Deterministic nonlinear features of cutaneous perfusion are lost in diabetic subjects with neuropathy

Abstract

The aim of this study was to analyze and compare the deterministic nonlinear structure of cutaneous laser Doppler flowmetry signals obtained from the forearm and foot of normal subjects and diabetic patients without neuropathy (D), with peripheral neuropathy (DPN) and with combined autonomic and peripheral neuropathy (DAN). Flow oscillations were evaluated under baseline conditions, after local warming of the skin to 44 °C and after warming plus iontophoresis of phenylephrine. The presence of nonlinearity was investigated by three complementary approaches: (i) attractor reconstruction, (ii) calculation of largest Lyapunov exponents (LLEs), and (iii) correlation dimension analysis. Conclusions were validated against surrogate stochastic time series generated by randomizing the Fourier phase of the raw data. In the control and D groups, the combination of phenylephrine and warming unmasked flowmotion with a prominent component at 0.1 Hz. Attractor reconstruction revealed toroidal structure and estimated LLEs were positive. LLEs decreased to zero and dimension estimates increased for surrogate data, consistent with loss of determinism. In diabetic subjects with neuropathy estimates of LLE were not significantly different from zero and dimensions were unaffected by phase randomization. Evidence for nonlinear structure was also obtained under baseline conditions in normal and D subjects, but was lost on warming alone. We conclude that deterministic control mechanisms contribute to cutaneous flowmotion, particularly when pseudo-quasiperiodic behavior is enhanced by phenylephrine. Nonlinear analysis of laser Doppler signals may provide previously unrecognized insights into the effects of diabetic neuropathy on perfusion because it can identify loss of complexity independently of the amplitude of the signals recorded.