Can we predict kidney involvement in patients with systemic lupus erythematosus? A retrospective cohort study with independent validation

Abstract Objectives To discern predictive factors for incident kidney involvement in patients with SLE. Methods Patients with SLE from the ‘Attikon’ Lupus cohort were monitored for LN, defined by kidney histology and/or classification criteria. Demographic and clinical characteristics at baseline were compared against patients who did not develop LN. LN-free Kaplan–Meier survival curves were generated. A multivariate Cox proportional hazards model was used to identify independent predictors of LN. Independent validation was performed in the University of Crete Lupus registry. Results Among the 570 patients in the derivation cohort, 59 exhibited LN as their initial presentation, while an additional 66 developed LN during the follow-up period (collectively, 21.9% incidence). In the latter group, baseline factors predictive of subsequent kidney involvement were male sex [multivariable-adjusted hazard ratio (aHR) 4.31; 95% CI: 1.82, 10.2], age of SLE diagnosis below 26 years (aHR 3.71; 95% CI: 1.84, 7.48), high anti-dsDNA titre (aHR 2.48; 95% CI: 1.03, 5.97) and low C3 and/or C4 (although not statistically significant, aHR 2.24; 95% CI: 0.83, 6.05; P = 0.11). A combination of these factors at time of diagnosis conferred an almost 90-fold risk compared with serologically inactive, older, female patients (aHR 88.77; 95% CI: 18.75, 420.41), signifying a very high-risk group. Independent validation in the Crete Lupus registry showed concordant results with the original cohort. Conclusion Male sex, younger age and serological activity at SLE diagnosis are strongly associated with subsequent kidney involvement. Vigilant surveillance and consideration of early use of disease-modifying drugs is warranted in these subsets of patients.