PREGNA-RISK: Pregnancy Risk Stratification in SLE/APS

Problem Statement

Pregnancy in SLE and APS carries 2-5× higher risk of adverse pregnancy outcomes (APO) including preeclampsia (16-30%), fetal loss (6-35%), preterm birth (21-54%), and intrauterine growth restriction (10-30%) compared to the general population (Østensen & Cetin, Best Pract Res Clin Rheumatol, 2015). Current management relies heavily on clinical judgment. Evidence-based quantitative tools integrating serological, clinical, and therapeutic variables are lacking.

Model Architecture

Composite Score

$$S = \max\left(\sum_{i=1}^{17} w_i \cdot x_i,\ 0\right), \quad x_i \in {0,1}$$

Where $w_i$ are evidence-derived weights (positive for risk factors, negative for protective factors) and $x_i$ indicates factor presence.

Risk Factors (17 total)

Positive risk factors:

• Active lupus nephritis (+25) — Bramham et al. JASN 2011
• Anti-dsDNA + hypocomplementemia (+15) — PROMISSE, Buyon et al. Ann Intern Med 2015
• Triple aPL positivity: LA+aCL+aβ2GPI (+20) — Pengo et al. J Thromb Haemost 2018
• Isolated LA (+12), high aCL >40 GPL (+8) — PROMISSE 2015, Lockshin 2012
• Prior APO history (+15) — Bramham 2011
• SLEDAI-2K >4 at conception (+10) — Clowse et al. Arthritis Rheum 2005
• eGFR <60 (+12), proteinuria >0.5g/24h (+10) — Smyth et al. CJASN 2010
• Hypertension (+8), thrombocytopenia <100k (+8) — PROMISSE 2015
• Age >35 (+5), BMI >30 (+5)

Protective factors:

• Hydroxychloroquine (−10) — Leroux et al. Autoimmun Rev 2015; EULAR 2017 recommendation
• Low-dose aspirin (−5) — ASPRE trial extrapolation
• Prophylactic LMWH (−8) — Mak et al. Rheumatology 2017
• Disease quiescence >6 months (−12) — Østensen 2015, EULAR 2017

Monte Carlo Uncertainty

Weight calibration carries epistemic uncertainty. We estimate 95% CI via 10,000 Monte Carlo simulations with uniform ±20% perturbation on each weight: $$w_i^{(k)} = w_i \cdot U(0.8, 1.2), \quad k = 1, \ldots, 10000$$ $$S^{(k)} = \max\left(\sum_i w_i^{(k)} x_i,\ 0\right)$$

Validation

Three clinical scenarios demonstrate expected stratification: (1) quiescent SLE on HCQ → Low; (2) APS with LA + prior loss on triple therapy → Low-Moderate (protective therapy offsets risk); (3) active nephritis + triple aPL + multiorgan → Very High (score 110, CI [99.7, 120.4]).

Clinical Implications

PREGNA-RISK enables systematic preconception counseling, risk-appropriate monitoring intensity, and standardized documentation. The Monte Carlo CI communicates calibration uncertainty to clinicians, supporting shared decision-making.

References

1. Buyon JP et al. Ann Intern Med. 2015;163(3):153-163 (PROMISSE)
2. Bramham K et al. J Am Soc Nephrol. 2011;22(11):2011-22
3. Clowse MEB et al. Arthritis Rheum. 2005;52:514-21
4. Leroux M et al. Autoimmun Rev. 2015;14(11):1013-20
5. Tektonidou MG et al. Ann Rheum Dis. 2019;78:1296-1304 (EULAR APS recommendations)
6. Pengo V et al. J Thromb Haemost. 2018;16:1642-1651
7. Smyth A et al. Clin J Am Soc Nephrol. 2010;5:2060-8
8. Østensen M, Cetin I. Best Pract Res Clin Rheumatol. 2015;29:734-747

RheumaAI • Frutero Club • DeSci