Incident Early- and Later-Onset Type 2 Diabetes and Risk of Early- and Later-Onset Cancer: Prospective Cohort Study

The global burden of diabetes, including cancers attributable to diabetes, is substantial (1–3). Early-onset (diagnosed before 40 years of age) type 2 diabetes generally manifests a more severe and aggressive disease phenotype characterized by exaggerated peripheral and hepatic insulin resistance (especially in those with overweight and obesity) (4–7), accelerated deterioration of β-cell function (4–8), aggressive disease course (4–8), and potentially higher risk for complications (4–9) and is increasing dramatically in prevalence and incidence over the past three decades (5,6,8,10). Also, there is a worrisome rising incidence of early-onset cancers across the globe, which is driven predominantly by obesity- and diabetes-related cancers (11–13).

Mounting evidence has established a close link, epidemiologically and biologically, between diabetes and increased cancer incidence (13–17). The American Diabetes Association (ADA) and the American Cancer Society consensus report, based on comprehensive review of the existing evidence, underscored that “patients with diabetes should be strongly encouraged by their health care professionals to undergo appropriate cancer screenings as recommended for all people of their age and sex” (14). To our knowledge, however, no previous study has evaluated the relationship between type 2 diabetes and risk of cancer according to their onset age and disentangled the impacts of type 2 diabetes onset age and duration. Important questions remaining unanswered entering the 2020s include: does type 2 diabetes onset age have an impact on cancer risk beyond type 2 diabetes duration? Is the effect of early-onset type 2 diabetes on cancer risk inherently stronger than that of later-onset type 2 diabetes?

We therefore investigated prospectively the associations between incident early- and later-onset type 2 diabetes and early- and later-onset cancer risk in two large longitudinal cohorts: the Nurses’ Health Study (NHS) and NHS II.

The NHS began in 1976, when 121,700 female registered nurses aged 30 to 55 years were enrolled (18). The NHS II was initiated in 1989, enrolling 116,429 female nurses aged between 25 and 42 years (18). Demographics were collected at cohort enrollment. Throughout follow-up in both cohorts, anthropometrics, lifestyle characteristics, diet, menstrual and reproductive history, family history, medical history, and disease diagnoses and outcomes were assessed biennially or quadrennially via self-administered biennial questionnaires, achieving a follow-up rate >90%. The Institutional Review Boards of the Brigham and Women’s Hospital (Boston, MA) and those of participating registries (as required) approved the study protocols. Written informed consent was obtained from participants to retrieve medical records. We used 1978 in the NHS and 1991 in NHS II as the baseline of this analysis, as this was the first follow-up cycle when incident type 2 diabetes cases were identified. Participants were excluded if they had a prior diagnosis of cancer (except nonmelanoma skin cancer) or prevalent diabetes before baseline.

Physician-diagnosed incident type 2 diabetes was first identified through the self-administered biennial questionnaires, followed by a supplementary questionnaire to further confirm their diagnosis by collecting information on symptoms, diagnostic tests, and hypoglycemic therapy (19). Cases identified before 1998 were confirmed if they met at least one of the National Diabetes Data Group criteria (20): 1) elevated glucose concentration (fasting plasma glucose ≥7.8 mmol/L, random plasma glucose ≥11.1 mmol/L, or 2-h plasma glucose ≥11.1 mmol/L after oral glucose load) and at least one related symptom (excessive thirst, polyuria, weight loss, or hunger); 2) no symptoms, but elevated glucose concentrations on at least two occasions; or 3) treatment with insulin or other hypoglycemic medications. Cases identified after 1998 were confirmed based on the ADA criteria (21), with the cutoff point for elevated fasting plasma glucose lowered to 7.0 mmol/L. Confirmation of cases identified after 2010 further considered HbA_1c ≥6.5% according to the updates in ADA criteria (22). A validation study in the NHS demonstrated that 98.4% of self-reported type 2 diabetes cases confirmed using the supplementary questionnaire were reconfirmed by medical record review (23). In primary analyses, we considered incident type 2 diabetes diagnosed at <40 years of age as early-onset type 2 diabetes (5,8); in sensitivity analyses, we considered <45 years as the cutoff age.

The following variables were selected a priori as covariates. In primary analyses, we considered age, cohort, follow-up cycle, race, cumulative average BMI, waist circumference, pack-years of smoking, physical activity, alcohol intake, Alternate Healthy Eating Index (AHEI) (24), total energy intake, multivitamin use, family history of diabetes, family history of cancer, history of lower endoscopy within the previous 10 years, fasting glucose screening within the previous 2 years, mammography screening within the previous 2 years, menopausal status, postmenopausal hormone use, and oral contraceptive use. In sensitivity analyses, we replaced cumulative average BMI with BMI at 18 years of age plus weight change or substituted BMI with 4-year lag for cumulative average BMI. We additionally considered antidiabetic medication use and BMI at 18 years old as potential effect modifiers.

In both cohorts, height, race, and body weight at 18 years of age (which we used to calculate BMI at 18 years old with height) were captured once at enrollment. Information on body weight (which we used to calculate cumulative average BMI with height), smoking behavior, multivitamin use, endoscopy screening (colonoscopy and/or sigmoidoscopy, starting from 1988 in the NHS and 1991 in NHS II), fasting glucose screening (starting from 1998 in the NHS and 1999 in NHS II), mammography screening (starting from 1988 in the NHS and 1989 in NHS II), menopausal status, postmenopausal hormone use, oral contraceptive use, and insulin and oral hypoglycemic medication use (starting from 1988 in the NHS and 1989 in NHS II) were biennially updated. Diet (measured via validated semiquantitative food frequency questionnaires), physical activity (frequencies of engaging in common recreational activities were captured, according to which MET scores were assigned and total physical activity in MET-hours per week was calculated), and family history were assessed quadrennially. Waist circumference was assessed in 1986, 1996, and 2000 in the NHS and in 1993 and 2005 in NHS II. Previous validation studies showed high accuracy and reliability of reported information on anthropometrics, lifestyles, diet, menopausal status, and disease outcomes in our cohorts of health care professionals (25–30).

Ascertainment of C-peptide and HbA_1c was elaborated in the Supplementary Methods. Briefly, blood samples were collected from subpopulations of the NHS (between 1989 and 1990) and NHS II (between 1996 and 1999). Plasma C-peptide and HbA_1c were measured based on different nested case-control studies for various outcomes. To minimize the potential influence of prediabetes on C-peptide and HbA_1c concentrations at blood draw, we excluded participants without diabetes who reported diabetes diagnosis within 4 years after blood sample collection.

Physician-diagnosed incident cancer cases were first reported on the biennial questionnaires or identified during follow-up of death. Cancer diagnoses were then confirmed by medical record and pathology report review (with permission) or linkage to state cancer registries (when medical records were unavailable) (18). We specifically considered incident early-onset (primary analyses: diagnosed before 50 years of age [11]; sensitivity analyses: diagnosed before 55 years of age) total cancer (except nonmelanoma skin cancers), diabetes-related cancer as defined by prior meta-analysis and umbrella review (13,31) (in aggregate), and obesity-related cancer as defined by the International Agency for Research on Cancer Working Group (32) (in aggregate), and major individual cancers in this study, the details of which are shown in the table and figure legends.

Participant deaths were ascertained via routine searches of the National Death Index and reports from next of kin or postal authorities, with completeness >98% (33). With permission from next of kin, cohort investigators further reviewed death certificates and medical records.

Person-years of follow-up accrued from the return date of the baseline questionnaire (NHS: 1978; NHS II: 1991) until the date of any cancer diagnosis reported (except nonmelanoma skin cancer), type 1 diabetes, death, date of return of last available follow-up questionnaire, or predefined follow-up completion (NHS: 2016; NHS II: 2017), whichever occurred earliest.

Pooled analyses in the combined data set were conducted because statistically significant heterogeneity by cohort was not detected (all P values for heterogeneity ≥0.11). As we found no evidence for violation of the proportional hazard assumption (tested using the likelihood ratio test to compare models with and without product terms between exposures and log-transformed follow-up time), age- and multivariable-adjusted hazard ratios (HRs) and 95% CI for early- and later-onset cancer incidence across diagnosis of type 2 diabetes by age were estimated using Cox proportional hazard models.

Multivariable analyses were stratified by cohort (NHS and NHS II), age (in months), and questionnaire cycle (each 2-year interval, to account for potential period effect) and adjusted for race (White, non-White), cumulative average BMI (continuous, in kilograms per meter squared), waist circumference (continuous, in centimeters), pack-years of smoking (continuous), physical activity (continuous, in MET-hours/week), alcohol intake (continuous, in grams per day), AHEI (without alcohol intake, continuous), total energy intake (continuous, in kilocalories per day), multivitamin use (yes or no), family history of diabetes (yes or no), family history of cancer (yes or no), history of lower endoscopy within the previous 10 years (yes or no), fasting glucose screening within the past 2 years (yes or no), mammography screening within the past 2 years (yes or no), menopausal status and postmenopausal hormone use (premenopausal, postmenopausal and never used hormone, or postmenopausal and past/current hormone use), and oral contraceptive use (yes or no). In sensitivity analyses, we replaced cumulative average BMI with BMI at 18 years of age (continuous, in kilograms per meter squared) plus weight change (continuous, in kilograms). We also considered substituting BMI with 4-year lag for cumulative average BMI.

We repeated these analyses to further evaluate whether the associations differed by status of insulin/oral hypoglycemic drug use among participants with type 2 diabetes. We performed stratified analyses by BMI at 18 years old (<21 vs. ≥21 kg/m²; the median value of BMI at 18 years of age in our study population) (34) to evaluate whether these factors modified the associations. Effect modification was assessed by adding interaction terms to the models and using likelihood ratio tests to determine statistical significance. Finally, to make the comparisons more equitable across type 2 diabetes and cancer onset age and to better disentangle the inherent difference in risk over time, we conducted further analyses based on refined timing by exploring the associations between type 2 diabetes diagnosed in 10-year age groups and subsequent cancer risk in the next 10-year periods.

In secondary analyses, we explored the associations between incident early-onset type 2 diabetes and risk of early- and later-onset cancers according to major individual cancer types (using less strict cutoff age). We also conducted cross-sectional analysis of plasma C-peptide and HbA_1c level according to type 2 diabetes onset age.

We performed data analyses using SAS statistical software (version 9.4 for UNIX; SAS Institute Inc., Cary, NC). All tests were two-sided, with P values <0.05 indicating statistical significance.

Data described in this study may be made available upon application to and with approval by the Channing Division of Network Medicine at Brigham and Women’s Hospital, Harvard Medical School. Further information, including the procedures to obtain and access data from the NHS and NHS II, is available at https://www.nurseshealthstudy.org/researchers (e-mail: [email protected]).

During up to 38 years of follow-up (6,747,869 person-years) among 228,073 eligible participants, we documented 18,290 cases of incident type 2 diabetes (early-onset: 661; later-onset: 17,629), including 440 insulin/oral hypoglycemic drug users among those with early-onset type 2 diabetes (insulin only: 36; oral hypoglycemic drug only: 253; both: 151) and 11,009 insulin/oral hypoglycemic drug users among those with later-onset type 2 diabetes (insulin only: 389; oral hypoglycemic drug only: 9,523; both: 1,097). We excluded 357 cases of type 1 diabetes. A total of 43,427 incident cancer cases (early-onset: 6,520; later-onset: 36,907) were documented, including 26,809 cases of diabetes-related cancer (early-onset: 1,959, later-onset: 24,850) (13,31) and 24,719 cases of obesity-related cancer (early-onset: 1,739, later-onset: 22,980) (32). Numbers of major individual cancers according to age at diagnosis and incident type 2 diabetes onset age are detailed in Supplementary Table 1. In the pooled study population, compared with participants without type 2 diabetes, those with incident type 2 diabetes diagnosed at <40 years of age tended to have a higher BMI and waist circumference, were engaged in less physical activity, consumed less alcohol, had a higher intake of total energy, and were more likely to have family history of diabetes. Largely similar differences in population characteristics were observed comparing participants diagnosed with type 2 diabetes at ≥40 years of age versus those without type 2 diabetes (Table 1).

In fully adjusted analyses (controlled for cumulative average BMI, waist circumference, and a wide spectrum of other covariates listed in the table footnote), incident early-onset type 2 diabetes (diagnosed at <40 years of age) was associated with increased risk of early-onset (diagnosed at <50 years of age) total cancer (HR [95% CI] 1.47 [1.06–2.04]), diabetes-related cancer (2.11 [1.38–3.23]), and obesity-related cancer (1.75 [1.08–2.82]) (Table 2).

In subgroup analyses, the risk elevations were observed among both participants with early-onset type 2 diabetes who reported diabetes medications (total cancer: 1.53 [1.02–2.30]; diabetes-related cancer: 2.01 [1.17–3.45]; obesity-related cancer: 1.81 [1.01–3.23]) and those without antidiabetic medication use (total cancer: 1.38 [0.80–2.38]; diabetes-related cancer: 2.30 [1.19–4.47]; obesity-related cancer: 1.67 [0.74–3.75]) (Supplementary Table 2).

In analyses stratified by BMI at 18 years of age, positive associations for total cancer (1.75 [1.20–2.56]; P for interaction: 0.03), diabetes-related cancer (2.43 [1.50–3.94]; P for interaction: 0.13), and obesity-related cancer (1.84 [1.05–3.22]; P for interaction: 0.42) were observed only in individuals with a BMI at 18 years of age ≥21 kg/m² (Fig. 1).

In fully adjusted analyses, incident later-onset type 2 diabetes was associated with higher risk of later-onset cancer (total cancer: 1.15 [1.11–1.20]; diabetes-related cancer: 1.17 [1.12–1.22]; obesity-related cancer: 1.18 [1.13–1.24]). The risk elevations were detected consistently in subgroup analyses according to antidiabetic medication use and were detected among individuals with lower BMI at 18 years of age (Fig. 1, Table 2, and Supplementary Table 2).

We observed no association of incident early-onset type 2 diabetes with later-onset cancer risk in primary analyses and across diabetes medication use subgroups; however, a higher risk of later-onset diabetes-related cancer (1.82 [1.08–3.06]; P for interaction: 0.03) and obesity-related cancer (1.68 [1.00–2.83]; P for interaction: 0.08) was detected specifically in those with higher BMI at 18 years old (Fig. 1, Table 2, and Supplementary Table 2).

In fully adjusted analyses based on refined timing, incident type 2 diabetes diagnosed at <40 years of age was associated with higher risk of cancer diagnosed between ≥40 and <50 years of age (total cancer: 1.66 [1.19–2.32]; diabetes-related cancer: 2.47 [1.61–3.78]; obesity-related cancer: 2.05 [1.26–3.31]). Keeping the 10-year age gap between the independent exposure and follow-up periods unchanged, the HRs attenuated substantially with aging (Table 3).

In sensitivity analyses replacing cumulative average BMI with BMI at 18 years of age plus weight change or substituting BMI with 4-year lag for cumulative average BMI in confounding control, the results remained largely similar (Supplementary Tables 3 and 4).

In sensitivity analyses using a less strict cutoff age, incident type 2 diabetes diagnosed at <45 years of age was associated with higher risk of cancer diagnosed at <55 years of age (total cancer: 1.20 [1.01 to 1.43]; diabetes-related cancer: 1.63 [1.31–2.04]; obesity-related cancer: 1.56 [1.24 to 1.97]), whereas no association was observed for cancer diagnosed at ≥55 years of age. In contrast, incident type 2 diabetes diagnosed at ≥45 years of age was associated with increased risk of cancer diagnosed at ≥55 years of age (total cancer: 1.13 [1.09–1.18]; diabetes-related cancer: 1.16 [1.10–1.21]; obesity-related cancer: 1.17 [1.11–1.23]) (Supplementary Table 5). Subgroup analyses and analyses according to refined timing based on this less strict cutoff age are presented in Supplementary Tables 6 and 7 and Supplementary Fig. 1.

In analyses of the association between incident early-onset type 2 diabetes and risk of early-onset cancer according to major individual cancer types (using less strict cutoff age), the risk elevation was predominantly driven by endometrial cancer (1.52 [0.99–2.31]) and kidney cancer (2.75 [1.33–5.67]) and was suggestively driven also by cancers of colorectum, pancreas, thyroid, and bladder (Supplementary Table 8).

In cross-sectional analyses of plasma C-peptide and HbA_1c level, participants with type 2 diabetes onset age <40 years and ≥40 to <50 years had lower C-peptide levels, whereas those aged ≥50 years at type 2 diabetes onset had higher C-peptide levels. Among individuals aged <40 years and ≥40 to <50 years, those without and with type 2 diabetes had generally similar C-peptide levels. HbA_1c level was consistently high across type 2 diabetes onset age (Supplementary Table 9). In further analyses of fasting and nonfasting plasma C-peptide level, nonfasting C-peptide level was not higher than fasting C-peptide level among individuals with type 2 diabetes onset age <40 years and ≥40 to <50 years. In contrast, nonfasting C-peptide level was substantially higher than fasting C-peptide level among those aged ≥50 years at type 2 diabetes onset (Supplementary Table 10).

Although early- and later-onset type 2 diabetes share similar pathogenetic pathways (e.g., β-cell dysfunction, insulin resistance, and obesity-driven mechanisms) (5,6), key traits of early-onset type 2 diabetes, including faster β-cell failure, exaggerated insulin insensitivity, and higher prevalence of obesity, suggest the potential of distinctions in their pathophysiology (5,6). The associations between early-onset type 2 diabetes and subsequent burden of major macrovascular and microvascular complications have been intensively investigated and well-characterized (5,7,9); however, there are scarce data on early-onset cancer.

In this large prospective cohort study, we reported significantly increased risk of early-onset (diagnosed at <50 years of age) total cancer (1.47-fold), diabetes-related cancer (2.11-fold), and obesity-related cancer (1.75-fold) associated with early-onset (diagnosed at <40 years of age) type 2 diabetes after rigorous control for cumulative average BMI (or BMI at 18 years of age plus weight change), waist circumference, and a wide spectrum of other potential confounders. Furthermore, the risk elevations were consistently observed in participants with and without antidiabetic medication use. The magnitude of our observed association is substantially higher than the previously reported risk of total cancer, diabetes-related cancer, and obesity-related cancer associated with type 2 diabetes overall (17,19). Our results highlight the urgent need for optimal preventive initiatives and screening strategies to mitigate the overlapping burden of early-onset type 2 diabetes and early-onset cancer. Special attention needs to be paid to early-onset diabetes- and obesity-related cancers in this context, especially for cancers of endometrium, kidney, colorectum, pancreas, thyroid, and bladder.

Plausible biological mechanisms driving diabetes-induced neoplastic processes include the role of hyperinsulinemia, hyperglycemia, chronic inflammation, and shared risk factors (especially obesity-related) between diabetes and cancer (14,15,35). Exact mechanisms driving the link between early-onset type 2 diabetes and early-onset cancer remain to be elucidated. The ADA and American Cancer Society Consensus Statement underscored “whether the association between diabetes and cancer is direct (due to hyperglycemia), whether diabetes is a marker of underlying biologic factors that alter cancer risk (due to insulin resistance and hyperinsulinemia), or whether the association between cancer and diabetes is indirect and due to common risk factors, such as obesity” as a key unanswered question (14).

In a prior large prospective cohort study comprehensively examining incident type 2 diabetes duration and cancer risk in the NHS and Health Professionals Follow-up Study, the risk elevation peaked at ˜8 years after type 2 diabetes onset and attenuated afterward (19), suggesting a greater role of insulin resistance and hyperinsulinemia than hyperglycemia. This evidence, in conjunction with our observation of lower C-peptide levels specifically in patients with early-onset type 2 diabetes (indicating rapid β-cell function decline [8,36]) (Supplementary Tables 9 and 10), suggested that the low insulin level due to rapid depletion of β-cells may explain the lack of strong association between early-onset type 2 diabetes and later-onset cancer risk in our primary analyses and across diabetes medication use subgroups (Table 2 and Supplementary Tables 2 and 4). Another recent Mendelian randomization analysis that reported a causal effect of higher fasting insulin, rather than glucose traits, on colorectal tumorigenesis also supported this hypothesis (37).

Of note, in our subgroup analyses according to BMI at 18 years of age, the observed excess risk of early-onset cancer associated with early-onset type 2 diabetes diagnosed at <40 years of age was restricted to individuals with larger BMI at 18 years old rather than their counterparts. Interestingly, in our further subgroup analyses investigating whether early-onset type 2 diabetes confers influence on risk of later-onset cancer according to BMI at 18 years of age, a higher risk of later-onset diabetes-related cancer (1.82-fold) and obesity-related (1.68-fold) cancer was detected only in those with higher BMI at 18 years old. These findings support that cancer-prevention efforts tailored for individuals with early-onset type 2 diabetes may need to focus on those with higher BMI in adolescence and emerging adulthood. Of note, the prevalence of BMI ≥21 kg/m² at 18 years old before the 1980s in our cohort participants (Supplementary Fig. 2) is substantially lower than that in the recent National Health and Nutrition Examination Survey cycles, further supporting the previously reported dramatically increasing trends in U.S. adolescent (38) and emerging adulthood (39) overweight/obesity prevalence and the importance of adiposity as a strong risk factor of early-onset type 2 diabetes (4–6,8).

Our analyses based on refined timing provided strong evidence supporting that the impact of early-onset type 2 diabetes on cancer risk may be inherently stronger than that of later-onset type 2 diabetes. Although the relative risk attenuates with aging, because both type 2 diabetes and cancer prevalence and incidence increase sharply with age, the absolute number of excess cases with increasing aging is substantial (14,40).

Collectively, our findings support that insulin and obesity-related mechanisms may synergistically contribute to the pathophysiologic processes linking early-onset type 2 diabetes with cancer risk. In our study, the risk elevations associated with early-onset type 2 diabetes seemed to be statistically independent of adiposity (based on the effect estimates before and after maximal adjustment for BMI and waist circumference), whereas there may be dependence mechanistically.

We reported slightly higher risk (1.15- to 1.18-fold) of later-onset (diagnosed at ≥50 years of age) total cancer, diabetes-related cancer, and obesity-related cancer associated with later-onset (diagnosed at ≥40 years of age) type 2 diabetes, whereas the risk elevations were detected only among those with lower BMI at 18 years of age. These findings further support the potential differences in pathophysiologic processes linking early- and later-onset diabetes with cancer risk.

Major strengths of our study include the prospective study design, large sample size, long-term follow-up, excellent follow-up rates attained, validated longitudinal assessments of incident type 2 diabetes and cancer diagnoses, and validated time-varying information on a wide spectrum of covariates that permitted rigorous control for confounding. The nature of our cohort participants (all health care professionals) further ensured internal validity and data quality.

Our study has several limitations. The possibility for residual confounding cannot be completely ruled out due to the observational study design. However, randomized trials cannot be used to address this research question, because randomization to time of diabetes onset is not feasible. Given the professional and racial/ethnic homogeneities of our study participants as well as the fact that our stratified analyses by BMI at 18 years of age and our analyses of major individual cancers have relatively limited power, future large investigations with high-quality databases and consortium efforts in more diverse populations are needed to validate current findings.

Incident early-onset type 2 diabetes was associated with increased risk of early-onset total cancer, diabetes-related cancer, and obesity-related cancer, especially in those with higher BMI at 18 years of age. Cancer-prevention efforts tailored for patients with early-onset type 2 diabetes may need to focus on those with higher adolescent or emerging adulthood BMI. The impact of early-onset type 2 diabetes on cancer risk may be inherently stronger than that of later-onset type 2 diabetes.

Acknowledgments. The authors thank all participants and staff of the NHS and NHS II for the contributions to this research. The authors also 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 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.

Funding. The NHS was supported by National Institutes of Health (NIH) grants UM1 CA186107 and P01 CA87969 to M.J.S. and A.H.E. The NHS II was supported by NIH grant U01 CA176726 to W.C.W. and A.H.E. This work was additionally supported by the American Cancer Society Mentored Research Scholar Grant MRSG-17-220-01-NEC and NIH grant R00 CA215314 to M.S.; NIH grant R37 CA246175 to Y.C.; Lustgarten Foundation dedicated laboratory program, Dana-Farber Cancer Institute Hale Family Center for Pancreatic Cancer Research, NIH grants U01 CA210171 and P50 CA127003, Pancreatic Cancer Action Network, Stand Up to Cancer, Noble Effort Fund, Wexler Family Fund, Promises for Purple, and Bob Parsons Fund to B.M.W.; and NIH grant R01 CA205406 and Project P Fund to K.N. Y.Z. was supported by the Irene M. & Fredrick J. Stare Nutrition Education Fund Doctoral Scholarship and Mayer Fund Doctoral Scholarship.

The funding sources played no role in the study design, data collection, data analysis, and interpretation of results or the decisions made in preparation and submission of the article.

Duality of Interest. B.M.W. declares research funding from Celgene and Eli Lilly and Company and consulting for BioLineRx, Celgene, G1 Therapeutics, and GRAIL, outside the submitted work. K.N. declares institutional research funding from Revolution Medicines, Pharmavite, Evergrande Group, and Janssen, outside the submitted work, and consulting/advisory board fees from Seagen, BiomX, Bicara Therapeutics, X-Biotix Therapeutics, and Redesign Health, outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.Z. and E.L.G. were involved in the conception, design, and conduct of the study and the analysis and interpretation of the results. Y.Z. wrote the first draft of the manuscript. Y.Z., M.S., Y.C., A.H.E., B.M.W., M.J.S., W.C.W., and K.N. obtained funding. M.S., A.H.E., M.J.S., W.C.W., and F.B.H. provided administrative, technical, and material support. All authors edited, reviewed, and approved the final version of the manuscript. Y.Z. and E.L.G. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in abstract form at the 9th Annual Dana-Farber/Harvard Cancer Center Celebration of Early Career Investigators in Cancer Research Symposium, 19 January 2022.