Background:

Pancreatic cancer risk is increasing in countries with high consumption of Western dietary patterns and rising obesity rates. We examined the hypothesis that specific dietary patterns reflecting hyperinsulinemia (empirical dietary index for hyperinsulinemia; EDIH), systemic inflammation (empirical dietary inflammatory pattern; EDIP), and postprandial glycemia [glycemic index (GI); glycemic load (GL)] are associated with pancreatic cancer risk, including the potential modifying role of type 2 diabetes (T2D) and body mass index (BMI).

Methods:

We calculated dietary scores from baseline (1993–1998) food frequency questionnaires among 129,241 women, 50–79 years-old in the Women's Health Initiative. We used multivariable-adjusted Cox regression to estimate HRs and 95% confidence intervals (95% CI) for pancreatic cancer risk.

Results:

During a median 19.9 years of follow-up, 850 pancreatic cancer cases were diagnosed. We observed no association between dietary scores and pancreatic cancer risk overall. However, risk was elevated among participants with longstanding T2D (present >3 years before pancreatic cancer diagnosis) for EDIH. For each 1 SD increment in dietary score, the HRs (95% CIs) were: EDIH, 1.33 (1.06–1.66); EDIP, 1.26 (0.98–1.63); GI, 1.26 (0.96–1.67); and GL, 1.23 (0.96–1.57); although interactions were not significant (all Pinteraction >0.05). Separately, we observed inverse associations between GI [0.86 (0.76–0.96), Pinteraction = 0.0068] and GL [0.83 (0.73–0.93), Pinteraction = 0.0075], with pancreatic cancer risk among normal-weight women.

Conclusions:

We observed no overall association between the dietary patterns evaluated and pancreatic cancer risk, although women with T2D appeared to have greater cancer risk.

Impact:

The elevated risk for hyperinsulinemic diets among women with longstanding T2D and the inverse association among normal-weight women warrant further examination.

Pancreatic cancer is the third leading cause of cancer-related deaths in the United States (1). Because of the nonspecific nature and late onset of symptoms, early detection is challenging, and most patients are diagnosed at an advanced cancer stage. Combined with biological factors promoting treatment resistance, pancreatic cancer has a poor prognosis, with a five-year survival rate of only 9% (1). Therefore, it is crucial to identify modifiable risk factors for prevention.

Diet is a modifiable factor that may influence pancreatic cancer risk (2). In contrast to the reductionist strategies of single nutrients or single foods, the dietary pattern approach accounts for the complex interactions between dietary variables and allows assessment of the cumulative effects of multiple dietary components on disease risk. Such efforts regarding pancreatic cancer risk are few (3), and have been conducted primarily as case–control studies, with inherent concerns of recall bias. Nevertheless, current literature suggests a greater risk with dietary patterns described as the Western dietary pattern rich in animal products while inverse associations have been noted for dietary patterns defined as “prudent” and rich in fruits, vegetables and fiber (3). Potential reverse causation by occult disease, which cannot be addressed in case–control studies, is a major limitation and it is imperative that additional studies of dietary patterns focus on large, prospective designs. It is also important to consider multiple strategies for defining dietary patterns. For example, one approach uses dietary guidelines or hypotheses (based on prevailing evidence) regarding a diet-disease relation to define a pattern, a priori, such as the healthy eating index. Another strategy is purely empirical (data-driven) and employs statistical approaches to group dietary variables into patterns, a posteriori, based on the explained variation in the diet. Our team utilized a hybrid approach to define empirical hypothesis-oriented dietary patterns that are data-driven yet based on a specific hypothesis (e.g., hyperinsulinemia, chronic systemic inflammation, etc.) relating diet with disease (4, 5). We hypothesize that dietary patterns associated with hyperinsulinemia or a chronic systemic inflammatory state may increase risk of pancreatic cancer.

Dietary patterns have been associated with risk of obesity (6) and T2D (7, 8), which interfaces with investigations of the association of dietary patterns with pancreatic cancer risk, yet the temporal relationships have not been clearly described. Further refinement in our understanding of the role of obesity and T2D in pancreatic cancer risk offers opportunities to define prevention strategies. The dietary glycemic index (GI) and dietary glycemic load (GL) are two dietary indices that are widely used for assessing the postprandial glycemic potential of the diet; however, these indices do not account for the intake of fat, protein, and the diverse array of phytochemicals that influence insulin secretion and glucose regulation (5). Our group previously developed the empirical dietary index for hyperinsulinemia (EDIH) score based on circulating C-peptide levels, for assessing the insulinemic potential of the dietary pattern (5), and the empirical dietary inflammatory pattern (EDIP) score, based on circulating inflammatory biomarkers, for evaluating the inflammatory potential of the dietary pattern (4). In this study, we calculated the EDIH, EDIP, GI, and GL scores to estimate the insulinemic, inflammatory, and glycemic potentials, respectively, of the diet and examined associations with risk of developing pancreatic cancer in the Women's Health Initiative (WHI). In addition, we investigated potential effect modification of these associations by T2D and BMI.

Study population

Between 1993 and 1998, a total of 161,808 postmenopausal women aged 50–79 years were enrolled in the WHI (9) at 40 clinical centers across the U.S. Women were enrolled into either an observational study (n = 93,676) or one or more of 4 overlapping clinical trials (n = 68,132). The institutional review boards at the Clinical Coordinating Center at the Fred Hutchinson Cancer Research Center (Seattle, WA) and at each Clinical Center approved the WHI protocol (10). The original WHI study completed data collection in 2005 but extension and ancillary studies have continued to collect long-term data. The current extension study is collecting annual health information from consenting WHI participants through 2020. Supplementary Table S1 contains a list of WHI investigators.

We sequentially excluded women with: implausible energy intake (<600 kcal/day and >5,000 kcal/day; n = 4,686) as these individuals may have filled out questionnaires incorrectly (11); extreme BMI (<15 or >50 kg/m2; n = 6,476); prevalent cancer (except non-melanoma skin cancer) at baseline (n = 11,840); baseline T2D (n = 7,768), as dietary modifications usually occur after disease diagnosis; baseline pancreatitis (n = 496); and those with missing information on pancreatic cancer status or those with a pancreatic cancer diagnosis and missing date of diagnosis (n = 1,154; Supplementary Fig. S1). Early symptoms of undiagnosed pancreatic cancer may alter one's dietary pattern and body weight; hence we applied a 4-year lag (12) between dietary assessment and pancreatic cancer ascertainment, and excluded those who were diagnosed with pancreatic cancer within 4 years from baseline (n = 147). Our final analytic sample included 129,241 women who had comparable baseline characteristics with the excluded participants for most variables (Supplementary Table S2), as well as with the entire WHI cohort (Supplementary Table S3).

Dietary assessment and calculation of dietary indices

Dietary scores were calculated using baseline habitual dietary data, assessed using the WHI food frequency questionnaire (FFQ), a 122-item semi-quantitative self-administered FFQ covering the dietary intake in the preceding three months (13). Nutrient intake from the FFQ was estimated using the University of Minnesota Nutrition Coordinating Center food and nutrient database (Nutrient data System for Research – NDSR; ref. 14). The measurement characteristics of the WHI FFQ were evaluated by comparing the FFQ nutrient intake estimates with those from four 24-hour dietary recalls and 4-day food records (13). The mean intake of most nutrients estimated from the FFQ was found to be comparable to corresponding intakes estimated from dietary recalls and records (13).

The development and validation of the EDIP and EDIH scores have been described (4, 5). Briefly, the EDIP is a weighted sum of 18 food groups most predictive of three circulating inflammatory biomarkers (IL6, CRP, TNFαR-2) measured from plasma, with more positive scores indicating more proinflammatory dietary patterns (4). EDIH is comprised of 18 food groups, selected from 39 food groups most predictive of plasma C-peptide concentrations, a marker of beta-cell secretory activity. More positive scores indicate hyperinsulinemic dietary patterns (5). The component foods of both scores are presented in Supplementary Table S4. A GI score estimates the quality of carbohydrates in the diet, and represents the percent incremental area under the 2-hour postprandial glucose response curve for consumption of a given carbohydrate-containing food relative to the corresponding area for consumption of a reference food (glucose or white bread) with equal amount of carbohydrates (15). The GL of each food is the product of the food's GI and the amount of carbohydrate in that food, summed across all foods for each individual (16).

Ascertainment of pancreatic cancer

The primary outcome, incident pancreatic cancer, was identified through medical record adjudication by study physicians following self-report of a diagnosis at semi-annual contact in the Clinical Trials (CT) and/or annual contact in the Observational Study (OS) and extension studies. A total of 850 pancreatic cancer cases were ascertained between 4 years from baseline and end of study on March 1, 2019. (17).

Assessment of covariates

Age, race/ethnicity, education, pack-years of cigarette smoking, family history of diabetes, gallbladder removal, and nonsteroidal anti-inflammatory drug (NSAID) use were assessed at baseline via self-administered questionnaires. Hormone use was the sum (yes = 1/no = 0 for each hormone) of 8 WHI hormone usage variables at baseline. Dietary supplement use was defined as the number of supplements taken and was the sum (yes = 1/no = 0 for each supplement) of 23 vitamin and/or mineral supplements (18). Physical activity was defined as total energy expended from recreational physical activity (MET-hours/week) and was assessed semi-annually (CT) or annually (OS; ref. 19). The Hormone Therapy study arm and Dietary Modification study arm to which the participants were randomized were also included as covariates. We calculated a comorbidity score by summing the presence (yes = 1/no = 0) of hypercholesterolemia, high blood pressure, heart disease, stroke, and rheumatoid/other arthritis at baseline. Details regarding covariates are presented in Supplementary Table S5.

The T2D status and duration variable (No T2D, recent onset, and longstanding T2D) was defined as follows. First, we ascertained a T2D status variable: at each contact, incident T2D was ascertained if participants self-reported that they had received T2D treatment (i.e., oral medications, insulin, and/or diabetes diet/exercise) and/or had been hospitalized for diabetes (20). This was validated using diabetes medication inventories (19). Participants were followed from enrollment until T2D diagnosis, death, loss to follow-up, or the end of the study on March 1, 2019 to define the time-to-T2D diagnosis. Next, a case of longstanding T2D was defined as diabetes diagnosed more than 3 years before pancreatic cancer diagnosis, whereas recent onset T2D as a diabetes diagnosis less than or equal to 3 years from a pancreatic cancer diagnosis. Body mass index [BMI = weight (kg)/height (m)2] was categorized as normal weight, 18.5–<25; overweight, 25–<30; and obese, 30–50.

Statistical analysis

We described participants' baseline characteristics using means ± SDs for continuous variables and frequencies for categorical variables, and adjusted dietary scores for total energy intake using the residual method (21). We created the dietary quintiles with cut-off points based on the entire final analytic sample. We used Cox proportional hazards regression to estimate HRs and 95% confidence intervals (CI) for the risk of developing pancreatic cancer in higher dietary index quintiles using the lowest quintiles as reference categories. Participants were followed from enrollment to pancreatic cancer diagnosis, death, loss to follow-up, or end of study on March 1, 2019. We calculated P values for linear trend across dietary index quintiles by assigning the quintile medians of each quintile to all participants in the corresponding quintile as an ordinal variable in the multivariable-adjusted models. In addition to the categorical analysis, we modeled the dietary indices as continuous variables (1-SD increment). We tested the proportional hazards (PH) assumption using the Schoenfeld residuals method and by running time-dependent covariate models. The multivariable adjusted models were stratified by hormone use, education, and age (covariates that violated the PH assumption), and further adjusted for family history of T2D, physical activity, race/ethnicity, pack-years of cigarette smoking, hormone therapy trial arms, NSAID use, supplement use, dietary modification trial arms, gallbladder removal status (22), and comorbidity score (Supplementary Table S5; refs. 3, 23). The multivariable plus BMI-adjusted models were further stratified by BMI category. In subgroup analyses, we used the likelihood ratio test to test for potential effect modification (24) by diabetes status and duration categories and by BMI categories, by comparing the models with and without the interaction terms. For subgroup analyses, the dietary indices were categorized into quartiles.

We calculated multivariable-adjusted incidence rates of pancreatic cancer in quintiles of each dietary index. For incidence rate analyses, we used the residual method (21) to adjust the dietary indices for the same covariates that were adjusted in the corresponding Cox regression models. We further estimated the incidence rate in dietary index quintiles within each T2D and BMI categories. We conducted all statistical analyses using SAS 9.4 (SAS Institute) and two-sided P < 0.05 was considered statistically significant.

Ethical approval

The institutional review boards (IRB) at the Clinical Coordinating Center at the Fred Hutchinson Cancer Research Center (Seattle, WA) and at each Clinical Center approved the WHI protocol (10).

Compared with those in the lowest quintiles, participants in the highest quintiles (reflecting higher potential of the dietary pattern to contribute to higher insulin, inflammation or postprandial glucose, respectively) for all four dietary indices (EDIH, EDIP, GI, and GL) had higher proportions of black/African Americans, lower proportions of non-Hispanic white, and higher prevalence of cholecystectomy. Participants classified in the highest quintiles of EDIH, EDIP, and GI had higher BMI and were less physically active as compared with those in the lowest quintiles. In contrast, participants with higher GL scores had lower BMI and reported more physical activity (Table 1).

Table 1.

Baseline characteristics of study participants in dietary patterns quintiles, WHI, n = 129,241.

EDIH scorea,bEDIP scorea,cDietary GIaDietary GLa
Characteristics Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 
Range of dietary indices (−10.52 to −0.79) (−0.26 to 0.16) (0.67 to 8.51) (−13.42 to −0.82) (−0.21 to 0.24) (0.73 to 6.90) (−13.57 to −0.76) (−0.22 to 0.25) (0.80 to 4.00) (−8.58 to −0.70) (−0.21 to 0.20) (0.74 to 8.07) 
Sample size 25,848 25,849 25,848 25,848 25,849 25,848 25,848 25,849 25,848 25,848 25,849 25,848 
Race/ethnicity, %            
Black/African American 3.8 6.5 12.3 2.9 5.7 14.8 4.3 5.8 13.3 6.1 7.3 8.6 
American Indian or Alaskan Native 0.3 0.5 0.5 0.3 0.4 0.5 0.5 0.3 0.5 0.4 0.4 0.4 
Hispanic/Latino 2.4 3.2 4.6 1.4 2.4 7.6 3.9 3.2 3.1 3.6 3.3 3.2 
Asian or Pacific Islander 2.0 2.9 2.5 1.2 2.1 4.9 2.5 3.0 1.8 1.6 2.9 3.0 
White (not of Hispanic origin) 90.1 85.6 78.6 92.9 88.0 70.5 87.5 86.3 80.1 86.9 84.6 83.5 
Other race groups 1.4 1.3 1.5 1.3 1.4 1.7 1.5 1.4 1.4 1.4 1.5 1.4 
Age, years, mean ± SD 63.1 ± 7.2 63.7 ± 7.2 61.9 ± 7.1 62.7 ± 7.0 63.5 ± 7.2 62.4 ± 7.3 63.6 ± 7.2 63.1 ± 7.2 62.3 ± 7.1 62.7 ± 7.0 63.4 ± 7.2 62.8 ± 7.4 
BMI, kg/m2, mean ± SD 26.0 ± 4.9 27.0 ± 5.3 29.4 ± 6.2 26.5 ± 5.0 27.1 ± 5.3 28.8 ± 6.2 26.7 ± 5.3 27.3 ± 5.5 28.1 ± 5.9 28.2 ± 5.8 27.3 ± 5.4 26.9 ± 5.5 
Underweight (15 ≤ BMI < 18.5), % 2.2 2.2 1.9 2.0 2.0 2.2 2.2 2.0 2.2 1.9 2.0 2.5 
Normal weight (18.5 ≤ BMI < 25), % 46.4 37.3 23.2 41.6 37.1 27.1 40.6 36.2 31.2 31.0 36.5 40.0 
Overweight (25 ≤ BMI < 30), % 33.3 35.9 32.9 35.1 35.4 33.0 34.5 35.1 33.9 34.4 35.3 33.2 
Obese (BMI ≥30), % 18.0 24.7 42.0 21.3 25.5 37.7 22.7 26.7 32.8 32.8 26.2 24.3 
Physical activity, MET-hours/week, mean ± SD 16.7 ± 15.6 12.8 ± 13.1 9.1 ± 11.3 15.3 ± 14.7 12.9 ± 13.1 9.9 ± 12.2 16.1 ± 15.2 12.8 ± 13.3 9.5 ± 11.7 11.3 ± 12.6 12.4 ± 13.0 14.9 ± 15.1 
Pack years of smoking, mean ± SD 10.8 ± 18.2 9.2 ± 17.3 10.6 ± 19.3 13.1 ± 20.1 9.2 ± 17.1 8.1 ± 16.8 10.1 ± 18.0 9.5 ± 17.5 10.9 ± 19.2 13.6 ± 21.0 9.0 ± 16.9 8.5 ± 16.7 
Current smoking, % 6.1 6.1 9.5 8.9 6.1 6.8 5.6 6.0 9.8 11.3 5.9 4.6 
Aspirin/NSAIDs use, % 13.9 13.5 13.1 14.4 13.8 12.4 13.8 13.5 13.6 13.9 13.6 13.1 
Statin use, % 2.0 2.4 2.1 1.9 2.3 2.3 2.1 2.2 2.3 1.6 2.5 2.7 
Hypercholestrolemia, % 12.0 13.9 13.5 11.6 13.9 14.6 12.4 13.7 14.2 10.9 13.5 15.9 
Educational level, %            
Less than high school 3.0 4.4 7.1 2.7 3.8 8.7 3.3 4.1 7.6 4.6 4.8 4.8 
High school/GED/some college 45.1 54.8 61.7 50.0 54.2 57.9 47.5 53.3 61.9 55.7 54.7 50.9 
≥4 years of college 51.0 40.1 30.0 46.6 41.4 32.6 48.4 41.9 29.7 39.0 39.7 43.5 
Total alcohol intake, alcohol servings/week4 4.8 ± 7.5 1.9 ± 3.7 1.6 ± 3.7 5.3 ± 7.8 2.0 ± 3.6 0.9 ± 2.6 3.6 ± 6.5 2.4 ± 4.4 1.7 ± 4.1 5.5 ± 8.1 1.8 ± 3.3 1.1 ± 2.6 
Gallbladder removed, % 9.1 11.6 15.2 9.7 11.4 14.4 9.7 11.7 14.4 11.3 11.8 12.5 
Total energy, kcal/day 1,832 ± 626 1,482 ± 556 1,829 ± 744 1,731 ± 620 1,548 ± 578 1,789 ± 761 1,591 ± 625 1,680 ± 642 1,605 ± 631 1,826 ± 713 1,489 ± 573 1,831 ± 646 
EDIH scorea,bEDIP scorea,cDietary GIaDietary GLa
Characteristics Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 Quintile 1 Quintile 3 Quintile 5 
Range of dietary indices (−10.52 to −0.79) (−0.26 to 0.16) (0.67 to 8.51) (−13.42 to −0.82) (−0.21 to 0.24) (0.73 to 6.90) (−13.57 to −0.76) (−0.22 to 0.25) (0.80 to 4.00) (−8.58 to −0.70) (−0.21 to 0.20) (0.74 to 8.07) 
Sample size 25,848 25,849 25,848 25,848 25,849 25,848 25,848 25,849 25,848 25,848 25,849 25,848 
Race/ethnicity, %            
Black/African American 3.8 6.5 12.3 2.9 5.7 14.8 4.3 5.8 13.3 6.1 7.3 8.6 
American Indian or Alaskan Native 0.3 0.5 0.5 0.3 0.4 0.5 0.5 0.3 0.5 0.4 0.4 0.4 
Hispanic/Latino 2.4 3.2 4.6 1.4 2.4 7.6 3.9 3.2 3.1 3.6 3.3 3.2 
Asian or Pacific Islander 2.0 2.9 2.5 1.2 2.1 4.9 2.5 3.0 1.8 1.6 2.9 3.0 
White (not of Hispanic origin) 90.1 85.6 78.6 92.9 88.0 70.5 87.5 86.3 80.1 86.9 84.6 83.5 
Other race groups 1.4 1.3 1.5 1.3 1.4 1.7 1.5 1.4 1.4 1.4 1.5 1.4 
Age, years, mean ± SD 63.1 ± 7.2 63.7 ± 7.2 61.9 ± 7.1 62.7 ± 7.0 63.5 ± 7.2 62.4 ± 7.3 63.6 ± 7.2 63.1 ± 7.2 62.3 ± 7.1 62.7 ± 7.0 63.4 ± 7.2 62.8 ± 7.4 
BMI, kg/m2, mean ± SD 26.0 ± 4.9 27.0 ± 5.3 29.4 ± 6.2 26.5 ± 5.0 27.1 ± 5.3 28.8 ± 6.2 26.7 ± 5.3 27.3 ± 5.5 28.1 ± 5.9 28.2 ± 5.8 27.3 ± 5.4 26.9 ± 5.5 
Underweight (15 ≤ BMI < 18.5), % 2.2 2.2 1.9 2.0 2.0 2.2 2.2 2.0 2.2 1.9 2.0 2.5 
Normal weight (18.5 ≤ BMI < 25), % 46.4 37.3 23.2 41.6 37.1 27.1 40.6 36.2 31.2 31.0 36.5 40.0 
Overweight (25 ≤ BMI < 30), % 33.3 35.9 32.9 35.1 35.4 33.0 34.5 35.1 33.9 34.4 35.3 33.2 
Obese (BMI ≥30), % 18.0 24.7 42.0 21.3 25.5 37.7 22.7 26.7 32.8 32.8 26.2 24.3 
Physical activity, MET-hours/week, mean ± SD 16.7 ± 15.6 12.8 ± 13.1 9.1 ± 11.3 15.3 ± 14.7 12.9 ± 13.1 9.9 ± 12.2 16.1 ± 15.2 12.8 ± 13.3 9.5 ± 11.7 11.3 ± 12.6 12.4 ± 13.0 14.9 ± 15.1 
Pack years of smoking, mean ± SD 10.8 ± 18.2 9.2 ± 17.3 10.6 ± 19.3 13.1 ± 20.1 9.2 ± 17.1 8.1 ± 16.8 10.1 ± 18.0 9.5 ± 17.5 10.9 ± 19.2 13.6 ± 21.0 9.0 ± 16.9 8.5 ± 16.7 
Current smoking, % 6.1 6.1 9.5 8.9 6.1 6.8 5.6 6.0 9.8 11.3 5.9 4.6 
Aspirin/NSAIDs use, % 13.9 13.5 13.1 14.4 13.8 12.4 13.8 13.5 13.6 13.9 13.6 13.1 
Statin use, % 2.0 2.4 2.1 1.9 2.3 2.3 2.1 2.2 2.3 1.6 2.5 2.7 
Hypercholestrolemia, % 12.0 13.9 13.5 11.6 13.9 14.6 12.4 13.7 14.2 10.9 13.5 15.9 
Educational level, %            
Less than high school 3.0 4.4 7.1 2.7 3.8 8.7 3.3 4.1 7.6 4.6 4.8 4.8 
High school/GED/some college 45.1 54.8 61.7 50.0 54.2 57.9 47.5 53.3 61.9 55.7 54.7 50.9 
≥4 years of college 51.0 40.1 30.0 46.6 41.4 32.6 48.4 41.9 29.7 39.0 39.7 43.5 
Total alcohol intake, alcohol servings/week4 4.8 ± 7.5 1.9 ± 3.7 1.6 ± 3.7 5.3 ± 7.8 2.0 ± 3.6 0.9 ± 2.6 3.6 ± 6.5 2.4 ± 4.4 1.7 ± 4.1 5.5 ± 8.1 1.8 ± 3.3 1.1 ± 2.6 
Gallbladder removed, % 9.1 11.6 15.2 9.7 11.4 14.4 9.7 11.7 14.4 11.3 11.8 12.5 
Total energy, kcal/day 1,832 ± 626 1,482 ± 556 1,829 ± 744 1,731 ± 620 1,548 ± 578 1,789 ± 761 1,591 ± 625 1,680 ± 642 1,605 ± 631 1,826 ± 713 1,489 ± 573 1,831 ± 646 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake using the residual method. Lower EDIP indicates anti-inflammatory diets while higher EDIP scores indicate proinflammatory diets. Lower EDIH indicates low insulinemic dietary patterns while a higher score indicates hyperinsulinemic diet. We used precomputed GI and GL (total carbohydrate) from WHI FFQ.

bThe EDIH component foods (servings/d) in the WHI were listed in Supplementary Table S3.

cThe EDIP component foods (servings/d) in the WHI were listed in Supplementary Table S3.

Compared with those in the lowest quintiles, participants in the highest quintiles of EDIH and EDIP had higher intakes of red meat, processed meat, sugar-sweetened beverages and lower intakes of whole grain, wine, fruit juice, dark-yellow vegetables, green-leafy vegetables, and coffee/tea intake. Participants who were in the higher quintiles of GI had higher red meat, processed meat, sugar-sweetened beverages, and refined grain intake and had lower wine, fruit juice, dark-yellow vegetables, green-leafy vegetables, and coffee/tea intake. Regarding nutrient intakes, participants in higher quintiles of EDIH, EDIP, and GI had lower total fiber and lycopene intake compared with the lowest quintiles (Table 2). In contrast, the trend of food and nutrient intakes in GL quintiles appeared inversely related to that for EDIH, which aligns with the inverse correlation between the two scores (Supplementary Table S6).

Table 2.

Distribution of dietary intakes across quintiles of the dietary indices.

EDIH scorea,bEDIP scorea,cDietary GIaDietary GLa
Q1Q3Q5Q1Q3Q5Q1Q3Q5Q1Q3Q5
Food/food groups, med servings/week (means ± SDs) 
Red meat 2.3 ± 2.2 3.0 ± 2.4 6.2 ± 4.3 3.2 ± 2.9 3.3 ± 2.8 4.7 ± 4.1 3.1 ± 3.1 3.7 ± 3.2 3.7 ± 3.2 5.6 ± 4.2 3.2 ± 2.6 2.5 ± 2.5 
Processed meat 1.2 ± 1.5 1.5 ± 1.7 3.2 ± 3.0 1.5 ± 1.8 1.6 ± 1.8 2.6 ± 2.9 1.5 ± 2.0 1.9 ± 2.1 2.1 ± 2.3 2.7 ± 2.8 1.6 ± 1.8 1.4 ± 1.9 
Sugar-sweetened beverages 0.4 ± 1.4 0.7 ± 1.7 3.2 ± 6.5 0.4 ± 1.4 0.7 ± 1.8 3.1 ± 6.4 0.3 ± 1.0 0.8 ± 2.2 2.8 ± 6.1 0.6 ± 1.6 0.8 ± 1.9 2.7 ± 6.5 
Tomatoes 4.1 ± 3.5 3.4 ± 3.1 4.1 ± 4.2 4.0 ± 3.4 3.5 ± 3.1 4.1 ± 4.4 4.3 ± 3.8 3.9 ± 3.5 2.9 ± 3.1 4 ± 3.6 3.4 ± 3.2 4.2 ± 4.0 
Refined grains 15.3 ± 9.3 12.3 ± 7.2 13.1 ± 8.1 12.0 ± 7.1 12.4 ± 7.2 16.5 ± 10.0 9.4 ± 6.1 13.6 ± 7.6 16.3 ± 9.3 11.1 ± 7.2 12.2 ± 6.9 17.8 ± 9.8 
Whole grainse 10.6 ± 6.3 9.1 ± 5.1 9.3 ± 5.1 10.1 ± 5.7 9.4 ± 5.2 9.4 ± 5.6 8.7 ± 5.1 9.9 ± 5.5 9.4 ± 5.6 8.6 ± 4.8 9.1 ± 4.9 11.7 ± 6.6 
Wine 3.6 ± 6.0 0.9 ± 1.9 0.5 ± 1.2 3.7 ± 6.1 0.9 ± 1.7 0.3 ± 0.9 2.4 ± 5.0 1.4 ± 2.9 0.7 ± 1.8 3 ± 5.7 1.1 ± 2.3 0.7 ± 1.7 
Fruit juice 5.0 ± 5.2 4.1 ± 4.1 3.7 ± 4.0 4.5 ± 4.9 4.3 ± 4.3 3.6 ± 4.0 4.3 ± 4.8 4.5 ± 4.5 3.3 ± 3.7 3.3 ± 3.6 3.9 ± 3.9 5.6 ± 5.7 
Yellow vegetables 6.7 ± 5.2 5.0 ± 3.7 4.8 ± 4.0 7.4 ± 5.7 5.1 ± 3.6 3.8 ± 3.1 5.8 ± 4.6 5.5 ± 4.2 4.4 ± 3.7 4.8 ± 3.9 4.9 ± 3.8 6.7 ± 5.2 
Green-leafy vegetables 8.3 ± 6.4 5.6 ± 4.3 4.9 ± 4.1 9.2 ± 6.8 5.6 ± 3.9 3.9 ± 3.3 8.1 ± 6.1 6 ± 4.5 4.1 ± 3.7 6.5 ± 5.2 5.7 ± 4.5 6.4 ± 5.5 
Coffee or tea 22.6 ± 14.9 13.8 ± 10.6 11.1 ± 10.6 28.4 ± 15.5 13.6 ± 7.8 6.5 ± 6.6 15.8 ± 13.1 15.3 ± 12.2 14.8 ± 12.5 17.3 ± 13.6 14.5 ± 11.9 14.8 ± 12.6 
Pizza 0.4 ± 0.6 0.3 ± 0.5 0.4 ± 0.6 0.5 ± 0.8 0.3 ± 0.4 0.3 ± 0.4 0.3 ± 0.4 0.4 ± 0.5 0.4 ± 0.6 0.4 ± 0.6 0.4 ± 0.5 0.4 ± 0.6 
Nutrient Intakes (means ± SDs) 
Total fiber, g/d 20.1 ± 7.7 15.2 ± 5.8 14.2 ± 6.2 17.9 ± 7.3 15.7 ± 6.4 15.2 ± 6.8 17 ± 7.2 16.7 ± 6.8 13.6 ± 5.9 13.7 ± 5.9 14.9 ± 5.8 21.2 ± 7.6 
Total carbohydrate, g/d 243.1 ± 84.2 187.0 ± 66.8 202.4 ± 87.0 213.1 ± 79.4 194.2 ± 72.6 217.7 ± 91.7 196.8 ± 78.7 207.9 ± 78.8 199.8 ± 80.1 176.2 ± 73.5 186 ± 64.3 272.6 ± 81.8 
Total protein, g/d 72.5 ± 27.7 62.1 ± 25.2 78 ± 33.3 71.7 ± 28.1 64.9 ± 26.0 72.9 ± 33.3 72.5 ± 30.9 69.6 ± 27.8 60.4 ± 25.2 79.6 ± 33.0 62.5 ± 25.8 70.5 ± 27.3 
BCAAd, g/d 12.9 ± 5.2 11.0 ± 4.6 13.8 ± 6.0 12.7 ± 5.2 11.6 ± 4.8 13 ± 6.0 13.2 ± 5.8 12.3 ± 5.0 10.6 ± 4.5 14.3 ± 6.0 11.1 ± 4.7 12.4 ± 5.0 
Total fat, g/d 59.5 ± 31.7 53.9 ± 27.7 78.5 ± 38.1 60.5 ± 31.7 56.8 ± 29.0 71.1 ± 38.8 54.8 ± 31.2 62.7 ± 33.0 62.6 ± 31.7 82.4 ± 38.5 55 ± 27.0 53.9 ± 30.0 
Saturated fat, g/d 20.2 ± 11.9 18.0 ± 9.9 26.5 ± 13.7 20.4 ± 11.7 19.1 ± 10.6 23.9 ± 13.8 18.5 ± 11.4 21.1 ± 12.0 20.9 ± 11.4 28 ± 14.1 18.5 ± 9.8 17.8 ± 10.9 
Total cholesterol, g/d 196.8 ± 113.6 193.4 ± 107.0 295.7 ± 160.8 216.4 ± 126.4 203.6 ± 116.0 254.8 ± 153.9 212.9 ± 139.1 222.6 ± 126.3 212.6 ± 124.1 304.3 ± 163.9 196.7 ± 102.6 183 ± 108.9 
Dietary calcium, mg/d 1,045.3 ± 526.6 768.2 ± 395.0 746.9 ± 413.1 911.1 ± 486.0 802.9 ± 421.9 810 ± 467.6 1013.1 ± 563.4 826.4 ± 405.9 641.6 ± 321.8 836.5 ± 475.0 769.1 ± 412.9 964 ± 487.1 
Lycopene, mcg/d 5,862.8 ± 3,685.0 4,679.5 ± 2,954.0 4,673.7 ± 3,080.2 5,554.6 ± 3,412.7 4,811.8 ± 3,006.8 4,677.7 ± 3,384.6 5,778.9 ± 3,629.0 5,152.7 ± 3,138.8 3,693.2 ± 2,471.8 4,750.6 ± 2,920.3 4,622 ± 2,924.0 5,898.6 ± 3,924.4 
Dietary magnesium (mg/day) 317.5 ± 98.5 237.1 ± 81.2 237.6 ± 94.7 295.5 ± 97.0 245.7 ± 87.7 241.4 ± 99.6 277.0 ± 102.8 263.2 ± 92.1 217.5 ± 81.6 252.3 ± 96.2 236.0 ± 86.1 306.2 ± 100.2 
Dietary Manganese (mg/day) 4.4 ± 1.5 3.0 ± 1.1 2.7 ± 1.2 4.2 ± 1.4 3.0 ± 1.2 2.7 ± 1.3 3.3 ± 1.4 3.4 ± 1.4 2.9 ± 1.3 3.1 ± 1.4 3.0 ± 1.2 4.0 ± 1.5 
Dietary vitamin D (mcg/day) 5.1 ± 3.7 4.1 ± 2.7 4.3 ± 2.8 4.4 ± 3.3 4.2 ± 2.8 4.6 ± 3.2 5.4 ± 4.2 4.4 ± 2.7 3.4 ± 2.0 4.8 ± 3.4 4.1 ± 2.8 4.6 ± 3.1 
EDIH scorea,bEDIP scorea,cDietary GIaDietary GLa
Q1Q3Q5Q1Q3Q5Q1Q3Q5Q1Q3Q5
Food/food groups, med servings/week (means ± SDs) 
Red meat 2.3 ± 2.2 3.0 ± 2.4 6.2 ± 4.3 3.2 ± 2.9 3.3 ± 2.8 4.7 ± 4.1 3.1 ± 3.1 3.7 ± 3.2 3.7 ± 3.2 5.6 ± 4.2 3.2 ± 2.6 2.5 ± 2.5 
Processed meat 1.2 ± 1.5 1.5 ± 1.7 3.2 ± 3.0 1.5 ± 1.8 1.6 ± 1.8 2.6 ± 2.9 1.5 ± 2.0 1.9 ± 2.1 2.1 ± 2.3 2.7 ± 2.8 1.6 ± 1.8 1.4 ± 1.9 
Sugar-sweetened beverages 0.4 ± 1.4 0.7 ± 1.7 3.2 ± 6.5 0.4 ± 1.4 0.7 ± 1.8 3.1 ± 6.4 0.3 ± 1.0 0.8 ± 2.2 2.8 ± 6.1 0.6 ± 1.6 0.8 ± 1.9 2.7 ± 6.5 
Tomatoes 4.1 ± 3.5 3.4 ± 3.1 4.1 ± 4.2 4.0 ± 3.4 3.5 ± 3.1 4.1 ± 4.4 4.3 ± 3.8 3.9 ± 3.5 2.9 ± 3.1 4 ± 3.6 3.4 ± 3.2 4.2 ± 4.0 
Refined grains 15.3 ± 9.3 12.3 ± 7.2 13.1 ± 8.1 12.0 ± 7.1 12.4 ± 7.2 16.5 ± 10.0 9.4 ± 6.1 13.6 ± 7.6 16.3 ± 9.3 11.1 ± 7.2 12.2 ± 6.9 17.8 ± 9.8 
Whole grainse 10.6 ± 6.3 9.1 ± 5.1 9.3 ± 5.1 10.1 ± 5.7 9.4 ± 5.2 9.4 ± 5.6 8.7 ± 5.1 9.9 ± 5.5 9.4 ± 5.6 8.6 ± 4.8 9.1 ± 4.9 11.7 ± 6.6 
Wine 3.6 ± 6.0 0.9 ± 1.9 0.5 ± 1.2 3.7 ± 6.1 0.9 ± 1.7 0.3 ± 0.9 2.4 ± 5.0 1.4 ± 2.9 0.7 ± 1.8 3 ± 5.7 1.1 ± 2.3 0.7 ± 1.7 
Fruit juice 5.0 ± 5.2 4.1 ± 4.1 3.7 ± 4.0 4.5 ± 4.9 4.3 ± 4.3 3.6 ± 4.0 4.3 ± 4.8 4.5 ± 4.5 3.3 ± 3.7 3.3 ± 3.6 3.9 ± 3.9 5.6 ± 5.7 
Yellow vegetables 6.7 ± 5.2 5.0 ± 3.7 4.8 ± 4.0 7.4 ± 5.7 5.1 ± 3.6 3.8 ± 3.1 5.8 ± 4.6 5.5 ± 4.2 4.4 ± 3.7 4.8 ± 3.9 4.9 ± 3.8 6.7 ± 5.2 
Green-leafy vegetables 8.3 ± 6.4 5.6 ± 4.3 4.9 ± 4.1 9.2 ± 6.8 5.6 ± 3.9 3.9 ± 3.3 8.1 ± 6.1 6 ± 4.5 4.1 ± 3.7 6.5 ± 5.2 5.7 ± 4.5 6.4 ± 5.5 
Coffee or tea 22.6 ± 14.9 13.8 ± 10.6 11.1 ± 10.6 28.4 ± 15.5 13.6 ± 7.8 6.5 ± 6.6 15.8 ± 13.1 15.3 ± 12.2 14.8 ± 12.5 17.3 ± 13.6 14.5 ± 11.9 14.8 ± 12.6 
Pizza 0.4 ± 0.6 0.3 ± 0.5 0.4 ± 0.6 0.5 ± 0.8 0.3 ± 0.4 0.3 ± 0.4 0.3 ± 0.4 0.4 ± 0.5 0.4 ± 0.6 0.4 ± 0.6 0.4 ± 0.5 0.4 ± 0.6 
Nutrient Intakes (means ± SDs) 
Total fiber, g/d 20.1 ± 7.7 15.2 ± 5.8 14.2 ± 6.2 17.9 ± 7.3 15.7 ± 6.4 15.2 ± 6.8 17 ± 7.2 16.7 ± 6.8 13.6 ± 5.9 13.7 ± 5.9 14.9 ± 5.8 21.2 ± 7.6 
Total carbohydrate, g/d 243.1 ± 84.2 187.0 ± 66.8 202.4 ± 87.0 213.1 ± 79.4 194.2 ± 72.6 217.7 ± 91.7 196.8 ± 78.7 207.9 ± 78.8 199.8 ± 80.1 176.2 ± 73.5 186 ± 64.3 272.6 ± 81.8 
Total protein, g/d 72.5 ± 27.7 62.1 ± 25.2 78 ± 33.3 71.7 ± 28.1 64.9 ± 26.0 72.9 ± 33.3 72.5 ± 30.9 69.6 ± 27.8 60.4 ± 25.2 79.6 ± 33.0 62.5 ± 25.8 70.5 ± 27.3 
BCAAd, g/d 12.9 ± 5.2 11.0 ± 4.6 13.8 ± 6.0 12.7 ± 5.2 11.6 ± 4.8 13 ± 6.0 13.2 ± 5.8 12.3 ± 5.0 10.6 ± 4.5 14.3 ± 6.0 11.1 ± 4.7 12.4 ± 5.0 
Total fat, g/d 59.5 ± 31.7 53.9 ± 27.7 78.5 ± 38.1 60.5 ± 31.7 56.8 ± 29.0 71.1 ± 38.8 54.8 ± 31.2 62.7 ± 33.0 62.6 ± 31.7 82.4 ± 38.5 55 ± 27.0 53.9 ± 30.0 
Saturated fat, g/d 20.2 ± 11.9 18.0 ± 9.9 26.5 ± 13.7 20.4 ± 11.7 19.1 ± 10.6 23.9 ± 13.8 18.5 ± 11.4 21.1 ± 12.0 20.9 ± 11.4 28 ± 14.1 18.5 ± 9.8 17.8 ± 10.9 
Total cholesterol, g/d 196.8 ± 113.6 193.4 ± 107.0 295.7 ± 160.8 216.4 ± 126.4 203.6 ± 116.0 254.8 ± 153.9 212.9 ± 139.1 222.6 ± 126.3 212.6 ± 124.1 304.3 ± 163.9 196.7 ± 102.6 183 ± 108.9 
Dietary calcium, mg/d 1,045.3 ± 526.6 768.2 ± 395.0 746.9 ± 413.1 911.1 ± 486.0 802.9 ± 421.9 810 ± 467.6 1013.1 ± 563.4 826.4 ± 405.9 641.6 ± 321.8 836.5 ± 475.0 769.1 ± 412.9 964 ± 487.1 
Lycopene, mcg/d 5,862.8 ± 3,685.0 4,679.5 ± 2,954.0 4,673.7 ± 3,080.2 5,554.6 ± 3,412.7 4,811.8 ± 3,006.8 4,677.7 ± 3,384.6 5,778.9 ± 3,629.0 5,152.7 ± 3,138.8 3,693.2 ± 2,471.8 4,750.6 ± 2,920.3 4,622 ± 2,924.0 5,898.6 ± 3,924.4 
Dietary magnesium (mg/day) 317.5 ± 98.5 237.1 ± 81.2 237.6 ± 94.7 295.5 ± 97.0 245.7 ± 87.7 241.4 ± 99.6 277.0 ± 102.8 263.2 ± 92.1 217.5 ± 81.6 252.3 ± 96.2 236.0 ± 86.1 306.2 ± 100.2 
Dietary Manganese (mg/day) 4.4 ± 1.5 3.0 ± 1.1 2.7 ± 1.2 4.2 ± 1.4 3.0 ± 1.2 2.7 ± 1.3 3.3 ± 1.4 3.4 ± 1.4 2.9 ± 1.3 3.1 ± 1.4 3.0 ± 1.2 4.0 ± 1.5 
Dietary vitamin D (mcg/day) 5.1 ± 3.7 4.1 ± 2.7 4.3 ± 2.8 4.4 ± 3.3 4.2 ± 2.8 4.6 ± 3.2 5.4 ± 4.2 4.4 ± 2.7 3.4 ± 2.0 4.8 ± 3.4 4.1 ± 2.8 4.6 ± 3.1 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake using the residual method. Lower EDIP indicates anti-inflammatory dietary patterns while higher EDIP scores indicate pro-inflammatory diets. Lower EDIH indicates low insulinemic dietary patterns while a higher score indicates hyperinsulinemic dietary patterns. We used pre-computed GI and GL (from total carbohydrates).

bThe EDIH component foods (servings/d) in the WHI were listed in Supplementary Table S3.

cThe EDIP component foods (servings/d) in the WHI were listed in Supplementary Table S3.

dBCAA, branched chain amino acids.

eWhole grain was calculated by taking the sum of dark bread, corn tortilla, popcorn, cooked cereal, corn/hominy.

Over a median of 19.9 years of follow-up, 850 incident cases of pancreatic cancer were ascertained. Table 3 presents the HRs and 95% confidence intervals (CIs) for the associations of each dietary index with pancreatic cancer risk. In multivariable-adjusted models, none of the four indices was associated with future development of pancreatic cancer, and the HRs (95% CI) for each 1 SD increment in dietary index were as follows: EDIH 1.03 (0.96–1.10); Ptrend = 0.83; EDIP 0.95 (0.89–1.02); Ptrend = 0.07; GI 0.96 (0.89–1.03); Ptrend = 0.28; GL 0.96 (0.89–1.03); Ptrend = 0.19.

Table 3.

HRs (95% CI) for the associations of dietary patterns with risk of developing pancreatic cancer.a

HRs for pancreatic cancer risk
Quintile 1Quintile 2Quintile 3Quintile 4Quintile 5P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 
Cases/noncases 170/25,678 174/25,674 183/25,666 174/25,674 149/25,699    
Age-adjusted 1 (Ref) 1.02 (0.83–1.26) 1.09 (0.88–1.34) 1.08 (0.87–1.33) 1.01 (0.81–1.25) 0.79 1.04 (0.97–1.11) 0.24 
Multivariable adjusted 1 (Ref) 1.02 (0.83–1.27) 1.09 (0.88–1.34) 1.06 (0.85–1.31) 0.96 (0.76–1.20) 0.88 1.03 (0.96–1.10) 0.43 
Multivariable + BMI adjusted 1 (Ref) 1.01 (0.82–1.25) 1.09 (0.88–1.34) 1.06 (0.85–1.31) 0.95 (0.75–1.19) 0.83 1.03 (0.96–1.10) 0.47 
Empirical dietary inflammatory pattern (EDIP) scoreb 
Cases/noncases 200/25,648 183/25,665 152/25,697 158/25,690 157/25,691    
Age-adjusted 1 (Ref) 0.91 (0.75–1.11) 0.75 (0.61–0.93) 0.81 (0.66–1.00) 0.88 (0.71–1.08) 0.071 0.96 (0.90–1.02) 0.16 
Multivariable adjusted 1 (Ref) 0.92 (0.75–1.12) 0.77 (0.62–0.95) 0.82 (0.66–1.01) 0.88 (0.71–1.09) 0.069 0.95 (0.89–1.02) 0.16 
Multivariable + BMI adjusted 1 (Ref) 0.91 (0.74–1.11) 0.76 (0.61–0.94) 0.82 (0.66–1.01) 0.87 (0.70–1.09) 0.066 0.95 (0.89–1.02) 0.15 
Dietary glycemic index (GI)c 
Cases/noncases 186/25,662 169/25,679 175/25,674 164/25,684 156/25,692    
Age-adjusted 1 (Ref) 0.91 (0.74–1.12) 0.95 (0.77–1.17) 0.91 (0.74–1.12) 0.92 (0.74–1.13) 0.45 0.97 (0.90–1.04) 0.37 
Multivariable adjusted 1 (Ref) 0.90 (0.73–1.11) 0.94 (0.76–1.15) 0.89 (0.72–1.10) 0.88 (0.71–1.10) 0.28 0.96 (0.89–1.03) 0.23 
Multivariable + BMI adjusted 1 (Ref) 0.90 (0.73–1.11) 0.93 (0.76–1.15) 0.89 (0.72–1.10) 0.88 (0.71–1.10) 0.28 0.96 (0.89–1.03) 0.23 
Dietary glycemic load (GL)c 
Cases/noncases 162/25,686 200/25,648 190/25,659 145/25,703 153/25,695    
Age-adjusted 1 (Ref) 1.20 (0.97–1.47) 1.12 (0.91–1.39) 0.85 (0.68–1.07) 0.92 (0.74–1.15) 0.064 0.94 (0.88–1.01) 0.10 
Multivariable adjusted 1 (Ref) 1.23 (1.00–1.51) 1.16 (0.94–1.43) 0.89 (0.71–1.12) 0.95 (0.76–1.20) 0.16 0.96 (0.89–1.03) 0.25 
Multivariable + BMI adjusted 1 (Ref) 1.23 (1.00–1.51) 1.17 (0.95–1.45) 0.89 (0.71–1.12) 0.95 (0.76–1.20) 0.19 0.96 (0.90–1.03) 0.28 
HRs for pancreatic cancer risk
Quintile 1Quintile 2Quintile 3Quintile 4Quintile 5P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 
Cases/noncases 170/25,678 174/25,674 183/25,666 174/25,674 149/25,699    
Age-adjusted 1 (Ref) 1.02 (0.83–1.26) 1.09 (0.88–1.34) 1.08 (0.87–1.33) 1.01 (0.81–1.25) 0.79 1.04 (0.97–1.11) 0.24 
Multivariable adjusted 1 (Ref) 1.02 (0.83–1.27) 1.09 (0.88–1.34) 1.06 (0.85–1.31) 0.96 (0.76–1.20) 0.88 1.03 (0.96–1.10) 0.43 
Multivariable + BMI adjusted 1 (Ref) 1.01 (0.82–1.25) 1.09 (0.88–1.34) 1.06 (0.85–1.31) 0.95 (0.75–1.19) 0.83 1.03 (0.96–1.10) 0.47 
Empirical dietary inflammatory pattern (EDIP) scoreb 
Cases/noncases 200/25,648 183/25,665 152/25,697 158/25,690 157/25,691    
Age-adjusted 1 (Ref) 0.91 (0.75–1.11) 0.75 (0.61–0.93) 0.81 (0.66–1.00) 0.88 (0.71–1.08) 0.071 0.96 (0.90–1.02) 0.16 
Multivariable adjusted 1 (Ref) 0.92 (0.75–1.12) 0.77 (0.62–0.95) 0.82 (0.66–1.01) 0.88 (0.71–1.09) 0.069 0.95 (0.89–1.02) 0.16 
Multivariable + BMI adjusted 1 (Ref) 0.91 (0.74–1.11) 0.76 (0.61–0.94) 0.82 (0.66–1.01) 0.87 (0.70–1.09) 0.066 0.95 (0.89–1.02) 0.15 
Dietary glycemic index (GI)c 
Cases/noncases 186/25,662 169/25,679 175/25,674 164/25,684 156/25,692    
Age-adjusted 1 (Ref) 0.91 (0.74–1.12) 0.95 (0.77–1.17) 0.91 (0.74–1.12) 0.92 (0.74–1.13) 0.45 0.97 (0.90–1.04) 0.37 
Multivariable adjusted 1 (Ref) 0.90 (0.73–1.11) 0.94 (0.76–1.15) 0.89 (0.72–1.10) 0.88 (0.71–1.10) 0.28 0.96 (0.89–1.03) 0.23 
Multivariable + BMI adjusted 1 (Ref) 0.90 (0.73–1.11) 0.93 (0.76–1.15) 0.89 (0.72–1.10) 0.88 (0.71–1.10) 0.28 0.96 (0.89–1.03) 0.23 
Dietary glycemic load (GL)c 
Cases/noncases 162/25,686 200/25,648 190/25,659 145/25,703 153/25,695    
Age-adjusted 1 (Ref) 1.20 (0.97–1.47) 1.12 (0.91–1.39) 0.85 (0.68–1.07) 0.92 (0.74–1.15) 0.064 0.94 (0.88–1.01) 0.10 
Multivariable adjusted 1 (Ref) 1.23 (1.00–1.51) 1.16 (0.94–1.43) 0.89 (0.71–1.12) 0.95 (0.76–1.20) 0.16 0.96 (0.89–1.03) 0.25 
Multivariable + BMI adjusted 1 (Ref) 1.23 (1.00–1.51) 1.17 (0.95–1.45) 0.89 (0.71–1.12) 0.95 (0.76–1.20) 0.19 0.96 (0.90–1.03) 0.28 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake using the residual method. The multivariable adjusted models were stratified by hormone use, education, and age, and further adjusted for family history of T2D, physical activity, race/ethnicity, pack-years of smoking, hormone therapy trial arms, NSAID use, supplement use, dietary modification trial arms, cholecystectomy status, and comorbidity score. The multivariable + BMI adjusted models were further stratified by BMI.

bLower EDIP scores indicate anti-inflammatory dietary patterns while higher EDIP scores indicate pro-inflammatory patterns. Lower EDIH indicates low insulinemic dietary patterns while higher scores indicate more hyperinsulinemic dietary patterns.

cGI and GL scores were calculated using total carbohydrates. Lower GI/GL scores indicate low glycemic diets while higher scores indicate hyperglycemic dietary patterns.

dP values for linear trend across dietary index quartiles were estimated by assigning the median dietary index value for each quintile to all participants in the corresponding quartile, as an ordinal variable. Models for linear trend were adjusted for all covariates listed in the corresponding models in footnote a.

Although there was no statistical evidence of interaction between the dietary indices and T2D categories (Pinteraction values; EDIH: 0.96, EDIP: 0.41, GI: 0.94, GL: 0.28; Table 4), HRs were modestly elevated among women with longstanding T2D. An increase in EDIH score by 1 SD was associated with a 33% higher risk of developing pancreatic cancer (HR, 1.33; 95% CI, 1.06–1.66; Ptrend = 0.01). Similarly, we observed increased, but statistically nonsignificant, associations for 1-SD increments in the other three dietary indices with risk of pancreatic cancer among women with longstanding diabetes (EDIP: HR 1.26; 95% CI, 0.98–1.63; Ptrend = 0.07; GI: HR, 1.26; 95% CI, 0.96–1.67; Ptrend = 0.10; GL: HR, 1.23; 95% CI, 0.96–1.57; Ptrend = 0.10). No associations were observed between any of the dietary indices and pancreatic cancer risk among women with recent onset diabetes or among those with no diabetes (Table 4).

Table 4.

HRs (95% CI) for the associations of dietary patterns with risk of developing pancreatic cancer in subgroups defined by diabetes status and duration.a,e

Quartile 1Quartile 2Quartile 3Quartile 4P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
No T2D        
Cases/noncases 171/27,755 177/27,406 173/26,796 148/25,659    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.06 (0.85–1.30) 1.09 (0.88–1.35) 1.04 (0.83–1.32) 0.67 1.02 (0.94–1.10) 0.68 
Cases/noncases 203/27,518 178/27,317 152/26,901 136/25,880    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.88 (0.72–1.08) 0.79 (0.64–0.97) 0.81 (0.65–1.02) 0.02 0.93 (0.86–1.00) 0.051 
Cases/noncases 184/27,248 170/26,822 167/26,819 148/26,727    
Dietary glycemic index (GI)c 1 (Ref) 0.94 (0.76–1.15) 0.93 (0.75–1.15) 0.87 (0.70–1.09) 0.23 0.94 (0.87–1.02) 0.14 
Cases/noncases 164/26,671 199/26,807 167/27,030 139/27,108    
Dietary glycemic load (GL)c 1 (Ref) 1.19 (0.96–1.46) 0.99 (0.79–1.23) 0.83 (0.66–1.05) 0.07 0.93 (0.86–1.01) 0.09 
Recent Onset T2D (T2D diagnosed ≤3 years before pancreatic cancer diagnosis) 
Cases/noncases 30/1,473 35/1,548 24/1,587 29/1,781    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.14 (0.69–1.90) 0.75 (0.43–1.31) 0.83 (0.47–1.44) 0.47 0.96 (0.80–1.14) 0.63 
Cases/noncases 32/1,525 27/1,551 25/1,612 34/1,701    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.86 (0.51–1.46) 0.69 (0.40–1.19) 0.88 (0.53–1.48) 0.57 0.95 (0.80–1.13) 0.58 
Cases/nNoncases 31/1,547 35/1,605 25/1,626 27/1,611    
Dietary glycemic index (GI)c 1 (Ref) 1.10 (0.67–1.80) 0.72 (0.42–1.25) 0.79 (0.46–1.37) 0.30 0.90 (0.74–1.10) 0.30 
Cases/noncases 28/1,537 39/1,607 26/1,602 25/1,643    
Dietary glycemic load (GL)c 1(ref) 1.40 (0.85–2.31) 0.78 (0.45–1.37) 0.73 (0.41–1.30) 0.10 0.93 (0.77–1.12) 0.46 
Longstanding T2D (T2D diagnosed > 3 years before pancreatic cancer diagnosis) 
Cases/noncases 11/2,853 11/3,116 12/3,705 28/4,645    
Empirical dietary index for hyperinsulinemic (EDIH) score b 1 (Ref) 0.87 (0.37–2.03) 0.75 (0.32–1.75) 1.47 (0.70–3.08) 0.21 1.33 (1.06–1.66) 0.01 
Cases/noncases 10/3,004 18/3,205 15/3,590 19/4,520    
Empirical dietary inflammatory pattern (EDIP) score b 1(ref) 1.74 (0.79–3.81) 1.31 (0.58–2.95) 1.33 (0.60–2.93) 0.71 1.26 (0.98–1.63) 0.07 
Cases/noncases 12/3,262 15/3,650 15/3,645 20/3,762    
Dietary glycemic index (GI)c 1 (Ref) 1.04 (0.48–2.2) 1.06 (0.49–2.31) 1.61 (0.76–3.37) 0.25 1.26 (0.96–1.67) 0.10 
Cases/noncases 20/3,877 11/3,627 17/3,446 14/3,369    
Dietary glycemic load (GL)c 1 (Ref) 0.62 (0.29–1.30) 1.10 (0.57–2.15) 1.18 (0.57–2.42) 0.62 1.23 (0.96–1.57) 0.10 
Quartile 1Quartile 2Quartile 3Quartile 4P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
No T2D        
Cases/noncases 171/27,755 177/27,406 173/26,796 148/25,659    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.06 (0.85–1.30) 1.09 (0.88–1.35) 1.04 (0.83–1.32) 0.67 1.02 (0.94–1.10) 0.68 
Cases/noncases 203/27,518 178/27,317 152/26,901 136/25,880    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.88 (0.72–1.08) 0.79 (0.64–0.97) 0.81 (0.65–1.02) 0.02 0.93 (0.86–1.00) 0.051 
Cases/noncases 184/27,248 170/26,822 167/26,819 148/26,727    
Dietary glycemic index (GI)c 1 (Ref) 0.94 (0.76–1.15) 0.93 (0.75–1.15) 0.87 (0.70–1.09) 0.23 0.94 (0.87–1.02) 0.14 
Cases/noncases 164/26,671 199/26,807 167/27,030 139/27,108    
Dietary glycemic load (GL)c 1 (Ref) 1.19 (0.96–1.46) 0.99 (0.79–1.23) 0.83 (0.66–1.05) 0.07 0.93 (0.86–1.01) 0.09 
Recent Onset T2D (T2D diagnosed ≤3 years before pancreatic cancer diagnosis) 
Cases/noncases 30/1,473 35/1,548 24/1,587 29/1,781    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.14 (0.69–1.90) 0.75 (0.43–1.31) 0.83 (0.47–1.44) 0.47 0.96 (0.80–1.14) 0.63 
Cases/noncases 32/1,525 27/1,551 25/1,612 34/1,701    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.86 (0.51–1.46) 0.69 (0.40–1.19) 0.88 (0.53–1.48) 0.57 0.95 (0.80–1.13) 0.58 
Cases/nNoncases 31/1,547 35/1,605 25/1,626 27/1,611    
Dietary glycemic index (GI)c 1 (Ref) 1.10 (0.67–1.80) 0.72 (0.42–1.25) 0.79 (0.46–1.37) 0.30 0.90 (0.74–1.10) 0.30 
Cases/noncases 28/1,537 39/1,607 26/1,602 25/1,643    
Dietary glycemic load (GL)c 1(ref) 1.40 (0.85–2.31) 0.78 (0.45–1.37) 0.73 (0.41–1.30) 0.10 0.93 (0.77–1.12) 0.46 
Longstanding T2D (T2D diagnosed > 3 years before pancreatic cancer diagnosis) 
Cases/noncases 11/2,853 11/3,116 12/3,705 28/4,645    
Empirical dietary index for hyperinsulinemic (EDIH) score b 1 (Ref) 0.87 (0.37–2.03) 0.75 (0.32–1.75) 1.47 (0.70–3.08) 0.21 1.33 (1.06–1.66) 0.01 
Cases/noncases 10/3,004 18/3,205 15/3,590 19/4,520    
Empirical dietary inflammatory pattern (EDIP) score b 1(ref) 1.74 (0.79–3.81) 1.31 (0.58–2.95) 1.33 (0.60–2.93) 0.71 1.26 (0.98–1.63) 0.07 
Cases/noncases 12/3,262 15/3,650 15/3,645 20/3,762    
Dietary glycemic index (GI)c 1 (Ref) 1.04 (0.48–2.2) 1.06 (0.49–2.31) 1.61 (0.76–3.37) 0.25 1.26 (0.96–1.67) 0.10 
Cases/noncases 20/3,877 11/3,627 17/3,446 14/3,369    
Dietary glycemic load (GL)c 1 (Ref) 0.62 (0.29–1.30) 1.10 (0.57–2.15) 1.18 (0.57–2.42) 0.62 1.23 (0.96–1.57) 0.10 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake using the residual method. The multivariable adjusted +BMI models were stratified by hormone use, education, BMI, and age, and further adjusted for family history of T2D, physical activity, race/ethnicity, pack-years of smoking, hormone therapy trial arms, NSAID use, supplement use, dietary modification trial arms, cholecystectomy status, and comorbidity score.

bLower EDIP scores indicate anti-inflammatory dietary patterns while higher EDIP scores indicate pro-inflammatory patterns. Lower EDIH indicates low insulinemic dietary patterns while a higher score indicates hyperinsulinemic patterns.

cGI and GL were computed using total carbohydrates. Lower GI/GL scores indicate low glycemic diets while higher scores indicate hyperglycemic diets.

dP values for linear trend across dietary index quartiles were estimated by assigning the median dietary index value for each quartile to all participants in the corresponding quartile, as an ordinal variable. Models for linear trend were adjusted for all covariates listed in the corresponding models in footnote a.

eWe tested for interaction using the likelihood ratio test, comparing the full model (with dietary score x diabetes terms) and reduced model (without the interaction terms). P values for interaction with each dietary index were as follows: EDIH: 0.96, EDIP: 0.41, GI: 0.94, GL: 0.28. There were 62 cases of pancreatic cancer in the longstanding T2D category and 118 in the recent onset T2D category.

The BMI subgroup analysis is presented in Table 5. In general, we observed no significant associations within BMI categories, although we found an inverse association between higher GI and GL scores and pancreatic cancer risk among normal-weight women (GI: HR, 0.86; 95% CI, 0.76–0.96; Ptrend = 0.009; Pinteraction = 0.007; GL HR, 0.83; 95% CI, 0.73, 0.93; Ptrend = 0.002; Pinteraction = 0.007).

Table 5.

HRs (95% CI) for the associations of dietary patterns with risk of developing pancreatic cancer in subgroups defined by body weight categories.a,e

HRs for pancreatic cancer risk
Quartile 1Quartile 2Quartile 3Quartile 4P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
Normal weight women (BMI: 18.5–24.9 kg/m2) 
Cases/noncases 89/14,696 93/12,945 72/10,817 50/7,954    
Empirical dietary index for hyperinsulinemic (EDIH)b score 1 (Ref) 1.20 (0.90–1.61) 1.16 (0.84–1.59) 1.16 (0.81–1.65) 0.37 1.04 (0.92–1.17) 0.57 
Cases/noncases 104/13,273 81/12,513 67/11,469 52/9,157    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.83 (0.62–1.12) 0.78 (0.57–1.07) 0.81 (0.57–1.14) 0.13 0.94 (0.84–1.05) 0.27 
Cases/noncases 104/12,986 78/12,146 68/11,141 54/10,139    
Dietary glycemic index (GI)c 1 (Ref) 0.80 (0.60–1.08) 0.77 (0.56–1.05) 0.69 (0.49–0.97) 0.025 0.86 (0.76–0.96) 0.009 
Cases/noncases 74/10,075 89/11,315 88/12,139 53/12,883    
Dietary glycemic load (GL)c 1 (Ref) 1.07 (0.78–1.46) 0.97 (0.71–1.33) 0.54 (0.38–0.78) 0.0008 0.83 (0.73–0.93) 0.002 
Overweight women (BMI: 25–29.9 kg/m2) 
Cases/noncases 74/10,775 79/11,404 84/11,566 67/10,701    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.00 (0.72–1.37) 1.06 (0.78–1.46) 1.00 (0.71–1.40) 0.92 1.00 (0.89–1.13) 0.93 
Cases/noncases 86/11,231 87/11,343 64/11,222 67/10,650    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.98 (0.72–1.32) 0.73 (0.52–1.01) 0.92 (0.66–1.27) 0.26 0.95 (0.85–1.06) 0.34 
Cases/noncases 73/11,027 88/11,221 79/11,294 64/10,904    
Dietary glycemic index (GI)c 1 (Ref) 1.18 (0.86–1.61) 1.05 (0.76–1.45) 0.93 (0.66–1.32) 0.5903 0.96 (0.85–1.08) 0.50 
Cases/noncases 69/11,082 98/11,367 63/11,265 74/10,732    
Dietary glycemic load (GL)c 1 (Ref) 1.33 (0.98–1.81) 0.87 (0.61–1.22) 1.10 (0.78–1.54) 0.87 1.00 (0.88–1.13) 0.95 
Obese women (BMI: ≥30 kg/m2) 
Cases/noncases 44/5,917 47/7,053 53/9,073 82/12,831    
Empirical dietary index for hyperinsulinemic (EDIH)b score 1 (Ref) 0.92 (0.61–1.38) 0.82 (0.55–1.23) 0.91 (0.63–1.33) 0.64 1.05 (0.93–1.19) 0.42 
Cases/noncases 51/6,910 51/7,597 58/8,732 66/11,635    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.92 (0.63–1.36) 0.95 (0.65–1.39) 0.85 (0.59–1.24) 0.44 0.99 (0.87–1.12) 0.85 
Cases/noncases 42/7,370 54/8,090 58/9,034 72/10,380    
Dietary glycemic index (GI)c 1 (Ref) 1.18 (0.79–1.77) 1.16 (0.78–1.73) 1.33 (0.90–1.97) 0.17 1.11 (0.96–1.27) 0.15 
Cases/noncases 67/10,320 59/8,760 52/8,054 48/7,740    
Dietary glycemic load (GL)c 1 (Ref) 1.08 (0.76–1.53) 1.04 (0.72–1.51) 1.05 (0.71–1.54) 0.83 1.08 (0.95–1.23) 0.22 
HRs for pancreatic cancer risk
Quartile 1Quartile 2Quartile 3Quartile 4P value for linear trenddPer 1-SD increment in dietary scoreP value for continuous dietary score
Normal weight women (BMI: 18.5–24.9 kg/m2) 
Cases/noncases 89/14,696 93/12,945 72/10,817 50/7,954    
Empirical dietary index for hyperinsulinemic (EDIH)b score 1 (Ref) 1.20 (0.90–1.61) 1.16 (0.84–1.59) 1.16 (0.81–1.65) 0.37 1.04 (0.92–1.17) 0.57 
Cases/noncases 104/13,273 81/12,513 67/11,469 52/9,157    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.83 (0.62–1.12) 0.78 (0.57–1.07) 0.81 (0.57–1.14) 0.13 0.94 (0.84–1.05) 0.27 
Cases/noncases 104/12,986 78/12,146 68/11,141 54/10,139    
Dietary glycemic index (GI)c 1 (Ref) 0.80 (0.60–1.08) 0.77 (0.56–1.05) 0.69 (0.49–0.97) 0.025 0.86 (0.76–0.96) 0.009 
Cases/noncases 74/10,075 89/11,315 88/12,139 53/12,883    
Dietary glycemic load (GL)c 1 (Ref) 1.07 (0.78–1.46) 0.97 (0.71–1.33) 0.54 (0.38–0.78) 0.0008 0.83 (0.73–0.93) 0.002 
Overweight women (BMI: 25–29.9 kg/m2) 
Cases/noncases 74/10,775 79/11,404 84/11,566 67/10,701    
Empirical dietary index for hyperinsulinemic (EDIH) scoreb 1 (Ref) 1.00 (0.72–1.37) 1.06 (0.78–1.46) 1.00 (0.71–1.40) 0.92 1.00 (0.89–1.13) 0.93 
Cases/noncases 86/11,231 87/11,343 64/11,222 67/10,650    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.98 (0.72–1.32) 0.73 (0.52–1.01) 0.92 (0.66–1.27) 0.26 0.95 (0.85–1.06) 0.34 
Cases/noncases 73/11,027 88/11,221 79/11,294 64/10,904    
Dietary glycemic index (GI)c 1 (Ref) 1.18 (0.86–1.61) 1.05 (0.76–1.45) 0.93 (0.66–1.32) 0.5903 0.96 (0.85–1.08) 0.50 
Cases/noncases 69/11,082 98/11,367 63/11,265 74/10,732    
Dietary glycemic load (GL)c 1 (Ref) 1.33 (0.98–1.81) 0.87 (0.61–1.22) 1.10 (0.78–1.54) 0.87 1.00 (0.88–1.13) 0.95 
Obese women (BMI: ≥30 kg/m2) 
Cases/noncases 44/5,917 47/7,053 53/9,073 82/12,831    
Empirical dietary index for hyperinsulinemic (EDIH)b score 1 (Ref) 0.92 (0.61–1.38) 0.82 (0.55–1.23) 0.91 (0.63–1.33) 0.64 1.05 (0.93–1.19) 0.42 
Cases/noncases 51/6,910 51/7,597 58/8,732 66/11,635    
Empirical dietary inflammatory pattern (EDIP) scoreb 1 (Ref) 0.92 (0.63–1.36) 0.95 (0.65–1.39) 0.85 (0.59–1.24) 0.44 0.99 (0.87–1.12) 0.85 
Cases/noncases 42/7,370 54/8,090 58/9,034 72/10,380    
Dietary glycemic index (GI)c 1 (Ref) 1.18 (0.79–1.77) 1.16 (0.78–1.73) 1.33 (0.90–1.97) 0.17 1.11 (0.96–1.27) 0.15 
Cases/noncases 67/10,320 59/8,760 52/8,054 48/7,740    
Dietary glycemic load (GL)c 1 (Ref) 1.08 (0.76–1.53) 1.04 (0.72–1.51) 1.05 (0.71–1.54) 0.83 1.08 (0.95–1.23) 0.22 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake using the residual method. The multivariable adjusted models were stratified by hormone use, education, and age, and further adjusted for family history of T2D, physical activity, race/ethnicity, pack-years of smoking, hormone therapy trial arms, NSAID use, supplement use, dietary modification trial arms, cholecystectomy status, and comorbidity score.

bLower EDIP scores indicate low inflammatory dietary patterns whereas higher EDIP scores indicate pro-inflammatory patterns. Lower EDIH indicates low insulinemic dietary pattern whereas higher scores indicate hyperinsulinemic dietary patterns.

cGI and GL were calculated using total carbohydrates. Lower GI/GL scores indicate low glycemic diets while higher scores indicate hyperglycemic dietary patterns.

dP values for linear trend across dietary index quartiles were estimated by assigning the median dietary index value for each quartile to all participants in the corresponding quartile, as an ordinal variable. Models for linear trend were adjusted for all covariates listed in the corresponding models in footnote a.

eWe tested for interaction using the likelihood ratio test, comparing the full (with dietary score x BMI terms) and reduced models (without interaction terms). P values for interaction for each dietary index were as follows: EDIH = 0.80; EDIP = 0.43; GI = 0.0068; GL = 0.0075.

Corresponding absolute risk estimates presented in Table 6 for the overall sample and in T2D and BMI subgroups, aligned well with the relative risks. For example, there was no excess absolute risk for any of the four dietary indices in the overall sample, whereas all four dietary indices resulted in modest excess risk of between 11 and 13 incident pancreatic cancer cases per 100,000 person-years among women with longstanding diabetes, but no excess risk in other subgroups.

Table 6.

Incidence rate of pancreatic cancer in dietary index quintiles overall, by diabetes status, and by body mass index category.a

Overall incidence rate of pancreatic cancer per 100,000 person-years
Overall study sampleQuintile 1Quintile 2Quintile 3Quintile 4Quintile 5Difference (Q5-Q1)b
Empirical dietary index for hyperinsulinemic (EDIH) score 36 35 41 38 35 −1 
Empirical dietary inflammatory pattern (EDIP) score 42 38 36 32 35 −7 
Dietary glycemic index (GI)c 39 36 38 37 34 −5 
Dietary glycemic load (GL)c 37 42 40 33 33 −4 
Incident pancreatic cancer cases per 100,000 person-years by diabetes status and duration 
NoT2D Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Difference (Q5-Q1)b 
Empirical dietary index for hyperinsulinemic (EDIH) score 34 33 40 36 33 −2 
Empirical dietary inflammatory pattern (EDIP) score 43 36 33 31 34 −9 
Dietary glycemic index (GI)c 40 31 36 37 31 −9 
Dietary glycemic load (GL)c 36 40 38 33 30 −7 
Recent onset T2D 
Empirical dietary index for hyperinsulinemic (EDIH) score 103 96 96 97 67 −36 
Empirical dietary inflammatory pattern (EDIP) score 102 100 99 71 86 −17 
Dietary glycemic index (GI)c 69 144 93 67 84 15 
Dietary glycemic load (GL)c 94 125 94 63 82 −12 
Longstanding T2D 
Empirical dietary index for hyperinsulinemic (EDIH) score 17 20 22 20 30 13 
Empirical dietary inflammatory pattern (EDIP) score 11 30 31 16 23 11 
Dietary glycemic index (GI)c 20 18 23 18 32 12 
Dietary glycemic load (GL)c 20 18 23 18 32 12 
Incident pancreatic cancer cases per 100,000 person-years by BMI categories 
Normal weight women (BMI: 18.5–24.9 kg/m2) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Difference (Q5-Q1)b 
Empirical dietary index for hyperinsulinemic (EDIH) score 33 32 42 37 35 
Empirical dietary inflammatory pattern (EDIP) score 42 34 35 32 36 −6 
Dietary glycemic index (GI)c 47 29 42 32 28 −19 
Dietary glycemic load (GL)c 44 41 44 26 24 −20 
Overweight women (BMI: 25–29.9 kg/m2) 
Empirical dietary index for hyperinsulinemic (EDIH) score 36 38 41 40 35 −1 
Empirical dietary inflammatory pattern (EDIP) score 47 40 35 33 35 −11 
Dietary glycemic index (GI)c 36 45 33 42 34 −2 
Dietary glycemic load (GL)c 31 51 33 39 35 
Obese women (BMI: ≥30 kg/m2) 
Empirical dietary index for hyperinsulinemic (EDIH) score 40 37 39 34 36 −4 
Empirical dietary inflammatory pattern (EDIP) score 37 43 42 32 33 −4 
Dietary glycemic index (GI)c 32 34 40 37 42 
Dietary glycemic load (GL)c 37 34 40 33 42 
Overall incidence rate of pancreatic cancer per 100,000 person-years
Overall study sampleQuintile 1Quintile 2Quintile 3Quintile 4Quintile 5Difference (Q5-Q1)b
Empirical dietary index for hyperinsulinemic (EDIH) score 36 35 41 38 35 −1 
Empirical dietary inflammatory pattern (EDIP) score 42 38 36 32 35 −7 
Dietary glycemic index (GI)c 39 36 38 37 34 −5 
Dietary glycemic load (GL)c 37 42 40 33 33 −4 
Incident pancreatic cancer cases per 100,000 person-years by diabetes status and duration 
NoT2D Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Difference (Q5-Q1)b 
Empirical dietary index for hyperinsulinemic (EDIH) score 34 33 40 36 33 −2 
Empirical dietary inflammatory pattern (EDIP) score 43 36 33 31 34 −9 
Dietary glycemic index (GI)c 40 31 36 37 31 −9 
Dietary glycemic load (GL)c 36 40 38 33 30 −7 
Recent onset T2D 
Empirical dietary index for hyperinsulinemic (EDIH) score 103 96 96 97 67 −36 
Empirical dietary inflammatory pattern (EDIP) score 102 100 99 71 86 −17 
Dietary glycemic index (GI)c 69 144 93 67 84 15 
Dietary glycemic load (GL)c 94 125 94 63 82 −12 
Longstanding T2D 
Empirical dietary index for hyperinsulinemic (EDIH) score 17 20 22 20 30 13 
Empirical dietary inflammatory pattern (EDIP) score 11 30 31 16 23 11 
Dietary glycemic index (GI)c 20 18 23 18 32 12 
Dietary glycemic load (GL)c 20 18 23 18 32 12 
Incident pancreatic cancer cases per 100,000 person-years by BMI categories 
Normal weight women (BMI: 18.5–24.9 kg/m2) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Difference (Q5-Q1)b 
Empirical dietary index for hyperinsulinemic (EDIH) score 33 32 42 37 35 
Empirical dietary inflammatory pattern (EDIP) score 42 34 35 32 36 −6 
Dietary glycemic index (GI)c 47 29 42 32 28 −19 
Dietary glycemic load (GL)c 44 41 44 26 24 −20 
Overweight women (BMI: 25–29.9 kg/m2) 
Empirical dietary index for hyperinsulinemic (EDIH) score 36 38 41 40 35 −1 
Empirical dietary inflammatory pattern (EDIP) score 47 40 35 33 35 −11 
Dietary glycemic index (GI)c 36 45 33 42 34 −2 
Dietary glycemic load (GL)c 31 51 33 39 35 
Obese women (BMI: ≥30 kg/m2) 
Empirical dietary index for hyperinsulinemic (EDIH) score 40 37 39 34 36 −4 
Empirical dietary inflammatory pattern (EDIP) score 37 43 42 32 33 −4 
Dietary glycemic index (GI)c 32 34 40 37 42 
Dietary glycemic load (GL)c 37 34 40 33 42 

aEDIP, EDIH, GI, and GL scores were adjusted for total energy intake, family history of T2D, physical activity, race/ethnicity, pack-years of smoking, hormone replacement therapy arms, NSAID, supplement, dietary modification trial arms cholecystectomy status, comorbidity score, hormone use, education, BMI, age.

bQ5-Q1: The excess incidence due to consuming a hyperinsulinemic, pro-inflammatory or hyperglycemic dietary.

cGI and GL were calculated using total carbohydrates.

We used several validated dietary indices to assess the association between habitual consumption of hyperinsulinemic (EDIH), proinflammatory (EDIP), and hyperglycemic (GI and GL) dietary patterns and future risk of pancreatic cancer in a large cohort of postmenopausal women. In the overall sample, we did not observe significant associations between these biologic domains of the diet and risk of pancreatic cancer. However, when stratified by diabetes categories, we observed a modestly elevated (although nonsignificant) risk of pancreatic cancer for higher scores of each dietary index among women with longstanding diabetes, and a corresponding excess absolute risk. We also observed significant inverse associations between dietary glycemic scores and pancreatic cancer risk among normal-weight women.

Previous epidemiologic studies of the association of dietary inflammatory potential and risk of developing pancreatic cancer have used a literature-derived nutrient-based dietary inflammatory index (DII) to assess the inflammatory potential of the diet and the results have been mixed (12, 25, 26). The DII, being nutrient-based, is heavily weighted toward nutritional supplements and, therefore, results based on the DII are difficult to directly compare with those obtained from the food-based EDIP score used in this study, as it is hard to uncover the influence of diet when mixed with supplements. Investigators found significant associations between higher DII scores, reflecting more proinflammatory diets, and pancreatic cancer risk in an Italian case–control study (25), a finding that was later confirmed by pooling data from six case–control studies in the Pancreatic Cancer Case-Control Consortium (PanC4) but not in the Pancreatic Cancer Cohort Consortium studies (PanScan; ref. 26). Also, when the DII was applied in a prospective study using data from the Prostate Lung, Colorectal and Ovarian (PLCO) cancer cohort, there was no association with pancreatic cancer risk (12), highlighting similar inconsistencies by study design that are evident when other dietary patterns have been examined in relation to pancreatic cancer risk (3). In addition, when effect modification by time was investigated, higher DII scores appeared to be inversely associated with pancreatic cancer risk in the first 4 years of follow-up and positively associated with pancreatic cancer risk when follow-up was at least 4 years (12). This highlights the potential reverse causation that we have addressed in this study by including a 4-year lag as our primary analytic approach, to separate diet assessment from pancreatic cancer diagnosis, thus improving the internal validity of our findings. In the only previous study of the EDIH score in relation to pancreatic cancer risk, there was no association among women in the Nurses' Health Study (NHS) and among men in the Health Professionals Follow-up Study (HPFS; ref. 27), consistent with our findings here.

Evidence regarding the glycemic potential of the diet in relation to pancreatic cancer risk has been mostly inconsistent. One meta-analysis that included both case–control (n = 11) and cohort (n = 9) studies observed no associations of pancreatic cancer with higher GI and GL scores (28). Another meta-analysis that included only cohort studies (n = 13) found no association between GI or GL and pancreatic cancer risk. The summary RR per 10 GI units was 1.02; 95% CI, 0.93–1.12, and per 50 GL units was 1.03; 95% CI, 0.93–1.14 (29). Furthermore, a previous prospective study conducted in the WHI, examined associations of GI and GL with risk of pancreatic cancer and included only 287 cases with a median of 8 years of follow-up (30). This study did not support an association between dietary patterns high in GI or GL and elevated pancreatic cancer risk, findings that we have verified in this study with almost three times the number of cases and longer follow-up.

We observed elevated, although nonsignificant, risk of pancreatic cancer for each of the four dietary indices among women with longstanding diabetes. To our knowledge, this is the first study to report on the association of dietary pattern and pancreatic cancer risk stratified by diabetes duration. This study suggests that the observed diet-pancreatic cancer association is influenced by coexisting chronic hyperglycemia, hyperinsulinemia, and inflammation resulting from the longstanding diabetes. Unlike recent onset T2D, which may be more related to pancreatic dysfunction associated with nascent pancreatic cancer not yet diagnosed, diet may directly influence the development of longstanding T2D (7, 8, 31). Longstanding T2D may then mediate pancreatic cancer development through prolonged insulin resistance, hyperinsulinemia, hyperglycemia, and progressive deterioration in beta-cell function, combined with a proinflammatory state (32). A recent prospective cohort study conducted in the NHS and HPFS cohorts, reported a nonlinear relationship between T2D duration and pancreatic cancer risk, where the risk peaked around 8 years after T2D diagnosis and gradually decreased afterwards (33). Also, the study found a higher C-peptide level (reflecting higher beta-cell secretory activity) among participants with prevalent T2D of ≤8 years, whereas HbA1c levels were found to be higher among those with prevalent T2D of up to 15 years (33). In this study, the median duration of T2D was 7.22 (mean 8.19 years) years for the longstanding T2D category. This may indicate that diet may influence pancreatic cancer development among those with longstanding diabetes via sustained hyperinsulinemia and insulin resistance.

Multiple studies suggest an interrelationship between obesity and type 2 diabetes (T2D), both characterized by insulin resistance, hyperinsulinemia, hyperglycemia, and the promotion of a chronic inflammatory state, which may promote greater risk of pancreatic cancer (34–37). Conceptually, two different types of associations between glucose dysregulation and pancreatic cancer likely exist (38, 39). First, developing diabetes mellitus in the months prior to a pancreatic cancer diagnosis is common and likely due to dysregulation of endocrine and exocrine functions of the pancreas due to the developing malignancy in the organ, often described as a paraneoplastic process and referred to as “pancreatogenic” diabetes (40). This scenario is supported by preclinical studies and the observation that recent onset diabetes immediately prior to detection of pancreatic cancer often resolves following successful treatment of the cancer (41–43). In contrast, obesity promotes the metabolic syndrome and sustained insulin hypersecretion leading to T2D (34), while also promoting chronic systemic inflammation (44). The hyperglycemia and hyperinsulinemia of obesity and T2D may also act upon premalignant and malignant pancreatic ductal epithelial cells to support cancer stem cell functions linked to epithelial—mesenchymal transition and the carcinogenesis cascade (45).

The finding suggesting a protective association between higher dietary GI and GL and pancreatic cancer risk in normal-weight women is intriguing. It may suggest that in the absence of obesity and insulin resistance, higher glycemic exposures do not elevate insulin, inflammation or glucose, the mechanisms proposed to drive cancer risk. In addition, this finding may suggests that the composition of the diet was low in fat, as lower fat intake has previously been shown to be associated with lower pancreatic cancer risk (46), although early evaluation in WHI did not show protection of a low-fat dietary pattern in normal-weight women nor in a recent meta-analysis (47, 48). Furthermore, higher GL scores were associated with lower fat intake in this study. Also, the inverse associations may be partially explained by the properties of the dietary indices, especially the GL, as we found that higher GL scores were associated with lower BMI and with higher physical activity and higher total fiber intake.

A strength of this study is the application of novel food-based empirical hypothesis-oriented dietary patterns in a large, multiethnic sample. The prospective design allowed us to account for potential reverse causation bias that is not possible in the case–control design. The large sample size and long duration of follow-up allowed us to conduct subgroup analyses though the overall incidence of pancreatic cancer cases among women with recent onset and longstanding diabetes was low and power may have been limited. We were able to calculate the absolute risk of pancreatic cancer, which aligned well with the relative risks, and is more reflective of the clinical utility of the dietary pattern. Also, the self-reported T2D had been validated against diabetes medication use (19). However, our study has limitations as well. Though the measurement characteristics of the FFQ were previously assessed, it is appreciated that there is measurement error in diet assessment (49, 50), and that dietary patterns may change during the subjects' lifetime, although our group has shown that dietary intake was relatively stable in WHI (51). Data regarding pancreatic cancer subgroups (e.g., adenocarcinoma or pancreatic neuroendocrine tumor) were unavailable, but considering the relative preponderance of pancreatic ductal adenocarcinoma compared to other types of pancreatic cancer, this is expected to have a small effect, if any. We adjusted for a large number of potential confounding variables in the estimation of both the relative and absolute risk, but potential residual confounding and confounding by unmeasured variables remain possible.

In summary, our study does not support an overall association between the insulinemic, inflammatory, or glycemic potential of diet and risk of developing pancreatic cancer in this large cohort of postmenopausal women in the United States. However, these dietary patterns may influence pancreatic cancer development among women with longstanding diabetes. Future studies are warranted to confirm these associations in a larger sample of patients with longstanding diabetes and a larger number of pancreatic cancer cases. Also, the finding of a protective association for GI and GL in normal weight women warrants additional investigation.

P.A. Hart reports grants from NIH during the conduct of the study. L.F. Tinker reports other support from NIH during the conduct of the study. L. Qi reports grants from NIH during the conduct of the study. No disclosures were reported by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Q. Jin: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. P.A. Hart: Resources, formal analysis, investigation, writing–review and editing. N. Shi: Formal analysis, investigation, writing–review and editing. J.J. Joseph: Formal analysis, investigation, writing–review and editing. M. Donneyong: Formal analysis, investigation, writing–review and editing. D.L. Conwell: Formal analysis, writing–review and editing. S.K. Clinton: Formal analysis, investigation, writing–review and editing. Z. Cruz-Monserrate: Formal analysis, investigation, writing–review and editing. T.M. Brasky: Resources, formal analysis, investigation, writing–review and editing. L.F. Tinker: Resources, formal analysis, investigation, writing–review and editing. S. Liu: Resources, formal analysis, investigation, writing–review and editing. A.H. Shadyab: Formal analysis, investigation, writing–review and editing. C.A. Thomson: Formal analysis, investigation, writing–review and editing. L. Qi: Formal analysis, investigation, writing–review and editing. T. Rohan: Formal analysis, investigation, writing–review and editing. F.K. Tabung: Conceptualization, data curation, formal analysis, supervision, investigation, methodology, writing–original draft, writing–review and editing.

The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Siegel
RL
,
Miller
KD
,
Jemal
A
. 
Cancer statistics, 2020
.
CA Cancer J Clin
2020
;
70
:
7
30
.
2.
Tognon
G
,
Nilsson
LM
,
Lissner
L
,
Johansson
I
,
Hallmans
G
,
Lindahl
B
, et al
The mediterranean diet score and mortality are inversely associated in adults living in the subarctic region
.
J Nutr
2012
;
142
:
1547
53
.
3.
Zheng
J
,
Guinter
MA
,
Merchant
AT
,
Wirth
MD
,
Zhang
J
,
Stolzenberg-Solomon
RZ
, et al
Dietary patterns and risk of pancreatic cancer: a systematic review
.
Nutr Rev
2017
;
75
:
883
908
.
4.
Tabung
FK
,
Smith-Warner
SA
,
Chavarro
JE
,
Wu
K
,
Fuchs
CS
,
Hu
FB
, et al
Development and validation of an empirical dietary inflammatory index
.
J Nutr
2016
;
146
:
1560
70
.
5.
Tabung
FK
,
Wang
W
,
Fung
TT
,
Hu
FB
,
Smith-Warner
SA
,
Chavarro
JE
, et al
Development and validation of empirical indices to assess the insulinaemic potential of diet and lifestyle
.
Br J Nutr
2016
:
1
12
.
6.
Tabung
FK
,
Satija
A
,
Fung
TT
,
Clinton
SK
,
Giovannucci
EL
. 
Long-term change in both dietary insulinemic and inflammatory potential is associated with weight gain in adult women and men
.
J Nutr
2019
;
149
:
804
15
.
7.
Lee DH
L
,
Jun
Li
,
Yanping
Liu
,
Gang
Wu
,
Kana
Bhupathiraju
,
Shilpa
Rimm
, et al
Dietary inflammatory and insulinemic potential and risk of type 2 diabetes: results from three prospective u.s. cohort studies
.
Diabetes Care
2020
;
43
:
2675
83
.
8.
Jin
Q
,
Shi
N
,
Aroke
D
,
Lee
DH
,
Joseph
JJ
,
Donneyong
M
, et al
Insulinemic and inflammatory dietary patterns show enhanced predictive potential for type 2 diabetes risk in postmenopausal women
.
Diabetes Care
2021
;
44
:
707
14
.
9.
Design of the Women's Health Initiative clinical trial and observational study. The Women's Health Initiative Study Group
.
Control Clin Trials
1998
;
19
:
61
109
.
10.
Ma
Y
,
Hebert
JR
,
Li
W
,
Bertone-Johnson
ER
,
Olendzki
B
,
Pagoto
SL
, et al
Association between dietary fiber and markers of systemic inflammation in the Women's Health Initiative Observational Study
.
Nutrition
2008
;
24
:
941
9
.
11.
Petimar
J
,
Smith-Warner
SA
,
Fung
TT
,
Rosner
B
,
Chan
AT
,
Hu
FB
, et al
Recommendation-based dietary indexes and risk of colorectal cancer in the Nurses' Health Study and Health Professionals Follow-up Study
.
Am J Clin Nutr
2018
;
108
:
1092
103
.
12.
Zheng
J
,
Merchant
AT
,
Wirth
MD
,
Zhang
J
,
Antwi
SO
,
Shoaibi
A
, et al
Inflammatory potential of diet and risk of pancreatic cancer in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial
.
Int J Cancer
2018
;
142
:
2461
70
.
13.
Patterson
RE
,
Kristal
AR
,
Tinker
LF
,
Carter
RA
,
Bolton
MP
., 
Agurs-Collins T. Measurement characteristics of the Women's Health Initiative food frequency questionnaire
.
Ann Epidemiol
1999
;
9
:
178
87
.
14.
Schakel
SF
,
Sievert
YA
,
Buzzard
IM
. 
Sources of data for developing and maintaining a nutrient database
.
J Am Diet Assoc
1988
;
88
:
1268
71
.
15.
Meinhold
CL
,
Dodd
KW
,
Jiao
L
,
Flood
A
,
Shikany
JM
,
Genkinger
JM
, et al
Available carbohydrates, glycemic load, and pancreatic cancer: is there a link?
Am J Epidemiol
2010
;
171
:
1174
82
.
16.
Liu
S
,
Willett
WC
,
Stampfer
MJ
,
Hu
FB
,
Franz
M
,
Sampson
L
, et al
A prospective study of dietary glycemic load, carbohydrate intake, and risk of coronary heart disease in US women
.
Am J Clin Nutr
2000
;
71
:
1455
61
.
17.
Wang
W
,
Fung
TT
,
Wang
M
,
Smith-Warner
SA
,
Giovannucci
EL
,
Tabung
FK
. 
Association of the insulinemic potential of diet and lifestyle with risk of digestive system cancers in men and women
.
JNCI Cancer Spectr
2018
;
2
:
pky080
.
18.
Patterson
RE
,
Levy
L
,
Tinker
LF
,
Kristal
AR
. 
Evaluation of a simplified vitamin supplement inventory developed for the Women's Health Initiative
.
Public Health Nutr
1999
;
2
:
273
6
.
19.
Margolis
KL
,
Lihong
Q
,
Brzyski
R
,
Bonds
DE
,
Howard
BV
,
Kempainen
S
, et al
Validity of diabetes self-reports in the Women's Health Initiative: comparison with medication inventories and fasting glucose measurements
.
Clin Trials
2008
;
5
:
240
7
.
20.
Liu
S
,
Tinker
L
,
Song
Y
,
Rifai
N
,
Bonds
DE
,
Cook
NR
, et al
A prospective study of inflammatory cytokines and diabetes mellitus in a multiethnic cohort of postmenopausal women
.
Arch Intern Med
2007
;
167
:
1676
85
.
21.
Willett
WC
,
Howe
GR
,
Kushi
LH
. 
Adjustment for total energy intake in epidemiologic studies
.
Am J Clin Nutr
1997
;
65
:
1220S
8S
;
discussion 9S-31S
.
22.
Schernhammer
ES
,
Michaud
DS
,
Leitzmann
MF
,
Giovannucci
E
,
Colditz
GA
,
Fuchs
CS
. 
Gallstones, cholecystectomy, and the risk for developing pancreatic cancer
.
Br J Cancer
2002
;
86
:
1081
4
.
23.
Jiao
L
,
Chen
L
,
White
DL
,
Tinker
L
,
Chlebowski
RT
,
Van Horn
LV
, et al
Low-fat dietary pattern and pancreatic cancer risk in the women's health initiative dietary modification randomized controlled trial
.
J Natl Cancer Inst
2018
;
110
:
49
56
.
24.
Aitchison
J
,
Silvey
SD
. 
Maximum-likelihood estimation of parameters subject to restraints
.
The Annals of Mathematical Statistics
1958
;
29
:
813
28
.
25.
Shivappa
N
,
Bosetti
C
,
Zucchetto
A
,
Serraino
D
,
La Vecchia
C
,
Hébert
JR
. 
Dietary inflammatory index and risk of pancreatic cancer in an italian case-control study
.
Br J Nutr
2015
;
113
:
292
8
.
26.
Antwi
SO
,
Bamlet
WR
,
Pedersen
KS
,
Chaffee
KG
,
Risch
HA
,
Shivappa
N
, et al
Pancreatic cancer risk is modulated by inflammatory potential of diet and ABO genotype: a consortia-based evaluation and replication study
.
Carcinogenesis
2018
;
39
:
1056
67
.
27.
Bao
Y
,
Nimptsch
K
,
Wolpin
BM
,
Michaud
DS
,
Brand-Miller
JC
,
Willett
WC
, et al
Dietary insulin load, dietary insulin index, and risk of pancreatic cancer
.
Am J Clin Nutr
2011
;
94
:
862
8
.
28.
Turati
F
,
Galeone
C
,
Augustin
LSA
,
La Vecchia
C
Glycemic index, glycemic load and cancer risk: an updated meta-analysis
.
Nutrients
2019
;
11
:
2342
.
29.
Aune
D
,
Chan
DS
,
Vieira
AR
,
Navarro Rosenblatt
DA
,
Vieira
R
,
Greenwood
DC
, et al
Dietary fructose, carbohydrates, glycemic indices and pancreatic cancer risk: a systematic review and meta-analysis of cohort studies
.
Ann Oncol
2012
;
23
:
2536
46
.
30.
Simon
MS
,
Shikany
JM
,
Neuhouser
ML
,
Rohan
T
,
Nirmal
K
,
Cui
Y
, et al
Glycemic index, glycemic load, and the risk of pancreatic cancer among postmenopausal women in the women's health initiative observational study and clinical trial
.
Cancer Causes Control
2010
;
21
:
2129
36
.
31.
Quoc Lam
B
,
Shrivastava
SK
,
Shrivastava
A
,
Shankar
S
,
Srivastava
RK
. 
The impact of obesity and diabetes mellitus on pancreatic cancer: molecular mechanisms and clinical perspectives
.
J Cell Mol Med
2020
;
24
:
7706
16
.
32.
Abbruzzese
JL
,
Andersen
DK
,
Borrebaeck
CAK
,
Chari
ST
,
Costello
E
,
Cruz-Monserrate
Z
, et al
The interface of pancreatic cancer with diabetes, obesity, and inflammation: research gaps and opportunities: summary of a national institute of diabetes and digestive and kidney diseases workshop
.
Pancreas
2018
;
47
:
516
25
.
33.
Hu
Y
,
Zhang
X
,
Ma
Y
,
Yuan
C
,
Wang
M
,
Wu
K
, et al
Incident type 2 diabetes duration and cancer risk: a prospective study in two US cohorts
.
J Natl Cancer Inst
2020
;
113
:
381
9
.
34.
Eibl
G
,
Cruz-Monserrate
Z
,
Korc
M
,
Petrov
MS
,
Goodarzi
MO
,
Fisher
WE
, et al
Diabetes mellitus and obesity as risk factors for pancreatic cancer
.
J Acad Nutr Diet
2018
;
118
:
555
67
.
35.
Ben
Q
,
Xu
M
,
Ning
X
,
Liu
J
,
Hong
S
,
Huang
W
, et al
Diabetes mellitus and risk of pancreatic cancer: a meta-analysis of cohort studies
.
Eur J Cancer
2011
;
47
:
1928
37
.
36.
Song
S
,
Wang
B
,
Zhang
X
,
Hao
L
,
Hu
X
,
Li
Z
, et al
Long-term diabetes mellitus is associated with an increased risk of pancreatic cancer: a meta-analysis
.
PLoS One
2015
;
10
:
e0134321
.
37.
Batabyal
P
,
Vander Hoorn
S
,
Christophi
C
,
Nikfarjam
M
. 
Association of diabetes mellitus and pancreatic adenocarcinoma: a meta-analysis of 88 studies
.
Ann Surg Oncol
2014
;
21
:
2453
62
.
38.
Molina-Montes
E
,
Coscia
C
,
Gomez-Rubio
P
,
Fernandez
A
,
Boenink
R
,
Rava
M
, et al
Deciphering the complex interplay between pancreatic cancer, diabetes mellitus subtypes and obesity/BMI through causal inference and mediation analyses
.
Gut
2020
;
70
:
319
29
.
39.
Yuan
C
,
Babic
A
,
Khalaf
N
,
Nowak
JA
,
Brais
LK
,
Rubinson
DA
, et al
Diabetes, weight change, and pancreatic cancer risk
.
JAMA Oncol
2020
;
6
:
e202948
.
40.
Hart
PA
,
Bellin
MD
,
Andersen
DK
,
Bradley
D
,
Cruz-Monserrate
Z
,
Forsmark
CE
, et al
Type 3c (pancreatogenic) diabetes mellitus secondary to chronic pancreatitis and pancreatic cancer
.
Lancet Gastroenterol Hepatol
2016
;
1
:
226
37
.
41.
Mizuno
S
,
Nakai
Y
,
Isayama
H
,
Yanai
A
,
Takahara
N
,
Miyabayashi
K
, et al
Risk factors and early signs of pancreatic cancer in diabetes: screening strategy based on diabetes onset age
.
J Gastroenterol
2013
;
48
:
238
46
.
42.
Pannala
R
,
Leirness
JB
,
Bamlet
WR
,
Basu
A
,
Petersen
GM
,
Chari
ST
. 
Prevalence and clinical profile of pancreatic cancer-associated diabetes mellitus
.
Gastroenterology
2008
;
134
:
981
7
.
43.
Gardner
TB
,
Hessami
N
,
Smith
KD
,
Ripple
GH
,
Barth
RJ
,
Klibansky
DA
, et al
The effect of neoadjuvant chemoradiation on pancreatic cancer-associated diabetes mellitus
.
Pancreas
2014
;
43
:
1018
21
.
44.
Tuomisto
AE
,
Makinen
MJ
,
Vayrynen
JP
. 
Systemic inflammation in colorectal cancer: underlying factors, effects, and prognostic significance
.
World J Gastroenterol
2019
;
25
:
4383
404
.
45.
Rahn
S
,
Zimmermann
V
,
Viol
F
,
Knaack
H
,
Stemmer
K
,
Peters
L
, et al
Diabetes as risk factor for pancreatic cancer: hyperglycemia promotes epithelial-mesenchymal-transition and stem cell properties in pancreatic ductal epithelial cells
.
Cancer Lett
2018
;
415
:
129
50
.
46.
Sanchez
GV
,
Weinstein
SJ
,
Stolzenberg-Solomon
RZ
. 
Is dietary fat, vitamin D, or folate associated with pancreatic cancer?
Mol Carcinog
2012
;
51
:
119
27
.
47.
Nadella
S
,
Burks
J
,
Al-Sabban
A
,
Inyang
G
,
Wang
J
,
Tucker
RD
, et al
Dietary fat stimulates pancreatic cancer growth and promotes fibrosis of the tumor microenvironment through the cholecystokinin receptor
.
Am J Physiol Gastrointest Liver Physiol
2018
;
315
:
G699
712
.
48.
Shen
QW
,
Yao
QY
. 
Total fat consumption and pancreatic cancer risk: a meta-analysis of epidemiologic studies
.
Eur J Cancer Prev
2015
;
24
:
278
85
.
49.
Neuhouser
ML
,
Tinker
L
,
Shaw
PA
,
Schoeller
D
,
Bingham
SA
,
Horn
LV
, et al
Use of recovery biomarkers to calibrate nutrient consumption self-reports in the Women's Health Initiative
.
Am J Epidemiol
2008
;
167
:
1247
59
.
50.
Prentice
RL
,
Mossavar-Rahmani
Y
,
Huang
Y
,
Van Horn
L
,
Beresford
SA
,
Caan
B
, et al
Evaluation and comparison of food records, recalls, and frequencies for energy and protein assessment by using recovery biomarkers
.
Am J Epidemiol
2011
;
174
:
591
603
.
51.
Tabung
FK
,
Steck
SE
,
Zhang
J
,
Ma
Y
,
Liese
AD
,
Tylavsky
FA
, et al
Longitudinal changes in the dietary inflammatory index: an assessment of the inflammatory potential of diet over time in postmenopausal women
.
Eur J Clin Nutr
2016
;
70
:
1374
80
.