Abstract
Cardiovascular Autonomic Neuropathy (CAN) is one of the most devastating complications of Diabetes Mellitus (DM) and presents high morbidity and mortality. Its association with diabetic kidney disease (DKD) worsens the condition even further. CAN diagnosis remains a challenge and is being based on reflex tests which are laborious, risky and difficult to perform. Heart Rate Variability (HRV) tests has been suggested as having high utility in diagnosing CAN, but this issue remains controversial. The aim is to evaluate the sensitivity and specificity of HRV tests to diagnose CAN in patients with type 2 diabetes mellitus (T2DM) and DKD with severely increased albuminuria. This is a cross-sectional study in patients with T2DM and DKD with severely increased albuminuria. A total of 48 subjects were recruited and underwent laboratory and neuropathy assessment. The diagnosis of CAN was first confirmed in 75% (36/48) of patients based on cardiovascular autonomic reflex tests (CARTs). HRV tests (VLF, LF, TP and SDNN) differed between groups with and without CAN (212 vs. 522 ms(2), p = 0.024; 57 vs. 332 ms(2), p = 0.025; 359.5 vs. 2733 ms(2), p = 0.007; 20 vs. 48 ms, p = 0.012), respectively. The best cut-off points based on ROC curve were < 1,117 ms(2), < 152.5 ms(2), < 1,891 ms(2) and < 46.5 ms, respectively. VLF and TP reached highest sensitivity values (97% and 92%) and F1 Score of 90%, while LF had best specificity (75%) and TP had best accuracy (85%). Our best model of serial algorithm using VLF as first screening test and TP in sequency obtained a sensitivity of 97% and accuracy of 90%, reducing in 90% the need to perform CARTs. Our findings suggest that it is possible to achieve high sensitivity and accuracy using an algorithm with VLF and TP parameters analyzed in series. It could enable a simpler and early diagnosis, avoiding CARTs complications.