Diet plays a key, but neglected role, for health as a modifiable factor. Unhealthy dietary patterns characterized by higher energy consumption, fat and sugar products, red and processed meats, and alcohol, which may influence health, and associated with an increased risk of adverse outcomes (1, 2). The dietary pattern effect on adverse health outcomes is partly through its influence on inflammatory pathways (3). Dietary Inflammatory Index (DII) is developed to characterize the overall inflammatory potential of an individual’s diet (4) and a higher DII score represents a pro-inflammatory diet. In terms of mental health, DII score is positively associated with depression symptoms risks in limited populations, such as patients with chronic diseases, female nurses, elderly over 55 years or U.S. adults based on only three cycles (2007–2012) of National Health and Nutrition Examination Survey (NHANES) (5–8). A higher DII score was additionally observed to be related with higher risks of various chronic inflammation-related health outcomes including cancer and cardiovascular disease (CVD) incidence and related mortality (9).

The U.S. cancer survivor population is projected to expand to 22.1 million by 2030 with a cancer incidence increment and treatment advancements (10). Cancer survivors are more likely to suffer common mental disorders such as depression (11, 12). Depressive symptoms reduce cancer survivors’ quality of life and treatment compliance, promote disease progression, and increase mortality risk (13–15). A previous study found that cancer survivors have poor dietary quality compared with subjects without cancer (16); however, a few studies investigate the dietary effect of diet quality in depressive symptoms. Further evidence is required to support evidence-based preventive interventions that will improve mental health and survival among cancer survivors.

The association between energy-adjusted DII (E-DII) score and depression is unclear in U.S. adults, especially in cancer survivors although E-DII score is positively associated with depression in limited populations (5–8). Fewer still have evaluated whether depression affected the relationship between E-DII score and mortality. Therefore, this study was performed using the data in a large prospective cohort of a nationally representative population, the 2007–2018 National Health and Nutrition Examination Survey (NHANES) database, aimed to (1) assess the association between energy-adjusted DII (E-DII) score and depression risk and (2) explore the E-DII score and depression on mortality risk impact in subjects without cancer as well as cancer survivors.

A total of 27,447 participants who met the criteria were included in the analysis, including 24,694 subjects without cancer and 2,753 cancer survivors (Figure 1). Table 1 shows the subjects’ baseline characteristics. Cancer survivors were more likely to be aged ≥65 years (51.1 vs 15.2%), female (56.3 vs 51.5%), non-Hispanic white (86.1 vs 65.7%), tobacco user (53.3 vs 43.1%), more than high-school educated (67.9 vs 61.4%), and have health insurance (94.9 vs 82.0%) compared to subjects without cancer. Cancer survivors were less likely to be without chronic comorbidity (26.5 vs 50.7%) and never married (5.9 vs 19.4%). There was no significant difference between subjects without cancer and cancer survivors for the distribution of E-DII score and depression.

We further evaluated the association of E-DII scores with depression by comparing the proportion of depression in the four quartiles of E-DII scores in subjects without cancer and cancer survivors, respectively (Table 2). The depression proportion increased as the E-DII quartile climbed, both in subjects without cancer and cancer survivors. The proportion of depression tripled from 3.9% in Q1 to 11.9% in Q4 in subjects without cancer (P trend < 0.001), and from 5.7% in Q1 to 15.2% in Q4 in cancer survivors (P trend < 0.001). The E-DII scores association with depression were assessed further by adjusting for potential confounders as several variables were found to be related to E-DII score depression (Supplementary Table 1) (5, 31). The ORs of having depression across the quartiles of E-DII scores are shown in Table 2. There was a significant upward trend in the odds of having depression with the increment of E-DII scores (P trend < 0.001), subjects in the Q4 of E-DII score were found to be at an increased risk of developing depression compared with those in Q1 (OR: 2.17, 95% CI: 1.75–2.70) for subjects without cancer, after adjusting for variables including age, gender, race/ethnicity, BMI, tobacco use, co-morbidity index, education level, marital status, and health insurance status. The significant upward associations were also observed after adjusting for the same covariate (P trend < 0.001) and the highest quartile of the E-DII score had highest risk of depression than those in the lowest quartile (OR: 1.78, 95% CI: 1.09–2.92) for cancer survivors. Subgroup analyses were done and classified by previous cancer diagnosis times, primary cancer site, and duration of diagnosis in cancer survivors to explore the association between E-DII score and depression (Supplementary Table 2). The magnitudes of associations were similar across the subgroups.

We finally evaluated the roles of E-DII score or depression on the risks of death of all-cause, and cancer-cause or CVD-cause in subjects without cancer and cancer survivors. A total of 2,496 deaths were identified and 225 deaths within the first 12 months of follow-up, 164 in subjects without cancer and 61 in cancer survivors, were excluded to reduce the possibility of reverse causation during a median follow-up time of 87 person-months (32). The subjects without cancer had 1,701 all-cause deaths, including 369 (22.7%) cancer deaths and 435 (24.4%) CVD deaths, while cancer survivors had 570 all-cause deaths, including 192 (33.9%) cancer deaths and 122 (22.3%) CVD deaths. As expected, the all-cause, cancer-cause, and CVD-cause deaths were higher in cancer survivors than in subjects without cancer (15.2 vs 4.8%, 5.2 vs 1.1%, 3.4 vs 1.2%). A significantly higher risk of all-cause (HR:1.90, 95%CI: 1.54–2.35) were observed for subjects with Q4 of E-DII score compared to those of Q1 in subjects without cancer after adjusting for the effects of depression and other variables related to survival, and similar association was also observed between E-DII and CVD-cause death (HR_{Q4VS Q1}: 2.21, 95%CI: 1.44–3.39) after considering the competitive risk of other causes of death as shown in Table 3. Meanwhile, depression was associated with a higher risk of all-cause mortality (HR: 1.29, 95% CI: 1.04–1.59) after accounting for the E-DII score influence and other variables, but not with cancer-cause or CVD-cause death in the multivariable-adjusted competing risk regression models. No significant correlations were found for E-DII score or depression with all-cause, cancer-cause, or CVD-cause death (Table 3) in cancer survivors.

This study examined the E-DII score relationship with depression and depression with mortality using data from NHANES 2007–2018, with special focus on cancer survivors. Our results demonstrate that subjects with a higher E-DII score had a higher depression risk, both in subjects without cancer or in cancer survivors. However, the pro-inflammatory diet effect characterized by higher E-DII scores and depression on survival was only observed in the subjects without cancer.

Our results are consisted with several community-based studies (5, 31, 33). We examined the association within the subjects without cancer and cancer survivors, respectively, using the data of NHANES 2007–2018. Results from the two groups both show a linear upward trend that the proportion of depression increases with the E-DII increment. A previous study only included NHANES 2007–2012 adult participants reported similar associations between DII scores without energy adjustment and depressive symptoms (OR_{Q5VS Q1}:2.26, 95%CI: 1.60–3.20) (8). These studies indicated that the relationship between dietary inflammation and depression was robust. In the current study, the median E-DII scores for the most pro-inflammatory quartile in in NHANES 2007–2012 and NHANES 2013–2018 were 2.1 and 2.2, respectively, whereas the upper quartile median E-DII score of was 0.6 in a previous study based on the Mediterranean diet (34), indicating that the dietary inflammation potential pattern of U.S. adults was higher and the secular E-DII score trend was consistent. There are currently no studies that explore the relationship between DII score and depression among cancer survivors. A similar association between E-DII score and depression was also observed in cancer survivors. A possible explanation is that insufficient nutrient intake is common in cancer survivors, such as digestive tumor (35, 36), which weakens the depressive symptoms caused by pro-inflammatory or anti-inflammatory diet by affecting the level of systemic inflammation. Several studies reported that the majority of survivors modified their diet intake after their cancer diagnosis (37–39) and changing unhealthy dietary patterns has been shown to lessen cancer treatment sequelae, possibly reducing the risk of recurrence for specific cancer types (40). The E-DII distribution score were not statistically significant between subjects without cancer and cancer survivors; however, we did observe a gradually decreased proportion from the most anti-inflammatory quartile to the most pro-inflammatory quartile. This little shift suggests that cancer survivors may change dietary patterns post cancer diagnosis, but this remains subject to speculation. Prospective research is warranted to explore the relationships between dietary changes and depressive symptoms in the future.

The mechanism between the positive association between DII score and depression is not clearly clarified. Cumulating systemic inflammatory level after consuming the pro-inflammatory diet may underlie the depression (21). Several evidences could explain the underlying association between systemic inflammation and depression, at least partly. First, systemic inflammation can produce overdose free radicals resulting in oxidative damage (41). The oxidative stress biomarkers are increased in depressive patients (42). A likely explain is that the brain is particularly susceptible to oxidative damage because of its high lipid content and energy consumption and relatively limited anti-oxidants defense (43). Second, systemic inflammation can increase the level of proinflammatory cytokines released from immune cells (44). Proinflammatory cytokines may induce the hyperactivity of the hypothalamus–pituitary–adrenal axis and an imbalance of tryptophan metabolism, which are all involved in the etiology of depression (45). Lastly, systemic inflammation may influence the brain-gut-microbiota axis, the intestinal microbiota and diet play an essential role in these gut-brain interactions and be involved in the pathogenesis of depression (46).

Previous studies reported higher E-DII scores and predict a higher risk of adverse health outcomes, including all-cause, cancer-cause, and CVD-cause mortality (47–51). Our study also observed the positive E-DII score association with all-cause and CVD cause mortality in subjects without cancer, but not associated with cancer-cause death. However, in cancer survivor subjects, reports on the correlation between E-DII score and mortality among different studies were controversial. The two studies, with median 13.3 and 14.6 years of follow-up, deemed that higher E-DII were associated with a shorter overall survival time in survivors of breast cancer (52, 53) while another study on colorectal cancer survivors (54),with a median follow-up time of 7 years, found no E-DII association with cancer prognosis. The inconsistency may result from a variety of cancer types with a different mortality and a diverse follow-up time. There are about 27.8 and 27.0% deaths of breast cancer survivors (52, 53), and about 15% deaths of colorectal cancer survivors (54). However, despite the median follow-up time being over 7 years in our study, only 4.8% of deaths in subjects without cancer and 15.2% of deaths in cancer survivors in our study. Further studies with extended follow-up time and sample size are warranted in the future. In the current study, we compared the subjects without cancer and cancer survivors who have a higher risk of death (HR: 3.59, 95%CI: 3.16–4.07), and cancer survivors had a higher co-morbidity index and are more likely to be tobacco users, which enhance the death risk and mask the relatively weak contribution of E-DII scores on mortality.

Several theories may explain the underlying mechanism between higher E-DII diets and mortality risk. High-DII diets accelerate the rate of telomere shortening, and shorter telomeres indicate increased mortality risk in the general population (55). Diets with high E-DII scores additionally contribute to increased levels of inflammatory cytokines, including C-reactive protein, interleukin 6, and tumor necrosis factor-α, which are associated with a higher chronic diseases and mortality risk (56, 57). The increased inflammatory cytokine level partly account for the mechanisms through high-DII diets increases the CVD-cause death risk. Cumulated inflammatory cytokines cause attraction and migration of inflammatory cells into vascular tissue (58), and enhance the cellular adhesion molecules expression, such as selectins and cadherins, which mediate white blood cell adhesion to the vascular endothelium (59). Moreover, cumulative inflammatory level produce overdose free radicals that cause oxidative damage (60), which are associated with CVD risk and mortality (61). Nonetheless, the exact mechanism needs further confirmation.

As expected, we also observed the association of depression with all-cause mortality in subjects without cancer after E-DII score adjustment and other covariates. Depression itself increases mortality (62) and depressive victims also suffer from other comorbidities such as asthma and CVD, which could worsen the overall health status and enhance the risk of mortality (63, 64). In addition, there were no significant interaction effects between E-DII score and depression in their association with all-cause, cancer-cause death, and CVD-cause death (Supplementary Table 3). These independent associations of E-DII score or depression with mortality suggest that their pathogenesis on mortality may differ and further studies are warranted.

Our study has strengths. We conducted the present analysis using the public data from NHANES, a nationally representative survey with high-quality measurements, which improves the generalizability of the results. Additionally, this is the first study to explore the relationship between E-DII score and depression among cancer survivors using NHANES, and this work also examine for the first time the association between the E-DII score, depression and mortality. Previous researchers have suggested that dietary supplementation with nutrients such as vitamin (65, 66), minerals (67) and polyunsaturated fatty acids (68, 69) could alleviate inflammation-related depression. The emerging field of nutritional epidemiology turns its attention to the relationship between dietary patterns and health outcomes, rather than focusing on specific nutrients or foods. The use of E-DII as an indicator to directly and reasonably connect the three of nutrition, inflammation, depression and mortality, may have clinical and public health significance for promoting good mental health and reducing death induced by diet. However, we acknowledge some limitations in this study. First, estimation of E-DII score effect on depression is based on data from cross-sectional design. This may limit the evidence grade on causal inference. Second, the E-DII score and depression score are calculated by self-report from 24 h dietary information and PHQ-9, which could not preclude the possibility of information biases. However, data on depression and dietary patterns were based on repeated and validated records and this could minimize this concern. Third, cancer-related measures such as tumor stage and therapy were not taken into account because of no relevant data. We adjusted for potential predictors of survival to minimize the potential effects of confounding factors. Fourth, reverse causation could be possible, as dietary consumption and nutritional status may be altered when people feel unwell. Future studies with high quality design are required to confirm the effect of the dietary inflammatory score on depression and mortality risk.

In summary, we have demonstrated that higher E-DII score is associated with an increased depression risk. Higher E-DII score and depressive symptoms are related to increased risks of all-cause mortality or CVD-cause mortality in subjects without cancer, but not in cancer survivors. Encouraging anti-inflammatory diet may be an effective way to prevent depression and reduce mortality risk may have clinical and public health implications.

Publicly available datasets were analyzed in this study. This data can be found here: https://wwwn.cdc.gov/nchs/nhanes/default.aspx.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.

JJ and ZJ conceived and designed the study. HH and YuW accessed the data. YaZ and DC performed statistical analysis and checked date analysis process. YuZ and YaW drafted the manuscript. JJ, ZJ, and XC revised the manuscript. All authors made critical comments on the manuscript and approved the final version for submission.