Background: Antiphospholipid syndrome (APS) causes recurrent thrombosis and pregnancy morbidity in the context of antiphospholipid antibodies (aPL). Nailfold capillaroscopy (NFC) enables objective assessment of peripheral microcirculation, yet an APS-specific NFC phenotype and its clinical correlates have not been established.

Objectives: To characterise AI-derived NFC patterns and quantitative metrics in APS and explore associations with aPL, vascular phenotype and pregnancy morbidity.

Methods: Cross-sectional study of patients with APS. Clinical history captured arterial and venous thrombosis, obstetric morbidity, and antithrombotic therapy. aPL profile included lupus anticoagulant (LAC), anticardiolipin (aCL) and anti-β2-glycoprotein I (aβ2GPI) IgG/IgM; single/double/triple positivity was recorded. In SLE-APS, activity and damage were assessed (SELENA-SLEDAI, SDI, DIAPS). NFC of bilateral ring and little fingers was analysed [1] using Capillary.io® (AI-based platform extracting quantitative microvascular parameters) [2], providing pattern classification and quantitative microvascular metrics (density, morphometry and lesion proportions). Associations were explored using χ2 tests for categorical variables (Monte Carlo exact p-values when appropriate) and ANOVA/Kruskal–Wallis for continuous variables (two-sided p<0.05).

Results: Fifty-one women with APS were included, comprising 7 (13.7%) with primary APS and 44 (86.3%) with SLE-associated APS. Mean age was 35.2±8.1 years and disease duration 6.7±5.4 years. Antithrombotic therapy was used in 74.5% (antiplatelet 33.3%, anticoagulation 58.8%). Clinical APS manifestations were vascular only 47.1%, obstetric only 17.6% and both 35.3%. Venous events were more frequent than arterial events (68.6% vs 47.1%). Among patients with pregnancy history (n=39), abortions occurred in 69.2%, with ≥3 abortions in 48.1%; abortion timing was early 44.4%, late 33.3% and mixed 22.2%. In SLE-APS with available indices (n=46), SELENA-SLEDAI was 12.2±9.9, SDI 2.6±1.9 and DIAPS 2.4±1.8. aPL profile showed LAC positivity in 80.4%, any aCL positivity in 58.8% and any aβ2GPI positivity in 52.9%; triple positivity was present in 31.4% (single 37.3%, double 31.4%). NFC pattern was normal in 54.9% and non-specific in 43.1%; an early scleroderma-like pattern was rare (2.0%). Across the cohort, AI-derived quantitative metrics showed a low capillary density (6.83±0.98/mm), alongside % enlarged capillaries 19.04±14.30; megacapillaries were rare (0.01±0.10). Microhaemorrhages were infrequent (total haemorrhages median 0, IQR 0–3). with mean loop diameter 18.4±2.4 µm, arterial limb width 12.6±1.9 µm, and venous limb width 15.7±2.3 µm. aβ2GPI IgG positivity was associated with NFC pattern (normal pattern 72.7% vs 41.4% in negatives; p=0.015). In contrast, no significant associations were observed for LAC or aCL (IgG/IgM) with NFC pattern or quantitative metrics. Arterial events were associated with a normal NFC pattern (70.8% vs 40.7%; p=0.049) and a more “normal-like” quantitative profile, including higher % normal capillaries (79.4±13.0 vs 66.9±12.6; p=0.001) and lower % enlarged (13.3±14.0 vs 24.1±12.8; p=0.001) and % tortuosity (6.6±4.0 vs 10.3±6.7; p=0.030), in a cohort with overall low capillary density. NFC metrics showed a graded association with the number of arterial events: % normal capillaries increased and % enlarged capillaries decreased from none to single to multiple events (p=0.005 for both); tortuosity also decreased (p=0.049). Venous events overall were not associated with NFC metrics; however, venous event type showed differences in capillary density (p=0.009). No significant associations were observed between NFC parameters and obstetric phenotype, number/timing of abortions, aPL multiplicity, or antithrombotic therapy and microhaemorrhages.

Conclusions: In women with APS, AI-assisted NFC most often showed a normal or non-specific pattern; however, the overall low capillary density suggests measurable microvascular involvement. Paradoxically, patients with arterial events—and those with increasing numbers of arterial events—displayed a more “normal-like” profile, driven by a higher proportion of normal capillaries and fewer enlarged capillaries, despite reduced density at cohort level. Pregnancy morbidity was not mirrored by nailfold microcirculation in this dataset. Larger, longitudinal studies with multivariable adjustment are needed to validate AI-derived NFC metrics as biomarkers for APS stratification.

REFERENCES: [1] El Miedany, Y., Ismail, S., Fawzy, M., Elgaafary, M. and Abu-Zaid, M.H., 2023. POS1300 nailfold capillaroscopy for standard clinical practice: are two fingers as accurate as eight?. Annals of the Rheumatic Diseases, 82, pp.997-998.

[2] Lledó Ibáñez GM, et al. CAPI-DETECT: Machine Learning in Capillaroscopy Reveals New Variables Influencing Diagnosis. CAPI-detect: machine learning in capillaroscopy reveals new variables influencing diagnosis. Rheumatology (Oxford). 2025 Feb 7:keaf073. doi: 10.1093/rheumatology/keaf073.

Acknowledgments: NIL.

Disclosure of Interests: None declared.