Biomarker studies
collection of 129 (97.5–150) months. 50 (72.4%) patients achieved PERR 24±3 months after the LN flare. Most patients in this study were white (42.0%) and identified as female (79.7%). The group of patients who achieved PERR at 24±3 months had more females (p=0.0152), shorter disease duration (5.1 vs 8.7 years, p=0.07, PERR vs non-PERR, respectively), more frequently experi- enced their first LN flare (66% vs 37%, p=0.12, PERR vs non-PERR, respectively) and had higher non-renal SLE Disease Activity Index 2000 scores (6 vs 4, p=0.01, PERR vs non-PERR, respectively). 53 (76.8%) of the patients had a KB at the time of the LN flare, with the most frequent LN classes for both groups being IV (n=17, 24.6%) and V (n=16, 23.2%). Of the 16 patients without a biopsy at flare, 14 had prior biopsies (12 proliferative/mixed, 1 class II, 1 class V). One had class IV LN on subsequent biopsy-proven LN flare, and one was ineligible for a biopsy due to antico- agulation and bleeding risk. Regarding treatment, most patients were treated with mycophenolate (58.0%), with a similar proportion of patients in each group. However, patients who achieved PERR 24±3 months after the LN flare received a higher dose of prednisone, which did not reach statistical significance (40 mg vs 30 mg, p=0.08, PERR vs non-PERR, respectively). When examining patient characteristics 24±3 months after their LN flare, those who did not achieve PERR had significantly higher anti-double-stranded DNA (dsDNA) abs (17 IU/mL vs 53 IU/mL, p=0.009, PERR vs non-PERR, respectively), higher serum creatinine (63.5 µmol/L vs 76 µmol/L, p=0.032, PERR vs non- PERR, respectively), lower serum albumin levels (44 g/L vs 34 g/L, p<0.0001, PERR vs non-PERR, respectively) and remained on higher prednisone doses (6.2 mg vs 15 mg, p<0.0001, PERR vs non-PERR, respectively) (table 1). UBs in our cohort Patients who achieved PERR (n=47) had signifi- cantly lower UBs levels 24±3 months after the LN flare (figure 1A) compared with patients who did not achieve PERR (n=18). The correlations between proteinuria and each UB were stronger across the entire population and among patients who did not achieve PERR (figure 1B,C) than among patients who achieved PERR (figure 1D). Association between UBs and adverse long-term renal outcomes In the whole cohort, 35 (50.7%) patients had a subsequent LN flare (33 (50.8%) of patients with available corrected UB), and 24 (34.8%) developed a 30% sustained decline in eGFR (23 (35.3%) of patients with available corrected UB). In univariable analysis, MCP-1 and adiponectin were linked to higher flare risk, and all UBs were associated with ≥ 30% eGFR decline. In multivariable analysis, after adjusting for possible confounding variables, patients with higher levels of MCP-1 and adiponectin 24±3 months after the LN flare had a higher risk of having a subse- quent LN flare (MCP-1: HR 1.12 (1.03–1.21), p=0.008;
Multivariable Cox proportional regression (HRs and 95% CIs) analysis was performed to determine the asso- ciation between UBs levels and adverse long-term renal outcomes, including subsequent LN flares and a sustained 30% decline in eGFR. For the outcome of time to subse- quent LN flares in the whole cohort, ESRD (eGFR ≤ 15 mL/ min/1.73 m 2 ) was considered as a competing risk and Fine and Gray’s subdistribution hazard model was used. Multivariable models were adjusted for clinically relevant confounders that were selected a priori based on literature review. 4 24 25 Univariable and multivariable Cox regressions were also performed using standardised biomarker values (mean-centred and scaled by SD), allowing HRs to repre- sent risk per one SD. The same models as in the unscaled analyses were applied. To assess consistency in patients with biopsy-proven LN at flare, univariable Cox regression was repeated in this subcohort. Bootstrap validation method with 100 iterations was used to assess the models’ overfitting. Collinearity was assessed using the variance inflation factor. The Cox proportional hazard assumption was checked using the ZPH function. 26 We performed time-dependent receiver operating characteristic (ROC) analysis on the UBs and variables associated with the adverse long-term renal outcomes in the multivariable analysis. Pairwise area under the curve (AUC) comparisons were performed using DeLong’s test with Holm-Bonferroni adjustment, applied to individual and combined biomarkers. Values of the AUC were inter- preted as having outstanding, excellent, good, fair and poor performance, corresponding to values of 1.0–0.91, 0.90–0.81, 0.80–0.71, 0.70–0.60 or <0.60, respectively. 27 Cut-offs for each predictive biomarker were determined by maximising the sum of sensitivity and specificity using Youden’s index. Sensitivity, specificity, positive predictive values (PPVs), negative predictive values (NPVs), positive likelihood ratio test (+LR) and negative likelihood ratio test (–LR) for the UBs and other predictors of subse- quent LN flares and sustained 30% decline in eGFR were estimated with 95% CIs. Sensitivity, specificity, PPV and NPV above 80% were considered good and above 90% excellent. LR above 10 for +LR and below 0.1 for –LR was considered to provide strong evidence for developing subsequent LN flares or sustained 30% decline in eGFR. 28 All statistical analyses were performed using R software V.4.2.2 and all p values were two-sided, with values <0.05 considered statistically significant. GraphPad Prism V.9 was used to generate graphs. Patient and public involvement Patients or the public were not involved in the design, conduct, reporting or plans for dissemination of our research.
RESULTS Cohort characteristics
69 patients were included in our study, with a median (IQR) follow-up time after the 24-month urinary sample
3
Baker R, et al . Lupus Science & Medicine 2026; 13 :e001724. doi:10.1136/lupus-2025-001724
Powered by FlippingBook