1541: MRI Digital Artery Volume Index (DAVIX) Predicts Future Digital Ulcers and Disease Progression in Systemic Sclerosis

Michael Hughes¹, Stefano Di Donato², Klodian Gjeloshi³, Giuseppina Abignano⁴, Fiammetta Danzo³, Giovanni Lettieri G⁴, Enrico De Lorenzis², Dominic Bertham⁵, Philip O’Connor², Olga Kubassova⁶ and Francesco Del Galdo², ¹Department of Rheumatology, Tameside and Glossop Integrated Care NHS Foundation Trust & The University of Manchester, Manchester, United Kingdom, ²Leeds Institute of Rheumatic and Musculoskeletal Medicine, Faculty of Medicine and Health and NIHR Biomedical Research Centre, University of Leeds, Leeds, United Kingdom, ³University of Campania ‘‘Luigi Vanvitelli’’, Napoli, Italy, ⁴Istituto Reumatologico Lucano (IReL), Azienda Ospedaliera Regionale San Carlo, Potenza, Italy, ⁵NIHR Leeds Biomedical Research Centre, Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom, ⁶Image Analysis Group, London, United Kingdom

Background/Purpose: Fibroproliferative vasculopathy in systemic sclerosis (SSc) is caused by neointima proliferation is a key pathologic feature of SSc which results in arterial vessel narrowing (e.g., through adventitial fibrosis) and contributes to microvascular injury and damage. The aim of our study was to determine the value of non-contrast MRI-based DAVIX in predicting digital ulcers (DUs) and worsening of patient reported outcomes (PROs)/clinical manifestations in SSc.

Methods: 91 consecutive patients with Raynaud's phenomenon were enrolled in this study.

At baseline, patients underwent an MRI scan of their dominant hand. We utilised a time-of-flight MR angiography (TOF) sequence which allows the visualization of vessels by the enhancement of flowing blood and the suppression of surrounding stationary tissue in a definite volume (Figure 1). Briefly, our method uses a region growing algorithm to semi-automatically detect the arteries and the finger anatomy (Figure 1). DAVIX of the dominant hand was calculated as %mean of the 4 fingers.

Data collected included (but was not limited to): pulmonary function tests, nailfold capillaroscopy, modified Rodnan Skin Score (mRSS), history/presence of DU, and sHAQ-DI.

The distribution of data was analysed with D'Agostino-Pearson normality test. Medians were compared by Mann-Whitney-Wilcoxon test, correlation with clinical manifestations was performed using Spearman's or Pearson test, as appropriate using Prism 7, GraphPad Software. Correlation between DAVIX and capillaroscopy was scored as a semi quantitative integer scale ranging from 0 to 4 (normal, non-specific, early, active, late).

Results: The majority (86%) were females, 62 fulfilled 2013 SSc classification criteria with 5.9 (7) years median (IQR) disease duration. Twenty-nine were patients with VEDOSS (mRSS Score < 9). Complete historical and prospective (12-month) follow-up data were available for 85 patients. Patients fulfilling SSc criteria had a lower median (IQR) DAVIX compared to those with VEDOSS [0.46% (0.48) vs 0.75% (0.54), P=0.0025] (Figure 2).

DAVIX was significantly lower in patients with DUs at baseline (P=0.01) and in those with DU history (P=0.008) (Figure 2). Further, in the group with no previous/current DU, DAVIX of patients who developed new DUs during follow-up was 0.23% vs. 0.66% (P=0.02) (Figure 2). ROC curve analysis (Figure 3) indicated that a DAVIX threshold < 0.36% conferred a 4 times higher risk of new DUs.

DAVIX correlated with disease duration (r=-0.415, P< 0.0001), baseline mRSS (r=-0.278, P=0.028), DLCO% (r=0.368, P=0.004), FVC/DLCO (r=-0.337, P=0.009), and capillaroscopy pattern (r=-0.447, P< 0.001). Additionally, DAVIX correlated with worsening HAQ-DI (r=-0.308, P=0.024), Raynaud (r=-0.270, P=0.048) and DU VAS (r=-0.291, P=0.044).

Conclusion: DAVIX is a promising surrogate outcome measure of vascular disease in SSc. The predicative value of DAVIX to predict worsening of PROs and future DUs may be useful for patient enrichment/stratification in clinical trials and offer insights into vascular disease activity in SSc disease progression.
Figure 1. Workflow of the finger and vessel segmentation. DYNAMIKA (IAG, Image Analysis Group) workflow. Left-sided images: 3D T1 VIBE images and 2D TOF images are uploaded. The target finger is identified and outlined with a bounding box. Middle images: Finger segmentation model applied to generate the volume of the finger and the vessels segmentation model is applied to generate the vessels volume. Right-sided images: 3D rendering of the segmented finger and blood vessels.

Figure 2. DAVIX in patients who fulfilled the ACR/EULAR 2013 SSc classification criteria versus patients presenting a score < 9 i.e., very early diagnosis of SSc (VEDOSS) (A). DAVIX in patients with baseline digital ulcers versus patients without baseline digital ulcers (B). DAVIX in patients with history of digital ulcers versus patients who never experienced digital ulcers disease (C). DAVIX in patients without baseline digital ulcers who developed digital ulcers at 12-month follow up versus patients without baseline digital ulcers who did not develop new ulcers after 12 months (D). DAVIX of the single fingers of the dominant hands of patients presenting with at least one digital ulcer at baseline: fingers with current ulceration vs fingers without current ulceration (E). DAVIX in patients presenting a stable CHF at 12-month follow up vs patients experiencing a worsening in the CHF after 12 months (F).

Figure 3. ROC curve with area under curve for the performance of the DAVIX index in the detection of new digital ulcers at 12 months. The 0.36% cut-off shows a specificity of 84% and a sensitivity of 67% in identifying patients who went on developing new digital ulcers. Patients with a dominant hand DAVIX © index inferior to 0.36% showed a significantly higher risk of developing new ulcers at 12 months (Incident Risk Ratio=4.55, Odds Ratio=7.31, Attributable fraction in the exposed=78%, P=0.005).
Disclosures: M. Hughes, Actelion pharmaceuticals, Eli Lilly, Pfizer; S. Di Donato, None; K. Gjeloshi, None; G. Abignano, None; F. Danzo, None; G. Lettieri G, None; E. De Lorenzis, None; D. Bertham, None; P. O’Connor, None; O. Kubassova, IAG Image Analysis Group; F. Del Galdo, AbbVie/Abbott, AstraZeneca, Boehringer-Ingelheim, Mitsubishi-Tanabe, Capella biosciences, Chemomab LTD, Kymab.