The study protocol for DPP is publicly available at https://dppos.bsc.gwu.edu/web/dppos/dpp  and the design and methods for both DPP and DPPOS are detailed elsewhere (NCT00004992, NCT00038727) [16, 21,22,23]. Briefly, DPP was a multicenter, randomized controlled clinical trial that recruited 3,234 participants (68% women, 45% from various ethnic/racial minority groups) from 27 clinical centers across the U.S. (1996-1999). Eligible participants were ≥25 years, had a body mass index (BMI) ≥24 kg/m2 (≥22 kg/m2 for Asian/Pacific Islanders), and had plasma glucose concentration between 5.3-6.9 mmol/L (95-125 mg/dL) in the fasting state and 7.8-11.0 mmol/L (140-199 mg/dL) two hours following a 75g oral glucose tolerance test (OGTT). Participants were excluded from this secondary analysis if they were missing dietary data (n=74), waist circumference (WC) or PA (n=4), or had energy outliers (n=9) at baseline. Outliers were defined as values more than two interquartile ranges above the 75th or below the 25th percentile on the logarithmic scale. The rate of missing data was low (~70% had dietary and visit data at 15 years) and did not differ among treatment groups; missing data were assumed to be missing at random. The final analytic cohort included 3,147 participants (see Supplemental Figure 1). Participants without year 1 dietary data available were excluded from analyses assessing Score change from baseline to year 1 (N=247).
DPP and DPPOS study designs
Participants in DPP were randomly assigned to receive an intensive lifestyle intervention (ILS), metformin (MET) or a placebo pill (PLB). ILS participants were offered an individualized 16-lesson curriculum over 24 weeks followed by monthly sessions through DPP. The curriculum focused on diet, exercise, and behavior change to a low-fat, low-calorie diet (<25% kcal from fat) and to perform ≥150 min/week of PA, with the primary goal to achieve ≥7% weight loss from baseline weight . MET participants were assigned to take blinded 850g metformin twice daily; PLB participants were assigned a matching placebo pill twice daily. Both the MET and PLB groups received written standard lifestyle recommendations and a one-on-one lifestyle session annually . Participants were followed for an average of 3.2 years.
Given the efficacy of ILS, DPP was terminated and participants’ groups were disclosed in July 2001; all participants were then offered the 16-session ILS curriculum  in group format through a Healthy Lifestyle Program (HELP) during a 6-month Bridge period  and invited to participate in the long-term follow-up study (DPPOS). DPPOS participants were offered group lifestyle HELP sessions every three months to reinforce weight and activity goals. ILS participants were offered an additional 2-4 booster lifestyle sessions twice annually. Metformin was continued unmasked in the MET group. Years of follow-up will be referred to as: years 0 (DPP baseline), 1 (DPP 1 year follow-up), and years 5, 6, 9, and 15 (DPPOS years 1, 2, 5, and 11). Protocols were approved by the local institutional review boards of participating study centers (Supplemental Table 1); all participants provided written informed consent.
Exposure: The 2018 WCRF/AICR Score
The 2018 WCRF/AICR Score is used to estimate alignment with the 2018 WCRF/AICR Cancer Prevention Recommendations . Eight recommendations operationalized within the standardized scoring system (Supplemental Table 2) address body weight, PA, fruit/vegetables and fiber, ultra-processed foods, red and processed meat, SSBs, and alcohol; the optional breastfeeding component was not included. Thus, total Scores ranged from 0-7 points, with a greater Score indicating greater alignment to the recommendations.
The body weight component of the Score is calculated based on BMI (kg/m2) and WC. BMI was estimated from participants’ height (cm) and weight (kg). Height was attained at years 0, 1, and 15; the most recently measured height was used to calculate BMI at each visit. Weight was measured twice annually and WC (cm) was measured annually by trained personnel in duplicate. If there was a discrepancy larger than 0.5 cm for height and WC or 0.2 kg for weight, a third measure was taken and the average of the three were reported.
PA was collected at every annual visit through 15 years using the Modifiable Activity Questionnaire (MAQ), a valid and reliable tool to assess adult moderate and vigorous PA (MVPA) [25, 26]. The 37 activities included in the questionnaire were considered to be MVPA based on guidance from the 2011 Compendium of Physical Activities. As detailed in Supplemental Table 2, participants were categorized as meeting the PA recommendation if they performed ≥150 min/week of MVPA (equivalent of 7.5 MET hours/week).
Study participants completed a modified version of the Insulin Resistance Atherosclerosis Study (IRAS) food frequency questionnaire (FFQ) in-person with trained personnel . The 117-item questionnaire captured dietary recalls over the past year and was administered at years 0, 1, 5, 6, 9, and 15. Nutrient and energy estimates were calculated using the DietSys Nutrient Analysis Program and Nutrition Data System (version 2.6/8A/23, Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN, USA) . Data were used to calculate the five dietary components of the Score (fruits/vegetables and fiber, fast foods, red and processed meat, SSBs, and alcohol), as well as energy. Details of how each component was estimated are included in Supplemental Table 2.
Self-reported age (years), sex (male/female), race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Hispanic, American Indian, Asian/Pacific Islander), education (years), smoking (never, current, former), family history of type 2 diabetes (yes/no), marital status (never married, living together, married, separated, divorced, widowed), and hormone therapy (in women, yes/no) were collected at year 0.
Fasting glucose and HbA1c were measured every six months and OGTTs were performed annually. The primary outcome was the development of diabetes based on the 1997 American Diabetes Association criteria: fasting plasma glucose ≥7 mmol/L (≥126 mg/dL) or 2-hour plasma glucose ≥11.1 mmol/L (≥200 mg/dL) after a 75g oral glucose load , confirmed by repeat test within six weeks. Participant outcomes were collected for DPP until July 31, 2001 and for DPPOS were used until January 2, 2014.
Descriptive statistics were used to examine characteristics of the study population. Comparisons between groups were computed using ANOVA for continuous variables and chi-squared tests for categorical variables. Cox proportional hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated for the association of the 2018 WCRF/AICR Score with incident diabetes over time, with person-years as the underlying time metric. The Score was modeled as a continuous variable (i.e., risk per 1-point increase). To examine if and how Scores changed over time (i.e., accounting for a time-dependent Score) and how they were associated with risk differences by group, the association between Score changes from years 0-1 on diabetes risk was examined through DPP (average 3 years follow-up) and through DPPOS (~15-years follow-up). Additionally, the association between time-dependent Score changes over 15 years and diabetes risk was examined. Treatment group, age, sex, race/ethnicity, and smoking were tested as potential effect modifiers; models were stratified as needed. Base models adjusted for age, sex, and baseline risk score. Multivariate models additionally adjusted for race/ethnicity, marital status, family history of type 2 diabetes, education, hormone therapy, and baseline energy intake.
Given the distribution of the data for fruit/vegetables and SSBs, sensitivity analyses were performed excluding participants with data outliers identified using the same approach described above to determine if they significantly affected estimates. Asian participant weight component cut-points were also adjusted in a second sensitivity analysis following World Health Organization (WHO) guidelines and WCRF/AICR Recommendations [6, 28]. Additionally, a sensitivity analysis was conducted to examine if findings differed by DPPOS lifestyle session attendance.
Lastly, models were run to explore the independent associations of each individual WCRF/AICR Score component. To further explore the effect of weight change and PA, models were run to assess associations with the body weight and PA components combined; the five nutrition components combined; and the Score excluding the weight component. All exploratory models adjusted for the other components in the Score and aforementioned covariates. SAS version 9.4 (SAS Institute, Inc., Cary, NC) was used for all analyses. Statistical tests were two-sided, with a significance level of 0.05.