HbA1c 5·7-6·4% and impaired fasting plasma glucose for diagnosis of prediabetes and risk of progression to diabetes in Japan (TOPICS 3): a longitudinal cohort study

Abstract

Background: The clinical relevance of the diagnostic criteria for prediabetes to prediction of progression to diabetes has been little studied. We aimed to compare the prevalence of prediabetes when assessed by the new glycated haemoglobin A(1c) (HbA(1c)) 5·7-6·4% criterion or by impaired fasting glucose, and assessed differences in progression rate to diabetes between these two criteria for prediabetes in a Japanese population.

Methods: Our longitudinal cohort study included 4670 men and 1571 women aged 24-82 years without diabetes at baseline (diabetes was defined as fasting plasma glucose ≥7·0 mmol/L, self-reported clinician-diagnosed diabetes, or HbA(1c) ≥6·5%) who attended Toranomon Hospital (Tokyo, Japan) for a routine health check between 1997 and 2003. Participants with a baseline diagnosis of prediabetes according to impaired fasting glucose (fasting plasma glucose 5·6-6·9 mmol/L) or HbA(1c) 5·7-6·4%, or both, were divided into four groups on the basis of baseline diagnosis of prediabetes. Rate of progression to diabetes was assessed annually.

Findings: Mean follow-up was 4·7 (SD 0·7) years. 412 (7%) of 6241 participants were diagnosed with prediabetes on the basis of the HbA(1c) 5·7-6·4% criterion. Screening by HbA(1c) alone missed 1270 (61%) of the 2092 prediabetic individuals diagnosed by a combination of impaired fasting glucose and HbA(1c) 5·7-6·4%. Overall cumulative probability of progression to diabetes did not differ significantly between participants with prediabetes discordantly diagnosed by either HbA(1c) or impaired fasting glucose alone (incidence was 7% for HbA(1c) alone [n=412 individuals and 30 incident cases] and 9% for impaired fasting glucose alone [n=1270, 108 cases]; log-rank test, p=0·3317). Multivariate-adjusted hazard ratios for incident diabetes were 6·16 (95% CI 4·33-8·77) for those diagnosed with prediabetes by impaired fasting glucose alone and 6·00 (3·76-9·56) for diagnosis by HbA(1c) alone, and were substantially increased to 31·9 (22·6-45·0) for diagnosis by both impaired fasting glucose and HbA(1c) compared with normoglycaemic individuals.

Interpretation: Diagnosis of prediabetes by both the new HbA(1c) criterion and impaired fasting glucose identified individuals with an increased risk of progression to diabetes. Although the new HbA(1c) criterion identified fewer individuals at high risk than did impaired fasting glucose, the predictive value for progression to diabetes assessed by HbA(1c) 5·7-6·4% was similar to that assessed by impaired fasting glucose alone. The two tests used together could efficiently target people who are most likely to develop diabetes and allow for early intervention.

Funding: Japan Society for the Promotion of Science; Ministry of Health Labor and Welfare, Japan.