Abstract
Background and Aims
We examined multiple dietary patterns in relation to total digestive system cancer (DSC) incidence and death.
Method
A total of 213,038 health professionals from the Health Professionals Follow-up Study (1986–2016) and the Nurses’ Health Study (NHS) (1986–2018) and the NHS II (1991–2017) with no cancer diagnosis at baseline were analyzed. DSC incidence and death were estimated using time varying Cox proportional hazards regression models.
Results
During up to 32 years of follow-up, 5724 DSC cases accrued. Adherence to eight healthy diet patterns were associated with a 7–13% lower risk of DSC, particularly for digestive tract cancers. An inverse association with gastrointestinal tract cancer was also shown for all pattern scores except alternative Mediterranean diet (AMED) and healthy Plant Dietary Index, with hazard ratios (HRs) between 0.84 and 0.89. Inverse associations were shown for the rEDIH (HR for 90th versus 10th percentile 0.64, 95% confidence interval (CI) 0.47 to 0.87) and the empirical dietary index associated with lower inflammation (rEDIP) (0.53, 95% CI 0.39 to 0.72) for stomach cancer, and for the rEDIP (0.58, 95% CI 0.37 to 0.92) for small intestine cancer. Among accessory cancers, the Alternate Healthy Eating Index-2010 (AHEI-2010), AMED, and diabetes risk reduction diet were associated with a 43–51% lower risk of liver cancer. The rEDIH, rEDIP, and the AHEI-2010 were inversely associated with risk of digestive system fatal cancer.
Conclusions
Adherence to healthy diets was associated with a lower risk of incident and fatal DSC, although the magnitude of the association varied slightly among the patterns.
Keywords: dietary pattern, digestive system cancer, gastrointestinal cancer, superior diet, prospective cohort
Introduction
Digestive system cancers account for approximately 40% of all cancer-related deaths and 30% of the global cancer incidence (1). According to epidemiologic studies, modifiable risk factors contribute to more than half of all digestive system cancers. Some modifiable factors such as obesity, physical activity, diabetes, and aspirin use influence multiple digestive system cancers (2). Diet may be one of broadly acting factors and could influence the entire digestive system cancer. Particularly, research on dietary patterns evaluating the whole diet could capture the overall effect of diet on chronic disease risk because of the constituents of multiple foods and nutrients.
Multiple dietary patterns have been associated with individual cancers that occur in the digestive system. Dietary recommendations for Americans such as the Mediterranean style diet(3), Alternate Healthy Eating Index (AHEI) (4), and the Dietary Approaches to Stop Hypertension (DASH) (5), and the diet recommended by the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) (6) have been suggested to be favorable for preventing individual cancers of the digestive system such as those of the esophagus, stomach, colorectum, liver, and pancreas. Some studies have shown that diets reflecting hyperinsulinemia or inflammation are associated with digestive system cancer(7) and individual cancers (8, 9). However, no study has systematically compared their relative efficacy superior for preventing total digestive system cancer and death in the same dataset.
In this context, this study aimed to examine how eight dietary patterns that were promoted in the recommended dietary guidelines and previously created for predicting chronic diseases or common biological pathways are associated with digestive system cancer incidence and mortality.
Methods
Study population
Data from three ongoing prospective cohorts were used for the study, which include the Health Professionals Follow-up Study (HPFS), Nurses’ Health Study (NHS), and Nurses’ Health Study II (NHS II). The HPFS enrolled 51529 male health professionals ages 40 to 75 in 1989. The NHS was established in 1976 and enrolled 121700 female nurses ages 30 to 55 at baseline. As a younger cohort, the NHS II enrolled 116429 female nurses ages 25 to 42 in 1989. In each cohort, participants completed the questionnaires for collecting and updating their lifestyle and medical history biennially. Participants provided dietary data using validated semi-quantitative food frequency questionnaires (FFQ) every four years. The follow-up rates were over 90% in all three cohorts. In this study, we used 1986 for the HPFS, 1984 for the NHS, and 1991 for the NHS II as the baseline when dietary data using an FFQ with >130 items were first assessed.
We excluded participants who had missing data on dietary pattern scores, those with implausible energy intake (<800 or >4200 kcal/d for men and <500 or >3500 kcal/d for women), those with a history of cancer (except non-melanoma skin cancer) or inflammatory bowel disease at baseline, and those with extreme BMI (<15 or > 50 kg/m2). Participants were censored when they were 80 years old to limit potential reverse causation. The final analysis included 213002 participants (46274 men and 166728 women) (see Figure S1, Supplemental Digital Content 1). The institutional review boards of the Brigham and Women’s Hospital, Harvard T.H. Chan School of Public Health, and participating registries have approved the study protocol.
Dietary assessment
Participants were asked to describe their food consumption frequency and portion size in the FFQ. The nutrient intake was calculated as the sum of the nutrient content of each contributing food multiplied by its consumption frequency. The validity and reliability of self-reported food and nutrient measures were evaluated. Comparing the estimates from FFQ with those from multiple one-week diet records, the average correlation coefficient for food was 0.66 in the NHS and 0.63 in the HPFS, and for nutrients was 0.53 in the NHS and 0.66 in the HPFS (10, 11).
Each dietary pattern was described in detail (see Table S1, Supplemental Digital Content 1) (12). Based on the US dietary guidelines, the AHEI-2010 score consists of 11 dietary components and scores zero to 10 points for each component based on portion size (13). The Alternate Mediterranean Diet (AMED) score includes nine dietary components and scores zero or one point for each component comparing with the median intake of population (14). The healthful plant-based diet index (hPDI) score gives positive weights to healthy plant foods and negative weights to animal and less healthy plant foods contains 18 food groups and each group is assigned from one to five points according to its intake quintile (15). The DASH score is comprised of eight components and scores one to five points based on its intake quintile (16). The Diabetes Risk Reduction Diet (DRRD) score consists of nine components associated with type 2 diabetes and scores one to five points for each component (17). Based on the dietary recommendation for preventing cancer, the 2018 WCRF/AICR dietary score includes six dietary components scoring zero to one point depending on the consumption level (18). To reflect the long-term dietary hyperinsulinemia potential, the empirical dietary index for hyperinsulinemia (EDIH) was developed to predict fasting plasma C-peptide (19). The empirical dietary inflammatory pattern (EDIP) was established to predict inflammation markers such as plasma interleukin-6, C-reactive protein, and tumor necrosis factor 𝛼 receptor (20). To facilitate better comparison across pattern scores, for which high scores are considered healthful, we reversed scores for EDIH and EDIP (termed rEDIH and rEDIP) to make higher levels as healthier. The correlation coefficient for dietary pattern scores ranges from 0.08 (rEDIP and WCRF/AICR) to 0.77 (AHEI and DRRD) in the pooled cohorts (see Figure S2, Supplemental Digital Content 1). The correlation coefficient for dietary pattern scores over 4-year follow-up period ranges from 0.66 to 0.78 and slightly attenuated from 0.51 to 0.59 over 12-year follow-up period.
Outcome definition
The outcomes were incidence of digestive system cancer, gastrointestinal tract cancer, accessory digestive system cancer, and death caused by digestive system cancer. Digestive system cancers were defined as cancers in the mouth and throat, gastrointestinal tract, and its accessory digestive system. Gastrointestinal tract cancers included cancers from the esophagus, stomach, small intestine, to colorectum. Accessory digestive system cancers included cancers in the pancreas, biliary tract, and liver. We also show results for digestive system cancers excluding colorectal cancer, which has been studied previously (21).
Written informed consent was obtained from the participants who reported new diagnosis of cancer to review their medical records and pathological reports. The confirmation rate was estimated to be over 90% for cancer (22). Confirmed cases were based on medical record review or rarely from death certificates if medical records could not be obtained. Deaths were identified through searches in the National Death Index and through the next-of-kin or postal office when questionnaires were mailed. Death ascertainment using National Death Index was reported to have a high sensitivity (98%) and specificity (100%) (23, 24). Physicians who were blinded to the exposure information reviewed the medical records to confirm the diagnosis or determine the cause of death after permission was obtained from the next-of-kin or other contact person.
Covariates assessment
We collected and updated information from baseline and biennial questionnaires for family history of cancer, type 2 diabetes, physical activity, body mass index (BMI), cigarette smoking (status, pack-years, and time since quitting), multivitamin use, regular aspirin use, regular non-steroidal anti-inflammatory drugs use, and postmenopausal hormone use for women. BMI was calculated from the height and weight reported every 2 years. Smoking status was updated biennially with pack-years smoked. Physical activity was converted using a metabolic equivalent of task (MET) accounting for types and intensity of physical activity. Regular use of aspirin was defined as use of 2 or more standard tablets (325 mg) per week and regular use of other nonsteroidal anti-inflammatory drug (NSAID) was defined as 2 or more times per week. Menopausal status and post-menopausal hormone use were obtained from the NHS and NHS II. Energy intake and alcohol intake were obtained from cumulative average of dietary measures based on the FFQs conducted every 4 years.
Statistical analysis
Person-time of follow-up accumulated from baseline until the occurrence of the outcome, death, or the end of follow-up (January 31, 2016 for HPFS; June 30, 2018 for NHS; June 30, 2017 for NHS II), whichever came first. Cumulative averages of dietary pattern scores (average of all available FFQs up to the time at risk) as the exposure were used for the main analysis to capture long-term intake. To limit the potential influence of outliers, pattern scores were winsorized at the 0.5 and 99.5 percentiles (25). The residual method was used to compute energy-adjusted scores by fitting each pattern score against the total energy intake (26). We used non-missing values from the preceding data cycle to fill in missing dietary variables and covariates.
We assessed the relationship of energy-adjusted pattern scores to each other using the Spearman correlation coefficients. Time-dependent Cox proportional hazards regression models were applied to estimate the risk of digestive system cancers and death across dietary pattern scores. We estimated the associations using each pattern score as a continuous variable standardized by its increment from the 10th to 90th percentile. The potential non-linear relationship between dietary pattern scores and digestive system cancer was examined by restricted cubic splines (27). We further estimated the risk of digestive system cancer according to the quintile of pattern scores and linear trend was tested by modeling the median value of each quintile as a continuous variable. The proportional hazards assumption was tested by adding an interaction term between each exposure variable and time scale age.
Analyses were performed in each cohort as well as the pooled data of three cohorts. We detected no statistically significant heterogeneity across cohort (all P heterogeneity ≥ 0.19). All the analyses were stratified by age in months and calendar year of questionnaire. In the pooled data, the model was additionally stratified by cohort. Multivariable models were adjusted for family history of cancer (yes or no), physical activity (<3.0, 3.0–8.9, 9.0–17.9, 18.0–26.9, 27.0–41.9, or ≥42 MET-h/week), cigarette smoking status (never, former quitting ≥10 y, former quitting <10 y, current), cigarette smoking packyears (0, 1–4, 5–14, 15–24, or ≥25 packyears), multivitamin use (yes or no), regular aspirin use (yes or no), regular non-steroidal anti-inflammatory drugs use (yes or no), postmenopausal hormone use (premenopausal, never, former, or current use) for women, and total energy intake (quintiles). The model was adjusted for alcohol consumption (<5.0, 5.0–14.9, or ≥15.0 g/d) for dietary patterns that did not include alcohol, such as DASH, hPDI, and DRRD.
Stratified analyses were conducted by age, BMI, sex, smoking status, alcohol, physical activity, and aspirin use. Potential interaction was assessed by the Wald test (binary variable) or likelihood ratio test (categorical variable). To better understand possible latency, we investigated dietary pattern scores at various intervals, including updated scores assessed at the most recent FFQ and patterns with different lags (4-year, 8-year, 12-year, or 16-year). For example, in a 4-y lagged analysis, the dietary pattern scores cumulatively updated until the 1990 FFQ (average of dietary pattern scores from 1986 and 1990 FFQ) was used as the exposure for the follow up period between 1994 and 1997. Longer time lags give greater weight to diet in the more distant past (for e.g., dietary effects mostly at earlier stages of carcinogenesis), and reduce the likelihood of reverse causation. In a sensitivity analysis, we excluded participants who had type 2 diabetes and major cardiovascular diseases (angina, transient ischemic attack, or coronary artery bypass graft surgery) at baseline and stopped updating the dietary information after these chronic diseases occurred. Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC). We set statistical significance at a two-sided P value <0.05.
Results
During up to 32 years of follow-up, we documented 5724 incident digestive system cancer cases from the pooled cohort (2034 cases in the HPFS, 2980 cases in NHS, and 710 cases in NHS II) among 213002 participants. Participants in the highest quintile of dietary pattern scores were older, more likely to be physically active, take multivitamins, use postmenopausal hormones (women), and were less likely to be a current smoker (except reversed EDIP [rEDIP]), have type 2 diabetes (except WCRF/AICR) (Table 1). Participants who scored high for reversed EDIH (rEDIH) and rEDIP were more likely to drink alcohol while those who scored high for WCRF/AICR were less likely to drink alcohol.
Table 1.
Age-standardized characteristics of study population in the lowest and highest quintiles of energy-adjusted dietary patterns during the follow-up in the pooled cohort*
| Quintile | Q1 | Q5 | Q1 | Q5 | Q1 | Q5 | Q1 | Q5 |
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| rEDIH | rEDIP | AHEI | AMED | |||||
|
| ||||||||
| Age†, year | 53.5 (11.3) | 58.6 (11.7) | 55.2 (12.1) | 56.4 (11.4) | 53.3 (11.9) | 59.2 (11.2) | 54.8 (11.7) | 57.1 (11.9) |
| Family history of cancer, % | 37.5 | 43.1 | 39.0 | 41.4 | 37.0 | 43.7 | 40.3 | 39.9 |
| Type 2 diabetes, % | 12.1 | 3.1 | 11.9 | 3.2 | 7.2 | 5.3 | 7.3 | 5.0 |
| Physical activity, MET-h/week | 17.2 (19) | 28 (26.3) | 18.4 (20.4) | 24.8 (24) | 15.1 (17.1) | 29.8 (26.7) | 15.7 (17.6) | 28 (25.5) |
| Body mass index, kg/m2 | 27.3 (5.5) | 23.9 (3.6) | 27.1 (5.5) | 24.3 (3.9) | 25.9 (5) | 24.6 (4.1) | 26.1 (5.1) | 24.6 (4) |
| Alcohol consumption, g/day | 4.4 (8.6) | 9.9 (12.1) | 3.1 (7) | 12.1 (13.5) | 5.9 (12.7) | 6.8 (7.3) | 5.8 (11.8) | 6.6 (7.7) |
| Current smoking, % | 21.3 | 16.9 | 17.6 | 22.4 | 24.4 | 13.3 | 24.5 | 14.1 |
| Regular aspirin use‡, % | 35.7 | 35.7 | 35.5 | 37.6 | 35.1 | 35.7 | 34.7 | 36.7 |
| Regular NSAIDs use§, % | 35.9 | 34.4 | 35.4 | 35.6 | 33.1 | 35.0 | 35.5 | 33.0 |
| Multivitamin use, % | 43.6 | 56.2 | 45.8 | 53.5 | 42.8 | 57.5 | 44.0 | 56.3 |
| Postmenopausal hormone use, % | 12.8 | 14.3 | 12.8 | 14.5 | 12.3 | 15.3 | 11.9 | 15.6 |
| Total energy intake, kcal/d | 1913 (533) | 1923 (487) | 1893 (529) | 1889 (495) | 1835 (485) | 1835 (496) | 1822 (514) | 1831 (444) |
|
| ||||||||
| DASH | hPDI | DRRD | WCRF/AICR | |||||
|
| ||||||||
| Age, year | 54.5 (11.4) | 57.4 (12.2) | 54.8 (11.9) | 57.3 (11.9) | 54.7 (11.8) | 57.3 (11.9) | 53.8 (11.6) | 57.9 (12.1) |
| Family history of cancer, % | 39.8 | 39.8 | 40.6 | 39.7 | 40.1 | 40.2 | 38.8 | 40.5 |
| Type 2 diabetes, % | 7.1 | 5.5 | 6.8 | 6.3 | 7.8 | 5.2 | 4.9 | 7.6 |
| Physical activity, MET-h/week | 15.3 (17.1) | 29.1 (26.6) | 16.5 (17.8) | 27.9 (26.5) | 15.8 (17.4) | 28.4 (26.3) | 16.5 (17.8) | 27.5 (26.6) |
| Body mass index, kg/m2 | 26 (5.1) | 24.6 (4.1) | 25.9 (5) | 24.9 (4.3) | 26.1 (5.2) | 24.7 (4.1) | 25.2 (4.5) | 25.3 (4.7) |
| Alcohol consumption, g/day | 6.7 (11.4) | 5.3 (7.9) | 5.2 (8.8) | 6.6 (10.1) | 5 (9.3) | 6.6 (9.4) | 9.4 (11.8) | 3.3 (6.5) |
| Current smoking, % | 28.0 | 11.9 | 20.6 | 16.0 | 21.3 | 15.1 | 26.0 | 12.5 |
| Regular aspirin use, % | 35.3 | 35.4 | 35.5 | 35.9 | 34.5 | 36.9 | 36.2 | 34.2 |
| Regular NSAIDs use, % | 35.9 | 31.6 | 34.7 | 33.4 | 34.6 | 33.8 | 35.4 | 31.5 |
| Multivitamin use, % | 41.3 | 58.2 | 45.5 | 55.1 | 43.5 | 56.8 | 44.7 | 56.1 |
| Postmenopausal hormone use, % | 11.9 | 15.3 | 12.0 | 15.4 | 12.1 | 15.7 | 12.7 | 14.8 |
| Total energy intake, kcal/d | 1840 (516) | 1843 (451) | 1841 (462) | 1833 (507) | 1824 (497) | 1826 (483) | 1804 (466) | 1801 (498) |
MET, metabolic equivalent for task; NSAIDs, nonsteroidal anti-inflammatory drugs; rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score; NHS, Nurses’ Health Study; HPFS, Health Professional Follow-up Study.
Values are means (standard deviations) for continuous variables and percentages for categorical variables if not specified otherwise.
All variables are standardized to the age distribution of the study population, except for age.
Regular users are defined as participants who take at least 2 tables of aspirin (325 mg/tablet) per week in the NHS and at least 2 times per week in the HPFS and NHSII.
Regular users are defined as participants who take at least 2 times per week.
Although the point estimates for the associations between dietary patterns and digestive system cancer varied by cohort, the general direction and magnitude ranking for the associations were the same in each cohort (see Table S2, Supplemental Digital Content 1). In age-adjusted and multivariable-adjusted model, inverse associations with digestive system cancer were observed for all pattern scores in the pooled cohort, with hazard ratios (HRs) between 0.87 and 0.93. (Table 2, Fig 1, see Table S3, Supplemental Digital Content 1). After adjusting for BMI, the association with digestive system cancer, particularly for rEDIH and rEDIP, was slightly attenuated, but the trend of association was similar. An inverse association with gastrointestinal tract cancer was also shown for all pattern scores except AMED and hPDI, with HRs between 0.84 and 0.89. The magnitudes of the hazard ratios were generally similar for digestive system cancer excluding colorectal cancer, but the confidence intervals slightly exceeded 1.00. None of dietary patterns was associated with total accessory digestive system cancer after adjusting for potential confounders. Inverse associations with digestive system cancer-related death was shown only for rEDIH (HR for 90th versus 10th percentile 0.84, 95% confidence interval (CI) 0.76 to 0.92), rEDIP (0.84, 95% CI 0.77 to 0.92) and AHEI-2010 (0.90, 95% CI 0.81 to 0.99). Also, inverse associations with individual cancers were observed for pattern scores (Table 3, Fig 1). An inverse association with stomach cancer was shown for the rEDIH (0.64, 95% CI 0.47 to 0.87) and rEDIP (0.53, 95% CI 0.39 to 0.72) and an inverse association with small intestine cancer was observed only for the rEDIP (0.58, 95% CI 0.37 to 0.92). The AHEI-2010, AMED, and DRRD scores had a 43–51% lower risk of liver cancer.
Table 2.
Association between cumulative average dietary patterns (comparing the 90th to 10th percentile) and digestive system cancer and its related death in the pooled cohort. Values are hazard ratios (95% confidence intervals) stated otherwise.
| Outcomes | Digestive system cancer | Digestive system cancer excluding CRC | Gastrointestinal tract cancer | Accessory digestive system cancer | Digestive system cancer death | |
|---|---|---|---|---|---|---|
|
| ||||||
| Cases | 5724 | 2376 | 4015 | 1291 | 3029 | |
| Person-year | 5238166 | 5255624 | 5245284 | 5272667 | 5267861 | |
|
| ||||||
| rEDIH | MV | 0.87 (0.81, 0.93) | 0.92 (0.83, 1.02) | 0.84 (0.77, 0.91) | 0.94 (0.81, 1.09) | 0.84 (0.76, 0.92) |
| MV+BMI | 0.92 (0.86, 0.99) | 0.99 (0.89, 1.11) | 0.89 (0.82, 0.97) | 1.02 (0.88, 1.19) | 0.90 (0.82, 0.99) | |
|
| ||||||
| rEDIP | MV | 0.90 (0.84, 0.96) | 0.92 (0.83, 1.02) | 0.87 (0.80, 0.94) | 0.92 (0.80, 1.06) | 0.84 (0.77, 0.92) |
| MV+BMI | 0.94 (0.88, 1.01) | 0.98 (0.88, 1.09) | 0.92 (0.84, 0.99) | 0.98 (0.85, 1.13) | 0.89 (0.81, 0.97) | |
|
| ||||||
| AHEI-2010 | MV | 0.91 (0.84, 0.97) | 0.93 (0.83, 1.05) | 0.89 (0.82, 0.98) | 1.02 (0.87, 1.19) | 0.90 (0.81, 0.99) |
| MV+BMI | 0.92 (0.85, 0.99) | 0.95 (0.85, 1.07) | 0.91 (0.83, 0.99) | 1.04 (0.89, 1.22) | 0.91 (0.83, 1.01) | |
|
| ||||||
| AMED | MV | 0.90 (0.84, 0.97) | 0.90 (0.80, 1.01) | 0.92 (0.84, 1.01) | 0.93 (0.80, 1.10) | 0.94 (0.84, 1.04) |
| MV+BMI | 0.92 (0.85, 0.99) | 0.92 (0.82, 1.03) | 0.94 (0.86, 1.03) | 0.96 (0.81, 1.12) | 0.96 (0.86, 1.06) | |
|
| ||||||
| DASH | MV | 0.90 (0.83, 0.97) | 1.01 (0.89, 1.13) | 0.86 (0.78, 0.94) | 1.09 (0.92, 1.28) | 0.97 (0.87, 1.08) |
| MV+BMI | 0.91 (0.84, 0.98) | 1.02 (0.91, 1.15) | 0.87 (0.79, 0.95) | 1.10 (0.94, 1.30) | 0.98 (0.88, 1.09) | |
|
| ||||||
| hPDI | MV | 0.93 (0.86, 1.00) | 0.90 (0.81, 1.01) | 0.95 (0.88, 1.04) | 0.90 (0.77, 1.04) | 0.91 (0.82, 1.00) |
| MV+BMI | 0.94 (0.87, 1.01) | 0.92 (0.82, 1.03) | 0.97 (0.89, 1.05) | 0.91 (0.78, 1.06) | 0.92 (0.83, 1.02) | |
|
| ||||||
| DRRD | MV | 0.90 (0.84, 0.97) | 0.98 (0.87, 1.10) | 0.87 (0.80, 0.95) | 1.04 (0.88, 1.21) | 0.91 (0.82, 1.01) |
| MV+BMI | 0.92 (0.85, 0.99) | 1.00 (0.89, 1.12) | 0.89 (0.81, 0.97) | 1.06 (0.90, 1.24) | 0.93 (0.84, 1.03) | |
|
| ||||||
| WCRF/AICR | MV | 0.92 (0.86, 0.99) | 1.03 (0.92, 1.15) | 0.89 (0.81, 0.97) | 1.10 (0.95, 1.29) | 1.03 (0.93, 1.13) |
| MV+BMI | 0.91 (0.85, 0.98) | 1.02 (0.91, 1.14) | 0.87 (0.80, 0.95) | 1.09 (0.94, 1.27) | 1.01 (0.92, 1.12) | |
rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score; CRC, colorectal cancer; MV, multivariable; BMI, body mass index.
Analyses were stratified by age, calendar year, and cohort. MV model was adjusted for family history of cancer, physical activity, cigarette smoking status, cigarette smoking packyears, multivitamin use, regular aspirin use, regular non-steroidal anti-inflammatory drugs use, postmenopausal hormone use for women, and total energy intake. For DASH, hPDI, and DRRD, alcohol consumption was additionally adjusted for.
Fig.1. Multivariable-adjusted association between cumulative average dietary patterns (comparing the 90th to 10th percentile) and digestive system cancer and its related death in the pooled cohort.
The models were adjusted for the same list of covariates as in Table 2. rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score
Table 3.
Association between cumulative average dietary patterns (comparing the 90th to 10th percentile) and individual digestive system cancers in the pooled cohort. Values are hazard ratios (95% confidence intervals) stated otherwise.
| Outcomes | Esophageal cancer | Stomach cancer | Small intestine cancer | Pancreatic cancer | Biliary tract cancer | Liver cancer | |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Cases | 267 | 273 | 117 | 951 | 200 | 145 | |
| Person-year | 5274809 | 5274852 | 5274894 | 5273677 | 5275147 | 5275299 | |
|
| |||||||
| rEDIH | Age | 1.15 (0.84, 1.57) | 0.59 (0.44, 0.81) | 0.86 (0.53, 1.38) | 0.95 (0.81, 1.13) | 0.76 (0.53, 1.09) | 0.71 (0.46, 1.08) |
| MV | 1.36 (0.99, 1.86) | 0.64 (0.47, 0.87) | 0.74 (0.46, 1.19) | 0.99 (0.84, 1.18) | 0.81 (0.56, 1.16) | 0.80 (0.52, 1.22) | |
|
| |||||||
| rEDIP | Age | 1.43 (1.06, 1.93) | 0.51 (0.38, 0.69) | 0.65 (0.41, 1.03) | 0.97 (0.82, 1.14) | 0.85 (0.59, 1.21) | 0.74 (0.49, 1.13) |
| MV | 1.37 (1.02, 1.84) | 0.53 (0.39, 0.72) | 0.58 (0.37, 0.92) | 0.95 (0.81, 1.13) | 0.88 (0.62, 1.26) | 0.78 (0.52, 1.18) | |
|
| |||||||
| AHEI-2010 | Age | 0.68 (0.49, 0.94) | 0.75 (0.54, 1.03) | 1.26 (0.77, 2.05) | 0.99 (0.83, 1.17) | 0.85 (0.59, 1.24) | 0.43 (0.27, 0.69) |
| MV | 0.94 (0.67, 1.33) | 0.85 (0.61, 1.19) | 1.08 (0.64, 1.81) | 1.12 (0.93, 1.35) | 0.98 (0.66, 1.46) | 0.50 (0.31, 0.81) | |
|
| |||||||
| AMED | Age | 0.71 (0.51, 0.99) | 0.84 (0.61, 1.16) | 1.05 (0.63, 1.73) | 0.87 (0.73, 1.04) | 0.83 (0.57, 1.22) | 0.46 (0.29, 0.72) |
| MV | 1.09 (0.76, 1.55) | 0.97 (0.70, 1.37) | 0.88 (0.52, 1.50) | 0.99 (0.82, 1.19) | 0.97 (0.65, 1.45) | 0.57 (0.35, 0.92) | |
|
| |||||||
| DASH | Age | 0.68 (0.49, 0.95) | 0.83 (0.60, 1.14) | 0.99 (0.60, 1.64) | 0.98 (0.82, 1.17) | 0.87 (0.59, 1.27) | 0.54 (0.34, 0.86) |
| MV | 1.17 (0.82, 1.68) | 0.95 (0.67, 1.34) | 0.84 (0.49, 1.42) | 1.17 (0.97, 1.42) | 1.03 (0.69, 1.55) | 0.70 (0.43, 1.14) | |
|
| |||||||
| hPDI | Age | 0.82 (0.59, 1.13) | 1.07 (0.78, 1.47) | 0.83 (0.51, 1.36) | 0.86 (0.73, 1.03) | 0.85 (0.59, 1.24) | 0.57 (0.36, 0.90) |
| MV | 0.98 (0.70, 1.38) | 1.20 (0.87, 1.66) | 0.72 (0.43, 1.19) | 0.93 (0.78, 1.11) | 0.95 (0.65, 1.40) | 0.64 (0.40, 1.03) | |
|
| |||||||
| DRRD | Age | 0.94 (0.67, 1.30) | 0.71 (0.52, 0.99) | 1.18 (0.72, 1.94) | 1.02 (0.86, 1.22) | 0.91 (0.62, 1.32) | 0.42 (0.27, 0.67) |
| MV | 1.21 (0.85, 1.73) | 0.81 (0.58, 1.13) | 1.02 (0.60, 1.71) | 1.14 (0.95, 1.38) | 1.04 (0.70, 1.54) | 0.49 (0.30, 0.79) | |
|
| |||||||
| WCRF/AICR | Age | 0.58 (0.42, 0.82) | 1.13 (0.83, 1.54) | 1.25 (0.77, 2.03) | 0.99 (0.84, 1.18) | 0.95 (0.65, 1.37) | 0.75 (0.48, 1.19) |
| MV | 0.83 (0.59, 1.17) | 1.24 (0.90, 1.71) | 1.17 (0.71, 1.95) | 1.14 (0.96, 1.36) | 1.05 (0.72, 1.55) | 0.88 (0.55, 1.40) | |
rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score; MV, multivariable
Analyses were stratified by age, calendar year, and cohort. MV model was adjusted for family history of cancer, physical activity, cigarette smoking status, cigarette smoking packyears, multivitamin use, regular aspirin use, regular non-steroidal anti-inflammatory drugs use, postmenopausal hormone use for women, and total energy intake. For DASH, hPDI, and DRRD, alcohol consumption was additionally adjusted for.
Similar findings were observed when continuous pattern score was used as the exposure. The spline analysis found no significant nonlinearity for all dietary pattern scores (see Figure S3, Supplemental Digital Content 1). Comparing quintile 5 to quintile 1 of pattern score, inverse association with digestive system cancer was observed for all pattern score except the hPDI and the WCRF/AICR score (see Table S4, Supplemental Digital Content 1).
In a subgroup analysis, the inverse relationships of all dietary patterns with digestive system cancer were observed regardless of age, BMI, sex, smoking status, physical activity, and regular use of aspirin (all P interactions > 0.05). However, the inverse association for the AHEI-2010 (P interaction=0.04) and the AMED (P interaction=0.04) was stronger in individuals who consumed less alcohol. (Fig 2, see Table S5, Supplemental Digital Content 1).
Fig. 2. Multivariable-adjusted associations of the cumulative average dietary patterns (comparing the 90th to 10th percentile) with digestive system cancer risk in subgroups.
Analyses details and corresponding estimates are provided in Table S5. The models were adjusted for the same list of covariates as in Table 2. rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score; BMI, body mass index; MET, metabolic equivalent for task.
P values for the Wald test of interaction term (pattern x age, pattern x BMI, pattern x sex, pattern x physical activity, pattern x aspirin use) were P>0.05 for all pattern scores. P values for the Wald test of interaction term (pattern x alcohol) was P>0.05 for all pattern scores except AHEI-2010 (P=0.05) and AMED (P=0.04). P values for the Likelihood ratio test of interaction terms (pattern x smoking status) were P>0.05 for all pattern scores.
Lagged analyses showed that the associations between dietary patterns and digestive system cancer were stronger when diet was assessed in recent periods (Fig 3, see Table S6, Supplemental Digital Content 1) In a 4-year lagged analysis, participants in the highest percentiles of all patterns except hPDI had a lower risk of digestive system cancer, but the associations attenuated more with longer lag periods.
Fig. 3. Multivariable-adjusted association between cumulative average dietary patterns (comparing the 90th to 10th percentile) and digestive system cancer according to different latency periods in the pooled cohort.
Analyses details and corresponding estimates are provided in Table S6. The models were adjusted for the same list of covariates as in Table 2. rEDIH, reversed Empirical dietary index for hyperinsulinemia; rEDIP, reversed Empirical dietary inflammation pattern; AHEI-2010, Alternative Healthy Eating Index-2010; AMED, Alternate Mediterranean Diet score; DASH, Dietary Approaches to Stop Hypertension score; hPDI, Healthful plant-based diet index; DRRD, Diabetes Risk Reduction Diet; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) dietary score.
As a sensitivity analysis, after excluding type 2 diabetes and major cardiovascular diseases at baseline and stop updating dietary variables after the diagnosis of these chronic diseases, similar trends were observed (see Table S7, Supplemental Digital Content 1).
Discussion
Over 30 years of follow-up, we compared two mechanism-based diets and six recommended dietary patterns based on dietary guidelines for the risk of digestive system cancer incidence and death using the same dataset and statistical analysis approaches. All healthy dietary pattern scores were associated with a lower risk of digestive system cancer in general. However, the CIs for the dietary patterns largely overlapped and no pattern was clearly superior statistically. Similar associations with gastrointestinal tract cancer were observed for all pattern scores except AMED and hPDI, but none of dietary patterns was associated with accessory digestive system cancers except liver cancer. Latency analyses showed associations were present within the prior 12 years but weakened with 12 or more years in the past.
In line with our study showing the inverse association of rEDIH with digestive system cancer and related death, a higher score of (non-reversed) EDIH was associated with a higher risk of colorectal cancer (28, 29), gastrointestinal tract cancer and digestive system cancer (7), and cancer mortality (30). Given the strong association between rEDIP and individual cancers in the digestive system, it is not surprising that this pattern score was strongly associated with risk of total digestive system cancer. Higher EDIP score was associated with a higher risk of individual cancers in esophagus (8), stomach (31), colorectum (32), and pancreas (33). The biological mechanism underpinning the association between EDIH and EDIP and digestive system cancer are likely due to biological effects (either synergistic or additive) of the various dietary components. Since EDIH and EDIP are developed to predict biomarkers related to biological pathways of hyperinsulinemia and chronic inflammation (34), the components of two diets could modulate this mechanism and thus reversed dietary patterns could be attractive strategies for potentially preventing total digestive system cancer and related death.
Similar to rEDIP and rEDIH, the DRRD, a diet targeting type 2 diabetes prevention based on knowledge from the literature (17), had an inverse association with digestive system cancer. The DRRD has been associated with risk of individual cancers in the digestive system such as liver (35), and pancreas (36), and cancer morality (37) although limited data are available. Possible mechanisms would be the potential anti-carcinogenic attributes of diet components through the improvement of insulin sensitivity, altered gut microbiota, reduced inflammation and oxidative stress (38–42). Our findings expand the current knowledge that greater adherence to DRRD may not only help prevent type 2 diabetes or individual cancers, but may also play a considerable role for the prevention of total digestive system cancer.
High quality diets such as the AHEI-2010, DASH, AMED, and hPDI, which originated from general or disease-specific dietary guideline, showed inverse associations with digestive system cancer. Similar to our study, a number of studies have found the inverse association between recommended dietary patterns and risk of cancers of the colorectum (5, 43–45), liver (46–48), and pancreas (49, 50). The current study suggests that the recommended dietary patterns might be advantageous for preventing overall digestive system cancer as well as specific diseases (especially cardiovascular diseases) targeted originally.
The WCRF/AIRC score had an inverse association with digestive system cancer, but the association is not as strong as that for the other patterns. The weak association with total digestive system cancer may be explained by the lack of association with accessory cancers. Compared with other pattern scores, the WCRF/AICR score had a similar association with gastrointestinal tract cancer, mostly driven by colorectal cancer (21). Because the 2018 WCRF/AICR recommendation included lifestyle components for cancer prevention, the WCRF/AICR score dependent on only dietary recommendation may not be the optimal strategy for cancer prevention. Our subgroup analysis supports the possibility that the association of WCRF/AICR score with digestive system cancer might be partially driven by factors such as BMI, smoking status, or aspirin use. Previous studies showed inverse association between the WCRF/AICR lifestyle score and cancer incidence were primarily driven by body weight and physical activity (18, 51).
The inverse association of pattern scores with digestive system cancer was attenuated after adjusting for BMI, particularly for rEDIH and rEDIP, which had the strongest inverse associations with BMI among the diets considered. We cannot tell from our study whether BMI is a mediator of the diets, or just a confounder, though the EDIH and EDIP do predict great weight gain prospectively suggesting it could be a mediator (52). Similarly, our stratified analysis showed that dietary intervention on digestive system cancer may be more favorable for people who had higher BMIs. Especially, the rEDIH and DRRD seems to be strong inverse associations for those with higher BMI. Diet related to insulin resistance and inflammation might have a greater impact among people at higher risk for insulin resistance resulting from a high BMI, which is related to the possible pathway of cancer.
Higher alcohol intake contributes to higher scores for the rEDIH and rEDIP. On the other hand, high alcohol increases risk for multiple digestive system cancers (53). Notably, the rEDIH and rEDIP had inverse associations only in those with alcohol intake in the range ≦14 g/d (less than about 1 drink/d). This suggests that higher rEDIH and rEDIP scores driven by higher alcohol intakes may not be beneficial for digestive system cancers, and benefits of these diets may be achieved by diets achieving high scores without alcohol or with low alcohol intakes.
Regular use of aspirin could prevent digestive tract cancers such as esophageal, stomach, hepato-biliary tract, colorectal, and pancreatic cancer (54). All dietary patterns except rEDIH had inverse associations only in non-regular users of aspirin. It could be expected that dietary modification would be more advantageous among people who lack of the benefits from regular use of aspirin.
The strengths of this study include the prospective cohort design with large number of cases, long follow-up period, repeated measurement of diet using validated instruments, detailed collection of lifestyle and medical data allowing for adjustment for potential confounders, high ascertainment of cancers with medical record confirmation, and comprehensive comparisons of multiple dietary patterns with digestive system cancer utilizing the same analytical approaches.
Several limitations should also be considered. Measurement error associated with FFQ might cause attenuated results. However, the FFQs in this study have been validated for measuring food and nutrient intake, as well as dietary patterns, and measurement error may have been lowered by repeated measures (55, 56). Although the major dietary patterns identified in our study are consistent with those generated in previous investigations, the generalizability of our findings may be limited due to a homogenous sample consisting of predominately White health professionals. However, results for diet related risk factors and cancer are remarkably similar for these cohorts and published studies in general (57). Potential confounders were considered for the analysis, but it is still possible that residual or unmeasured confounding remains due to the observational study design. In light of the limitations with dietary pattern methodology, the findings should be interpreted together with results from single dietary component studies. In this study, we did not perform conventional statistical adjustments for multiple comparisons across eight dietary patterns because neither the exposures nor the outcomes are not independent. This needs to be confirmed in the future studies.
In conclusion, great adherence to healthy diets was associated with a lower risk of total digestive system cancer and related death although the magnitude of association was slightly different between diets. The association appears to be driven by digestive tract cancers rather than accessory digestive system cancers. All dietary patterns recommended generally or disease-specifically may be beneficial for preventing digestive system cancer. Further research is required to confirm the underlying mechanisms between healthy dietary patterns and risk of digestive system cancer.
Supplementary Material
Study Highlight.
What is known
Multiple dietary patterns have been suggested to lower risk of individual cancers.
No study has systematically compared their relative efficacy superior for preventing total digestive system cancer and death in the same dataset.
What is new here
Greater adherence to healthy diets was associated with a lower risk of incident and fatal digestive system cancer in general, although the magnitude of the association varied slightly among the different patterns.
The inverse association between healthy dietary patterns and digestive system cancer appears to be driven by digestive tract cancers rather than accessory cancers.
Acknowledgement
The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries (NPCR) and/or the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Idaho, Indiana, Iowa, Kentucky, Louisiana, Massachusetts, Maine, Maryland, Michigan, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico, Rhode Island, Seattle SEER Registry, South Carolina, Tennessee, Texas, Utah, Virginia, West Virginia, Wyoming.
Financial support:
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (grant number 2021R1A2C1003211 to Dr. Jihye Kim) and the National Institutes of Health (NIH) (grant numbers UM1 CA186107, P01 CA87969, U01 CA176726, and U01 CA167552). Dr. Edward Giovannucci is supported by an American Cancer Society Clinical Research Professor award (CRP-23-1014041). The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH.
Footnotes
Conflict of Interest
Potential competing Interests: Authors declare that there is no conflict of interest.
References
- 1.Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209–249. [DOI] [PubMed] [Google Scholar]
- 2.Arnold M, Abnet CC, Neale RE, et al. Global Burden of 5 Major Types of Gastrointestinal Cancer. Gastroenterology 2020;159:335–349 e315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Morze J, Danielewicz A, Przybylowicz K, et al. An updated systematic review and meta-analysis on adherence to mediterranean diet and risk of cancer. Eur J Nutr 2021;60:1561–1586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Schwingshackl L, Bogensberger B, Hoffmann G. Diet Quality as Assessed by the Healthy Eating Index, Alternate Healthy Eating Index, Dietary Approaches to Stop Hypertension Score, and Health Outcomes: An Updated Systematic Review and Meta-Analysis of Cohort Studies. J Acad Nutr Diet 2018;118:74–100 e111. [DOI] [PubMed] [Google Scholar]
- 5.Tangestani H, Salari-Moghaddam A, Ghalandari H, Emamat H. Adherence to the Dietary Approaches to Stop Hypertension (DASH) dietary pattern reduces the risk of colorectal cancer: A systematic review and meta-analysis. Clin Nutr 2020;39:2975–2981. [DOI] [PubMed] [Google Scholar]
- 6.Jankovic N, Geelen A, Winkels RM, et al. Adherence to the WCRF/AICR Dietary Recommendations for Cancer Prevention and Risk of Cancer in Elderly from Europe and the United States: A Meta-Analysis within the CHANCES Project. Cancer Epidemiol Biomarkers Prev 2017;26:136–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang W, Fung TT, Wang M, et al. 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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chen QJ, Ou L, Li K, Ou FR. Meta-analysis of the relationship between Dietary Inflammatory Index and esophageal cancer risk. Medicine (Baltimore) 2020;99:e23539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Shivappa N, Godos J, Hebert JR, et al. Dietary Inflammatory Index and Colorectal Cancer Risk-A Meta-Analysis. Nutrients 2017;9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Al-Shaar L, Yuan C, Rosner B, et al. Reproducibility and Validity of a Semiquantitative Food Frequency Questionnaire in Men Assessed by Multiple Methods. Am J Epidemiol 2021;190:1122–1132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yuan C, Spiegelman D, Rimm EB, et al. Validity of a Dietary Questionnaire Assessed by Comparison With Multiple Weighed Dietary Records or 24-Hour Recalls. Am J Epidemiol 2017;185:570–584. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wang P, Song M, Eliassen AH, et al. Optimal dietary patterns for prevention of chronic disease. Nat Med 2023;29:719–728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr 2012;142:1009–1018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fung TT, Hu FB, McCullough ML, et al. Diet quality is associated with the risk of estrogen receptor-negative breast cancer in postmenopausal women. J Nutr 2006;136:466–472. [DOI] [PubMed] [Google Scholar]
- 15.Satija A, Bhupathiraju SN, Rimm EB, et al. Plant-based dietary patterns and incidence of type 2 diabetes in US men and women: results from three prospective cohort studies. PLoS Med 2016;13:e1002039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fung TT, Chiuve SE, McCullough ML, et al. Adherence to a DASH-style diet and risk of coronary heart disease and stroke in women. Arch Intern Med 2008;168:713–720. [DOI] [PubMed] [Google Scholar]
- 17.Rhee JJ, Mattei J, Hughes MD, Hu FB, Willett WC. Dietary diabetes risk reduction score, race and ethnicity, and risk of type 2 diabetes in women. Diabetes Care 2015;38:596–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shams-White MM, Brockton NT, Mitrou P, et al. Operationalizing the 2018 World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) Cancer Prevention Recommendations: A Standardized Scoring System. Nutrients 2019;11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tabung FK, Wang W, Fung TT, et al. Development and validation of empirical indices to assess the insulinaemic potential of diet and lifestyle. Br J Nutr 2016:1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tabung FK, Smith-Warner SA, Chavarro JE, et al. Development and Validation of an Empirical Dietary Inflammatory Index. J Nutr 2016;146:1560–1570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wang P, Song M, Eliassen AH, Wang M, Giovannucci EL. A comprehensive evaluation of dietary patterns with risk of colorectal cancer International Journal of Epidemiology 2022:In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Colditz GA, Martin P, Stampfer MJ, et al. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol 1986;123:894–900. [DOI] [PubMed] [Google Scholar]
- 23.Rich-Edwards JW, Corsano KA, Stampfer MJ. Test of the National Death Index and Equifax Nationwide Death Search. Am J Epidemiol 1994;140:1016–1019. [DOI] [PubMed] [Google Scholar]
- 24.Stampfer MJ, Willett WC, Speizer FE, et al. Test of the National Death Index. Am J Epidemiol 1984;119:837–839. [DOI] [PubMed] [Google Scholar]
- 25.Mozaffarian D, Hao T, Rimm EB, Willett WC, Hu FB. Changes in diet and lifestyle and long-term weight gain in women and men. N Engl J Med 2011;364:2392–2404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27. [DOI] [PubMed] [Google Scholar]
- 27.Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 2010;29:1037–1057. [DOI] [PubMed] [Google Scholar]
- 28.Tabung FK, Noonan A, Lee DH, et al. Post-diagnosis dietary insulinemic potential and survival outcomes among colorectal cancer patients. BMC Cancer 2020;20:817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tabung FK, Wang W, Fung TT, et al. Association of dietary insulinemic potential and colorectal cancer risk in men and women. Am J Clin Nutr 2018;108:363–370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wan Y, Tabung FK, Lee DH, et al. Dietary Insulinemic Potential and Risk of Total and Cause-Specific Mortality in the Nurses’ Health Study and the Health Professionals Follow-up Study. Diabetes Care 2022;45:451–459. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liang Y, Jiao H, Qu L, Liu H. Positive association between dietary inflammatory index and gastric cancer risk: A systematic review and meta-analysis. Nutr Cancer 2020;72:1290–1296. [DOI] [PubMed] [Google Scholar]
- 32.Syed Soffian SS, Mohammed Nawi A, Hod R, et al. Meta-Analysis of the Association between Dietary Inflammatory Index (DII) and Colorectal Cancer. Nutrients 2022;14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Guo Z, Hong Y, Cheng Y. Dietary inflammatory index and pancreatic cancer risk: a systematic review and dose-response meta-analysis. Public Health Nutr 2021;24:6427–6435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Steck SE, Murphy EA. Dietary patterns and cancer risk. Nat Rev Cancer 2020;20:125–138. [DOI] [PubMed] [Google Scholar]
- 35.Luo X, Sui J, Yang W, et al. Type 2 Diabetes Prevention Diet and Hepatocellular Carcinoma Risk in US Men and Women. Am J Gastroenterol 2019;114:1870–1877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Huang Y, Liu F, Chen AM, et al. Type 2 diabetes prevention diet and the risk of pancreatic cancer: A large prospective multicenter study. Clin Nutr 2021;40:5595–5604. [DOI] [PubMed] [Google Scholar]
- 37.Wang CR, Hu TY, Hao FB, et al. Type 2 Diabetes-Prevention Diet and All-Cause and Cause-Specific Mortality: A Prospective Study. Am J Epidemiol 2022;191:472–486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Awika JM, Rose DJ, Simsek S. Complementary effects of cereal and pulse polyphenols and dietary fiber on chronic inflammation and gut health. Food Funct 2018;9:1389–1409. [DOI] [PubMed] [Google Scholar]
- 39.Lundsgaard AM, Holm JB, Sjoberg KA, et al. Mechanisms Preserving Insulin Action during High Dietary Fat Intake. Cell Metab 2019;29:229. [DOI] [PubMed] [Google Scholar]
- 40.McMacken M, Shah S. A plant-based diet for the prevention and treatment of type 2 diabetes. J Geriatr Cardiol 2017;14:342–354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Tong Y, Gao H, Qi Q, et al. High fat diet, gut microbiome and gastrointestinal cancer. Theranostics 2021;11:5889–5910. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Weickert MO, Pfeiffer AFH. Impact of Dietary Fiber Consumption on Insulin Resistance and the Prevention of Type 2 Diabetes. J Nutr 2018;148:7–12. [DOI] [PubMed] [Google Scholar]
- 43.Park SY, Boushey CJ, Wilkens LR, Haiman CA, Le Marchand L. High-Quality Diets Associate With Reduced Risk of Colorectal Cancer: Analyses of Diet Quality Indexes in the Multiethnic Cohort. Gastroenterology 2017;153:386–394 e382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kim J, Boushey CJ, Wilkens LR, et al. Plant-based dietary patterns defined by a priori indices and colorectal cancer risk by sex and race/ethnicity: The Multiethnic Cohort Study. BMC Medicine 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Petimar J, Smith-Warner SA, Fung TT, 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;108:1092–1103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Ma Y, Yang W, Simon TG, et al. Dietary Patterns and Risk of Hepatocellular Carcinoma Among U.S. Men and Women. Hepatology 2019;70:577–586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Bogumil D, Park SY, Le Marchand L, et al. High-Quality Diets Are Associated With Reduced Risk of Hepatocellular Carcinoma and Chronic Liver Disease: The Multiethnic Cohort. Hepatol Commun 2019;3:437–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Li WQ, Park Y, McGlynn KA, et al. Index-based dietary patterns and risk of incident hepatocellular carcinoma and mortality from chronic liver disease in a prospective study. Hepatology 2014;60:588–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Luu HN, Paragomi P, Jin A, et al. Quality Diet Index and Risk of Pancreatic Cancer: Findings from the Singapore Chinese Health Study. Cancer Epidemiol Biomarkers Prev 2021;30:2068–2078. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Zheng J, Guinter MA, Merchant AT, et al. Dietary patterns and risk of pancreatic cancer: a systematic review. Nutr Rev 2017;75:883–908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Kaluza J, Harris HR, Hakansson N, Wolk A. Adherence to the WCRF/AICR 2018 recommendations for cancer prevention and risk of cancer: prospective cohort studies of men and women. Br J Cancer 2020;122:1562–1570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.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;149:804–815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Bagnardi V, Rota M, Botteri E, et al. Alcohol consumption and site-specific cancer risk: a comprehensive dose-response meta-analysis. Br J Cancer 2015;112:580–593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Bosetti C, Santucci C, Gallus S, Martinetti M, La Vecchia C. Aspirin and the risk of colorectal and other digestive tract cancers: an updated meta-analysis through 2019. Ann Oncol 2020;31:558–568. [DOI] [PubMed] [Google Scholar]
- 55.Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985;122:51–65. [DOI] [PubMed] [Google Scholar]
- 56.Rimm EB, Giovannucci EL, Stampfer MJ, et al. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 1992;135:1114–1126; discussion 1127–1136. [DOI] [PubMed] [Google Scholar]
- 57.Wang P, Giovannucci EL. Are exposure-disease relationships assessed in cohorts of health professionals generalizable?: a comparative analysis based on WCRF/AICR systematic literature reviews. Cancer Causes Control 2022. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.