To evaluate the use of novel regression equations to investigate the profile of conduction slowing in diabetic distal symmetric polyneuropathy (DSP).

Conduction slowing beyond what is expected from pure axonal loss have been reported in diabetic DSP and have been attributed to an additional demyelinating component. Conventional nerve conduction studies do not allow for a clear distinction between pure axonal loss and loss of large fibers with contributing demyelination. 

Electrodiagnostic data from CIDP patients was plotted, transformed and linear regression analysis was performed to create confidence intervals to determine abnormal nerve conduction ranges of distal latency (DL), conduction velocity (CV) and F latency for compound muscle action potential (CMAP). These ranges were then used to evaluate abnormal nerve conduction values in 219 patients with diabetic DSP and 219 non-diabetic axonal DSP.

The mean CV was significantly slower in diabetic DSP than in non-diabetic DSP groups for all tested nerves. There was a significantly higher number of patients fulfilling the regression equation above criteria in the diabetic group compared to the axonal non-diabetic group (47.0% vs. 23.3%; p < 0.0001). There are significantly more patients with more than 2 motor nerves with CV in the demyelinating range in the diabetic DSP group compared to the axonal non-diabetic group (25.6% vs. 7.8%; p < 0.0001). Furthermore, there are significantly more patients that fulfilled an additional criteria of at least one motor nerve with the corresponding F response in the demyelinating range by AAN criteria in the diabetic DSP group compared to the axonal non-diabetic group (21.0% vs. 4.1%; p < 0.0001).

Regression analysis identified conduction slowing in diabetic DSP, beyond what is expected from exclusive axonal loss. Combining regression analysis for conduction slowing with adequate clinical evaluation may improve the identification of demyelination in diabetic DSP.