Hypothesis

Serial sample entropy (SampEn) of heart rate variability (HRV) measured via continuous wearable photoplethysmography declines significantly 3–6 weeks before clinical flare in ANCA-associated vasculitis (AAV), reflecting autonomic dysfunction driven by subclinical endothelial inflammation and complement-mediated microvascular injury.

Rationale

AAV targets small vessels with neutrophil-mediated endothelial damage. The autonomic nervous system is exquisitely sensitive to systemic microvascular inflammation: perivascular nerve fibers are damaged early, and complement activation (C5a-mediated) impairs baroreflex sensitivity before clinical symptoms emerge. HRV entropy — specifically SampEn computed over 5-minute nocturnal windows — captures loss of autonomic complexity that linear HRV metrics (SDNN, RMSSD) miss.

Preliminary evidence from SLE cohorts shows HRV reduction preceding flares by 2–4 weeks (Thanou et al., Lupus 2020). AAV, with its direct vascular tropism, should exhibit an even stronger and earlier autonomic signal.

Proposed Study Design

• Design: Prospective observational cohort, 24 months
• Population: 120 AAV patients (GPA + MPA) in remission (BVAS = 0), on stable maintenance therapy
• Intervention: Continuous wrist-worn PPG device (e.g., Empatica E4) recording interbeat intervals
• Primary endpoint: SampEn slope change ≥1.5 SD below individual baseline preceding confirmed flare (BVAS ≥ 3)
• Analysis: Mixed-effects logistic regression with time-varying SampEn covariates; AUROC for 3-week and 6-week prediction windows; Bonferroni-corrected subgroup analysis by ANCA type (PR3 vs MPO)
• Confounders: Medication changes, infections, physical activity (accelerometer-controlled), sleep quality

Testable Predictions

1. SampEn decline ≥1.5 SD precedes clinical flare (BVAS ≥ 3) in ≥65% of events with AUROC ≥ 0.78
2. PR3-ANCA patients show earlier autonomic signal (4–6 weeks) vs MPO-ANCA (2–4 weeks) due to higher endothelial activation
3. Combining SampEn with serial CRP produces AUROC ≥ 0.85, superior to CRP alone (p < 0.01)
4. The signal is independent of glucocorticoid dose changes (confirmed via propensity-adjusted models)

Limitations

• PPG-derived HRV is less precise than ECG; motion artifacts may reduce usable nocturnal windows to ~70%
• 120 patients may be underpowered for PR3/MPO subgroup analysis if flare rate is <15%/year
• Autonomic neuropathy from prior cyclophosphamide may attenuate baseline HRV, reducing dynamic range
• Single-center design limits generalizability; multi-site validation needed
• Confounding by subclinical infection (which also reduces HRV) requires rigorous adjudication

Clinical Significance

If validated, a passive wearable-based early warning system could trigger preemptive ANCA titer and imaging evaluation weeks before organ damage, enabling treat-to-target escalation in a disease where delayed treatment causes irreversible renal and pulmonary injury. This approach requires no additional blood draws and integrates with existing digital health infrastructure.

LES AI • DeSci Rheumatology