Annualized longitudinal change in corneal small nerve fiber measures and associated factors in diabetes: a systematic review and meta-analysis

Background Corneal confocal microscopy (CCM) enables repeated in vivo assessment of corneal small nerve fibers in diabetic peripheral neuropathy. The annual rate of change in CCM measures remains uncertain, as do the clinical factors associated with subsequent nerve loss. Methods PubMed, Embase, Web of Science, and Scopus were searched through July 19, 2026 for longitudinal studies with CCM assessment at two or more time points. The primary outcome was annualized absolute change in corneal nerve fiber length (CNFL), calculated as follow-up minus baseline divided by years of follow-up. Secondary outcomes were corneal nerve fiber density (CNFD), corneal nerve branch density (CNBD), and inferior whorl length (IWL). Random-effects meta-analyses used the generic inverse-variance method. When paired-change variance was unavailable, the primary analysis assumed a within-person correlation of r=0.5, with r=0.2 and r=0.8 examined in sensitivity analyses. Associated factors were synthesized narratively. Results Nineteen longitudinal reports met the review criteria; a reliable count of unique participants could not be established. Five reports contributed to the CNFL reference meta-analysis (sum of analysis n=232), with possible participant overlap. The pooled annualized CNFL change was -0.45 mm/mm 2 /year (95% CI -1.02 to 0.12; P = 0.12; I 2 = 89%). Sensitivity analyses showed that omitting Ferdousi 2020 changed the pooled point estimate to +0.042, while all conventional 95% CIs crossed 0. Four reports contributed to each of the CNFD and CNBD reference analyses (sum n=195 per outcome). Pooled annualized change was -0.09 fibers/mm 2 /year for CNFD (95% CI -0.67 to 0.50; P = 0.77; I 2 = 43%) and -1.70 branches/mm 2 /year for CNBD (95% CI -4.54 to 1.15; P = 0.24; I 2 = 89%). IWL was available from one analyzable study (n=30; annualized change −3.50 mm/mm 2 /year, 95% CI −4.59 to −2.41), without pooling. Narrative evidence suggested possible associations with age, glycemic exposure, HDL cholesterol, physical activity, diabetes duration, and glucose-lowering medications associated with weight gain. Conclusions Current longitudinal evidence does not define a stable annual rate of CCM progression that can be generalized across diabetes populations. CNFL and CNBD estimates favored decline but remained imprecise, while average CNFD change was close to zero. Standardized longitudinal reporting and comparable analyses of associated factors are needed. The four-report CNFD reference analysis excluded Ponirakis 2022; an assumption-dependent five-report sensitivity analysis did not resolve the uncertainty. Systematic review registration https://www.crd.york.ac.uk/prospero/ , identifier CRD420261492037.