Insulin resistance is a fundamental component of Type 2 diabetes mellitus (T2DM) etiology and is related to a wide range of diseases such as hypertension, hyperlipidemia and polycystic ovarian disease [1
]. Diet is a modifiable factor that can prevent or predispose to insulin resistance. Within the dietary factor, the effect of carbohydrate intake on insulin resistance has been scarcely studied in adolescents. Observational studies have found higher insulin sensitivity in youth with diets high in dietary fiber [2
] and whole-grain intake [3
]. On the contrary, high total sugar intake has been linked to increased insulin resistance in girls and adolescents [4
], although this association remains controversial [5
The effect of carbohydrates on serum glucose and insulin concentration is mainly determined by the amount of carbohydrates consumed and their absorption rate [6
]. In order to include these two determinants of glycemic response, two indices have been developed: glycemic index (GI) [7
] and glycemic load (GL) [8
]. Diets high in GI and GL have been associated with a higher risk of T2DM in adults [9
]. High GI diets during puberty have been associated with increased insulin resistance in adulthood [10
]. Conversely, low GI diets can produce significant decreases in the homeostasis model assessment of insulin resistance (HOMA-IR) in children and adolescents [11
These findings suggest that the quantity, type and quality of carbohydrates during adolescence might affect health status in the short and long term, conditioning the development of chronic degenerative diseases. However, only a few studies have evaluated the effect that dietary carbohydrates exert on insulin resistance in Mexican adolescents [12
]. Thus, the aim of the present study was to assess the association between carbohydrate nutrition variables and insulin resistance in adolescents from marginalized areas of Chiapas, México.
General characteristics of the sample according to energy-adjusted dietary fiber intake are described in Table 1
. Study participants with the highest dietary fiber intake were more likely to be males, to live in rural areas, to have a mother who speaks an indigenous language, to have normal weight, to have less body fat percentage, compared to participants with the lowest dietary fiber intake. Subjects in the top category of dietary fiber intake also had higher total carbohydrate intake, lower fat intake (from all types), lower total sugar intake and lower dietary GI than those in the bottom category of fiber intake.
shows the results for biochemical measurements and the prevalence of insulin resistance and abnormal fasting insulin levels, according to categories of carbohydrate intake variables. In our study, the prevalence of insulin resistance according to two different cut-off points was: HOMA-IR > 3.16 (Overall: 21%; Females: 25%; Males: 17%) and >2.97 (Overall 23%; Females: 28%; Males: 18%). Similarly, abnormal insulin levels were found in 22% of the sample (Females: 27%; Males: 16%). Fasting serum insulin and HOMA-IR median values were higher for subjects with the lowest dietary fiber intake. On the contrary, adolescents with a high dietary fiber intake showed significantly lower prevalences of insulin resistance (9.7%) and abnormal insulin levels (8.3%) than those with the lowest fiber intake (34.3% and 35.6%, respectively). Fasting insulin levels and HOMA-IR values were lower in adolescents with a moderate sugar intake (second tertile), when compared to high sugar intake group. Nevertheless, a higher prevalence of HOMA-IR > 2.97 and hyperinsulinemia was observed for individuals in the highest sugar intake category. Median fasting glucose in adolescents with high dietary GI was 86 mg/dL, and it was significantly higher than the low GI group.
The ORs and 95% CIs for insulin resistance (HOMA-IR > 3.16) according to dietary carbohydrates (total carbohydrate, dietary fiber, total sugars, dietary GL and dietary GI) are presented in Table 3
. We found a statistically significant interaction between sex and total carbohydrate intake; no significant interactions were observed for dietary fiber, total sugars, dietary GI or GL. Female adolescents in the top category of carbohydrate intake had lower odds of insulin resistance than those in the bottom category (Model 1: OR = 0.19; 95% CI: 0.05–0.73; p
for trend = 0.009). Conversely the direction of the effect changed after the adjustment for dietary fiber and MUFAs intake, showing non-significant higher odds of insulin resistance in girls with high total carbohydrate intake (Model 4: OR = 1.08; 95% CI: 0.19–6.20; p
for trend = 0.839). Furthermore, high dietary fiber intake was associated with lower odds of HOMA-IR > 3.16 (Model 1: OR = 0.21; 95% CI: 0.08–0.52); p
for trend = 0.001). However, this association was no longer significant after adjustment for sex, age, BF%, energy-adjusted MUFAs intake and total energy intake (kcal/day) (Table 3
). Hosmer–Lemeshow tests and individual classification criteria (sensitivity, specificity) of the models are shown in Supplementary Table S1
Similar results were found when the second cut-off (HOMA-IR > 2.97) for insulin resistance was used (Table 4
). Adolescents with high fiber intake had lower odds of insulin resistance (HOMA-IR > 2.97) (Model 1: OR = 0.24; 95% CI: 0.10–0.58); p
for trend = 0.001), which remained statistically significant after adjustment for sex, age, BF% and energy-adjusted SFAs intake. No statistically significant associations were observed between total carbohydrates, total sugars, dietary GI or dietary GL and HOMA-IR > 2.97. Goodness of fit assessment and quality criteria of the models are shown in Supplementary Table S2
presents the ORs and 95% CIs for elevated fasting insulin concentration (≥14.38 μU/mL) according to categories of carbohydrate intake variables. Lower odds of hyperinsulinemia were also observed for subjects with high dietary fiber intake (Model 1: OR = 0.16; 95% CI: 0.06–0.43); p
for trend <0.001). On the contrary, the probability of elevated fasting insulin concentrations was two times higher for adolescents with high dietary GI, when compared to adolescents with a low dietary GI (Model 1: OR = 2.32; 95% CI: 1.02–5.26); p
for trend = 0.042). However, this association was attenuated, and statistical significance was lost after adjustment for sex, age, mother’s language, weight status and energy-adjusted dietary fiber intake. Supplementary Table S3
shows the results for Hosmer–Lemeshow tests and individual classification criteria (sensitivity, specificity) of the models.
The median of HOMA-IR in our study was 1.8 units, which is similar to values previously reported for similar aged adolescents in México [36
]. The prevalence of insulin resistance in our study (21%–23%) was lower than the prevalence found for overweight/obese adolescents from Tuxtla Gutiérrez, the capital of Chiapas (40%) but higher than normal-weight adolescents in Chiapas (16%) [38
]. Another study conducted in 292 adolescents (12–15 years) from Coahuila, México, found a higher proportion (46%) of insulin resistance [37
]. Contrast between studies might be explained to weight status, since almost half of the sample in the previous study was classified as overweight or obese, whereas in our study, overweight and obesity prevalence was 27%.
We found that high dietary fiber intake was associated with lower odds of insulin resistance in adolescents from Chiapas, México. This finding adds to the growing evidence of the relevance of dietary fiber for health in adolescence. Previous research in other countries have found similar results. For instance, an investigation carried out in 754 adolescents from Georgia, USA, found that fiber intakes (insoluble and soluble) were inversely associated with fasting insulin and HOMA-IR values [39
]. A prospective study of American female adolescents (16–17 years) has also shown that increases in dietary soluble fiber during 3 years were related to reductions in HOMA-IR values and insulin levels [40
]. Similarly, it has been demonstrated that a 10 g/MJ increase in dietary fiber reduced more than 1 SD of HOMA-IR among Danish girls [4
Nevertheless, evidence is not consistent and our results differ from studies conducted in children and adolescents from Canada [41
] and eight European cities [42
], where no association between dietary fiber intake and insulin levels was found. Differences between studies might be attributed to lower exposure levels—in the Canadian study, fiber intake was around 13 g/day [41
]; in the European study, the age range was 12.5–17.5 years and mean dietary fiber consumption was 20 g/day [42
]. In México, an investigation carried out in 246 adolescents (12–14 years) from private and public schools located in Tuxtla Gutiérrez, Chiapas, did not find a relationship between carbohydrates or fiber intakes and HOMA-IR [12
]. Nevertheless, in Tuxtla Gutiérrez, the mean dietary fiber intake was considerably lower (10 g/day) than in our study (26.6 g/day). The latter figure is similar to dietary fiber intake reported for adolescents from rural population at a national level (28 g/day) [43
]. This could be attributed to the sample characteristics, since our research included urban and rural population from marginalized areas, where maize and beans are the two most important staple foods of the diet.
The possible protective role of dietary fiber on insulin resistance could be explained by different mechanisms involving different sites of action [44
]. It has been demonstrated that dietary fiber increases the sensitivity of peripheral tissues to physiologic concentrations of insulin [45
]. In the small intestine, fiber acts as a mechanical barrier and increases intraluminal viscosity, delays intestinal transit and reduces glucose absorption and insulin secretion. In addition, dietary fiber increases satiety, contributing to disrupt mechanisms leading to obesity. In the large intestine, fiber fermentation results in the production of short-chain fatty acids, which have been associated with the improvement of insulin sensitivity [44
]. Also, fermentable fiber intake (oligofructose) may alter microbiota composition, reducing gram-negative bacterial content [46
]. The gram-negative bacterial lipopolysaccharides have been associated with increased insulin levels in mice [47
]. Therefore, dietary fiber could reduce insulin resistance by relieving endotoxemia [44
In our study, no significant associations were observed for the rest of the carbohydrate variables. Likewise, other studies conducted in youth have not found a link between total carbohydrate intake [10
], total sugar intake [10
], dietary GI [48
] or dietary GL [10
] and insulin resistance. However, experimental evidence of such associations is controversial. It has been demonstrated that a high-carbohydrate/low-fat diet increased insulin sensitivity in normal-weight adolescents [49
], but in obese subjects (13–17 years) the same diet increased insulin secretion without improvements on insulin sensitivity [50
Regarding total sugar intake, a clinical trial conducted in weight-stable, physically active adolescents showed that a moderate intake of fructose and glucose beverages for a period of 2 weeks had no effect on insulin levels [51
]. In contrast, a short-term (9 days) restriction of sugar and fructose (within an isocaloric diet) has shown to be effective in decreasing liver fat, visceral fat and improving insulin dynamics (sensitivity, secretion, and clearance) in obese children [52
]. A prospective study conducted in 120 overweight Latino children in California showed that high sugar consumption was correlated to lower acute insulin responses and a reduced disposition index (index of β-cell function) [48
]. The evidence suggests that the detrimental effect of high sugar intake on insulin responses could be more pronounced in subjects with overweight or obesity. However, in our study, no significant interaction was found for total sugar intake and weight status. Our results showed higher odds for insulin resistance for subjects in the highest tertile of total sugar intake, although it did not reach statistical significance. Furthermore, there is evidence from clinical trials that low-GI/low GL diets reduce HOMA index in children and adolescents [11
]. Our results showed a non-significant trend of higher odds of insulin resistance for the highest categories of dietary GI and GL in the multivariate model.
The inconsistency of results could be explained by some methodological limitations of our study. The use of a single administration of a 24 h recall did not allow us to account for day-to-day variations of habitual dietary intake. As the precision of the method improves with the number of recalls administered in the same study subject [53
], it would have been preferable to apply the recall at least two or three times. However, the fieldwork was conducted in different municipalities with difficult access and we were not able to repeat the dietary assessment. The 24 h dietary recall has other disadvantages, the reported data relies on the study subject’s recent memory and depends on the interviewer’s capacity for describing ingredients, food preparation and dishes [53
]. However, to warrant the accuracy of dietary data, a group of dietitians was specifically trained for this study, the 24 h recall method was applied using tablets with data collection and codification in real-time [13
]. Furthermore, we assessed the plausibility of caloric intake by age, weight and height and implausible reporters of energy intake were excluded from the analysis. Misreporting may alter diet and disease associations [54
]. Previous studies have demonstrated that excluding implausible reporters yielded results that were more consistent with expectations, when the association between dietary factors and weight status was evaluated [55
]. Another limitation in our study is that physical activity and pubertal stage data were not available, thus, we were not able to adjust the models for these variables as potential confounders. Recent studies provide evidence that regular physical activity lowers the risk of insulin resistance and improves insulin sensitivity when individuals comply with physical activity guidelines [56
]. As regards to the pubertal stage, evidence shows that transition from Tanner stage I to Tanner stage III is related to reduced insulin sensitivity, with increments in fasting glucose and insulin levels [57
]. Nevertheless, models were adjusted for sex and age to reduce this source of error. Another potential weakness of our study is that GI values were assigned to local foods using data from previous studies of Mexican traditional dishes [23
] and the International tables of GI and GL values [25
], where most values were published for Australian or American foods. This might produce misclassification of foods to GI and GL categories, and probably causing bias towards the null. Finally, the cross-sectional design of our study allowed us to assess only a statistical association. Thus, causality cannot be inferred from the results obtained from the analysis.
A strength of this study is the assignment of GI values through a systematic protocol published for 24 h recalls [22
], and for most of the foods with available carbohydrate content (59.6%), the GI values of a closely related food or an exact match in the GI databases were assigned. Also, we used two cut-off points for HOMA-IR as surrogates of insulin resistance, which were previously validated for adolescents [29
] and have been demonstrated as a valid predictor of Metabolic Syndrome in a similar population group [27
]. Furthermore, our study was conducted in a sample of adolescents from marginalized communities in the South of México, comprising a high proportion of indigenous population (44% of the adolescents’ mothers spoke Mayan language) where few studies have evaluated the association between diet and insulin resistance.
In conclusion, this study provides evidence that high dietary fiber intake is associated with lower odds of insulin resistance in adolescents from Chiapas, México. However, further longitudinal research on carbohydrate nutrition and insulin resistance is required to effectively develop and promote dietary recommendations that may help to prevent Type 2 diabetes incidence and other chronic diseases among this vulnerable population.