Response to Letter to the Editor From Pattan et al: "U-shaped Association Between Dietary Zinc Intake and New-onset Diabetes: A Nationwide Cohort Study in China".

To the Editor:

We thank Professor Pattan et al (1) for their insightful comments.

The intake of micronutrients, such as zinc, on the risk of diabetes, is an important and interesting topic that remains to be elucidated. Previous cross-sectional studies (2) had provided some clues, but this type of design cannot determine temporal and causal relations. Prospective studies (3, 4), however, reported inconsistent conclusions. Our recent study (5), in a relatively large-scale, nationally prospective cohort, observed a U-shaped association between dietary zinc intake and new-onset diabetes in the general population of Chinese adults, with an inflection point at about 9.1 mg/day and minimal risk at 8.9 to 12.2 mg/day of dietary zinc intake. However, some important issues in this field need to be further discussed.

First, although a previous study (2) suggested that serum zinc concentration was highly correlated with daily zinc intake, serum zinc concentration was not available in the China Health and Nutrition Survey (CHNS) study. As such, we could not examine the correlation between serum zinc concentration and dietary zinc consumption, as well as the relationship between serum zinc concentration and diabetes. Although the absorption of zinc was a complex process and may be influenced by several factors, we have further adjusted not only the intakes of micronutrients including vitamin A, riboflavin, niacin, vitamin C, copper, magnesium, and iron but also the intakes of major food groups of zinc, including aquatic, nut, red meat, and whole grain to reduce the possible confounding effects. Besides, CHNS does not have detailed information on dietary supplement use. However, data from the 2010 to 2012 China Nutrition and Health Surveillance (6), a nationally representative in China, showed that only 0.21% of the Chinese population reported using zinc supplements. Because of the low supplement proportion of zinc, we speculate that our results may not be materially changed by the dietary supplement use.

Second, the associations between dietary zinc intake and the risk of diabetes are inconsistent in previous studies. These inconsistent findings might be due to differences in the target population, dietary zinc intake levels, and sample size (5). We reported a U-shaped association between dietary zinc intake and new-onset diabetes among Chinese adults (n = 16 257) (5). The inflection point we reported in Table 2 was determined using the likelihood-ratio test and bootstrap resampling method by the R package segmented. We found that the dietary zinc intake and new-onset diabetes association was different by different dietary zinc intake. The risk of new-onset diabetes was significantly lower in participants with zinc intake < 9.1 mg/day and higher in those with zinc intake ≥ 9.1 mg/ day. In Table 3, we divided the dietary zinc intake into deciles and found that the deciles 2 through 8 (dietary zinc intake: 8.9-<12.2 mg/day) had a lower risk of new-onset diabetes. These all indicated a U-shaped association between dietary zinc intake and new-onset diabetes. Consistently, although maybe due to the relatively lower sample size (n = 1056), the comparisons were not significant, a previous conducted in China also showed a potential U-shaped relation between dietary zinc intake and hyperglycemia (7).

Third, the information about new-onset diabetes in CHNS was documented based on the physicians’ diagnoses at each follow-up survey, and the fasting blood glucose and hemoglobin A1c were assessed in the 2009 survey round only. Therefore, diabetes incidence might be underestimated. However, when evaluating the dietary zinc intake levels in the 2009 survey, we found no significant difference in dietary zinc intake between the undiagnosed diabetes cases and the laboratory tests identified cases. Thus, the misclassification might occur virtually independently of dietary zinc intake and might not cause obvious bias in our study. Moreover, similar results were detected when the new-onset diabetes cases were defined only by physicians’ diagnoses (dietary zinc intake < 9.1 mg/day, per mg/day: hazard ratio, 0.74; 95% CI, 0.57-0.96; dietary zinc intake ≥ 9.1 mg/day, per mg/day: hazard ratio, 1.12; 95% CI, 1.09-1.15). However, repeated measurements of fasting glucose and hemoglobin A1c would have allowed more accurate findings.

Overall, our study suggested a U-shaped association between dietary zinc intake and new-onset diabetes in general Chinese adults, with an inflection point at about 9.1 mg/day. We highly agree with Pattan et al that future studies considering more potential confounding are needed to confirm our findings. Further, both the lower and upper reference range of serum zinc concentration may need to be clearly defined, especially for glucose metabolism.