OBJECTIVE

The deleterious effects of trans fatty acids (TFAs) on cardiovascular health are well established; however, their impact on type 2 diabetes remains poorly understood. In particular, little is known about the impact of specific TFA types on type 2 diabetes etiology. We aimed to explore the associations between different types of TFAs (total, ruminant, industry produced [iTFAs], and corresponding specific isomers) and risk of type 2 diabetes.

RESEARCH DESIGN AND METHODS

A total of 105,551 participants age >18 years from the French NutriNet-Santé cohort (2009–2021) were included (mean baseline age 42.7 years; SD 14.6 years); 79.2% were women. Dietary intake data, including usual TFA intake, were collected using repeated 24-h dietary records (n = 5.7; SD 3.1). Associations between sex-specific quartile of dietary TFAs and diabetes risk were assessed using multivariable Cox models.

RESULTS

Total TFA intake was associated with higher type 2 diabetes risk (hazard ratio [HR]quartile 4 vs. 1 1.38; 95% CI 1.11–1.73; Ptrend < 0.001; n = 969 incident cases). This association, specifically observed for iTFAs (HR 1.45; 95% CI 1.15–1.83; Ptrend < 0.001), was mainly driven by elaidic acid (HR 1.37; 95% CI 1.09–1.72; Ptrend < 0.001) and linolelaidic acid (HR 1.29; 95% CI 1.04–1.58; Ptrend = 0.07). In contrast, ruminant TFAs were not significantly associated with risk of type 2 diabetes.

CONCLUSIONS

In this large prospective cohort, higher intakes of total and iTFAs were associated with increased type 2 diabetes risk. These findings support the World Health Organization’s recommendation to eliminate iTFAs from the food supply worldwide. Consumers should be advised to limit the consumption of food products containing partially hydrogenated oils (main vector of iTFAs). This may contribute to lowering the substantial global burden of type 2 diabetes.

In the last decade, harmful effects of trans fatty acids (TFAs) on cardiometabolic health have raised many concerns. According to the World Health Organization (WHO), industry-produced TFA (iTFA) intake leads to >500,000 deaths resulting from cardiovascular diseases worldwide every year (1). Because they are still widely used in many countries, this has placed iTFAs at the core of major public health policy debates. Indeed, it is now well established that iTFAs raise LDL and lower HDL cholesterol levels, thereby increasing the risk of cardiovascular diseases and death resulting from ischemic causes (2). This has also been illustrated by a study analyzing hospital admissions for myocardial infarction and stroke before and after TFA restrictions in New York (3). Consequently, the elimination of iTFAs from the global food supply has been identified as one of the priority targets of the WHO’s strategic plan (1). Specifically, the WHO calls on governments to use the REPLACE (Review, Promote, Legislate, Assess, Create, Enforce) action package (4) to eliminate iTFAs from the food supply. As a response, the EU Commission made a step in that direction in 2019 by adopting a regulation that sets a 2% legal limit on the amount of trans fat in processed foods (5). TFAs are usually classified according to two main sources: industry produced and natural. iTFAs are produced during manufacturing by partial hydrogenation of vegetable or fish oils containing unsaturated fatty acids and heating and/or frying of the oil at a very high temperature. iTFAs are found in many semisolid and solid fats (e.g., shortening, margarine, spreads) and fats used in a wide range of commercial baked foods (e.g., biscuits, cakes, crackers, fast foods, deep-fried products). In contrast, lower levels of TFAs are naturally produced in the rumen of ruminant animals and occur in dairy products and meat (rTFAs).

Although the effects of TFAs on cardiovascular health are well established (2), their impact on type 2 diabetes remains poorly understood (6). Findings from epidemiological studies on the association between circulating TFA biomarkers (68) as well as dietary intake of TFAs (917) and type 2 diabetes have shown conflicting results. In particular, some studies investigating the role of TFA biomarkers in type 2 diabetes risk have found higher risks associated with some TFA isomers (trans-palmitoleic acid 16:1n-9t; conjugated linoleic acid [CLA] t11c9) (6,8), whereas others have suggested no association (elaidic acid 18:1n-9t; linolelaidic acid 18:2t) (7,8) or inverse associations for high levels of these biomarkers (CLA t10c12; trans-palmitoleic acid 16:1n-7t) (6,7). Similarly, some studies on dietary intake of TFAs have reported no association with type 2 diabetes risk (913), while others have suggested potential adverse effects (1417). Of note, almost all dietary intake studies have used food frequency questionnaires (FFQs) including a limited number of items, and thus a lower degree of precision than repeated dietary records or recalls. Lastly, to our knowledge, none of these studies have distinguished between the different TFA sources (ruminant, industry produced), and only one examined two specific isomers of TFAs (16).

In this context, our study aimed to examine the associations between total and specific types of dietary TFAs (ruminant, industry produced, and corresponding specific isomers) and the risk of type 2 diabetes in the large-scale prospective French cohort NutriNet-Santé using detailed dietary data.

Study Population

The NutriNet-Santé study is an ongoing web-based cohort launched in 2009 in France with the objective of studying the associations between nutrition and health, as well as the determinants of dietary behaviors and nutritional status. The NutriNet-Santé study, registered at ClinicalTrials.gov as NCT03335644, is conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee of INSERM (institutional review board 0000388FWA00005831; Paris, France) and by the National Commission on Informatics and Liberty (Commission Nationale de l’Informatique et des Libertés 908450 and 909216; Paris, France). Participants provided electronic informed consent. Details about this cohort have been previously described (18). Participants were volunteers age >18 years with access to the Internet, continuously recruited from the general French population. Questionnaires are completed online on a dedicated website (www.etude-nutrinet-sante.fr), and participants are followed using a secure online platform.

Data Collection for Dietary Factors and Covariates

At baseline, participants completed a set of five questionnaires related to sociodemographic and lifestyle characteristics (e.g., sex, educational level, smoking status, alcohol intake, marital status), anthropometric data (19) (height, weight), health status (personal and family history of diseases, drug treatment), physical activity (validated 7-day International Physical Activity Questionnaire [IPAQ] (20)), and dietary intake.

At baseline and every 6 months (to vary the season of completion), participants were asked to fill in a series of three nonconsecutive web-based 24-h dietary records randomly assigned over a 2-week period (2 weekdays and 1 weekend day in order to account for variability in diet during the week). The NutriNet-Santé web-based self-administered 24-h dietary records have been tested and validated against an interview by a trained dietitian (21) and against blood and urinary biomarkers (22,23).

Participants declared all foods and drinks consumed for the three main meals (breakfast, lunch, dinner) and any additional eating occasion. For each item, the participant indicated whether it was industry produced or homemade. For an industry-produced food/beverage, the participant declared the commercial name/brand, whereas for an artisanal or homemade product, the participant had the opportunity to provide a recipe. The tool features a comprehensive user guide and a built-in control system (with visual cues and prompts), both of which help minimize the chance of forgetting consumed items. Respondents estimated portion sizes using validated photographs (24,25) or direct declaration of quantities if known. For each item, the participant also specified if it was accompanied by salt, sugar, butter, and so on, and if it was organic or not.

Mean dietary intake from 24-h dietary records available during the first 2 years of each participant’s follow-up were averaged and considered as baseline usual dietary intake in this prospective analysis. At least two dietary records were mandatory to enter the study. To assess daily intakes of alcohol, micronutrients, macronutrients, and energy (i.e., dietary covariates), consumption data were linked to the NutriNet-Santé food composition database, which contains >3,500 food and beverage items (26). Amounts consumed from composite dishes were estimated using French recipes validated by nutrition professionals. Dietary underreporting was identified by the basal metabolic rate and Goldberg cutoff using the Black method (27), and energy underreporters were excluded.

Assessment of Dietary Intake of TFAs

In order to estimate the intake of TFAs, each food item of the NutriNet-Santé database was matched with the U.S. National Nutrient Database for Standard Reference (developed by the U.S. Department of Agriculture [USDA]) (28,29). When an exact match was not found in the composition table, the value of the food, the closest equivalent, was selected. For composite food dishes, the TFA amount was estimated using validated French recipes by summing TFA amounts contained in the different ingredients composing the dish. Among the 3,716 items in the database, 3,468 (93%) had a nonnull value for lipids and 963 (26%) contained TFAs after merging with the USDA composition table. As expected, several food categories did not contain any TFAs (e.g., water, raw fruits, raw vegetables). A detailed decision tree for TFA composition assignment is presented in Supplementary Fig. 3. This matching procedure allowed us to estimate the intake of overall TFAs and specific TFA isomers and categorize them according to their natural ruminant (from meat and dairy products) or industry-produced origin. A list of all TFA-related exposure variables studied is presented in Table 1.

Table 1

Baseline characteristics of overall study population and per quartile of TFA intake, NutriNet-Santé cohort, France, 2009–2021 (N = 105,551)

CharacteristicAll participants (N = 105,551)Sex-specific quartile of total TFA intakeP*
1 (n = 26,359)2 (n = 26,403)3 (n = 26,402)4 (n = 26,387)
Age, years, mean (SD) 42.6 (14.56) 44.0 (14.94) 42.8 (14.66) 41.6 (14.27) 41.2 (14.22)  
Female sex, n (%) 83,559 (79.16) 20,870 (79.18) 20,900 (79.16) 20,904 (79.18) 20,885 (79.15) — 
Follow-up time, years, mean (SD) 6.7 (3.82) 6.2 (3.87) 6.9 (3.79) 7.1 (3.75) 6.8 (3.80)  
BMI, kg/m2, mean (SD) 23.7 (4.39) 23.4 (4.26) 23.5 (4.17) 23.6 (4.26) 24.2 (4.81) <0.0001 
Family history of type 2 diabetes, n (%) 14,587 (13.82) 3,790 (14.38) 3,637 (13.77) 3,472 (13.15) 3,688 (13.98) 0.0007 
Higher education, n (%)      <0.0001 
 No 17,314 (16.40) 4,718 (17.90) 4,273 (16.18) 3,856 (14.60) 4,467 (16.93)  
 Yes, <2 years 16,994 (16.10) 4,456 (16.91) 4,127 (15.63) 4,112 (15.57) 4,299 (16.29)  
 Yes, ≥2 years 71,243 (67.50) 17,185 (65.20) 18,003 (68.19) 18,434 (69.82) 17,621 (66.78)  
Smoking status, n (%)      <0.0001 
 Current 15,144 (14.35) 4,280 (16.24) 3,709 (14.05) 3,517 (13.32) 3,638 (13.79)  
 Former 42,644 (40.40) 11,050 (41.92) 10,906 (41.31) 10,396 (39.38) 10,292 (39.00)  
 Never 47,763 (45.25) 11,029 (41.84) 11,788 (44.65) 12,489 (47.30) 12,457 (47.21)  
IPAQ physical activity level, n (%)      <0.0001 
 High 29,888 (28.32) 8,366 (31.74) 7,501 (28.41) 6,959 (26.36) 7,062 (26.76)  
 Moderate 53,435 (50.62) 13,064 (49.56) 13,534 (51.26) 13,623 (51.60) 13,214 (50.08)  
 Low 22,228 (21.06) 4,929 (18.7) 5,368 (20.33) 5,820 (22.04) 6,111 (23.16)  
Prevalence of dyslipidemia, n (%) 8,526 (8.08) 2,290 (8.69) 2,200 (8.33) 1,904 (7.21) 2,132 (8.08) <0.0001 
Prevalence of cardiovascular disease, n (%) 1,663 (1.58) 496 (1.88) 421 (1.59) 313 (1.19) 433 (1.64) <0.0001 
Marital status, n (%)      <0.0001 
 Married 49,572 (46.96) 11,723 (44.47) 12,434 (47.09) 12,499 (47.34) 12,916 (48.95)  
 In a relationship 27,379 (25.94) 6,462 (24.52) 6,846 (25.93) 7,320 (27.73) 6,751 (25.58)  
 Divorced 8,751 (8.29) 2,549 (9.67) 2,161 (8.18) 1,958 (7.42) 2,083 (7.89)  
 Widowed 2,165 (2.05) 578 (2.19) 548 (2.08) 495 (1.87) 544 (2.06)  
 Single 17,684 (16.75) 5,047 (19.15) 4,414 (16.72) 4,130 (15.64) 4,093 (15.51)  
Energy intake without alcohol, kcal/day, mean (SD) 1,900.5 (469.08) 1,643.7 (404.22) 1,815.3 (386.95) 1,958.7 (402.2) 2,183.9 (498.46) <0.0001 
Alcohol intake, g/day, mean (SD) 7.8 (11.79) 7.9 (12.90) 8.0 (11.61) 7.7 (11.17) 7.3 (11.38) <0.0001 
SFA intake, g/day, mean (SD) 33.2 (12.11) 23.3 (8.26) 30.4 (7.96) 35.9 (8.69) 43.1 (12.96) <0.0001 
Sodium intake, mg/day, mean (SD) 2,710.1 (881.76) 2,412.4 (842.81) 2,633.3 (804.11) 2,785.7 (823.92) 3,008.7 (940.81) <0.0001 
Simple carbohydrates, g/day, mean (SD) 92.8 (33.17) 82.5 (33.21) 89.6 (29.86) 95.3 (30.75) 104.0 (34.82) <0.0001 
Whole foods, g/day, mean (SD) 34.3 (46.16) 41.5 (53.56) 33.8 (44.48) 30.6 (40.51) 31.2 (44.29) <0.0001 
Fruits and vegetables, g/day, mean (SD) 406.7 (222.20) 455.7 (271.04) 407.2 (210.38) 385.2 (195.49) 378.6 (194.79) <0.0001 
Red and processed meat, g/day, mean (SD) 101.6 (60.95) 89.0 (63.19) 99.3 (57.26) 104.6 (56.99) 113.4 (63.45) <0.0001 
Milk and dairy products, g/day, mean (SD) 82.3 (121.84) 52.6 (96.24) 74.8 (112.37) 91.2 (124.55) 110.7 (141.96) <0.0001 
Total TFA intake, g/day, mean (SD) 1.09 (0.65) 0.46 (0.16) 0.83 (0.10) 1.15 (0.13) 1.92 (0.71) <0.0001 
 iTFAs 1.00 (0.62) 0.42 (0.15) 0.75 (0.09) 1.05 (0.13) 1.78 (0.70) <0.0001 
  Elaidic acid (18:1 t) 0.95 (0.62) 0.38 (0.14) 0.71 (0.10) 1.00 (0.13) 1.72 (0.70) <0.0001 
  Linolelaidic acid (18:2 t,t)§ 5.05 (8.44) 4.85 (9.46) 4.72 (7.62) 4.96 (7.67) 5.66 (8.82) <0.0001 
  Isomer 18:2 t not further defined 0.04 (0.03) 0.03 (0.02) 0.04 (0.02) 0.05 (0.02) 0.06 (0.03) <0.0001 
  Transdocosenoic acid (22:1 t)§ 0.19 (0.37) 0.2 (0.42) 0.18 (0.35) 0.18 (0.33) 0.18 (0.36) <0.0001 
 rTFAs 0.09 (0.05) 0.05 (0.02) 0.08 (0.02) 0.10 (0.03) 0.14 (0.06) <0.0001 
  Conjugated linoleic acid (18:2 CLAs) 0.07 (0.04) 0.03 (0.01) 0.06 (0.02) 0.08 (0.02) 0.10 (0.05) <0.0001 
  Trans-vaccenic acid (18:1-11 t or 18:1t n-7) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.02 (0.02) <0.0001 
  Trans-palmitoleic acid (16:1 t) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.02 (0.01) <0.0001 
CharacteristicAll participants (N = 105,551)Sex-specific quartile of total TFA intakeP*
1 (n = 26,359)2 (n = 26,403)3 (n = 26,402)4 (n = 26,387)
Age, years, mean (SD) 42.6 (14.56) 44.0 (14.94) 42.8 (14.66) 41.6 (14.27) 41.2 (14.22)  
Female sex, n (%) 83,559 (79.16) 20,870 (79.18) 20,900 (79.16) 20,904 (79.18) 20,885 (79.15) — 
Follow-up time, years, mean (SD) 6.7 (3.82) 6.2 (3.87) 6.9 (3.79) 7.1 (3.75) 6.8 (3.80)  
BMI, kg/m2, mean (SD) 23.7 (4.39) 23.4 (4.26) 23.5 (4.17) 23.6 (4.26) 24.2 (4.81) <0.0001 
Family history of type 2 diabetes, n (%) 14,587 (13.82) 3,790 (14.38) 3,637 (13.77) 3,472 (13.15) 3,688 (13.98) 0.0007 
Higher education, n (%)      <0.0001 
 No 17,314 (16.40) 4,718 (17.90) 4,273 (16.18) 3,856 (14.60) 4,467 (16.93)  
 Yes, <2 years 16,994 (16.10) 4,456 (16.91) 4,127 (15.63) 4,112 (15.57) 4,299 (16.29)  
 Yes, ≥2 years 71,243 (67.50) 17,185 (65.20) 18,003 (68.19) 18,434 (69.82) 17,621 (66.78)  
Smoking status, n (%)      <0.0001 
 Current 15,144 (14.35) 4,280 (16.24) 3,709 (14.05) 3,517 (13.32) 3,638 (13.79)  
 Former 42,644 (40.40) 11,050 (41.92) 10,906 (41.31) 10,396 (39.38) 10,292 (39.00)  
 Never 47,763 (45.25) 11,029 (41.84) 11,788 (44.65) 12,489 (47.30) 12,457 (47.21)  
IPAQ physical activity level, n (%)      <0.0001 
 High 29,888 (28.32) 8,366 (31.74) 7,501 (28.41) 6,959 (26.36) 7,062 (26.76)  
 Moderate 53,435 (50.62) 13,064 (49.56) 13,534 (51.26) 13,623 (51.60) 13,214 (50.08)  
 Low 22,228 (21.06) 4,929 (18.7) 5,368 (20.33) 5,820 (22.04) 6,111 (23.16)  
Prevalence of dyslipidemia, n (%) 8,526 (8.08) 2,290 (8.69) 2,200 (8.33) 1,904 (7.21) 2,132 (8.08) <0.0001 
Prevalence of cardiovascular disease, n (%) 1,663 (1.58) 496 (1.88) 421 (1.59) 313 (1.19) 433 (1.64) <0.0001 
Marital status, n (%)      <0.0001 
 Married 49,572 (46.96) 11,723 (44.47) 12,434 (47.09) 12,499 (47.34) 12,916 (48.95)  
 In a relationship 27,379 (25.94) 6,462 (24.52) 6,846 (25.93) 7,320 (27.73) 6,751 (25.58)  
 Divorced 8,751 (8.29) 2,549 (9.67) 2,161 (8.18) 1,958 (7.42) 2,083 (7.89)  
 Widowed 2,165 (2.05) 578 (2.19) 548 (2.08) 495 (1.87) 544 (2.06)  
 Single 17,684 (16.75) 5,047 (19.15) 4,414 (16.72) 4,130 (15.64) 4,093 (15.51)  
Energy intake without alcohol, kcal/day, mean (SD) 1,900.5 (469.08) 1,643.7 (404.22) 1,815.3 (386.95) 1,958.7 (402.2) 2,183.9 (498.46) <0.0001 
Alcohol intake, g/day, mean (SD) 7.8 (11.79) 7.9 (12.90) 8.0 (11.61) 7.7 (11.17) 7.3 (11.38) <0.0001 
SFA intake, g/day, mean (SD) 33.2 (12.11) 23.3 (8.26) 30.4 (7.96) 35.9 (8.69) 43.1 (12.96) <0.0001 
Sodium intake, mg/day, mean (SD) 2,710.1 (881.76) 2,412.4 (842.81) 2,633.3 (804.11) 2,785.7 (823.92) 3,008.7 (940.81) <0.0001 
Simple carbohydrates, g/day, mean (SD) 92.8 (33.17) 82.5 (33.21) 89.6 (29.86) 95.3 (30.75) 104.0 (34.82) <0.0001 
Whole foods, g/day, mean (SD) 34.3 (46.16) 41.5 (53.56) 33.8 (44.48) 30.6 (40.51) 31.2 (44.29) <0.0001 
Fruits and vegetables, g/day, mean (SD) 406.7 (222.20) 455.7 (271.04) 407.2 (210.38) 385.2 (195.49) 378.6 (194.79) <0.0001 
Red and processed meat, g/day, mean (SD) 101.6 (60.95) 89.0 (63.19) 99.3 (57.26) 104.6 (56.99) 113.4 (63.45) <0.0001 
Milk and dairy products, g/day, mean (SD) 82.3 (121.84) 52.6 (96.24) 74.8 (112.37) 91.2 (124.55) 110.7 (141.96) <0.0001 
Total TFA intake, g/day, mean (SD) 1.09 (0.65) 0.46 (0.16) 0.83 (0.10) 1.15 (0.13) 1.92 (0.71) <0.0001 
 iTFAs 1.00 (0.62) 0.42 (0.15) 0.75 (0.09) 1.05 (0.13) 1.78 (0.70) <0.0001 
  Elaidic acid (18:1 t) 0.95 (0.62) 0.38 (0.14) 0.71 (0.10) 1.00 (0.13) 1.72 (0.70) <0.0001 
  Linolelaidic acid (18:2 t,t)§ 5.05 (8.44) 4.85 (9.46) 4.72 (7.62) 4.96 (7.67) 5.66 (8.82) <0.0001 
  Isomer 18:2 t not further defined 0.04 (0.03) 0.03 (0.02) 0.04 (0.02) 0.05 (0.02) 0.06 (0.03) <0.0001 
  Transdocosenoic acid (22:1 t)§ 0.19 (0.37) 0.2 (0.42) 0.18 (0.35) 0.18 (0.33) 0.18 (0.36) <0.0001 
 rTFAs 0.09 (0.05) 0.05 (0.02) 0.08 (0.02) 0.10 (0.03) 0.14 (0.06) <0.0001 
  Conjugated linoleic acid (18:2 CLAs) 0.07 (0.04) 0.03 (0.01) 0.06 (0.02) 0.08 (0.02) 0.10 (0.05) <0.0001 
  Trans-vaccenic acid (18:1-11 t or 18:1t n-7) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.02 (0.02) <0.0001 
  Trans-palmitoleic acid (16:1 t) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.01 (0.01) 0.02 (0.01) <0.0001 
*

P values for crude comparison between sex-specific quartiles of total TFA intake, by ANOVA or χ2 test where appropriate.

Sex-specific quartiles of total TFA intake; sex-specific cutoffs for quartiles were 0.76, 1.10, and 1.54 g/day in men and 0.66, 0.95, and 1.30 g/day in women.

1 kcal = 4.18 kJ = 0.00418 MJ.

§

mg/day.

Case Ascertainment

Participants’ health status was reported at enrollment and reassessed every 6 months using a checkup questionnaire. At any time during follow-up, participants could also declare a health event, a new treatment, or a medical examination via a dedicated and secure online platform. In addition, data were linked to medico-administrative databases of the national health insurance system (SNIIRAM [Système National d’Informations Inter-Régimes de l’Assurance Maladie]), providing detailed information about the reimbursement of medication and medical consultations. Details on type 2 diabetes case ascertainment (ICD-10 code E11) are provided in Supplementary Method 1, including Anatomical Therapeutic Chemical Classification codes related to type 2 diabetes medication. Mortality cases were identified using a linkage to CépiDC, the French national mortality registry. In this study, all incident type 2 diabetes diagnoses occurring between inclusion and October 2021 were considered cases.

Statistical Analysis

For this study, we included participants from the NutriNet-Santé cohort who completed at least two 24-h dietary records during their first 2 years of follow-up, were considered normo-energy reporters, and did not have prevalent type 1 or 2 diabetes diagnosed at baseline. Baseline characteristics of the study population were described across sex-specific quartiles of TFA intake and compared using ANOVAs for continuous variables or χ2 tests for categorical variables.

Associations between TFA intakes (overall and each type) and type 2 diabetes risk were investigated using Cox proportional hazards models with age as the primary time scale. Participants contributed person-time from their inclusion in the cohort until the date of type 2 diabetes diagnosis, date of last completed questionnaire, date of death, or end point date (October 2021), whichever occurred first. Type 1 diabetes was censored at the date of diagnosis. Cause-specific models were computed so that death during follow-up was handled as a competing event. Hazard ratios (HRs) and 95% CIs were estimated, with quartile 1 settled as the reference category. P for trend was obtained by using the median intake in each quartile. The proportional hazards assumption of the Cox model was tested and confirmed with the rescaled Schoenfeld-type residuals method (Supplementary Fig. 2AC). The linearity assumption was verified using restricted cubic spline functions (30) (Fig. 2). The multiple imputation by chained equations method was used to impute covariates with missing values (Supplementary Method 1). Main models were adjusted for age (time scale), sex, BMI, physical activity, smoking status, educational level, number of 24-h dietary records, family history of type 2 diabetes, prevalence of cardiovascular disease (including coronary and cerebrovascular diseases), prevalence of dyslipidemia, marital status, and daily intakes of energy without alcohol, alcohol, sodium, saturated fatty acids (SFAs), sugar, whole-grain foods, red and processed meat, fruits, and vegetables. In all models, TFAs were mutually adjusted for types of TFAs other than the one studied. A more parsimonious model adjusted only for age, sex, and energy intake was also tested (Table 2), and an intermediate model adjusted only for clinical factors (i.e., excluding dietary factors) (Supplementary Table 8). A series of sensitivity analyses were performed to assess the robustness of the findings by excluding type 2 diabetes cases occurring during the first 2 years of follow-up (to challenge the potential reverse causality bias), restricting the study population to participants with at least four 24-h dietary records during the first 2 years, excluding incident cases of type 1 diabetes from the study population, and further adjusting for dietary patterns (healthy and Western, derived by principal component analysis) (Supplementary Method 3) instead of fruit, vegetable, and meat intakes and for fiber instead of whole-grain intake. Main models were also further adjusted for the proportion of ultraprocessed food (as percentage of daily consumption in grams) in the total diet. We also tested a model restricted to nonsmokers in order to eliminate any residual confounding by tobacco smoking. We also computed the FSA-NPS DI (Food Standard Agency Nutrient Profiling System Dietary Index) (underlying the French food label NutriScore), a diet quality scoring system that includes SFA intake as a metric (31), and calculated the Pearson correlation coefficient between TFAs and 1) SFAs and 2) FSA-NPS DI score. We also performed substitution analyses by entering both SFAs and TFAs into the model. HRs and 95% CIs for substituting TFAs for SFAs were estimated using the difference in coefficients obtained from this model.

Table 2

Associations between TFA intake and type 2 diabetes risk, NutriNet-Santé cohort, France, 2009–2021 (N = 105,551)

Sex-specific quartile
1234Ptrend
Parsimonious model*      
 Total TFA 199/26,359 221/26,403 242/26,402 307/26,387  
   1.02 (0.84–1.25) 1.18 (0.96–1.43) 1.53 (1.25–1.86) <0.001 
 iTFAs 199/26,364 218/26,401 245/26,399 307/26,387  
   0.99 (0.81–1.21) 1.15 (0.93–1.42) 1.43 (1.14–1.79) <0.001 
  Elaidic acid (18:1 t) 202/26,367 219/26,397 242/26,399 306/26,388  
    0.96 (0.78–1.17) 1.08 (0.88–1.32) 1.31 (1.06–1.61) 0.002 
  Linolelaidic acid (18:2 t,t) 157/26,344 221/26,404 295/26,417 296/26,386  
    1.14 (0.93–1.41) 1.50 (1.23–1.83) 1.41 (1.15–1.72) 0.001 
  Isomer 18:2 t not further defined 179/26,362 230/26,397 274/26,416 286/26,376  
    1.14 (0.93–1.39) 1.44 (1.18–1.75) 1.71 (1.39–2.09) <0.001 
  Trans-docosenoic acid (22:1 t) 185/28,654 254/25,596 237/25,687 293/25,614  
    1.06 (0.87–1.29) 0.96 (0.79–1.17) 1.03 (0.85–1.25) 0.8 
 rTFAs 195/26,364 256/26,394 256/26,409 262/26,384  
   1.14 (0.94–1.38) 1.15 (0.94–1.40) 1.32 (1.06–1.63) 0.02 
  Conjugated linoleic acid (18:2 CLAs) 205/26,363 260/26,399 260/26,410 244/26,379  
    1.14 (0.95–1.38) 1.15 (0.95–1.40) 1.16 (0.93–1.44) 0.2 
  Trans-vaccenic acid (18:1–11 t or 18:1t n-7) 174/26,321 256/26,416 278/26,432 261/26,382  
    1.05 (0.86–1.28) 1.08 (0.89–1.31) 1.12 (0.91–1.36) 0.3 
  Trans-palmitoleic acid (16:1 t) 195/26,374 232/26,393 267/26,408 275/26,376  
    0.98 (0.81–1.19) 1.13 (0.93–1.37) 1.28 (1.05–1.55) 0.002 
Main model      
 Total TFAs 199/26,359 221/26,403 242/26,402 307/26,387  
   0.96 (0.78–1.17) 1.14 (0.92–1.41) 1.38 (1.11–1.73) <0.001 
 iTFAs 199/26,364 218/26,401 245/26,399 307/26,387  
   0.97 (0.79–1.19) 1.20 (0.97–1.49) 1.45 (1.15–1.83) <0.001 
  Elaidic acid (18:1 t) 202/26,367 219/26,397 242/26,399 306/26,388  
    0.96 (0.78–1.18) 1.15 (0.93–1.42) 1.37 (1.09–1.72) <0.001 
  Linolelaidic acid (18:2 t,t) 157/26,344 221/26,404 295/26,417 296/26,386  
    1.18 (0.95–1.47) 1.37 (1.11–1.69) 1.29 (1.04–1.58) 0.07 
  Isomer 18:2 t not further defined 179/26,362 230/26,397 274/26,416 286/26,376  
    1.07 (0.87–1.31) 1.28 (1.05–1.57) 1.31 (1.06–1.62) 0.006 
  Trans-docosenoic acid (22:1 t) 185/28,654 254/25,596 237/25,687 293/25,614  
    1.08 (0.88–1.32) 0.97 (0.78–1.19) 1.00 (0.82–1.23) 0.8 
 rTFAs 195/26,364 256/26,394 256/26,409 262/26,384  
   1.06 (0.87–1.29) 1.11 (0.90–1.38) 1.25 (0.96–1.61) 0.09 
  Conjugated linoleic acid (18:2 CLAs) 205/26,363 260/26,399 260/26,410 244/26,379  
    1.08 (0.89–1.31) 1.12 (0.90–1.39) 1.09 (0.83–1.42) 0.6 
  Trans-vaccenic acid (18:1–11 t or 18:1t n-7) 174/26,321 256/26,416 278/26,432 261/26,382  
    1.14 (0.93–1.40) 1.18 (0.96–1.45) 1.18 (0.95–1.46) 0.2 
  Trans-palmitoleic acid (16:1 t) 195/26,374 232/26,393 267/26,408 275/26,376  
    1.00 (0.82–1.21) 1.04 (0.86–1.27) 1.20 (0.98–1.46) 0.03 
Sex-specific quartile
1234Ptrend
Parsimonious model*      
 Total TFA 199/26,359 221/26,403 242/26,402 307/26,387  
   1.02 (0.84–1.25) 1.18 (0.96–1.43) 1.53 (1.25–1.86) <0.001 
 iTFAs 199/26,364 218/26,401 245/26,399 307/26,387  
   0.99 (0.81–1.21) 1.15 (0.93–1.42) 1.43 (1.14–1.79) <0.001 
  Elaidic acid (18:1 t) 202/26,367 219/26,397 242/26,399 306/26,388  
    0.96 (0.78–1.17) 1.08 (0.88–1.32) 1.31 (1.06–1.61) 0.002 
  Linolelaidic acid (18:2 t,t) 157/26,344 221/26,404 295/26,417 296/26,386  
    1.14 (0.93–1.41) 1.50 (1.23–1.83) 1.41 (1.15–1.72) 0.001 
  Isomer 18:2 t not further defined 179/26,362 230/26,397 274/26,416 286/26,376  
    1.14 (0.93–1.39) 1.44 (1.18–1.75) 1.71 (1.39–2.09) <0.001 
  Trans-docosenoic acid (22:1 t) 185/28,654 254/25,596 237/25,687 293/25,614  
    1.06 (0.87–1.29) 0.96 (0.79–1.17) 1.03 (0.85–1.25) 0.8 
 rTFAs 195/26,364 256/26,394 256/26,409 262/26,384  
   1.14 (0.94–1.38) 1.15 (0.94–1.40) 1.32 (1.06–1.63) 0.02 
  Conjugated linoleic acid (18:2 CLAs) 205/26,363 260/26,399 260/26,410 244/26,379  
    1.14 (0.95–1.38) 1.15 (0.95–1.40) 1.16 (0.93–1.44) 0.2 
  Trans-vaccenic acid (18:1–11 t or 18:1t n-7) 174/26,321 256/26,416 278/26,432 261/26,382  
    1.05 (0.86–1.28) 1.08 (0.89–1.31) 1.12 (0.91–1.36) 0.3 
  Trans-palmitoleic acid (16:1 t) 195/26,374 232/26,393 267/26,408 275/26,376  
    0.98 (0.81–1.19) 1.13 (0.93–1.37) 1.28 (1.05–1.55) 0.002 
Main model      
 Total TFAs 199/26,359 221/26,403 242/26,402 307/26,387  
   0.96 (0.78–1.17) 1.14 (0.92–1.41) 1.38 (1.11–1.73) <0.001 
 iTFAs 199/26,364 218/26,401 245/26,399 307/26,387  
   0.97 (0.79–1.19) 1.20 (0.97–1.49) 1.45 (1.15–1.83) <0.001 
  Elaidic acid (18:1 t) 202/26,367 219/26,397 242/26,399 306/26,388  
    0.96 (0.78–1.18) 1.15 (0.93–1.42) 1.37 (1.09–1.72) <0.001 
  Linolelaidic acid (18:2 t,t) 157/26,344 221/26,404 295/26,417 296/26,386  
    1.18 (0.95–1.47) 1.37 (1.11–1.69) 1.29 (1.04–1.58) 0.07 
  Isomer 18:2 t not further defined 179/26,362 230/26,397 274/26,416 286/26,376  
    1.07 (0.87–1.31) 1.28 (1.05–1.57) 1.31 (1.06–1.62) 0.006 
  Trans-docosenoic acid (22:1 t) 185/28,654 254/25,596 237/25,687 293/25,614  
    1.08 (0.88–1.32) 0.97 (0.78–1.19) 1.00 (0.82–1.23) 0.8 
 rTFAs 195/26,364 256/26,394 256/26,409 262/26,384  
   1.06 (0.87–1.29) 1.11 (0.90–1.38) 1.25 (0.96–1.61) 0.09 
  Conjugated linoleic acid (18:2 CLAs) 205/26,363 260/26,399 260/26,410 244/26,379  
    1.08 (0.89–1.31) 1.12 (0.90–1.39) 1.09 (0.83–1.42) 0.6 
  Trans-vaccenic acid (18:1–11 t or 18:1t n-7) 174/26,321 256/26,416 278/26,432 261/26,382  
    1.14 (0.93–1.40) 1.18 (0.96–1.45) 1.18 (0.95–1.46) 0.2 
  Trans-palmitoleic acid (16:1 t) 195/26,374 232/26,393 267/26,408 275/26,376  
    1.00 (0.82–1.21) 1.04 (0.86–1.27) 1.20 (0.98–1.46) 0.03 

Values are given as incident cases/participants and cause-specific HRs (95% CIs). During overall follow-up, 1,111 competing deaths occurred. Cause-specific HRs for death in the fourth quartile compared with the first quartile were 1.04 (0.85–1.28) for total TFA intake, 1.07 (0.85–1.35) for rTFA intake, and 1.08 (0.88–1.31) for iTFA intake.

*

Parsimonious model: multivariable Cox proportional hazards model adjusted for (= main model) age (time scale), sex, and baseline intakes of energy without alcohol (continuous, kcal/day).

Main model: multivariable Cox proportional hazards model adjusted for (= main model) age (time scale), sex, BMI (continuous, kg/m2), physical activity (categorical IPAQ variable: high, moderate, or low), smoking status (categorical: never, former, or current smoker), educational level (categorical: less than high school degree, <2 years after high school degree, or ≥2 years after high school degree), number of 24-h dietary records (continuous), family history of type 2 diabetes (categorical: yes or no), prevalence of cardiovascular disease (categorical: yes or no), prevalence of dyslipidemia (categorical: yes or no), marital status (categorical: married, in a relationship, divorced, widowed, or single), and baseline intakes of energy without alcohol (continuous, kcal/day), alcohol (continuous, g/day), sodium (continuous, g/day), SFAs (continuous, g/day), sugar (continuous, g/day), whole-grain foods (continuous, g/day), red and processed meat (continuous, g/day), fruits, and vegetables (continuous, g/day).

All tests were two sided, and P < 0.05 was considered statistically significant. The statistical analysis software SAS (version 9.4) was used for analyses.

Data and Resource Availability

The results of the current study will be disseminated to the NutriNet-Santé participants through the cohort website, public seminars, and a press release.

A total of 105,551 participants (79.2% women) were included in the analyses (detailed flowchart in Supplementary Fig. 1). Mean age at baseline was 42.6 years (SD 14.6 years). Table 1 shows the baseline characteristics of the study population. Compared with those in the first quartile, participants in the highest quartile tended to be younger, more educated, and nonsmokers. They had a higher BMI, a lower physical activity level, and were less likely to have a family history of type 2 diabetes. They had a lower intake of alcohol and a higher intake of energy. Mean daily intake was 1.09 g/day (SD 0.65 g/day) for total TFAs, 0.09 g/day (SD 0.05 g/day) for rTFAs, and 1.00 g/day (SD 0.62 g/day) for iTFAs, corresponding to 92% (+SD) of TFA intake. The relative contribution of each food group to TFA intake is shown in Fig. 1 for total TFAs, in Supplementary Fig. 4AE for iTFAs, and Supplementary Fig. 5AD for rTFAs and their corresponding isomers. Margarine (50%), biscuits/cakes/pastries (13%), and cheese (9%) were the main contributors to total TFAs as well as iTFAs (margarine 38%; biscuits/cakes/pastries 31%; cheese 7%), and biscuits/cakes/pastries (37%), cheese (28%), and delicatessen (8%) were the main contributors to rTFAs.

Figure 1

Contribution of food groups to total TFA intake, NutriNet-Santé cohort, France, 2009–2021 (N = 105,551).

Figure 1

Contribution of food groups to total TFA intake, NutriNet-Santé cohort, France, 2009–2021 (N = 105,551).

Close modal

During follow-up (median follow-up time 7 years), 969 incident type 2 diabetes cases were reported. Mean age at diagnosis was 59.2 years (SD 11.2 years). The proportional hazards assumptions of Cox models were met (Supplementary Fig. 2AC), as was the assumption of linear dose response (Fig. 2).

Figure 2

Spline plots for the linearity assumption of the association between total TFAs (A), iTFAs (B), and rTFAs (C) and risk of incident type 2 diabetes using the restricted cubic spline SAS macro (30). CL, confidence limit.

Figure 2

Spline plots for the linearity assumption of the association between total TFAs (A), iTFAs (B), and rTFAs (C) and risk of incident type 2 diabetes using the restricted cubic spline SAS macro (30). CL, confidence limit.

Close modal

Table 2 shows the HRs for the associations between dietary intake of TFAs and risk of type 2 diabetes. Total TFA intake was significantly associated with a higher type 2 diabetes risk (HRquartile 4 vs. 1 1.38; 95% CI 1.11–1.73; Ptrend < 0.001) (Table 2). In particular, a direct association was observed for iTFAs (HR 1.45; 95% CI 1.15–1.83; Ptrend < 0.001), specifically for elaidic acid 18:1n-9trans (HR 1.37; 95% CI 1.09–1.72; Ptrend < 0.001), linolelaidic acid (HR 1.29; 95% CI 1.04–1.58; Ptrend = 0.07), and the 18:2t isomer (HR 1.31; 95% CI 1.06–1.62; Ptrend = 0.006). Overall, these findings remained similar throughout all sensitivity analyses (Supplementary Tables 18). In addition, the sensitivity analysis further adjusting for principal component analysis–derived healthy and Western dietary patterns and restricting the study population to participants with at least four 24-h dietary records, higher intake of the rTFA trans-palmitoleic acid isomer was associated with a higher risk of type 2 diabetes (HRquartile 4 vs. 1 1.37; 95% CI 1.10–1.70; Ptrend = 0.007 and HRquartile 4 vs. 1 1.27; 95% CI 1.05–1.55; Ptrend = 0.006) (Supplementary Tables 2 and 3, respectively). The Pearson correlation coefficient between TFAs and SFAs was 0.57, and between TFAs and the FSA-NPS DI score, it was 0.33, which did not support a spontaneous substitution of TFAs for SFAs by consumers. Substitution analyses did not suggest a benefit for substituting TFAs for SFAs in type 2 diabetes risk (HR 1.03; 95% CI 0.93–6.27).

In this large prospective cohort study of 105,551 participants, higher dietary intake of total TFAs was associated with higher risk of type 2 diabetes. In particular, a positive association was found for iTFAs, where the association was mainly driven by elaidic acid, linolelaidic acid, and other 18:2t isomers not further defined. In contrast, rTFAs were not significantly associated with risk of type 2 diabetes.

Our results are consistent with findings from four prospective studies investigating the association of TFA intake with type 2 diabetes incidence (1417). Particularly, in the Cardiovascular Health Study (16), the iTFA linolelaidic acid (18:2t isomer) was positively associated with incident type 2 diabetes, with a 41% increased risk in the upper quartile compared with the lowest, in line with our results. In the Nurse’s Health Study (14), the authors found a significantly higher risk of type 2 diabetes with higher intake of total TFAs, with a 20% increased risk in the third and fourth quintiles and 39% greater risk in the fifth quintile. Another study conducted in the same population (15) suggested a significant relative risk of 1.39 (95% CI 1.15–1.67; P = 0.0006) for a 2% increase in energy from TFAs. Similarly, higher intake of TFAs among men in the Health Professionals Follow-up Study was significantly associated with a 28% higher risk of type 2 diabetes (17). In contrast, the other prospective studies examining the relationship between dietary TFA intake and the risk of incident type 2 diabetes reported no association (913), consistent with a recent meta-analysis including seven prospective cohorts among the aforementioned (32). The reasons underlying these conflicting results remain unclear. However, they may be due to various factors including levels of TFA intake, differences in food composition tables used, and in dietary assessment methods. Indeed, most of these studies used FFQs with a lower degree of precision than the repeated dietary records used in the current study. In addition, the lack of specification of TFA isomers of interest, as well as their natural or industry-produced origin, in most of the studies may explain the inconsistency in the results.

Interestingly, this inconsistency in results from studies on dietary TFA intake and risk of type 2 diabetes is also observed among studies based on plasma circulating TFAs (68). As an example, although high levels of the plasma phospholipid iTFA isomer 18:1t were significantly associated with higher risk of type 2 diabetes in the EPIC (European Prospective Investigation into Cancer and Nutrition)-Potsdam cohort study (8) and the Cardiovascular Health Study (16), a protective association was found in FORCE (Pooled Analysis of 12 Prospective Cohort Studies in the Fatty Acids and Outcomes Research Consortium) (6). The study by Prada et al. (8) on circulating TFAs on the risk of type 2 diabetes in the EPIC-Potsdam cohort also found a protective association between CLA isomers of rTFAs and the risk of type 2 diabetes, whereas no association was found for CLA in the Melbourne Collaborative Cohort Study (7). Comparing results from dietary intake versus biomarkers of TFAs in association with health outcomes is not straightforward. Indeed, the two approaches are complementary; biomarkers provide a more objective assessment of the exposure, but they reflect indistinctly both intake and individual metabolism of TFAs. In contrast, studies based on TFA intake more directly reflect dietary consumption (i.e., the modifiable risk factor of interest) and are generally adjusted for a wide range of variables reflecting diet quality and other dietary exposures.

At a molecular level, differences in biological mechanisms between rTFAs and iTFAs have been examined using preclinical models in animals and cultured cells. Evidence from both preclinical and clinical studies shows that iTFAs and rTFAs can behave similarly or differentially, depending on the biological pathway and clinical parameters under investigation (33). In numerous examples of bioactive molecules, a slight change in molecular structure has a profound impact on biological properties. Indeed, the position of the trans double bond could affect the extent to which fatty acids are taken up, sensed, metabolized, and incorporated into cellular organelles and membranes. Such potential mechanisms are still poorly demonstrated and warrant further investigation.

Biological pathways that might account for an effect of TFAs on the incidence of type 2 diabetes are yet not well established. Animal experimental models have shown that TFAs may reduce insulin sensitivity, known to play an important role in the pathogenesis of type 2 diabetes, through downregulation of peroxisome proliferator-activated receptor γ and lipoprotein lipase (34,35). In vitro and in vivo, greater TFA exposure induces genetic expression of sterol regulatory element-binding proteins and suppression of triglyceride transfer protein expression, leading to a stimulation of hepatic de novo lipogenesis and generation of nonalcoholic steatohepatitis-like lesions, conditions closely linked to insulin resistance (36).

Our study has several strengths, including its prospective design and a detailed and up-to-date assessment of dietary intake. Multiple 24-h dietary records (including >3,500 different food items) allowed more accurate quantification of regular TFA intake than FFQs with aggregated food groups or household purchasing data. Indeed, regarding industry-produced foods/beverages, the commercial names/brands were collected in the NutriNet-Santé 24-h dietary records, but brand-specific TFA composition was not available from manufacturers. Therefore, average TFA values per specific generic food items had to be applied. However, the level of detail in the NutriNet-Santé nomenclature is extensive (one of the most extensive globally for this type of epidemiological study) and led to a refined assessment of TFA exposure. For instance, the typical FFQ item “chocolate candy bar,” the NutriNet-Santé nomenclature presents four distinct items (the most relevant for the French market: “cookie chocolate candy bar–KitKat or similar,” “non-cookie chocolate candy bar–Mars or similar,” “chocolate candy bar with peanuts–Snickers or similar,” and “chocolate candy bar with coconut–Bounty or similar”) that could be selected by the participants and that corresponded to different TFA contents. Similarly, for a generic FFQ item like “cheese,” 112 distinct items were proposed in NutriNet-Santé, with distinct TFA composition values. However, our study also has several limitations. Firstly, as in most observational cohorts, participants were volunteers recruited from the general population. As a consequence, and as generally observed for this type of study, participants in the NutriNet-Santé cohort were more often women, with health-conscious behaviors and higher socioprofessional and educational levels than the general French population (37). This may have limited the generalizability of the findings and may have resulted in an overall lower exposure to TFAs, with less contrast between extreme categories, reducing statistical power in particular because of a lower consumption of ruminant meat. However, the geographical distribution of participants was close to that of metropolitan France (38), and the Internet-based recruitment was well suited to France, where 93% of households have access to the Internet (39). In addition, the mean TFA intake assessed in our study (1.09 g/day) was relatively similar to those observed in Austria (1.0 g/day), Finland (1.0 g/day), and Belgium (0.9 g/day) (40). Moreover, in etiological studies, diversity of the studied exposure (total TFA intake was contrasted between quartiles 1 and 4; mean 0.46 g/day; SD 0.16 g/day versus 1.92 g/day; SD 0.71 g/day) and the capacity to account for a broad range of associated behaviors (more than representativeness in itself) are the important features. Besides, the 30–35% proportion of energy intake from ultraprocessed foods (main sources of iTFAs) was similar in NutriNet-Santé (41) and in a representative sample of the French population (ENNS [Etude Nationale Nutrition Santé] survey [37]). Secondly, to determine TFA composition, we used USDA data (28), which are not specific to the French market. Indeed, to our knowledge, this database compiles the most complete data on TFA composition, including details on different sources and isomers. Potential classification bias resulting from differences in food composition between the U.S. and France cannot be excluded. Specific biomarkers of TFAs were not available in this cohort. However, two points should be noted. 1) Two validation studies performed in the NutriNet-Santé web-based cohort investigated the validity of the web-based self-administered dietary record tool against biomarkers of nutritional status (22,23). The first study showed high correlations between intakes of fish, fruits, and vegetables, β-carotene, vitamin C, and n-3 polyunsaturated fatty acids, as estimated by the NutriNet-Santé 24-h dietary record and plasma concentrations of the corresponding nutrients. The second study showed that the web-based dietary record tool used in the NutriNet-Santé cohort performs well in estimating protein and vitamin K intakes, and fairly well in estimating sodium intake, as measured by 24-h urinary biomarkers. 2) In addition, in the European EPIC cohort, based on an FFQ (known to be generally less precise than repeated dietary records [42]) and using the same USDA food composition table as in the current study, good correlations were observed between dietary TFA intake and plasmatic biomarkers of TFAs (e.g., Spearman correlation coefficient between plasmatic and dietary iTFAs and for the specific iTFA elaidic acid 0.526 and 0.528, respectively) (43). Because plasmatic TFAs not only reflect dietary intake but also endogenous metabolism, higher correlations were not expected. Besides, consistent associations with various outcomes have been observed for dietary and plasmatic TFAs in the EPIC cohort (e.g., for epithelial ovarian cancer [44]). Next, even though we used a multisource case ascertainment approach, exhaustiveness could not be guaranteed. Approximately 20% of type 2 diabetes cases are estimated to be underdiagnosed in France (45). This probably resulted in a loss of statistical power, and, because of the prospective design, the resulting potential misclassification bias was most likely nondifferential and rather resulted in an underestimation of the associations. Lastly, causation could not be established from this single observational study, and because of the observational design of this study, residual confounding from unmeasured behavioral factors or imprecision in the measure of included covariates cannot be ruled out. Once again, this most probably tended to underestimate the strength of the associations. However, the possibility that this bias may have led to an overestimation of some associations cannot be totally excluded. However, this was most likely limited because of the adjustment for a large number of potential confounding factors.

To conclude, results from this large perspective cohort suggest that dietary intake of several types of TFAs from industry-produced sources are associated with increased type 2 diabetes risk. These epidemiological observations are supported by consistent mechanistic plausibility from experimental data. These findings add to the growing body of evidence on the role of TFAs in the risk of type 2 diabetes, and supports the WHO’s recommendation to eliminate iTFAs from the food supply worldwide. In the meantime, consumers and patients should be advised to avoid the consumption of food products containing partially hydrogenated oils. This may contribute to lowering the substantial global burden of type 2 diabetes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.21543030.

This article is featured in a podcast available at diabetesjournals.org/journals/pages/diabetes-core-update-podcasts.

Acknowledgments. The authors thank Alexandre De-Sa and Rebecca Lutchia (dietitians), Younes Esseddik (information technology manager), Thi Hong Van Duong, Régis Gatibelza, Jagatjit Mohinder, and Aladi Timera (computer scientists), Fabien Szabo de Edelenyi (manager), Julien Allegre, Nathalie Arnault, Laurent Bourhis, and Nicolas Dechamp (data managers/statisticians), Merveille Kouam (health event validator), and Maria Gomes (NutriNet-Santé volunteer support) (all from Paris Nord University) for their technical contribution to the NutriNet-Santé study. The authors also thank all the volunteers of the NutriNet-Santé cohort.

Funding. The NutriNet-Santé study was supported by the following public institutions: Ministère de la Santé, Santé Publique France, INSERM, Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement, Conservatoire National des Arts et Métiers; and Université Sorbonne Paris Nord (Université Paris 13). This project was awarded the NACRe (French Network for Nutrition and Cancer Research) partnership label. G.W.-F. was supported by a grant from the CARPEM (Cancer Research for Personalized Medicine) Institute, funded by the French National Cancer Institute. L.S. and C.D. were supported by a grant from the French National Cancer Institute. M.T. received funding from the European Research Council under the EU Horizon 2020 research and innovation program (grant agreement 864219) and a research prize from the Bettencourt Foundation (2021).

Researchers were independent from funders. Funders had no role in the study design; collection, analysis, or interpretation of data; writing of the report, or decision to submit the article for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Where authors are identified as personnel of the International Agency for Research on Cancer (IARC)/WHO, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policy, or views of the IARC/WHO.

Author Contributions. G.W.-F. performed the statistical analysis and drafted the manuscript. G.W.-F., V.C., I.H., M.D.-T., and M.T. designed the research. L.K.F., C.J., E.K.-G., C.A., N.D.-P., P.G., S.H., M.D.-T., and M.T. collected the data. M.T. supervised the writing of the manuscript. G.W.-F., A.B., V.C., I.H., J.-M.B., C.D., B.S., L.S., L.K.F., C.J., E.K.-G., C.A., N.D.-P., P.G., S.H., M.D.-T., and M.T. contributed to data interpretation, revised each manuscript draft for important intellectual content, and read and approved the final manuscript. M.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
World Health Organization
.
Countdown to 2023: WHO Report on Global Trans-Fat Elimination 2020
.
Accessed 10 January 2022. Available from https://www.who.int/publications/i/item/9789240010178
2.
Chowdhury
R
,
Warnakula
S
,
Kunutsor
S
, et al
.
Association of dietary, circulating, and supplement fatty acids with coronary risk: a systematic review and meta-analysis
.
Ann Intern Med
2014
;
160
:
398
406
3.
Brandt
EJ
,
Myerson
R
,
Perraillon
MC
,
Polonsky
TS
.
Hospital admissions for myocardial infarction and stroke before and after the trans-fatty acid restrictions in New York
.
JAMA Cardiol
2017
;
2
:
627
634
4.
World Health Organization
.
REPLACE Trans Fat: An Action Package to Eliminate Industrially-Produced Trans Fat From the Global Food Supply: Trans Fat Free by 2023
.
Accessed 10 January 2022. Available from https://apps.who.int/iris/handle/10665/331303
5.
European Union
.
Commission Regulation (EU) 2019/649 of 24 April 2019 Amending Annex III to Regulation (EC) No 1925/2006 of the European Parliament and of the Council As Regards Trans Fat, Other Than Trans Fat Naturally Occurring in Fat of Animal Origin (text with EEA relevance)
.
6.
Lai
HTM
,
Imamura
F
,
Korat
AVA
, et al.;
Fatty Acids and Outcomes Research Consortium (FORCE)
.
Trans fatty acid biomarkers and incident type 2 diabetes: pooled analysis of 12 prospective cohort studies in the Fatty Acids and Outcomes Research Consortium (FORCE)
.
Diabetes Care
2022
;
45
:
854
863
7.
Hodge
AM
,
English
DR
,
O’Dea
K
, et al
.
Plasma phospholipid and dietary fatty acids as predictors of type 2 diabetes: interpreting the role of linoleic acid
.
Am J Clin Nutr
2007
;
86
:
189
197
8.
Prada
M
,
Wittenbecher
C
,
Eichelmann
F
, et al
.
Plasma industrial and ruminant trans fatty acids and incident type 2 diabetes in the EPIC-Potsdam cohort
.
Diabetes Care
2022
;
45
:
845
853
9.
Gaeini
Z
,
Bahadoran
Z
,
Mirmiran
P
,
Djazayery
A
.
The association between dietary fat pattern and the risk of type 2 diabetes
.
Prev Nutr Food Sci
2019
;
24
:
1
7
10.
Meyer
KA
,
Kushi
LH
,
Jacobs
DR
Jr
,
Folsom
AR
.
Dietary fat and incidence of type 2 diabetes in older Iowa women
.
Diabetes Care
2001
;
24
:
1528
1535
11.
Song
Y
,
Manson
JE
,
Buring
JE
,
Liu
S
.
A prospective study of red meat consumption and type 2 diabetes in middle-aged and elderly women: the Women’s Health Study
.
Diabetes Care
2004
;
27
:
2108
2115
12.
Dow
C
,
Mangin
M
,
Balkau
B
, et al
.
Fatty acid consumption and incident type 2 diabetes: an 18-year follow-up in the female E3N (Etude Epidémiologique auprès des femmes de la Mutuelle Générale de l’Education Nationale) prospective cohort study
.
Br J Nutr
.
15 November 2016 [Epub ahead of print]. DOI: 10.1017/S0007114516003883
13.
Guasch-Ferré
M
,
Becerra-Tomás
N
,
Ruiz-Canela
M
, et al
.
Total and subtypes of dietary fat intake and risk of type 2 diabetes mellitus in the Prevención con Dieta Mediterránea (PREDIMED) study
.
Am J Clin Nutr
2017
;
105
:
723
735
14.
Hu
FB
,
Manson
JE
,
Stampfer
MJ
, et al
.
Diet, lifestyle, and the risk of type 2 diabetes mellitus in women
.
N Engl J Med
2001
;
345
:
790
797
15.
Salmerón
J
,
Hu
FB
,
Manson
JE
, et al
.
Dietary fat intake and risk of type 2 diabetes in women
.
Am J Clin Nutr
2001
;
73
:
1019
1026
16.
Wang
Q
,
Imamura
F
,
Ma
W
, et al
.
Circulating and dietary trans fatty acids and incident type 2 diabetes in older adults: the Cardiovascular Health Study
.
Diabetes Care
2015
;
38
:
1099
1107
17.
van Dam
RM
,
Willett
WC
,
Rimm
EB
,
Stampfer
MJ
,
Hu
FB
.
Dietary fat and meat intake in relation to risk of type 2 diabetes in men
.
Diabetes Care
2002
;
25
:
417
424
18.
Hercberg
S
,
Castetbon
K
,
Czernichow
S
, et al
.
The Nutrinet-Santé Study: a web-based prospective study on the relationship between nutrition and health and determinants of dietary patterns and nutritional status
.
BMC Public Health
2010
;
10
:
242
19.
Lassale
C
,
Péneau
S
,
Touvier
M
, et al
.
Validity of web-based self-reported weight and height: results of the Nutrinet-Santé study
.
J Med Internet Res
2013
;
15
:
e152
20.
Craig
CL
,
Marshall
AL
,
Sjöström
M
, et al
.
International physical activity questionnaire: 12-country reliability and validity
.
Med Sci Sports Exerc
2003
;
35
:
1381
1395
21.
Touvier
M
,
Kesse-Guyot
E
,
Méjean
C
, et al
.
Comparison between an interactive web-based self-administered 24 h dietary record and an interview by a dietitian for large-scale epidemiological studies
.
Br J Nutr
2011
;
105
:
1055
1064
22.
Lassale
C
,
Castetbon
K
,
Laporte
F
, et al
.
Correlations between fruit, vegetables, fish, vitamins, and fatty acids estimated by web-based nonconsecutive dietary records and respective biomarkers of nutritional status
.
J Acad Nutr Diet
2016
;
116
:
427
438.e5
23.
Lassale
C
,
Castetbon
K
,
Laporte
F
, et al
.
Validation of a Web-based, self-administered, non-consecutive-day dietary record tool against urinary biomarkers
.
Br J Nutr
2015
;
113
:
953
962
24.
le Moullec
N
,
Deheeger
M
,
Preziosi
P
, et al
.
Validation du manuel-photos utilisé pour l’enquête alimentaire de l’étude SU.VI.MAX
.
Cah Nutr Diét
1996
;
31
:
158
164
25.
Hercberg
S
,
Deheeger
M
,
Preziosi
P
.
Portions Alimentaires: Manuel Photos pour l’Estimation des Quantités
.
Paris, France
,
Polytechnica
,
2002
26.
Etude NutriNet-Santé
.
Table de Composition des Aliments
.
Paris, France
,
Economica
,
2013
27.
Black
AE
.
Critical evaluation of energy intake using the Goldberg cut-off for energy intake:basal metabolic rate. A practical guide to its calculation, use and limitations
.
Int J Obes Relat Metab Disord
2000
;
24
:
1119
1130
28.
U.S. Department of Agriculture, Agricultural Research Service
.
Nutrient Data Laboratory
.
USDA National Nutrient Database for Standard Reference, Release 28. Version Current: September 2015. Accessed 3 February 2022. Available from https://www.ars.usda.gov/northeast-area/beltsville-md-bhnrc/beltsville-human-nutrition-research-center/methods-and-application-of-food-composition-laboratory/mafcl-site-pages/sr11-sr28/
29.
Van Puyvelde
H
,
Perez-Cornago
A
,
Casagrande
C
, et al
.
Comparing calculated nutrient intakes using different food composition databases: results from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort
.
Nutrients
2020
;
12
:
2906
30.
Desquilbet
L
,
Mariotti
F
.
Dose-response analyses using restricted cubic spline functions in public health research
.
Stat Med
2010
;
29
:
1037
1057
31.
Deschasaux
M
,
Huybrechts
I
,
Julia
C
, et al
.
Association between nutritional profiles of foods underlying Nutri-Score front-of-pack labels and mortality: EPIC cohort study in 10 European countries
.
BMJ
2020
;
370
:
m3173
32.
Neuenschwander
M
,
Barbaresko
J
,
Pischke
CR
, et al
.
Intake of dietary fats and fatty acids and the incidence of type 2 diabetes: a systematic review and dose-response meta-analysis of prospective observational studies
.
PLoS Med
2020
;
17
:
e1003347
33.
Oteng
A-B
,
Kersten
S
.
Mechanisms of action of trans fatty acids
.
Adv Nutr
2020
;
11
:
697
708
34.
Saravanan
N
,
Haseeb
A
,
Ehtesham
NZ
,
Ghafoorunissa
.
Differential effects of dietary saturated and trans-fatty acids on expression of genes associated with insulin sensitivity in rat adipose tissue
.
Eur J Endocrinol
2005
;
153
:
159
165
35.
Willson
TM
,
Lambert
MH
,
Kliewer
SA
.
Peroxisome proliferator-activated receptor gamma and metabolic disease
.
Annu Rev Biochem
2001
;
70
:
341
367
36.
Obara
N
,
Fukushima
K
,
Ueno
Y
, et al
.
Possible involvement and the mechanisms of excess trans-fatty acid consumption in severe NAFLD in mice
.
J Hepatol
2010
;
53
:
326
334
37.
Andreeva
VA
,
Deschamps
V
,
Salanave
B
, et al
.
Comparison of dietary intakes between a large online cohort study (Etude NutriNet-Santé) and a nationally representative cross-sectional study (Etude Nationale Nutrition Santé) in France: addressing the issue of generalizability in e-epidemiology
.
Am J Epidemiol
2016
;
184
:
660
669
38.
Andreeva
VA
,
Salanave
B
,
Castetbon
K
, et al
.
Comparison of the sociodemographic characteristics of the large NutriNet-Santé e-cohort with French Census data: the issue of volunteer bias revisited
.
J Epidemiol Community Health
2015
;
69
:
893
898
39.
Institut National de la Statistique et des Études Économiques (Insee)
.
Accès et Utilisation de l’Internet dans l’Union Européenne
.
Données Annuelles de 2003 à 2021. Accessed 14 February 2022. Available from https://www.insee.fr/fr/statistiques/2385835
40.
Wanders
AJ
,
Zock
PL
,
Brouwer
IA
.
Trans fat intake and its dietary sources in general populations worldwide: a systematic review
.
Nutrients
2017
;
9
:
840
41.
Fiolet
T
,
Srour
B
,
Sellem
L
, et al
.
Consumption of ultra-processed foods and cancer risk: results from NutriNet-Santé prospective cohort
.
BMJ
2018
;
360
:
k322
42.
Willett
W
.
Nutritional Epidemiology
. 2nd ed.
New York, NY
,
Oxford University Press
,
1998
43.
Aglago
EK
,
Murphy
N
,
Huybrechts
I
, et al
.
Dietary intake and plasma phospholipid concentrations of saturated, monounsaturated and trans fatty acids and colorectal cancer risk in the European Prospective Investigation into Cancer and Nutrition cohort
.
Int J Cancer
2021
;
149
:
865
882
44.
Yammine
S
,
Huybrechts
I
,
Biessy
C
, et al
.
Dietary and circulating fatty acids and ovarian cancer risk in the European prospective investigation into cancer and nutrition
.
Cancer Epidemiol Biomarkers Prev
2020
;
29
:
1739
1749
45.
Fafot-Campagna
A
,
Romon
I
,
Fosse
S
,
Roudier
C
;
Institut de Veille Sanitaire
.
Prévalence et incidence du diabète, et mortalité liée au diabète en France
.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.