Response to Comment on Andersen et al. Risk-Factor Trajectories Preceding Diabetic Polyneuropathy: ADDITION-Denmark. Diabetes Care 2018;41:1955–1962

We thank Rigalleau et al. (1) for their interest in our study (2) on risk-factor trajectories preceding a clinical diagnosis of diabetic polyneuropathy (DPN) 13 years after a diagnosis of type 2 diabetes by screening. Rigalleau et al. showed that in patients with long-standing diabetes, retrospectively analyzed strong declines in HbA_1c levels over 6 years were associated with higher indices of DPN (vibration perception thresholds and skin autofluorescence) (1). Strong decline was defined as the quintile of patients experiencing the greatest decline in HbA_1c in the 6 years preceding DPN assessments. Associations were studied in linear regression models adjusted for sex, age, diabetes duration, BMI, blood lipids, and arterial hypertension.

In our study (2), we hypothesized that the velocity of risk-factor changes during the course of diabetes might reflect distinct pathophysiological mechanisms and that the effect of change in risk factors over time may influence the development of DPN over and above the effect of risk-factor levels. We reported higher baseline levels of HbA_1c and steeper increases in HbA_1c over time being independently associated with a subsequent diagnosis of DPN. The effects of HbA_1c levels and change on the risk of DPN were estimated using multivariate logistic regression adjusting the effect of HbA_1c levels for the effect of HbA_1c change and vice versa (in addition to adjustments for sex, age, diabetes duration, height, and trial randomization group). Our findings might indicate a protective effect of stabilizing blood glucose in addition to lowering glucose levels to prevent DPN.

Rigalleau et al. (1) asked if an increased risk of DPN was present in participants of the ADDITION-Denmark study who experienced the greatest decline in HbA_1c preceding DPN assessment. We analyzed this question by categorizing our study cohort of 452 participants according to slopes of change in HbA_1c since diabetes diagnosis. We defined decliners as the participants in the lowest quintile of HbA_1c slopes and nondecliners as the remaining participants. It should be noted that, given the much more narrow HbA_1c range in our study, the decliners in our analysis would probably have been defined as nondecliners in the population studied by Rigalleau et al. In multivariate logistic regression models adjusting for sex, age, diabetes duration, height, and trial randomization group, a higher odds ratio of 1.74 (95% CI 1.01; 3.00) for DPN after 13 years was seen for decliners compared with nondecliners. However, the increased risk of DPN for decliners was attenuated markedly to odds ratio 1.14 (95% CI 0.52; 2.51) and was no longer statistically significant when adjusting for baseline levels of HbA_1c. Hence, we found that the apparent DPN risk elevation seen in those with the steepest HbA_1c decline is largely explained by the fact that those with the steepest decline tend to have the highest HbA_1c starting levels (in other words, they have been exposed to poorer glycemic control). In our opinion, studies of the potential impact of changes in blood glucose on the development of DPN should always take into account the baseline burden of hyperglycemia, as reflected by initial levels of HbA_1c.