Introduction

The global burden of diabetes has increased significantly in recent years and is projected to increase to 642 million people by 20401. Diabetes is prone to multiple complications and a poor prognosis. Diabetes increases the risk of all-cause mortality two- to three-fold and reduces life expectancy by nearly 20 years2. Prediabetes is the precursor stage of diabetes, and is defined as a condition in which glucose metabolism is impaired but does not yet meet the criteria for a diabetes diagnosis3. The prevalence of prediabetes is high, affecting one-third of adults in the U.S. and an estimated 720 million people globally4.

Evidence from longitudinal studies and Mendelian randomization (MR) studies supports the existence of a bidirectional association and possible causal relationship between major depressive disorder (MDD) and type 2 diabetes mellitus (T2DM), with T2DM leading to more severe depression, which in turn is associated with a worse prognosis in patients with T2DM5. Microvascular dysfunction caused by increased oxidative stress and inflammation is one of the key underlying mechanisms6,7. Notably, this microvascular disease process may begin to emerge in people with prediabetes8. Research and early intervention strategies in people with prediabetes may help to prevent the progression of diabetes.

Diet is a common and essential modifiable factor. There have been many studies on the potential role of diet in reducing depression risk. For example, polyphenols, which have potent anti-inflammatory and antioxidant properties, may have antidepressant effects9. Anti-inflammatory diets, such as the Mediterranean diet, may reduce depression risk, whereas pro-inflammatory diets increase this risk10. There is also evidence that the intake of several common dietary antioxidants is negatively correlated with depression risk in the general population11. However, few studies have focused on the role of dietary antioxidants in people with prediabetes.

In this study, which was based on a nationally representative sample, a cross-sectional design was first used to assess the associations of the intake of six dietary antioxidants (including vitamin A, vitamin C, vitamin E, zinc, selenium, and total carotenoids) with depression risk in people with prediabetes. A longitudinal design was then used to assess the associations of the six dietary antioxidants with all-cause mortality in people with prediabetes. In addition, the composite dietary antioxidant index (CDAI) was used to reflect the combined exposure to the six dietary antioxidants. This study was designed to provide further insight into the underlying mechanisms and clues for developing public health strategies for people with prediabetes.

Methods

Study design and population

This study used data from the 2005–2018 National Health and Nutrition Examination Survey (NHANES). The NHANES is a continuous survey that selects a representative group of U.S. populations through complex, multistage probability sampling designed to assess the health and nutritional status of adults and children in the U.S. The survey combines interviews and physical examinations and was approved by the National Center for Health Statistics Institutional Review Board (NCHS IRB/ERB). Informed consent was obtained from all participants prior to inclusion in the survey. All methods were performed in accordance with relevant guidelines and regulations. Detailed information is available online (https://www.cdc.gov/nchs/nhanes/index.htm).

39,749 U.S. adults (age ≥ 20) participated in the 2005–2018 NHANES. According to the most recent guidelines from the American Diabetes Association (ADA)12, prediabetes is defined as any of the following: (1) impaired fasting glucose (IFG) (fasting plasma glucose 5.6–7.0 mmol/l), (2) impaired glucose tolerance (IGT) (2 h glucose during a 75 g OGTT 7.8–11.1 mmol/l), (3) hemoglobin A1c (HbA1c) 5.7–6.5%, (4) self-reported, and participants with known diabetes (self-reported or use of insulin or oral hypoglycemic agents) were further excluded. Accordingly, 12,384 adult U.S. individuals with prediabetes were identified in this study, which is close to the prevalence of prediabetes. Subjects missing critical information were further excluded. Finally, 8789 subjects (representing 52,578,820 adults with prediabetes in the U.S.) were included in the analysis. A flowchart of the process is shown in Fig. 1.

Fig. 1
figure 1

Flow chart of the study selection process.

Full size image

Dietary assessment

Individual dietary antioxidant intake and total energy intake were assessed by averaging consumption over two non-consecutive 24-h dietary recalls. Recall status was restricted to reliable to improve the precision of intake assessment. We studied six dietary antioxidant intakes: vitamin A, vitamin C, vitamin E, zinc, selenium, and total carotenoids. We used a modified version of CDAI developed by Wright et al. to assess the combined exposure of dietary antioxidant intakes13. The CDAI is calculated based on the intake of these six dietary antioxidants. In short, the intake of each antioxidant nutrient was standardized by subtracting the mean and dividing by the standard deviation. The standardized intakes are then summed to calculate the CDAI.

Measurement of depression

Depression assessment was conducted using the Patient Health Questionnaire-9 (PHQ-9), a nine-item instrument noted for its excellent reliability, as evidenced by a Cronbach’s alpha of 0.8914. A PHQ-9 score of 10 or higher was defined as indicative of clinically relevant depression according to the criteria outlined in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)14,15. The PHQ-9 was administered during face-to-face interviews at the Mobile Examination Center (MEC), with trained interviewers using the Computer Assisted Personal Interviewing (CAPI) system.

Determination of mortality outcomes

The mortality status of NHANES participants is associated with the National Death Index (NDI) from the NCHS. This linkage allows access to the Public-use Linked Mortality Files (LMF) through December 31, 2019, which provides information on survival status and underlying causes of death. The follow-up time for each individual was defined by calculating the time difference from the NHANES examination date to the last known survival or censoring event.

Study covariates

Potential confounders, namely, age, sex (male or female), race (categorized as non-Hispanic white, non-Hispanic black, Hispanic and other races), education level (less than high school, high school, or more than high school), marital status (married/living with partner, widowed/divorced/separated, or never married), total energy intake, body mass index (BMI, calculated as weight divided by height squared), serum cotinine concentration (reflecting tobacco exposure), alcohol consumption, estimated glomerular filtration rate (eGFR) and disease history (including hypertension, hyperlipidemia, cancer, and cardiovascular disease) were ascertained using questionnaires or physiological and biochemical tests. Alcohol consumption was categorized into no (had < 12 alcoholic drinks during the past 12 months) and yes (had ≥ 12 alcoholic drinks during the past 12 months). Cardiovascular disease (CVD) includes self-reported coronary heart disease, angina, heart attack, or stroke. The formula used for eGFR was the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation16.

Statistical analysis

In accordance with NHANES recommendations and to ensure that the sample was representative of the population, the study applied sampling weights to weight all analyses. Participants were grouped by quartiles of the CDAI when analyzing baseline characteristics. The Kolmogorov–Smirnov test was used to test the distribution of continuous variables. Continuous variables were expressed as mean (standard error) if they satisfied a normal distribution and as median (interquartile range) otherwise, and categorical variables were presented as unweighted counts (weighted percentages). Comparative analyses of baseline characteristics across these groups were performed using weighted linear regression for continuous variables and the Rao-Scott chi-square (χ2) test for categorical variables.

Dietary antioxidant intake and CDAI were categorized into four groups according to their quartiles from low to high: Q1, Q2, Q3, and Q4, respectively, with Q1 as the reference group. The CDAI contains relatively limited components of antioxidant food items that significantly impact energy supply, so there is no need to adjust for energy intake17. However, an adjustment for energy intake was still made to account for the possibility that some people with prediabetes or obese subjects may restrict caloric intake. A weighted multivariate logistic regression model was used to assess the odds ratio (OR) and 95% confidence interval (CI) of the association between dietary antioxidant intake and depression risk in people with prediabetes. Weighted multivariate Cox regression analyses were used to assess the hazard ratio (HR) and 95% CI for the association between dietary antioxidant intake and all-cause mortality in people with prediabetes. Covariates were not adjusted in the crude model. Model 1 was adjusted for age, sex, race, education level, marital status, and total energy intake. Model 2 was further adjusted for BMI, serum cotinine concentration, alcohol consumption, eGFR, and disease history (including hypertension, hyperlipidemia, cancer, and CVD). We also used CDAI to assess the joint association of multiple antioxidant nutrients with depression risk and all-cause mortality in people with prediabetes. In addition, possible nonlinear associations between dietary antioxidant intake as well as CDAI with depression risk and all-cause mortality in people with prediabetes were assessed using the restricted cubic spline (RCS) model. To further compare the association between dietary antioxidant intake with depression risk and all-cause mortality in different populations and to assess the potential impact of prediabetes on this association, we performed subgroup analyses of the general population stratified by glucose metabolic status. 39,749 adult subjects from NHANES 2005–2018 were categorized into 3 groups: HC (participants without diabetes or prediabetes), pre-DM (participants with prediabetes), and DM (participants with diabetes). Subjects missing key information were further excluded. The final numbers in each group were 12,283 (HC), 8789 (pre-DM), and 4812 (DM). Likelihood ratio test was performed to calculate interactions between these subgroups.

Data were not complete for all covariates, as shown in Table S1 of the supplementary material. To ensure the reliability of the study results, the multiple imputations by chained equation were applied to missing covariates using the MICE package in R to avoid potential bias. All statistical analyses and visualizations in this study were performed using R 4.3.3 (http://www.R-project.org). A two-sided P value < 0.05 was considered significant.

Results

Table 1 shows the baseline characteristics of the study population. The study sample consisted of 8789 adult participants with prediabetes, 7.64% of whom were diagnosed with depression. Fig S1 in the Supplementary Material shows the distribution of serum HbA1c, fasting plasma glucose (FPG), and 2-h glucose (OGTT) in the study population. Based on the weighted results, the mean age of the participants was 53 years, and 52.25% were male. Most of the participants were white, married or living with partner, and had a high school diploma or higher. The four subgroups by quartiles of CDAI differed significantly in the baseline characteristics of sex, race, marital status, education level, serum cotinine, alcohol consumption, total energy intake, eGFR, HbA1c, and cardiovascular disease. Notably, the CDAI Q4 group had fewer depressed individuals than other groups.

Table 1 Characteristics of the study population.
Full size table

Table 2 shows the associations of the quartiles of six dietary antioxidant intakes and CDAI with depression risk in people with prediabetes. The negative association of higher levels of vitamin C, zinc, and selenium intake, as well as CDAI with depression risk, remained significant during stepwise adjustment for potential confounders, and all P for trend tests were significant. The corresponding ORs for Q4 were 0.67 (95% CI 0.48–0.93), 0.65 (95% CI 0.43–0.98), 0.42 (95% CI 0.28–0.63), and 0.58 (95% CI 0.39–0.85), respectively, when compared to their Q1 levels. In addition, Participants with vitamin E (OR 0.57, 95% CI 0.39–0.81) intake at Q3 level had a reduced depression risk compared to those in Q1. Notably, the RCS results showed a nonlinear J-shape association between dietary vitamin C, vitamin E, zinc, selenium intake, CDAI, and depression risk, with inflection points of 128, 11, 15, 155, and 3, respectively (Figs. 2 and Fig. 4a).

Table 2 The results of weighted multivariate logistic regression analysis for the association between dietary antioxidant intake and depression risk in people with prediabetes.
Full size table
Fig. 2
figure 2

Restricted cubic spline (RCS) model of the association between dietary antioxidant intake and depression risk in people with prediabetes. Abbreviations: OR odds ratio, CI confidence interval. Adjusted for: age, sex, race, education level, marital status, total energy intake, BMI, serum cotinine concentration, alcohol consumption, eGFR, hypertension, hyperlipidemia, cancer and CVD.

Full size image

Table 3 shows the associations of the quartiles of six dietary antioxidant intakes and CDAI with all-cause mortality in people with prediabetes. In model 2 adjusted for all covariates, higher levels of dietary vitamin E, selenium, total carotenoids intake, and CDAI were significantly associated with decreased all-cause mortality risk, and all trend tests were significant. The corresponding HRs at the Q4 level were 0.62 (95% CI 0.45–0.86), 0.72 (95% CI 0.52–0.99), 0.68 (95% CI 0.55–0.84), and 0.65 (95% CI 0.47–0.89), compared to participants at the Q1 level. In addition, participants with vitamin A intake at the Q2 level (HR 0.70, 95% CI 0.53–0.92) and vitamin C intake at the Q3 level (HR 0.73, 95% CI 0.57–0.94) had significantly lower all-cause mortality risk, as compared to participants at the Q1 level. The RCS analysis revealed a non-linear association between CDAI and the intake of vitamins A, C, E, and zinc with all-cause mortality risk. A J-shape association was found between vitamin C and zinc intake and all-cause mortality risk, with inflection points of 117 and 15, respectively. Notably, after an increase in the CDAI of more than 3, the decrease in all-cause mortality risk significantly slowed as the CDAI continued to increase (Fig. 3 and Fig. 4b).

Table 3 The results of weighted multivariate cox regression analysis for the association between dietary antioxidant intake and all-cause mortality risk in people with prediabetes.
Full size table
Fig. 3
figure 3

Restricted cubic spline (RCS) model of the association between dietary antioxidant intake and all-cause mortality risk in people with prediabetes. Abbreviations: HR hazard ratio, CI confidence interval. Adjusted for: age, sex, race, education level, marital status, total energy intake, BMI, serum cotinine concentration, alcohol consumption, eGFR, hypertension, hyperlipidemia, cancer and CVD.

Full size image
Fig. 4
figure 4

Restricted cubic spline (RCS) model of the association between CDAI with depression risk and all-cause mortality in people with prediabetes. Abbreviations: CDAI composite dietary antioxidant index, OR odds ratio, HR hazard ratio, CI confidence interval. Adjusted for: age, sex, race, education level, marital status, total energy intake, BMI, serum cotinine concentration, alcohol consumption, eGFR, hypertension, hyperlipidemia, cancer and CVD.

Full size image

Table 4 shows the results after applying the multiple imputations to the missing covariates. The associations of the CDAI with depression risk and all-cause mortality in adults with prediabetes remained stable and were significant for the test of trend.

Table 4 Association of CDAI with risk of depression and all-cause mortality in people with prediabetes after applying the multiple imputations to missing covariates.
Full size table

Fig S2 shows the results of the general population-based subgroup analysis. Among 3 subgroups, the reduction in depression risk was most significant in individuals with prediabetes when the CDAI reached the Q2 level (P for interaction < 0.001), and the reduction in all-cause mortality appeared to be apparent as well (although P for interaction = 0.755). However, the healthy population and people with diabetes need higher levels of antioxidants to benefit. Notably, when vitamin E intake was at the Q3 level, individuals with prediabetes had a more significant reduction in depression risk and all-cause mortality compared with the HC and DM groups (both P for interactions < 0.05). In addition, when selenium intake was at the Q2 level and zinc intake was at the Q3 level, the reduction in depression risk was most apparent in participants with prediabetes among the three subgroups (both P for interactions < 0.05).

Discussion

This cross-sectional and longitudinal analysis based on a representative sample of adult individuals with prediabetes in the U.S. showed a J-shape association between dietary antioxidants (including vitamin C, vitamin E, zinc, and selenium) intake and CDAI with depression risk, and inflection points that minimized the depression risk were identified. Moreover, appropriate intake of dietary antioxidants was associated with reduced all-cause mortality risk in people with prediabetes. A J-shape association was found between vitamin C and zinc intake with all-cause mortality risk. The all-cause mortality risk decreased with increasing CDAI, and the decline rate began to slow significantly at approximately 3.

To our knowledge, this is the first study to examine the association between dietary antioxidant intake with depression risk and all-cause mortality in people with prediabetes. Regarding dietary antioxidant vitamins, adequate intake of vitamins C and E has been shown to be associated with reduced depression risk18,19, which is consistent with our findings. In contrast, the association between dietary vitamin A intake and depression is not well documented, with a negative association found only among females in a prospective cohort20. In our study of people with prediabetes, the association was not significant. Zinc and selenium are both essential trace elements. Current evidence from observational and randomized controlled trials supports the potential benefits of zinc in reducing depression risk and alleviating depression21. Possible mechanisms underlying the antidepressant effects of zinc include anti-inflammatory, antioxidant, and antagonistic effects on the glutamatergic N-methyl-d-aspartate (NMDA) receptor22. The relationship between selenium and depression has been less well studied, and there are conflicting results23. Notably, several studies have found that both high and low levels of selenium are associated with an increased risk of depression24,25. In other words, there is an "optimal range" of selenium associations with depression risk, which is in line with our finding of a J-shape association. Selenium may exert its antidepressant effects through oxidative and inflammatory pathways and through the modulation of various neurotransmitter systems22. The relationship between dietary total carotenoid intake and depression is controversial. A cross-sectional analysis of Japanese older adults aged 65–75 years (n = 500) revealed that high dietary total carotenoid intake was significantly associated with a lower prevalence of depressive symptoms26. Another cross-sectional study (n = 278) of U.S. older adults aged 60 years or older revealed that dietary β-cryptoxanthin intake was significantly and negatively associated with depressive symptoms27. Carotenoids may protect nerve tissue from depression through their antioxidant capacity28. However, a study based on a longitudinal cohort of Australian adults aged 55–85 years (n = 2305) did not find a significant association between dietary total carotenoid intake and depressive symptoms29. These studies were conducted on older populations, which are more likely to have comorbidities, which may affect the reliability of the results. Our study did not find the beneficial effects of carotenoids in people with prediabetes. Some studies have found that high dietary antioxidant intake is associated with reduced risk of all-cause mortality and cardiovascular disease (CVD)-related mortality in adults with diabetes2. This is consistent with our findings.

In addition, our study revealed non-linear J-shape associations, suggesting that the protective effects of dietary antioxidants may change with increasing intake. On the one hand, it has been shown that vitamin C30, vitamin E31, zinc32, selenium33, and carotenoids34 have both antioxidant and pro-oxidant properties and that they may trigger oxidative stress when consumed at high levels. On the other hand, excessive scavenging of reactive oxide species (ROS) may impair some essential intracellular signaling and metabolic functions associated with ROS35. These may lead to increased depression risk and poorer prognosis, which explains the J-shape association we observed.

Various nutrients often coexist and may interact. We used the CDAI to estimate the combined exposure to six dietary antioxidants. Our study showed that in people with prediabetes, depression risk and all-cause mortality declined with increasing CDAI, with depression risk minimizing around 3, while the decline in all-cause mortality risk began to slow significantly. Previous studies have used dietary total antioxidant capacity (DTAC) to reflect the combined effects of multiple dietary antioxidants. A cross-sectional study (n = 265) revealed a significant negative association between DTAC and depression in females with type 2 diabetes36. Several other cross-sectional studies in different populations have reported a negative association between DTAC and depression37,38,39. A longitudinal study reported that higher DTAC was associated with lower all-cause and CVD-related mortality in the Singapore Chinese population40. These results are consistent with our findings. In conclusion, our findings support the idea that intake of moderate amounts of dietary antioxidants reduces depression risk and all-cause mortality in people with prediabetes and that the greatest benefit can be obtained by keeping the CDAI near 3.

Much of the current research has focused on patients with diabetes, and relatively little has been done on people with prediabetes. The mechanisms of diabetes mellitus comorbid with depression are unclear, and potential pathophysiological mechanisms include insulin resistance, disruption of the Hypothalamic–Pituitary–Adrenal (HPA) axis, alterations in brain structure and function, oxidative stress (OS), inflammation, etc.,41. OS is one of the major pathogenic factors in diabetes mellitus, and is characterized by excessive production and accumulation of ROS and dysfunction of the antioxidant system. OS damages the vascular endothelium of patients with diabetes, causing endothelial dysfunction and, ultimately, a variety of complications, including macrovascular and microvascular lesions42. The microvascular system is involved in the regulation of many brain processes, and when these processes are impaired, there is a predisposition to stroke, cognitive dysfunction, and depression6. Previous studies have shown a high prevalence of prediabetes and a high risk of progression to T2DM3. Cerebral microvascular dysfunction6 and OS7 are also evident in adults with prediabetes, which may explain why the intake of moderate amounts of dietary antioxidants may help reduce depression risk and all-cause mortality in people with prediabetes.

The results of the subgroup analyses suggest that appropriate dietary antioxidant intake exerts its protective effect against depression and all-cause mortality regardless of whether the glucose metabolic state is already abnormal. This protective effect seems to be more apparent in participants with prediabetes compared to healthy population and participants already suffering from diabetes. Levels of OS are usually normal in healthy population, and the available evidence suggests that antioxidants do not provide adequate or comprehensive protection against OS and injury under “normal” conditions43. It can also be observed that only a lower level of CDAI is needed for a significant benefit in participants with prediabetes, whereas participants with diabetes require a higher level of CDAI for a significant benefit. One possible explanation is that hyperglycemia upregulates markers of chronic inflammation and increases ROS production7. Compared to participants with prediabetes, participants with diabetes are exposed to higher blood glucose and higher levels of OS, which results in more severe damage to their microvascular system. Prediabetes may represent a time window for intervention in which only moderate amounts of dietary antioxidants are needed to provide benefits. This would also avoid the adverse effects of excessive antioxidant intake.

Strengths and limitations

This study has several significant strengths: (1) the use of a nationally representative sample of U.S. adults, with a large sample size and standardized data collection procedures, which enables the findings to be generalized to a broader population (2) the focus on people with prediabetes, and the findings may be potentially valuable for the development of public health strategies (3) use of the CDAI to assess combined exposure to six dietary antioxidants (4) the assessment of nonlinear associations through the RCS and the identification of some inflection points; and (5) use of multiple imputations to improve the reliability of results.

The following potential limitations exist in this study: (1) The cross-sectional design limits the ability to establish a causal relationship between dietary antioxidant intake and depression risk in people with prediabetes (2) Dietary data in NHANES were derived from 24-h dietary recalls; which may be subject to recall bias.

Conclusion

The results of this cross-sectional and longitudinal study suggest that moderate dietary antioxidant intake is negatively associated with depression risk and all-cause mortality in people with prediabetes. In addition, when using the CDAI to estimate combined exposure to six dietary antioxidants, a CDAI near 3 provides the greatest benefit. The benefits of dietary antioxidants appear to be more apparent in people with prediabetes compared to healthy population and people with diabetes. However, large prospective studies are needed to confirm these results further.