Regression Equation Analysis Enhances Detection of Conduction Slowing Beyond Exclusive Axonal Loss in Diabetic Distal Symmetrical Polyneuropathy
Kelly Saverino1, Zhao Zhong Chong1, Joseph Wardell1, Kazim Jaffry1, Tejas Patel1, Abu Nasar1, Ankit Pahwa2, Howard Sander3, Nizar Souayah1
1Rutgers New Jersey Medical School, 2SMR Consulting, 3Department of Neurology, NYU Grossman School of Medicine
Objective:
Objective: To evaluate the ability of regression analysis of conduction slowing to identify superimposed demyelination beyond fiber loss in diabetic distal symmetrical polyneuropathy (DSP).
Background:
Conduction slowing caused beyond pure axonal loss has been reported in DSP and was attributed to a demyelinating component. However, standard electrodiagnostic examination does not clearly differentiate between pure large fiber loss and large fiber loss with minor degrees of demyelination.
Design/Methods:
We previously used regression equation analysis to develop confidence intervals that assess the range of conduction slowing from primary demyelination in chronic inflammatory demyelinating polyneuropathy, and then used the equations to investigate the conduction slowing profile in ALS. This current study uses regression equation analysis to assess conduction slowing in 219 diabetic DSP patients and 219 axonal non-diabetic DSP and 95 ALS patients.
Results:
Conduction velocity (CV) is significantly slower in diabetic DSP than in non-diabetic DSP in all tested nerves. More patients fulfill the regression equation criteria in the diabetic group compared to the non-diabetic group (47.0% vs. 23.3%).There is significantly higher number of patients who have more than 2 motor nerves with CV slowing fulfilling the AAN or regression criteria for demyelination in the diabetic DSP group vs. the axonal non-diabetic and ALS groups (25.6% vs. 7.8% vs 0; p < 0.0001). In subgroups with at least one nerve with CV slowing in the American Academy of Neurology (AAN) demyelination range, the estimated likelihood of having more than two motor nerves with CV slowing in the demyelination range by AAN or regression equations criteria is significantly higher in the diabetic DSP group (0.73) compared to non-diabetic DSP group (0.52, p < 0.05), suggesting more diffuse demyelination in the diabetic group.
Conclusions:
Regression analysis can identify conduction slowing in patients with diabetic DSP suggestive of demyelination, which is not detected by conventional electrodiagnostic testing.