Winning Abstracts from the 2011 Medical Student Abstract Competition: In-Vivo Corneal Confocal Microscopy as a Diagnostic Test for Neuropathy in Type 1 Diabetes - A Cohort Study
Author: Ausma Ahmed, University of Ottawa, Class of 2013
Introduction: The morbidity of type 1 diabetes includes abnormality of small nerve fiber morphology occurring early in the course of diabetic sensorimotor polyneuropathy (DSP). This can be detected by examination of intra-epidermal nerve fibers by invasive skin biopsy. Alternatively, small nerve fibers can be examined non-invasively by a novel in-vivo Corneal Confocal Microscopy (CCM) method. We aimed to determine the CCM parameter and threshold value with optimal operating characteristics for DSP identification.
Methods: At the Toronto General Hospital 89 individuals with type 1 diabetes and a broad spectrum of neuropathy severity were selected. Concurrent with clinical and electrophysiological examination for the determination of DSP (the gold standard), CCM was performed to determine corneal nerve fiber length (CNFL), corneal nerve fiber density (CNFD), corneal nerve branch density (CNBD), and fiber tortuosity (TC). Receiver operating characteristic (ROC) curve analysis was used to determine the parameter with the largest area under the curve and the threshold values with optimal sensitivity and specificity.
Results: Of the 89 participants with type 1 diabetes, 33 (37%) were classified as DSP cases and 56 (63%) as non-neuropathic controls. DSP was associated with older age, longer diabetes duration, greater body mass index (kg/m2), higher blood pressure and higher glycosylated hemoglobin A1c (%). CNFL was significantly lower in those with DSP as compared to those without (1278 ± 411 vs. 1928 ± 491 µm/mm2, P < 0.001) as was CNFD (28 ± 9 vs. 39 ± 10 fibers/mm2, P < 0.001) and CNBD (18 ± 12 vs. 29 ± 16 branches/mm2, P < 0.001). Differences in TC were not significant. Area under the ROC curve was greatest for CNFL (0.88, p=0.0001) compared to all other parameters. The optimal CNFL threshold value was 1612 µm/mm2, with 85% sensitivity and 84% specificity, and positive and negative likelihood ratios of 5.3 and 0.18, respectively. An alternate approach in which a threshold value that maximizes sensitivity (=1823 µm/mm2, sensitivity 91%, negative likelihood ratio 0.16) and a separate threshold that maximizes specificity (=1320 µm/mm2, specificity 93%, positive likelihood ratio 8.5) permitted correct classification of 60 (67%) of subjects.
Conclusion: Among all of the corneal parameters, corneal nerve fiber length had the best capacity to discriminate DSP. A single, optimal threshold offers clinically acceptable operating characteristics, however, an alternative strategy that uses separate thresholds to respectively rule in and rule out DSP has excellent performance while maximizing classified subjects. We hypothesize that values between these thresholds indicate incipient nerve injury that represents those individuals at future risk of neuropathy. Together, this data suggests that CCM is a potential diagnostic test for DSP in clinical practice.