SPECTROVAS: The evaluation of biospectroscopy as a potential biomarker of disease activity in pauci-immune small vessel vasculitis

Abstract

Anti-neutrophil cytoplasmic autoantibody (ANCA)-associated vasculitis (AAV) characterises an autoimmune disorder that results in inflammation and necrosis of small- and medium-sized blood vessels, causing potential multi-organ and life threatening disease. Immunosuppressive treatment strategies are effective with improved patient survival, but carry a significant risk of treatment-related toxicity and long-term patient morbidity that often results from the sequelae of relapsing disease and required re-exposure to therapy. The current lack of a reliable biomarker of disease activity in multisystem AAV poses a significant clinical unmet need when determining relapsing or persisting disease. Biospectroscopy offers a highly versatile, non-destructive and cost-effective means of analysing a given biological sample to determine its chemical composition; in effect providing a surrogate of its metabolomic profile. This thesis aimed to evaluate the role of biospectroscopy as a candidate biomarker of disease activity in AAV with application to plasma, serum, urine and renal tissue samples.

For both initial biofluid studies, paired blood and urine samples were collected within a single UK centre from patients with active disease, disease remission, disease controls and healthy controls. Three key biofluids were evaluated; plasma, serum and urine, with subsequent chemometric analysis and blind predictive model validation. Considering ATR-FTIR, plasma proved to be the most conducive biofluid, with 100% sensitivity (F-score 92.3%) for disease remission and 85.7% specificity (F-score 92.3%) for active disease. This was independent of organ system involvement and current ANCA status. Considering Raman spectroscopy, plasma and serum samples demonstrated equal ability to discriminate disease activity: F-score 80% for plasma (specificity 93.3%, sensitivity 70%, AUC 0.95) and 80% for serum (specificity 80%, sensitivity 80%, AUC 0.92). Both techniques exhibited similar findings on analysis of paired remission samples following successful remission-induction therapy.

For renal tissue samples, consecutive patients with active ANCA-associated glomerulonephritis (AAGN) and those in disease remission were recruited from a single UK centre. In those with active disease, renal tissue and a paired urine sample were collected. Amongst those in remission at the time of recruitment, archived renal tissue samples taken at the time of initial diagnosis were attained. Using histological data, spectral analysis from unstained tissue samples was able to discriminate disease activity with a high degree of accuracy according to >25% interstitial fibrosis and tubular atrophy (F-score 95%, sensitivity 100%, specificity 90%, AUC 0.98), necrotising glomerular lesions (F-score 100%, sensitivity 100%, specificity 100%, AUC 1) and interstitial infiltrate (F-score 100%, sensitivity 100%, specificity 100%, AUC 0.97). Corresponding spectrochemical changes in paired urine samples was limited.

In this body of work, we confirm for the first time that both ATR-FTIR and Raman spectroscopy offer a novel and functional candidate biomarker in AAV, distinguishing active from quiescent disease with a high degree of accuracy using plasma and serum, as well as the application of machine learning in conjunction with Raman spectroscopy as an innovative low-cost technique for the automated computational detection of disease activity in AAGN. The promising technique of biospectroscopy is clinically translatable and warrants future larger study with longitudinal data and more varied pathology, potentially aiding earlier intervention and individualisation of treatment.