Articles 1 eas oe . Associations of the glycaemic index and the glycaemicload + “_@® with risk of type 2 diabetes in 127 594 people from 20 countries (PURE): a prospective cohort study A Jenkins, Mahshid Dehghan, Kristie Srichaikul, Sumathy Rangarajan, Andrew Mente, Viswanathan Mohan, Victoria Miller, David Sumathi Swaminathan, Rosnah Ismail, Maria Luz Diaz, Rekha M Ravindran, Katarzyna Zatonska, Ahmad Bahonar, Yuksel Altuntas, Rasha Khatib, Patricio Lopez-Jaramillo, Afzalhussein Yusufali, Karen Yeates, Jephat Chifamba, Romaina Iqbal, Rita Yusuf, Elizabeth Catherina Swart, Hu Bo, Guoliang Han, Xiaocong Li, Khalid F Alhabib, Annika Rosengren, Alvaro Avezum, Fernando Lanas, Salim Yusuf, on behalf and Rural Epidemiology (PURE) study investigators* of the Prospective Urban Summary Background The association between the glycaemic index and the glycaemic load with type 2 diabetes incidence is Lancet Diabetes Endocrinol 2024 controversial. We aimed to evaluate this association in an international cohort with diverse glycaemic index and published online glycaemic load diets. Methods April 5, 2024 https://doi.org/10.1016/ Py A sons '$2213-8587(24)00069-X The PURE study is a prospective cohort study of 127594 adults aged 35-70 years from 20 high-income, See Online/Comment scorns middle-income, and low-income countries. Diet was assessed at baseline using country-specific validated food jc) frequency questionnaires. The glycaemic index and the glycaemic load were estimated on the basis of the intake of _s9713-2587(24)00003-7 seven categories of carbohydrate-containing foods. Participants were categorised into quintiles of glycaemic index — ~Members|isted and in the appendix glycaemic load. The primary outcome was incident type 2 diabetes. Multivariable Cox Frailty models with random _ (pp 5-6) intercepts for study centre were used to calculate hazard ratios (HRs). Population Health Research Institute, McMaster University, Findings During a median follow-up of 11-8 years (IQR 9-0-13-0), 7326 (5-7%) incident cases of type 2 diabetes aime apenas no, Spangarjan seAMentePhO, occurred. In multivariable adjusted analyses, a diet with a higher glycaemic index was significantly associated with a higher risk of diabetes (quintile 5 vs quintile 1; HR 1-15 [95% CI 1-03-1.29). Participants in the highest quintile of _ProfSVusuF0Phi; Department the glycaemic load had a higher risk of incident type 2 diabetes compared with those in the lowest quintile (HR 1-21, of Medicine, McMaster 95% BMI C11-06-1-37). The glycaemic index was more strongly associated with diabetes among individuals with a higher (eee sii por snes, 1-10 (quintile 5 vs quintile 1; HR 1-23 [95% CI 1-08-1-41)) than those with a lower BMI (quintile 5 vs quintile 1; pepartmentof Nutitional [0-87-1- 39]; p interaction=0- 030). Sciences and Medicine, Faculty of Medicine, Temerty Interpretation Diets with a high glycaemic index and a high glycaemic load were associated with a higher risk of YnvesibefTorenio. glycaemic pAJenkinsMDPhD); incident type 2 diabetes in a multinational cohort spanning five continents. Our findings suggest that consuming low (prof index and low glycaemic load diets might prevent the development of type 2 diabetes. Lika Shing Knowledge Institute, St Michael's Hospital, Funding Full funding sources are listed at the end ofthe Article. Copyright © 2024 Published by Elsevier Ltd. All rights reserved. er ecanae K Stichaikul MD);Toronto 3D knowledge Synthesisand Clinical Trials Unit, Toronto, Introduction The glycaemic index ranks carbohydrate-containing QvConate Prof Alenkins): In 2021, there were an estimated 537 million cases of foods based on the postprandial blood glucose response.” Modification Centre, type 2 diabetes, with approximately three-quarters of cases The glycaemic load describes both the quality and StMichael’s Hospital, Toronto, occurring and middle-income countries.’ quantity of carbohydrate in awoe food, and is computed as ON. Canadaof Endocrinology (ProfD A Jenkins) . Division Over the inpastlow-income three decades, diabetes prevalence has the product of the glycaemic index and the amount jjsrabotam, and markedly increased in all regions of the world and is of carbohydrate available in a serving.” Several meta- Hospital, Toronto,st Michae!t projected ON, Canada to reach 643 million by 2030 without effective analyses of prospective cohort studies have reported that _ (Prof AJenkins); Department interventions. Preventing the onset of global priority, given its serious consequences, diabetesis includa major _ load diets arewithassociated a high glycaemic glycaemic omasterusivesiy ofHealth Research Methods, with a index higherandriska high of diabetes." ing major burden of disease, and health-care costs.’ Randomised controlled trials have shown that lifestyle A systematic review by Reynolds and colleagues," which Hamilton, ON, canada largely informed the recent WHO guideline on carbo- _(AMente), Department interventions, including dietary modifications, are effec- hydrate intake," concluded that, compared with the Pisbstelosy MaisDiabetes of tive with ata both and controlling evidenceindex for dietary and whole gly- _p,Mohan‘s diabetes high preventing glycaemic index and a highdiabetes.“’ glycaemic Diets load caemic and thefibre glycaemic load grains, might the be less specialties have entre, chennai, suggested to be associated with increase in useful measures of carbohydrate quality. In contrast, !ndia(Y insulinbeenresistance MohanDSc);StJohn’s cell function, which asuming 2024 meta-analysis of large cohorts found that con- seh ato Acdenyof in turn might increaseandtheimpaired a low glycaemic diet had similar beneficial risk ofdiabetes."” www.thelancet.com/diabetes-endocrinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X 1 l Articles Health Sciences, Bangalore, India (Prof Medgar SSwaminathanP teuiyefiedene hD); Researchi Evidence beforethis n context. study Universiti Kebangsaan, Kuala AWHO-sponsored meta-analysis on carbohydrate quality Lumpur Malaysia (ProfRismall PH Estudios . 127594 participants from 20 countries who did not have a previous history of diabetes. Consistent with most concluded thatthere was low quality evidence forthe effects of observational studies, but in contrast to 2023 dietary the glycaemicindex on the risk ofdiabetes. In contrast, a2024 recommendations for carbohydrate intake in adults, we vspesaria, Agenting meta-analysis of large cohort studies found that the association found that consuming a diet with a high glycaemic index and (Luz Diaz MD); Department of Health Sciences, between the glycaemic index and diabetes was similartothe _a high glycaemic load was associated with an increased risk of widely accepted associations for fibre and whole grain intake. diabetes. We also found that the glycaemic index was most stele henntunshoy We systematically searched PubMed fromJan 1, 1960, to strongly associated with the risk oftype 2 diabetes among Health Action byPeople, July 1, 2023, for relevant articles on the glycaemic index and the _ individuals with a higher BMI. Trivandrum, Kerala ndia glycaemic load, using the search terms "glycemic index’, implications capetnentetihaam alltheavailableevidence — gh/cemicload", “carbohydrate quality’, and “diabetes”. Most The prevalenceofof diabetes is markedly increasing, particularly Medical University, Wroclaw, studies that evaluated the association ofthe glycaemic index in low-income and middle-income countries, andthe (KZatonskamp);, andthe glycaemic load and diabetes were done in North identification of modifiable risk factors is urgently needed. enCardiovasc ‘America or Europe, and a few studies were done in China, Japan, findings suggestthat consuming a low glycaemic index diet Teearh incite ohana might lower the risk ofdiabetes, especially among people with esearch Our Poland University of Medical Sciences, ian, an SaherHD): Turkey, Faculty of Medicine, Added value of this study Weandinvestigated the association of the glycaemic load and risk of type 2 diabetes in the glycaemic index a higher BMI and in countries where high carbohydrate diets are traditionally consumed. Istanbul Sisli Hamidiye Etfal Health Training Research chlcvindgyand associations on the risk of diabetes as consuming a diet Procedures Metabolism, sisliIstabul, high in fibre or whole grains.” Given the uncertainty Data on demographic factors, socioeconomic status (eg, Turkiye (rot AtwntssMO): regarding the importance ofthe glycaemic index and the _ education, employment, income, and household wealth Institute, Milwaukee, wi,usa glycaemic load on diabetes prevention, we examined index), lifestyle (eg, physical activity, smoking, and diet), PhO); Institute (Rhatib Community ofthe associations between the glycaemic index and the health history and comorbidities (eg, hypertension, and PublicHealth, glycaemic load with incident type 2 diabetes using data cardiovascular disease, diabetes, and cancer), medication palestine thatgeMedica! te te fromProspective 127594 individuals countries enrolled history questionnaires of type 2 diabetes were colResearchinstitute, Urban andfrom Rural20 Epidemiology (PURE)in _ use, lectedand usingfamily standardised and procedures. School, Universidadde study. We hypothesised that higher glycaemic index and Physical assessments included standardised measureSantander, Bucaramanga, glycaemic load diets, by virtue of their postprandial ments of height, weight, waist and hip circumference, (Prof topesjornata Pho glycaemic effect," would be associated with higher and blood pressure. Education was categorised as low (no Tamani Foundation, incidence of diabetes. or primary school education), intermediate (secondary Matemnwe, Zanzibar, Tanzania school education), or high (completion of trade school, (AYusufali MD);Divisionof ~Mlethods college, or university). Data on household ownership Nein Kingston, nanttameny, ON, canada The design and of the PURE study have been of motorbike) was used to create television, a household Studydesign andmethods participants 14 assets and(eg, land electricity, car, computer, or (Profk Yeates MD); Department described previously.” The PURE study is a prospec- wealth index.” Data on physical activity were collected of Biomedical Sciences, tive cohort study done in seven geographic regions _ using the long form of the International Physical Activity UniversityofZimbabwe (North America and Europe, South America, the Questionnaire, and computed as the total of transporta- (/chifamba DPhil); Department Middle East, South Asia, China, Southeast Asia, and tion, occupation, housework, and recreational physical of Community HealthSciences, Africa). Adults aged 35-70 years at baseline were activity reported in metabolic equivalent minutes per Agevrerteath Khan universityKarchi enrolled from 631 urban and rural communities using week. Total physical activity was categorised as low Pepuavonsna country-specific recruitment and sampling methods (<600 metabolic equivalent minutes per week), moderate Development, Independent (appendix p 7). In this analysis, we include data on (600 to <3000 metabolic equivalent minutes per week), or University, Dhaka, Bangladesh individuals from 20 countries: four high-income high (£3000 metabolic equivalent minutes per week). (Prof vss PRD): Department countries (Canada, Saudi Arabia, Sweden, and the Smoking status and alcohol intake were categorised as Universityofthe Westerncape, United Arab Emirates), 11 middle-income coun- never, former, or current. Belville South Africa tries (Argentina, Brazil, China, Chile, Colombia, Iran, __ Follow-up visits were done via telephone or in-person (Prof€ Catherina Swart PhD); Malaysia, occupied Palestine territory, Poland, South visits at least every 3 years. Standard case-report forms iomenaca Researchand Africa, and Tiirkiye), and five low-income countries _ were used to collect incident type Canterfortetiovessie 2 diabetes cases during (Bangladesh,inTheeach India, Tanzania, and follow-up. analysis included andall outcome events that PekingDiseases, Zimbabwe). studystudywasPakistan, approved by participants local ethics occurred asThisof19April 3, 2023. Union FuwaiHospital, Medical College committees centre, and all At baseline, country-specific nine region-specific andchinese Academy of yrovided written informed consent. 2 (in India) validated food frequency questionnaires www.thelancet.com/diabetes-endocrinology Published online April 5, 2024. https://doi.org/10.1016/S2213-8587(24)00069-X Articles 1 (FFQ) were used to collect habitual food and beverage Plasma was immediately analysed for glucose locally or Medical Sciences, Beijing, China intake. The food lists for the FFQs included between stored at 20°C to -70°C and shipped in temperature- (Prof #Bo PhD, G Han MM, 98beverage, to 220 food items per questionnaire. For each food or controlled containers for central measurements. Plasma *¥inc odes tana aha4 a standard portion size was used. Participants glucose was measured by standardised enzymatic Cardiac Center, College of were asked how often on average during the previous year__ methods using hexokinase or glucose oxidase. Medicine, ng Saud University they had consumed a portion of each food item with Medica Cy King a0 possible frequencies ranging from never to six or more _ Statistical analysis habia Peat ance MBBS) times per day. The daily intake of each food or beverage We summarised continuous variables as means (SD) _pepartmentof Molecolar and was calculated by multiplying the daily frequency of intake and categorical variables as percentages. We calculated Clinical Medicine, institute of by the portion size. Daily nutrient intakes were calculated _ the hazard ratios (HR) for incident type 2 diabetes using Medicine, Sahlgrenska using the US Department of Agriculture food-composition multivariable shared frailty Cox regression with random _(avenyunsitirense composition tables for local dishes. Details about the simultaneously adjustsfor clustering for country by study and centre region).(which The Gothenburg, (ProfARosengren SwedenMD); development the appendix (pand11).validation of thewithFFQs are provided random-effectIn our termanalysis, was assumed normally IntemationalResearch enter Participants implausible baselinein distributed. the origintoandbe start times EUNISA, ‘Séo Paulo, SP Brazil database (release 18 and 21) and country-specific food- _ intercepts to account energy intake (<500 or >3500 kcal per day for women and__ were the date of recruitment. The end time was all out- (prof Avezum MD); <800 or >4200 kcal per day for men) were excluded from come events that occurred as ofApril 3, 2023. Participants _ Universidad the dela Frontera, study. The main exposures were the glycaemic index and the who died (11875 [9- 3%] of 127594) or were lost to follow- Temuco, Chile (Prof Lanas MD) up (150 [<1%] of 127594) were censored in the analysis. _Covespondenceto glycaemic the load. Details of the methods used to calculate Estimates of HRs and 95% Cis are presented for quintiles {to's MillePopulation Heath glycaemic index and the glycaemic load have been _ using the lowest glycaemic index or the lowest glycaemic University, Hamilton, published.” International glycaemic index tables and load group as the reference. The proportionality ont8.2x2, canada databases of tropical foods were used to obtain the assumption of the Cox model was assessed by visual vittoriamiller@phri.ca glycaemic index for carbohydrate-containing foods _ inspection of initial stratified Kaplan—Meier curves by the _S¢¢ Online forappendix occurring at least four times on the FFQs.” The mean of _ variables in the model, and by log (-log survival) versus the same foods from two or more studies was used when __ log time plots after fitting Cox models. The log-linearity possible, and in cases where no data were found, the glycaemic index value for a similar food was used. We used the bread scale that uses 50 g of carbohydrate from white bread as the control (glycaemic index value of 100). We assumption was assessed by visual inspection of the Martingale residuals plot. Model 1 adjusted for age, sex, and centre. Model 2 further adjusted for urban or rural location, education level, country-income category of carbohydrate-containing food and beverage groups: non-legume starchy foods (glycaemic index-93), sugarsweetened beverages (87), fruit (69), fruit juice (68), non- activity, waist-to-hip ratio, family history of diabetes, blood pressure medication use, and daily energy intake. Model 3 included the covariates included in model 2 with on weighted means of the individual food and beverage grain intake (g per day; for glycaemic index exposure categories of carbohydrate-containing food and beverage groups are provided in the appendix (p 15). On the bread scale, we considered a glycaemic index value of less than 79 as low, 79-99 as moderate, and more than 100 as high. The selection of covariates was done a priori based on expert knowledge of clinically meaningful risk factors for diabetes, as well as a reviewing previous studies to identify suspected or established factors associated beverage groups by the net carbohydrate consumed in each group to estimate each participant's mean glycaemic index. To calculate each participant's glycaemic load, the 80% of participants had complete covariate data, with missingness ranging from 0-4% for education level to 98% for family history of diabetes (appendix p 60). We assigned a glycaemic index value to seven categories (economic region), wealth index, smoking status, physical starchy vegetables (54), legumes (42), and dairy (38), based further adjustment for fibre intake (g per day), whole groups (appendix pp 17-24). The definitions for the only), and dietary protein intake (percent energy per day). We weighted the glycaemic indexes of the seven food and _ with either the exposure or outcome.” Approximately glycaemic index was multiplied by the net carbohydrate (grams per day) intake and divided by 100.* Outcomes The primary outcome was incident type 2 diabetes, used a multiple imputation procedure with ten rounds of imputation incorporating all covariates in the primary analysis (demographic, lifestyle, dietary covariates, and study centre) to impute missing data for covariates. To check the robustness of our multiple imputation, we also defined as a diagnosis of diabetes by a physician, oral _report complete-case analyses. In subgroup analyses, we assessed the associations fasting plasma glucose level of 7-0 mmol/L or more _ between the glycaemic index and the glycaemic load and antidiabetic agents or insulin use, or a documented (126 mg/dL) during follow-up in individuals without a history of diabetes at baseline. Venipuncture was done by diabetes by economic region (high-income countries, middle-income countries, or low-income countries) and centrifuged at the local site within 2 h of collection.” America, the Middle East, South Asia, China, Southeast health professionals, and the blood sample was _ by geographic region (North America and Europe, South www.thelancet.com/diabetes-endocrinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X 3 i Articles Q1(n=25519) Q2(n=25519) Q3(n=25519) Q4(n=25519) Q5 (n=25518) 512(97) 504(98) 50:0(9:9) 497 (98) 49-5 (96) 10114 (39-6%) 15405 (60-4%) 10706 (42.0%) 14813 (58.0%) 10133 (39.7%) 15386 (603%) 10959 (42-9%) 14560 (571%) 11.792 (46.2%) 13726 (53-8%) Urban residence 16 613 (65.1%) 16 383 (64-2%) 15286 (59:9%) 11688 (45:8%) 6533 (25-6%) Low education level 9289 (36-4%) 10386 (40.7%) 10361 (40-6%) 10922 (428%) 13882 (54.4%) Current smoker 4415 (17.3%) 4695 (18-4%) 5155 (20.2%) 5920 (23.2%) 7068 (27.7%) Low physical activity 3904 (15.3%) 4440 (17-4%) 4466 (17.5%) 4747 (186%) 5231 (205%) Demographics Age, years Sex Male Female Wealth index BMI, kg/m? 066 (089) 0:33 (097) -0.084 (0:93) -038 (089) -0.67 (082) 272(53) 088 (0.089) 26.8 (5-4) 0-88 (0.088) 25:6 (5:1) 087 (0-084) 24-6 (47) 0-86 (0.079) 23:8 (45) 0-86 (0.077) Family history of 7783 (305%) 6890 (270%) 4925 (193%) 3292 (12.9%) 2246 (8-8) Blood pressure 4517 (17.7%) 3675 (14-4%) 2782 (109%) 1990 (7.8%) 1480 (5-8%) North America and South America Europe 8375 (328%) 7478 (293%) 4562 (17-9%) 6637 (26-0%) 930 (36%) 4556 (17-9%) 116 (0.5%) 1555 (61%) 14 (0.05%) 485 (1.9%) Middle East 5079 (19-9%) 4869 (19.1%) 1201 (47%) 141 (0.6%) 12 (0.05%) South Asia 3442 (135%) 4914 (19.3%) 6019 (23-6%) 6356 (24.9%) 5961 (23.4%) China 584 (23%) 2953 (11.6%) 9514 (373%) 13.839 (54.2%) 15.240 (59-7%) Southeast Asia 296 (1.2%) 786 (3.1%) 2045 (8.0%) 2340 (9.2%) 2092 (82%) Dairy, g per day 303 (172-518) 223 (86-322) 105 (21-240) 32 (0-137) 0(0-31) Fruit, g per day 256 (135-443) 192 (97-318) 128 (59-225) 86 (40-152) 44 (21-83) Waist-to-hip ratio diabetes medication use Geographic region Africa 265 (1.0%) 798 (3.1%) 1254 (4-9%) 1172 (4-6%) 1714 (67%) Dietary factors Fruit juice, g perday 51 (10-214) 20 (0-107) 2 (0-33) 0(0-16) 0(03) Vegetables, g per day 337 (198-508) 250 (152-339) 236 (135-257) 247 (415-255) 180 (58-250) Starchy foods,gperday 294 (210-405) 410 (307-543) 567 (418-753) 742 (570-959) 948 (732-1207) Unprocessed red meat, gper day 69 (31-115) 64 (21-116) 43 (42-100) 319-79) 20(5-54) 4(0-14) Legumes, g per day Soft drinks, g per day 62 (29-123) 33(0-51) 45 (22-90) 7(0-51) 43 (20-79) 0(0-23) 33 (14-60) 0(0-5) 143-34) 0(0-0) Poultry, g per day 24 (10-60) 20 (6-51) 133-34) 70-21) Seafood, g per day 333-30) 20-27) 110-31) 10 (0-32) 6 (0-24) Energy, kcal per day 2085 (1637-2600) 2046 (1577-2587) 1908 (1480-2454) 1897 (1524-2393) 1964 (1529-2498) Fibre, g per day 25 (18-35) 22 (15-31) 18 (11-25) 15 (10-25) 12.4(82-26-4) Dietary protein, 168 (147-189) 16.0 (137-182) 1555 (13.0-17.6) 14-4 (123-463) 126 (111-143) Dietary fat, percentage of energy 289 (24-2-33-4) 28.0 (23-0-328) 243 (187-298) 18-2 (13-4-25-6) 12.6 (88-204) Carbohydrate, 543 (483-60-4) 562 (50-2-62-4) 603 (536-663) 66:8 (593-729) 741 (658-793) percentage of energy percentage of energy Glycaemic index 76-1 (738-777) 81,7 (80-4-82:9) 85-9 (850-867) 88-7 (88-1-89.2) 907 (90-2-91-4) Glycaemic load, g per 191-9 (141-8-258-7) 214-3 (162-5-280-2) 259-6 (195-8-337:3) 314-4 (244-6-4037) 3857 (2985-4871) day Data are mean (SD), n (2%), or median (IQR). Total study population=127594. Excluded participants with implausible energy (<500 kcal per day and >3500 keal per day for diabetes status. The daily intake of rut juice was not measured in China and Bangladesh; data for Bangladesh are included under south Asia. The glycaemic index isthe measure ofindex how ofmuch 50 g of carbohydrate from a specific food raises the blood glucose level. To adjust the reference standard from the bread scale low glycaemic alycaemic white bread = 100 on the bread scale and 70 on the glucose scale. On the bread scale, we considered a glycaemic index value of <79toastheglucoses cale:index, ‘women and <800 kcal per day and >4200 kcal per day for men), incomplete data on glycaemic index and glycaemic load, with baseline diabetes and missing baseline 79-99 as moderate glycaemic index, and =100 as high glycaemic index. Q1-5=quintile 1-5, Table 1: Participant characteristics by quintiles of glycaemic index 4 www thelancet.com/diabetes-endocrinology Published online April 5, 2024. https://doi.org/10.1016/S2213-8587(24)00069-X Articles 1 Qu (n=25519) Q2 (n=25519) Q3 (n=25519) Q4(n=25519) Q5 (n=25518) Ptrend HR perunit increment* Glycaemic index 764 (738-777) Peoplewithdiabetes 1673 (6-6%) Model 1, HR (95% Cl) 1.0 (ref) 817 (80-4-82.9) 1669 (6.5%) 0-99 (0-92-1-06) 85,9 (850-867) 1410 (5:5%) 1-03 (0-95-1-13) 88:7 (88-1-89.2) 1364 (53%) 1.06 (0-96-1-13) 907(902-91-4) 1210 (47%) 0-97 (087-107) 0-93, 1.01 (0-95-1-07) Model 2, HR (95% Cl) 1.0 (ref) 0.95 (089-102) 1.01 (093-141) 1.40 (0:99-1.21) 1.06 (096-118) 048 —1.04(0.98-1:10) Model 3, HR (95% Cl) 1.0 (ref) 0.98 (0-91-1.04) 11.05 (096-115) 1.16 (1.04-1.29) 1.15 (103-129) 0.010 1.09 (1.02-1.16) Glycaemic load Peoplewithdiabetes 136.5 (110-6-155-9) 204-1 (188-6-2195) 264-9 (250-4-281.0) 344-2 (320-6-370-4) 4748 (4232-5563) 1482 (5-8%) 1477 (58%) 1443 (57%) 1424 (5.6%) 1500 (5-9%) Model 1, HR (95% Cl) 1.0 (ref) 100 (0-92-1.07) 1.08 (1.00-1-17) 1.09 (101-119) 1.05 (0-95-15) 018 1.00 (0-99-1.01) Model 2, HR (95% Cl) 1.0 (ref) 1.01 (0-94-1-09) 1.41 (1.01-1.21) 1:13 (1.02-1.25) 1.12 (0-99-27) 0.070 1.01 (0-99-1-02) Model 3, HR (95% Cl) 10 (ref) 1-04 (096-112) 1.15 (1.05-1.26) 1:20 (107-133) 1.21 (106-137) 0.0060 1.01 (1-00-1.03) (IQR) or n (%), unless stated otherwise. Total study population=127594. Model 1: adjusted for age, sex, centre (random effect).Model2: adjusted rural are medianeducation, forage, sex, centre (random effect), urban or country-income category, wealth index, current smoker, physical activity, waist-to-hip ratio, family history of adjustedlocation, for age, sex, centre (random effect), urban or rural location, education, country-income category, wealth index, current smoker, diabetes, physical bloodpressure activity, waist-to-hip medication ratio,use, family andhistory energyofintake. Model3: diabetes, blood pressure medication use, energy intake, dietary fibre intake, whole grain intake (glycaemic index exposure only), and protein intake (percentage of energy). Q1-5=quintile 1-5.HR=hazard ratio. *HR per 10-unit glycaemic index increment and per 50-unit glycaemic load increment. Data Table 2: Association of glycaemic index and glycaemic load and diabetes, in the overall PURE cohort Asia, and Africa). We also did subgroup analyses by BMI__Baseline characteristics are shown in table 1. Overall, using region-specific median cutoffs to define low versus _ the median glycaemic index was 85-9 (IQR 804-892), high BMI, and subgroup analyses by waist-to-hip ratio _ with the consumption of highest glycaemic index diets using region-specific and sex-specific median cutoffs to define low versus high waist-to-hip ratio. Tests of heterogeneity were done with the / statistic between observed in China (88-9 [86-8-90-3]), followed by southeast Asia (88-2 [85-9-89-9]) and Africa (88-0 [84-8-90-5]; appendix p 60). The median glycaemic excluding participants who reported diabetes in the first 2 years after recruitment, excluding participants with baseline cardiovascular disease, further adjusting for (appendix p 60). Compared with people who consumed diets with the lowest glycaemic index, those who consumed a diet with the highest glycaemic index were unprocessed red meat, fish, poultry, total dietary fat, smoking, had lower levels of physical activity, and had categories ofthe glycaemic index and the glycaemic load _ load varied from 177-1 (136-0-227-0) in North America and these subgroups. We did sensitivity analyses by and Europe to 346-0 (204-0-466-0) in south Asia non-carbohydrate containing foods and nutrients (eg, _ less educated, resided in rural areas, had higher rates of saturated fat, or animal protein), and adjusting for potential confounders following the disjunctive cause criterion.” We also did a competing risk regression analysis with death as the competing risk. Lastly, we did a random effects meta-analysis to include the results of our study with the results of the Reynolds and colleagues’ meta-analysis.* All reported p values are two-tailed, with values less than 0-05 indicating statistical significance. All statistical analyses were done with Stata version 15 and RStudio version 4.0.2. alower mean BMI. They also had lower intakes of dairy, fruit, vegetables, legumes, fruit juices, unprocessed red meat and poultry, and higher intake of starchy foods (table 1). Table 2 shows the association of the glycaemic index with incident type 2 diabetes. After multivariate adjustment for demographic, lifestyle, health history, and dietary factors (energy, fibre, and whole grain intakes), higher glycaemic index was associated with incident type 2 diabetes and participants in the highest quintile of the glycaemic index had a 15% (95% CI 3-29) greater risk Role of the funding source of diabetes compared with those in the lowest quintile. In comparing highest versus lowest quintiles of the data collection, data analysis, data interpretation, or writing of the report. glycaemic load in the fully adjusted model, we observed a positive association with incident type 2 diabetes Results between the glycaemic index and (HRThe1-21association [95% CI 1-06-1-37}). available in 145895 participants. After excluding participants with implausible energy (n=6827) and baseline diabetes (n=11474), 127594 participants without known glycaemic index was positively associated with diabetes among participants from low-income countries (tertile 3 vs tertile 1; HR 1-32 [95% CI 1-11-1-56}), and a similar ‘The funder ofthe study had no role in the study design, Data on the glycaemic index and the glycaemic load were _ diabetes varied by economic region (appendix p 62). The diabetes were included in the analysis. During a median _ direction of association was found among participants follow-up of 11-8 years (IQR 9-0-13-0), 7326 (5-796) from middle-income countries (tertile 3 vs tertile 1; 1-10 incident cases of type 2 diabetes occurred. {0-98-1-24]; appendix p 62). Among the participants www.thelancet.com/diabetes-endocrinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X 5 l Articles HR (95% C) Low BMI (n=63899) a 26% (95% Cl 15-37) greater risk of diabetes (n=624614 participants and 25 370 diabetes cases).” Our findings for the glycaemic index and the glycaemic load Quinte? +00(00-209) | and diabetes are also in keeping with most studies of Quintile 3 vs quintile1 1.08 (0:90-1:30) Quintle2 vequintles ¥07(092425) , , a tooworaas, | example, the Shanghai Women's Health Study,” the 1:10 (087-139) Japan Public Health Center-based Prospective Study,” Quntle tre gone Quintle 5 vsquintle2 High BM(n-63658) and the Tehran Lipid and Glucose Study" found that Quintiles Quintile 2 vs quintile Quintile 3 vs quintile 1 Quintile 4vsquintile Quintile 1oS 5 vs quintile 1 asian and Middle Eastern populations." For 100(100-100) | consuming diets with a high glycaemic index and 0.95 (088-204) | a high glycaemic load was positively associated with the 1.06 (095-117) risk of diabetes incidence. —_ a5 0 as 147(104-132) = 123 (108-141) To Figure: Low Association of the glycaemic index and diabetes, by BMI and high BMI defined using region-specific median BMI cutoffs. forRegion-specific medianforBMIAfrica;cutoffs:26:3 22-4kg/m’kg/m’for southeast for south Asia; China; 24-9 kg/m’ Asia; 24:3 26-4 kg/m? kg/m’ for We found that the association ofthe glycaemic index and diabetes was stronger among individuals with a higher BMI. Similar findings have been reported by the Shanghai Cohort Women’sStudy,” HealthandStudy,* Collaborative amongthemenMelbourne enrolled . . in the Japan Prospective Cohort Study.*Public A 2021Health PURECenter-based study also reported that North America and Europe; 27-3 kg/m* for South America; and 28:2 kg/m* BMI (<25 vs =25 kg/m?) significantly modified the forlocation, the Middle East. Adjusted for age, category, sex, centre (randomindex,effect), urbansmoker, orrural aaciation of the glycaemic index cardiovascular physica activity, education, country-income . with waist-to-hip ratio, family historywealth of diabetes,current blood pressure «disease and all-cause mortality.” Notably, this finduse, energy intake, dietary fibre intake, whole grain intake, and medication protein intake (percentage of energy). p interaction=0.030. ing was consistent when BMI was defined using the from high-income countries, a positive association was seen for the middle tertile compared with the lowest tertile (1-29 [1-05-1-57]), but not the highest tertile region-specific cutoffs described in our current study (appendix p 69). Our findings on diabetes and studies on cardiovascular disease earlier from PURE,” together show that consuming (n=66) in the highest tertile. The associations by geographic region are in the appendix (p 63). When we examined the association of the glycaemic index with diabetes by BMI, we found a stronger reconsider current recommendations on the usefulness of the glycaemic index as a measure of carbohydrate quality." Alpha-glucoside hydrolase inhibitors (ie, Acarbose) phar- association in the higher BMI group (quintile 5 vs macologically slows the rate of carbohydrate absorption p interaction=0-030; figure). In analyses stratified by waist-to-hip ratio, similar associations were found in the high and low waist-to-hip ratio groups (appendix p 68). are also consistent with the STOP-NIDDM trial” and the ACE trial,“ which have shown that Acarbose reduces the incidence of diabetes. compared with the lowest tertile (0-34[0-05-2-47)), a highof diabetes, glycaemiccardiovascular index diet is associated with due likely mortality. a higher riskThese to the small number of events (n=1) and participants findings provide robust disease, evidenceand to quintile 1; HR 1-23 [95% CI 1-08-1-41)) than in the group __ reducing the postprandial glycaemic response, and alters with a lower BMI (quintile 5 vs quintile 1; 1-10 [0-87-1.39]; the glycaemic index ofthe diet.* As expected, our findings tions the glycaemic index the andpositive the glycaemic Discussion Severalbetween mechanisms might explain associaIn this large multinational prospective cohort study, we _ load and diabetes risk.** In experimental studies, higher found that diets with a high glycaemic index and a high _ glycaemic index diets have been associated with circulat- glycaemic load were associated with greater risk of incident type 2 diabetes after adjustment for ing free fatty acids,” lower insulin sensitivity, impaired pancreatic B-cell and intestinal K cell function,“ elevated demographic, lifestyle, and dietary factors. The glycaemic index was most strongly associated with diabetes among individuals with a higher BMI. visceral adiposity, and poor glycaemic control.“ In addition, postprandial hyperglycaemia has been shown to increase oxidative stress, endothelial dysfunction, and Several prospective studies of mostly USA and _ the release of proinflammatory factors.°”" between a positive association validated, The strengths of this analysis use of country-specific FFQs, include the longtheduration European populations have found the glycaemic index and the glycaemic load and diabetes.*" In a 2019 dose-response meta-analysis _ of follow-up, the large sample size, and the inclusion of glycaemic index was associated with a 27% (95% CI _ income countries, which provides information on a more of observational studies, each ten-unit increase in the _ participants from high-income, middle-income, and low- 15-40) greater risk of diabetes (n=346465 participants and 18063 diabetes cases) and each 80 g per day increase in the glycaemic load was associated with 6 diverse range of glycaemic index and glycaemic load than studies done in North American and European populations only. www.thelancet.com/diabetes-endocrinology Published online April 5, 2024. https://doi.org/10.1016/S2213-8587(24)00069-X Articles 1 Potential limitations should also be considered. First, _ interpreting the data and drafting the manuscript. MD coordinated diet was measured at baseline and intakes might have _ the entire nutrition component of the PURE study. SR coordinated the changed over time, particularly for some lower-income — W2tlwide study and reviewed and commented on drafts,Allother authors coordinated the study in their respective countries, provided comments on and middle-income countries that have had an economic Grafts ofthe manuscript, and approved the final manuscript. The transition. Second, diets were assessed via self-report, and corresponding author attests that all listed authors meet authorship criteria imprecision in measurement might lead to exposure and that of individual foods and beverages vary within the pedarationofinterests “cove . °€ no others meeting the citeria have been omitted. SY is the misclassification, attenuating measures of association Tu cript. all authors approvedthe final manuscript towards the null. Third, the glycaemic index values responsibility for the decision to submit for publication. and had final ‘ ae nan guarantor. VMi, MD, and SR verified the underlying data reported in the seven glycaemic index categories, which might have — yMi reports research grant from the Canadian Institutes of Health. weighted the findings towards the null hypothesis. Research during the conduct of the study. DAJ reports research grants Nevertheless, the categorical approach used mean _ fom the Soy Nutrition Institute and the Canadian Institutes of Health index Sh it-kindof California, supplies forWalnut trials asCouncil a esearch support fom the glycaemicic inindex values from international glycaemic y Almond Board of California, the Peanut tables weighted on the basis of the frequency of occurrence Institute, Barilla, Unilever, Unico, Primo, Loblaw Companies, Quaker in the country-specific FFQs, which broadly reflects the _ (Pepsico), Pristine Gourmet, Bunge, Kellogg Canada, and WhiteWave foods most consumed by all cultures in the PURE study, Fourth, ‘ : . 24s: payment or honoraria for lectures or presentations from Nutritional Fundamentals for Health-Nutramedica, Saint Barnabas the possibility given theof observational we cannot rule out — yfedical RutgersGlycemic New Jersey School, Conference, The UniversityAtlantic of unmeasured design, confounding and residual Chicago, Center 2020 China IndexMedical International Pain Conference, Academy such ofof Lifethe International Longeae Learning;Caalndene and travel supportOcay from ae eine . we extensively adjusted for Hee lutrition,een DAJ is a co-chair well as smoking.. However, Carbohydrate Quality confounding due to measurement error in confounders, measured diabetes risk factors and the findings were Consortium and has been invited by the International Diabetes Federation robust to adjustment. Fifth, the HR hasa built-in selection to join the committee on diabetes treatment and to take the lead in bias, and we reported a single HR for the duration of our —_ writing the dietary guidelines for the treatment ofdiabetes. His wife, study's follow-up, but the magnitude of the HR might — Alexandra change dependi fol 2 SixthSixth, , due due my Research for thehaveFood Industry his two daughters, Wendy Jenkins and depending on the the k lengthoff follow-up.” jenkins, published hange LJenkins sa directorand partner ofINQUIS Clinical to the relatively small number of participants and events the foods described here (Jenkinsa vegetarian book AE,that Jenkins promotesAL,the use of WM, Jenkins within regions, the power to detect region-specific associations was limited, and the regional subgroup Bnydson C. The portfolio dit for cardiovascular disease risk reduction. London: Elsevier, 2019) and his sister, Caroline Brydson, received funding analyses . ° through a grant from the St Michael's Hospital Foundation to develop Lastly, diabetes might diagnosis also be might subject have to sparse-data cookbook for oneResearch of his studies. ARreports been missedbias.or 4state, the Swedish Council, and the funding Swedish from HearttheandSwedish Lung misclassified in some participants. Fasting blood glucose — Foundation. AA reports consulting fees from NovoNordisk; payment or was available for most participants and was used to sore ioratenlenen NovoNordisk; cstavdl andtteparticipation BeepsSoien diabetes (>7-0 mmol/L). However, we found Cardiology Meeting fromesses adiagnosis on data Board; high level of agreement between those with both self- _ safety monitoring board or advisory board for OPTIMAL-Stroke, reported diabetes and fasting blood glucose measures.* OPTIMAL Diabetes, and IMPACT-BP. All other authors declare no In this large, multinational study of adults free of Competing interests diabetes at baseline, consuming a diet with a high we 5 Datasharing Data from the PURE study are not available for public use. For the glycaemic index and a high glycaemic load were PU pr mds srotcl sce pwwheicn/pure associated with higher incidence of type 2 diabetes. Our Acknowledg findings provide evidence to support the importance of Ty.’ tudy was supported by the Population Health Research Institut, low glycaemic index and glycaemic load diets to prevent — Hamilton Health Sciences Research Institute, the Canadian Institutes develop , including . a \cknowledgments diabetes of Ontario, Strate, world the Middle regi fEast,the [*of Patient Health Orientedh, Researchidof theStrok Canadian id: Institutes f of Health By not previously studied (eg,inAfrica, liabetes development, including in the regions of the of Health Research, Heart and Stroke Foundation South America, ‘| South Asia, Southeast Asia). A Our study Support ResearchUnit, through Patient-Oriented Research and thethe Ontario Ontario Strategy Ministryforof Health and Long-Term suggests that increasing the intake of low glycaemic Care; by unrestricted grants from several pharmaceutical companies, index foods, such as legumes, fruit, fruit juice, non- _ with major contributions from AstraZeneca (Canada), Sanofi-Aventis starchy and reducing and Canada), BoehringerandIngelheim (Germany and Canada), trientvegetables, foods and withdairy,2 hich eh ‘c theindex indintakeandof‘q _ (France _Setvier,and additional contributions glycaemic Novartis andGlaxoSmithKline; King Pharma and frombyvarious national or local from nutrient-poor foods with a high high glycaemic load varieties of non-legume starchy _ organisations in participating countries as follows: Argentina: foods might have a substantial effect on preventing — Fundacién Estudios Clinicos Latino America; Bangladesh: diabetes incidence globally, particularly among people !dependent University, Bangladesh, and Mitra and Associates; Brazil: ‘tri where here hihigh — pairy UnileverCouncil Healthof Institute; Dairy Agency Farmers of Canada, National withith aa hicher Aigher BMI andand ini countries the USA, Canada: Public Health carbohydrate Contributors diets are traditionally consumed. of Canada, and Champlain Cardiovascular Disease Prevention Network; Chile: drafted and revised the manuscript. DAJ, MD, and KS assisted with alaysia: Ministryof Science, Technology, and Innovationof Malaysia, SY conceived and initiated the PURE study, supervised its conduct, and reviewed and commented on the draft. VMi did the analyses, and Universidad de La Frontera; China: National Center for Cardiovascular Diseases and ThinkTank Research Center for Health Development nae ene NOT www.thelancet.com/diabetes-endocrinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X 7 l Articles Ministry of Higher Education of Malaysia, Universiti Teknologi 16 MARA, and Universiti Kebangsaan Malaysia; Palestinian territory: and International Development Research Centre, Canada; Philippines: UN Relief and Works Agency for Palestine Refugees in the Near East Geneva: World Health Organization, 2023 §, et al. Association ofglycaemic —-17_Jenkins DJA, Willett WC, Yusuf index and glycaemic load with type 2 diabetes, cardiovascular disease, Philippine Council for Health Research and Development; Poland: Polish Ministry ofScience and Higher Education and Wroclaw Medical University; Saudi Arabia: Saudi Heart Association, Dr Mohammad Alfagih Hospital, The Deanship of Scientific Research at King Saud University numberof Coronary RG-1436-013), Hamza Serafi(Research Chair forgroup Research HeartRiyadh, Disease,Saleh Umm AlQura University, Makkah, Saudi Arabia; South Africa: North-West University, Netherlands Programme for Alternative Development, ee eeeeee eee nh eileen J Africa, a Sciences: q Sweden: Swedish government,, sSwedish Heart and Health, WHO. Carbohydrate intake for adults and children: WHO guideline. cancer, and all-cause mortality: a meta-analysis of mega cohorts of more than 100000 participants. Lancet Diabetes Endocrinol 2024; 12: 107-18. 18 Esfahani A, Wong JM, Mirrahimi A, Srichaikul K, Jenkins DJ, glycemic index: physiological significance. JKendallaM Am Col Nutrhe2009; 28 (suppl): 19 Teo Chowepidemiology CK VazM, Rangarajan S: YusufS. Theprospective urbanK, rural (PURE) study: examining the impact of societal influences on chronic noncommunicable diseases in low., middle-, and hig: tries. Am Heart J 2009; Health SouthAfrica SugarAs ociation, andFaculty ofCommunityand 20 Yusuf bal $,Rangarajangh-incomec S, Teo K, et al.ountCardiovascular ries. Am HeartriskJ and events in 17 low-, middle-, and high-income countries. N Engl J Med 2014; Victoria's V and QueenTiirkiye: Metabolic Foundation of Freemasons, and AFA Insurance; 21 Gupta R, Islam S, Mony 371: 818-27 Award salary support for Medical Sciences and Dubai receives and discretionary fundingHealth from Authority. the CanadaDAJ Research J Prev Cardiol 2015; 22: 1261-71. 22. load, JenkinsandDJA, Dehghan M,disease MenteandA, etmortality. al. Glycemic index, cardiovascular N Engl J Medglycemic 2021; Chair endowment of the federal government of Canada. SY is supported by the Marion W Burke Chair in Cardiovascular Disease of the Heart and Stroke Foundation of Ontario. 23 Working Life, and Welfare, King Gustaf United ArabSyndrome SanofiAventis; and Emirates:Society, SheikhAstraZeneca, Hamdan Binand Rashid Al Maktoum Refe o erences P, et al. Socioeconomic factors and use of secondary preventive therapies for cardiovascular diseases in South Asia: the PURE study. Eur 384: 1312-22. Atkinson FS, Foster-Powell K, Brand-Miller JC. International tables of glycemic index and glycemic load values: 2008. Diabetes Care 2008; 31: 2281-83. 24 Salmerén J, Ascherio A, Rimm EB, etal. Dietary fiber, glycemic eae een ene he load, andRM,riskMohan ofNIDDM in men. Diabetes Care 1997; 20: 545-50. 25 Anjana V, Rangarajan S, et al. Contrasting associations : s prevalence estimates for 2019 and projections for 2030 and 2045: between diabetes andcardiovascularmorality rates in low, made ‘ results from the Intemational Diabetes Federation Diabetes Atlas, and high-income countries: cohort study data from 143567 individuals Oth edition. Diabetes Res Clin Pract 2019; 157: 107843. in 21 countries in the PURE study. Diabetes Care 2020; 43: 3094-101. JL.Analysis of incomplete multivariate data Ist ed. Boca 3 Bommer C, Heesemann E, Sagalova V, etyears: al. Thea cost-of global illness economic 26 Raton, Schafer burden ofdiabetes in adults aged 20-79 study. FL: CRC Press, 1997 Lancet Diabetes Endocrinol 2017; 5: 423-30. 27 VanderWeele TJ. Principles of confounder selection. Eur J Epidemiol 2 4 ae Sacedi P, Petersohn I, Salpea P, et al. Global and regional diabetes Tuomilehto J, Lindstrém J, Eriksson JG, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001; 344: 1343-50. 5 incidence Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the N Engl J Medoftype with lifestyle intervention or metformin. 2002:2 diabetes 346: 393-403. 6 — Taheri S, Zaghloul H, Chagoury O, et al. Effect lifestyle intervention on bodyweight and glycaemia in earlyof intensive type 2 diabetes (DIADEM-1); an open-label, parallel-group, randomised controlled 7 trial. Lancet Diabetes Endocrinol 2020; 8: 477-89. Lean ME], Leslie WS, Barres AC, et al, Primary care-led weight management for remission of type 2 diabetes (DIRECT 2019; 34: 211-19. 28 Villegas R, Liu S, Gao YT, et al. Prospective study of dietary carbohydrates, glycemic index, glycemic load, and incidence of type2diabetesmellitusin middle-aged Chinese women, ren Intern Meg #167: 2510-10. Witteman JC, FeskensEJ. Glycemic index and glycemic load and their association 29 van Woudenbergh GJ, Kuijsten A, Sijbrands EJ, Hofman A, with C-reactive 201; 2protein 01: and incident type liabetes. J Nutr Metab 30. Sluijs I, van der Schouw YT, van der A DL, et al. Carbohydrate quantity and quality and risk of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition-Netherlands 8 — anLudwigopen-label, trial. Lancet 2018; 391: 541-51.relatin, n (PINE va aeA, eee DS. Thecluster-randomised glycemic index: physiological mechanisms ‘opping BN,StudyErberamdoe E, Grandinetti Verheus teM, vol Kolonel LN LN, te obesity, “hiabotessuid cardiowasevlar divecse, TAMA 2002;7 8 Maskarinec Dietaryinfiber,a multiethnic magnesium,cohort and inglycemic alter risk of type 2G.diabetes Hawaii.load J Nutr ° 2010; 140: 68-74. Aston TM.Glycaemic index and metabolic disease risk 32. Patel AV, McCullough ML, Pavluck AL, Jacobs EJ, Thun MJ, 10. Jenkins Dj, ‘Wolever TM, Taylor RH, et al. Glycemic index offoods: Calle ee Glycemic load,glycemic index, and carbohydrate intake in 3 physiologicalbasis B for 4carbohydrateetal, exchange. Am Clin Nur relation Cancer Catves to pancreatic Control cancerr 2007: 18:isk 287-94. ina large US cohort. 11 poe 7 33. Teymoori F, Farhadnejad H, Moslehi N, Mirmiran P, Mokhtari E, Monro JA, shaw M Glycemic impact glycemic glucose equivalents, Azizi ofdietaryClininsulin and glycemic indices Nutr 2021; 40: 2138-44. with theF. The risk association oftype 2 diabetes. an ney‘Gli, Nutr 2008; 37: DS 43... mclions,al 9 Br 2 load theses G.q Taylor R, , Livesey HF, etal. etal. Dietary lycemic index and and the riskof prospective oftype 2 diabetes: systematicNutrients review2019; and 11:updated meta-analyses cohort astudies. 1280. a bs5, veo Kuretani Ke a Dietary glycemic index, glycemic te the Tapan Public Healt Conterbased Protpective Stat:- Nut 2013: 12: 165. 13 Reynolds,Mant J, Cunmmings J, Winter N, Mete£, Ze Morenga lGatbohydratequality 2 and . human health:series of . systematic 7 - 35, Sakurai M, Nakamura K, Miura K, et al. Dietary glycemic index and 14 Hardy DS, Garvin JT, Xu H. Carbohydrate quality, glycemic index, 36 He F. Diets with a low glycaemic load have favourable effects on alycemic load and cardiometabolic qseeresponse metaanalysis. NutrrisksMetab intheCardiownse Dis 20 , 4 US, EuropeandAsia 15 Reynolds A, Mann risk oftype 2 diabetes mellitus in middle-aged Japanese men, Metabolism 2012; 61: 47-55. prediabetes progression and regression:aprospective cohortstudy. J Hum Nutr Diet 2018; 31: 292-300. Carbohydrate qualityJ,Cummings Wintera series N, Meteof Esystematic , Te Morenga L. 2” Hodge aMog English GlycemicCareindev and20% and human J,health: ee reviews ns ne peopeiesGS. TSK of type 2 diabetes. Diabetes and meta-analyses. Lancet 2019; 393: 434-45. 8 www-thelancet.com/diabetes-endoctinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X Articles 1 38 acarbose Jenkins DJA, Taylor RH, Goff DV, et al. Scope and specificity of 46 orChiavaroli L, Lee D, Ahmed A, et al. Effect oflow glycaemic index in slowing carbohydrate absorption in man. Diabetes 1981; load dietary patterns on glycaemic control and cardiometabolic 30: 951-54. risk factors in diabetes: systematic review and meta-analysis of 39 Chiasson J-L, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M. Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial. Lancet 2002; 359: 2072-7. 40 randomised controlled trials. BMJ 2021; 374: n1651. 47 Hu, Block G, Norkus EP, Morrow JD, Dietrich M, Hudes M. Relations of glycemic index and glycemic load with plasma oxidative stress markers. AmJ Clin Nutr 2006; 84: 70-76, 48. Ceriello A, Quagliaro L, Piconi L, et al. Effect of postprandial hypertriglyceridemia and hyperglycemia on circulating adhesion molecules and oxidative stress generation and the possible role of Holman RR, Coleman RL, Chan JCN, et al. Effects of acarbose on cardiovascular and diabetes outcomes in patients with coronary heart disease and impaired glucose tolerance (ACE): a randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol quiz 266-67, 2017; 5: 877-86. 41 Gerstein HC, Coleman RL, Scott CAB, et al. Impact of acarbose on incident diabetes and regression to normoglycemia in people with coronary heart disease and impaired glucose tolerance: insights 42 Willett W, Manson J, Liu $. Glycemic index, glycemic load, and risk of type 2 diabetes. Am J Clin Nutr 2002; 76: 274S-80S. simvastatin treatment. Diabetes 2004; 53: 701-10. 49 from the ACE trial. Diabetes Care 2020; 43: 2242-47, 295: 1681-87. 50 43. Wang Y, Kaneko T, Wang PY, Sato A. Decreased carbohydrate intake is more important than increased fat intake in the glucose intolerance by a low-carbohydrate /high-fat diet. Diabetes Res Clin Pract 2002; 55: 61-63, 45 responses in obese, prediabetic humans. Am J Clin Nutr 2010; 92: 1359-68. de Assis Costa J, de Cassia Gongalves Alfenas R. The consumption Liu S, Manson JE, Buring JE, Stampfer MJ, Willett WC, Ridker PM. Relation between a diet with a high glycemic load and plasma concentrations of high-sensitivity C-reactive protein in middle-aged J Clin Nutr 2002; 75: 492-98, women. Am 51 44 Solomon TP, Haus JM, Kelly KR, et al. A low-glycemic index diet combined with exercise reduces insulin resistance, postprandial hyperinsulinemia, and glucose-dependent insulinotropic polypeptide Monnier L, Mas E, Ginet C, et al. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 2006; Wolever TMS, Gibbs AL, Mehling C, et al. The Canadian trial of carbohydrates in diabetes (CCD), a 1-y controlled trial of low-glycemic-index dietary carbohydrate in type 2 diabetes: no effect on glycated hemoglobin but reduction in C-reactive J Clin Nutr 2008; 87: 114-25. protein. Am 52. Hernan MA. The hazards of hazard ratios. Epidemiology 2010; 21: 13-15. ofexcess low glycemic meals reduces abdominal obesity in subjects with body weight. Nutr Hosp 2012; 27: 1178-83. www.thelancet.com/diabetes-endocrinology Published online April 5, 2024 https://doi.org/10.1016/S2213-8587(24)00069-X 9