Diagnostic performance of a multivariate model of normal epidermal nerve fiber (enf) density for skin-biopsy diagnosis of small-fiber polyneuropathy (sfpn)

Abstract

Background: Distal-leg skin biopsies are an endorsed objective test for confirming SFPN. ENF densities ≤5th centile of the population distribution are considered diagnostic of SFPN. Most laboratories use a single threshold (e.g., 3.8 ENF/linear mm) to determine normality of ENF results. The value of factoring demographics into diagnostic thresholds is untested. Objective: To develop and test a model of normal ENF density that incorporates demographics. Materials and Methods: With IRB permission, we obtained distal-leg skin biopsies from 373 normal volunteers (8-92 years) including 42 children. PGP9.5-immunolabeled ENF were measured using standard clinical methods. Results: Young people aged 8-23 had far more ENF than older adults (426 vs. 227/mm2; p < 0.001). Females had more ENF than males (314 vs. 247/mm2; p < 0.001) and Asians had more than age-matched Whites, Blacks, and Hispanics (336 vs. 237/mm2; p < 0.001). 13 subjects ≤23 years with repeat biopsies at different ages lost 46 ENF/mm2 on average per year, whereas older subjects (n = 9) lost 13 ENF/mm2 per year. We developed and compared a multivariate model of ENF density incorporating age, gender, and race to the single diagnostic threshold. Had we applied the single threshold to all 105 biopsies from patients ≤40 years that our lab interpreted as having SFPN in 2012-2013 using the multivariate model, 75% would have received false negative (normal) diagnoses. Conclusions: Different models yield contradictory interpretations of the same biopsies. Incorporating demographic covariates improves diagnostic sensitivity, especially for young patients. Repeat biopsies document rapid reduction in epidermal innervation during young adulthood.