BackgroundThe fluctuation of hemoglobin A1c (HbA1c) and changes in health habits over time was not considered in previous studies. The aim of this study was to evaluate the time-sequenced association between malignancy incidence and HbA1c with a longitudinal study design using repeated measurements of HbA1c.MethodsA retrospective longitudinal study was conducted at a large teaching hospital in Tokyo, Japan, from 2005 to 2016. All participants who underwent voluntary health check-ups at the hospital were included. Our outcomes were the development of malignancy. We compared these outcomes using HbA1c categories. Longitudinal analyses were conducted with a mixed effects model in which time-dependent HbA1c measurements were applied to consider fluctuations in HbA1c levels, adjusted for covariates.ResultsA total of 77,385 nondiabetic participants were included in the study; the mean age was 44.7 and 49.4% of participants were male. During a median follow-up of 1588 (interquartile range 730–2946) days, 4506 (5.8%) participants developed malignancies. The relationship between future malignancies and HbA1c was U-shaped; both the lower HbA1c groups (OR 1.31, 95% CI 1.17–1.46 for < 5.0%) and the higher HbA1c group (OR 1.87, 95% CI 1.03–3.39 for ≥ 7.5%) had significantly higher odds ratios compared to the 5.5–5.9%. The lowest HbA1c was associated with higher odds of breast cancer (OR 1.5, 95% CI 1.21–1.86) and female genital cancer (OR 1.57, 95% CI 1.04–2.37).ConclusionsOur study found a U-shaped association between HbA1c and future malignancies among nondiabetic people but did not find additional risk at the prediabetic level. Low HbA1c may be associated with the incidence of breast cancer and female genital cancer.