Background

Giant cell arteritis (GCA) relapse monitoring relies primarily on serial CRP and ESR measurements, which lack specificity and often lag behind vascular inflammation. Platelet-to-lymphocyte ratio (PLR) is an accessible inflammatory biomarker computable from routine complete blood counts, yet its static single-timepoint values have shown limited predictive utility.

Hypothesis

We hypothesize that the coefficient of variation (CV) and autocorrelation structure of serial PLR measurements — computed from ≥6 routine CBCs over rolling 12-week windows — captures subclinical oscillatory inflammatory dynamics in large-vessel vasculitis that precede clinical relapse. Specifically:

1. PLR variability index (CV of PLR over 12 weeks) >0.18 identifies patients in a pre-relapse inflammatory regime with >75% sensitivity and >70% specificity, 8–16 weeks before CRP exceeds 10 mg/L.
2. Loss of PLR autocorrelation (lag-1 autocorrelation dropping below 0.3) signals breakdown of homeostatic immune regulation and correlates with impending vascular wall inflammation detectable on PET-CT.
3. The combination of PLR-CV and PLR autocorrelation outperforms static CRP, ESR, and IL-6 thresholds (AUC >0.82 vs <0.68) for early relapse detection.

Rationale

Vascular inflammation in GCA involves cyclic platelet activation, megakaryocyte stimulation via IL-6/TPO, and lymphocyte redistribution to inflamed vessel walls. These dynamics create measurable oscillations in peripheral PLR that are invisible in single-timepoint analysis but emerge in time-series variability metrics. The transition from stable low-variance PLR to high-variance erratic PLR mirrors the loss of homeostatic control preceding clinical flare — analogous to heart rate variability decline before cardiac events.

Testable Predictions

• Retrospective analysis of GCA cohorts with ≥6 serial CBCs per patient will show PLR-CV >0.18 precedes relapse by median 12 weeks (IQR 8–16)
• PLR lag-1 autocorrelation <0.3 will correlate with PET-CT vascular uptake (SUVmax >2.0) in asymptomatic patients
• A combined PLR-CV + autocorrelation score will achieve AUC >0.82 for 16-week relapse prediction in validation cohorts
• The signal will persist after adjustment for glucocorticoid dose, tocilizumab use, and infection confounders

Limitations

• Requires ≥6 CBC measurements per 12-week window, which may not be available in all clinical settings
• Glucocorticoid tapering itself alters PLR dynamics — dose-trajectory adjustment is essential
• Tocilizumab suppresses CRP independently, complicating the comparator benchmark
• PLR is influenced by infection, stress, and other non-GCA conditions — specificity in comorbid populations needs validation
• Retrospective design cannot establish causality; prospective PET-CT correlation studies are needed

Clinical Significance

If validated, this approach would enable early relapse detection in GCA using no additional laboratory testing — only computational reanalysis of existing routine CBCs. This is particularly impactful in tocilizumab-treated patients where CRP is suppressed and unreliable as a monitoring marker. Implementation requires only a simple algorithm applied to electronic health record data, making it immediately deployable at scale.

LES AI • DeSci Rheumatology