPR associated with survival, rs11585329 and rs12025906 were positively, and rs3790431 was negatively associated with LEPR expression in blood cells (Supplementary Table S5).

In this large study of patients with pancreatic cancer, higher adiponectin levels in the years preceding diagnosis were associated with increased mortality among women with pancreatic cancer. Furthermore, we observed associations between several SNPs within ADIPOR1 and LEPR genes with survival of patients with pancreatic cancer.

Higher circulating levels of leptin and lower circulating levels of adiponectin have previously been associated with increased risk of pancreatic cancer in several epidemiological studies (12–16). However, prior studies have not evaluated whether long-term exposure to these adipokines in the prediagnostic period is associated with survival of patients with pancreatic cancer. Because leptin and adiponectin interact with several pancreatic cancer-associated pathways potentially modifying the tumor microenvironment (5), we hypothesized that altered levels of those hormones in the period leading to diagnosis would not only be associated with the risk of pancreatic cancer, but also with tumor features and patient survival.

In this study, we observed an inverse association between circulating adiponectin and patient survival in women, but not in men. We also observed a statistically significant association between survival of female patients and 2 polymorphisms affecting the expression of ADIPOR1 in blood cells. Although these SNPs were not associated with survival in men, the estimates were similar, and the lack of statistical significance may be due to a smaller sample size.

We and others have previously reported an inverse association between circulating adiponectin levels and pancreatic cancer incidence (13, 16). The positive association between adiponectin and mortality in the current study might, therefore, seem inconsistent. A similar pattern of associations was previously observed for colorectal cancer, where higher adiponectin levels were associated with lower risk and higher mortality of patients with colorectal cancer (32, 46). In the context of pancreatic cancer, there are several potential explanations for the observed association.

First, adiponectin levels are inversely correlated with weight (7), and higher adiponectin could be marking patients experiencing a more severe cancer-associated weight loss due to a more aggressive disease. However, patients diagnosed with pancreatic cancer within 1 year from blood collection were excluded from this analysis. Furthermore, the association between adiponectin levels and survival persisted when blood samples were analyzed from more than 10 years before diagnosis, when patients are unlikely to have occult weight loss due to pancreatic cancer. Individuals with higher adiponectin levels in prediagnostic period have a lower BMI and may therefore have lower skeletal muscle mass, which has been positively associated with patient survival (47–49). Another potential explanation for our observations is that higher adiponectin levels in the prediagnostic period alter the tumor microenvironment and contribute to development of more aggressive tumors. Although adiponectin is generally considered an antitumorigenic hormone with a negative effect on tumor cell proliferation, metastasis, and inflammation (50), studies have also shown a role of adiponectin in stimulating tumor angiogenesis (51) and blunting of antigen-specific T-cell responses (52). Alternatively, tumors developing in individuals with and despite high adiponectin levels might be different from those developing in low adiponectin individuals. For example, high adiponectin levels might be protective of more indolent tumors, but not of more aggressive tumors that ultimately lead to shorter patient survival. In our previously published study conducted in same patient population included in the current analysis, adiponectin levels >4.4 μg/mL were associated with decreased risk of pancreatic cancer (13). Interestingly, similar threshold for adiponectin levels was observed in this analysis, where patients with levels above ≥4.9 had shorter survival. It is therefore possible that high adiponectin levels indirectly lead to selection or enrichment of more aggressive tumors.

The association between adiponectin levels with pancreatic cancer survival was observed only in women. Epidemiological studies have previously reported sex-specific associations between adiponectin and several outcomes, including diabetes (53) and metabolic syndrome (54, 55). There are several potential explanations for these differences. Similar to previous studies (29), adiponectin levels in our study were significantly higher in women than in men, and higher adiponectin levels might be required for activating potential pathways associated with shorter survival. However, we used the cutoff value of adiponectin levels associated with survival in women (≥4.9 μg/mL) and observed no association with mortality in men. Alternatively, these sex-specific associations could be due to interaction between sex hormones and adipokine signaling. In our study, the association between adiponectin levels and survival was limited to women using HRT, supporting this hypothesis. Sex hormones, more specifically estrogen, interact with adiponectin signaling on several levels. Estrogen alters expression of both adiponectin (56) and adiponectin receptors (57), and modulates the effect of adiponectin on proliferation and apoptosis of breast cancer cells (58). Finally, the observed difference might be due to a smaller number of male compared with female patients.

In this study, we observed an association between 4 polymorphisms in LEPR with patient survival. Furthermore, the association between rs11585329 and patient survival was observed only in men. We have previously reported sex-specific associations between leptin and LEPR polymorphisms and pancreatic cancer risk (12). In this study, higher leptin levels were associated with increased risk of pancreatic cancer in men, but not in women. We also reported an association between rs10493380 in LEPR and increased risk of pancreatic cancer in women (12). The results of the current analysis further strengthen the evidence that leptin signaling might be associated with pancreatic cancer in a sex-specific manner. Furthermore, the current analysis also suggests that the adiponectin and leptin pathways are associated not only with risk of developing pancreatic cancer, but also might be driving or selecting for tumors with different patient outcomes.

Strengths of the current study include a large number of patients with pancreatic cancer with prospectively collected blood samples allowing the investigation of adiponectin signaling in the early stages of pancreatic tumor development. We were able to comprehensively evaluate several confounders associated with adiponectin levels and pancreatic cancer survival, most importantly BMI and diabetes (3, 37, 38, 59). Limitations of this study include adipokine measurements at a single time point before diagnosis. However, studies reported relatively stable leptin and adiponectin levels in healthy people. Repeated adiponectin and leptin measurements 1 year apart showed correlation coefficients of 0.85 and 0.74, respectively (60, 61). Our analysis includes a smaller number of male participants, and the observed lack of association in men could be due to limited statistical power. Next, we were not able to account for potential confounding by treatment type because this information was not available in our participants. However, it is unlikely that cancer treatments differ by circulating adiponectin levels measured in the years preceding diagnosis. The majority of participants in this study were White, and further studies are needed to validate adipokines in relation to survival of patients with pancreatic cancer of different races and ethnicities.

In conclusion, higher adiponectin levels in prediagnostic period and SNPs affecting expression of ADIPOR1 and LEPR were associated with patient survival in a sex-specific manner. Further epidemiological and experimental studies are necessary to determine mechanisms behind the observed associations and to further elucidate the complex and sex-specific relationship between adiponectin and leptin with pancreatic cancer.

J. Wactawski-Wende reports grants from NIH/NHLBI during the conduct of the study. P. Kraft reports grants from NIH during the conduct of the study. J.E. Buring reports grants from NIH during the conduct of the study. M.J. Stampfer reports grants from NIH during the conduct of the study. K. Ng reports grants and non-financial support from Pharmavite, and grants from Evergrande Group, Janssen, and Revolution Medicines, as well as personal fees from Seagen, GlaxoSmithKline, Pfizer, CytomX, and Bayer outside the submitted work. C.S. Fuchs reports other support from Genentech/Roche outside the submitted work. B.M. Wolpin reports grants from Celgene, Eli Lilly, Novartis, and Revolution Medicines, as well as personal fees from Mirati, GRAIL, Ipsen, and Third Rock Ventures outside the submitted work. No disclosures were reported by the other authors.

A. Babic: Conceptualization, formal analysis, funding acquisition, investigation, writing–original draft, writing–review and editing. Q.-L. Wang: Investigation, writing–review and editing. A.A. Lee: Investigation, writing–review and editing. C. Yuan: Investigation, methodology, writing–review and editing. N. Rifai: Data curation, methodology, writing–review and editing. J. Luo: Resources, writing–review and editing. F.K. Tabung: Resources, writing–review and editing. A.H. Shadyab: Resources, writing–review and editing. J. Wactawski-Wende: Resources, writing–review and editing. N. Saquib: Resources, writing–review and editing. J. Kim: Data curation, investigation, writing–review and editing. P. Kraft: Resources, data curation, writing–original draft. H.D. Sesso: Resources, writing–review and editing. J.E. Buring: Resources, writing–review and editing. E.L. Giovannucci: Resources, investigation, writing–review and editing. J.E. Manson: Resources, investigation, writing–review and editing. M.J. Stampfer: Resources, investigation, writing–review and editing. K. Ng: Writing–review and editing. C.S. Fuchs: Resources, funding acquisition, investigation, writing–review and editing. B.M. Wolpin: Conceptualization, resources, investigation, writing–review and editing.

The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention's National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Idaho, Indiana, Iowa, Kentucky, Louisiana, Massachusetts, Maine, Maryland, Michigan, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico, Rhode Island, Seattle SEER Registry, South Carolina, Tennessee, Texas, Utah, Virginia, West Virginia, and Wyoming. The authors assume full responsibility for analyses and interpretation of these data. HPFS is supported by NIH grant U01 167552. NHS is supported by NIH grants UM1 CA186107, P01 CA87969, and R01 CA49449. PHS is supported by NIH grants CA 97193, CA 34944, CA 40360, HL 26490, and HL 34595. The WHI program is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C. The WHS is supported by grants CA047988, HL043851, HL080467, HL099355, and UM1 CA182913 from the NIH. A. Babic is supported by the NIH K07 CA222159 and Bob Parsons Memorial Fellowship. B.M. Wolpin is supported by the Hale Family Center for Pancreatic Cancer Research, Lustgarten Foundation Dedicated Laboratory program, NIH grant U01 CA210171, NIH grant P50 CA127003, a Stand Up To Cancer-Lustgarten Foundation Pancreatic Cancer Interception Translational Cancer Research grant (grant number: SU2C-AACR-DT25–17), Pancreatic Cancer Action Network, Noble Effort Fund, Wexler Family Fund, Parsons Pancreatic Cancer Early Detection Fund, and Promises for Purple. The indicated Stand Up To Cancer grant is administered by the AACR, scientific partner of Stand Up To Cancer.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.