Renal Remission Status and Longterm Renal Survival in Patients with Lupus Nephritis: A Retrospective Cohort Analysis

Study design

This was an observational study (GSK Study WEUKBRE6068) involving retrospective analysis of the prospectively collected Hopkins Lupus Cohort study data. The study assessed longterm outcomes of LN based on the degree of renal remission following standard-of-care therapy as measured at 24 months posttreatment.

Eligible patients with LN were identified from the Hopkins Lupus Cohort records, and patients’ remission status was categorized retrospectively into ordinal (CR, PR, or no remission) or binary (response or no response) renal remission categories at 24 months after LN diagnosis (date of positive biopsy), based on available laboratory measures recorded during standard clinical care. Patients were followed for longterm renal survival from Month 24 after LN diagnosis until the end of followup [because of outcome event, loss to followup, or the end of the study dataset (December 2013); Figure 1].

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Study design. ^aDegree of remission status achieved at Month 24 after LN biopsy data was used as the key exposure variable in this study. Renal remission status was also defined retrospectively at other study timepoints (Month 6 and Month 36). ^bClosest date to Day 0 (LN biopsy date) within period 3 months prior to Day 0 to 3 months after Day 0. ^cClosest date to Day 730 (Month 24) within a period of 3 months prior to Month 24 to 3 months after Month 24. ^dClosest date to Day 1095 (Month 36) within a period of 3 months prior to 3 months after Month 36. LN: lupus nephritis.

Because data were collected in a real-world observational setting, renal functional assessments were not conducted at strict timepoints or regular intervals. Therefore, a designated window of ± 3 months was defined around the Day 0 window for measurement of baseline renal function and model covariates and the Month 24 postbiopsy date. The result most proximal to the Month 24 timepoint was selected. It was also anticipated that few patients would have renal function screening at exactly 6 months and 36 months postbiopsy. Therefore, for the Month 36 assessments, a window of ± 3 months was applied, to allow calculation of remission status. A hierarchy was set for defining which laboratory test to use for each remission measure if more than 1 type of measure was available (Supplementary Table 1, available with the online version of this article). If more than 1 test was available during an assessment window, an algorithm based on clinical preferences was developed to select the most appropriate test.

The Hopkins Lupus Cohort study is approved on an annual basis by the Johns Hopkins University School of Medicine Institutional Review Board (Research Project Notification: NA_00039294), and informed written consent is obtained from all participants.

Patients

Eligible patients were ≥ 18 years of age at diagnosis, had SLE according to the revised American College of Rheumatology (ACR) or the Systemic Lupus International Collaborating Clinics (SLICC) criteria^14,15, plus a biopsy record of International Society of Nephrology (ISN) class III, IV, V, or mixed LN. Patients had to have made at least 1 visit to the Johns Hopkins Lupus Center in the 3-month period following biopsy and at least 1 clinic visit during the followup period (initiation of induction therapy plus 24 mos) to allow assessment of longterm renal outcomes. Each patient was required to have sufficient laboratory data for assessment of renal remission at 24 months. Patients with ESRD at the initiation of induction therapy were excluded.

Remission

Ordinal and binary remission categories were defined according to criteria in the ALMS (Clinicaltrials.gov identifier: NCT00377637)⁹ and BLISS-LN trials (Clinicaltrials.gov identifier: NCT01639339; GSK study 114054; Supplementary Table 1, available with the online version of this article). These criteria were modified (mALMS and mBLISS-LN) to exclude urinary sediment data following findings indicating that these data had low sensitivity and poor predictive value for LN outcomes^16,17. For the current study, mALMS and mBLISS-LN criteria were applied at 24 months.

For the ordinal remission criteria, in accordance with mALMS criteria, CR was defined as occurrence of a return to normal serum creatinine and urine protein ≤ 0.5 g/day. PR was defined as one of either a return to normal serum creatinine, or urine protein ≤ 0.5 g/day. As per mBLISS-LN criteria, CR (ordinal criteria) was defined as occurrence of estimated creatinine clearance within the normal range, and urinary protein:creatinine ratio < 0.5. Creatinine clearance was estimated according to the Cockcroft-Gault formula¹⁸. PR was defined as occurrence of creatinine clearance of no more than 10% below the baseline value or within normal range and ≥ 50% decrease in urinary protein:creatinine ratio to < 1.0 (if the baseline ratio were ≤ 3.0) or < 3.0 (if the baseline ratio were > 0.3).

For the binary remission criteria (remission or no remission), in accordance with mALMS criteria, remission was defined as a decrease in urine protein:creatinine ratio to < 3 in patients with baseline nephritic range proteinuria (≥ 3 urine protein:creatinine ratio), and decrease in the urine protein:creatinine ratio by ≥ 50% in patients with subnephrotic proteinuria (< 3 urine protein:creatinine ratio). In accordance with mBLISS-LN criteria, remission was defined as meeting either the CR or the PR ordinal criteria (no remission defined as not meeting the response criteria).

Endpoints

The primary endpoint was longterm renal survival defined as survival without ESRD or mortality. ESRD was recorded prospectively in the Hopkins Lupus cohort at each clinic visit if patients had received renal replacement therapy (dialysis or transplantation), as part of the SLICC/ACR Damage Index¹⁹. The primary predictor of interest was renal remission status (ordinal criteria: CR, PR, or no remission) according to mBLISS-LN and mALMS criteria at 24 months after the biopsy date (Figure 1). Secondary endpoints included the incidence of chronic renal insufficiency and serum creatinine over time. Chronic renal insufficiency was defined as new kidney damage or new occurrence of creatinine clearance < 60 ml/min on at least 2 consecutive measurement occasions ≥ 3 months apart. In posthoc analyses, proteinuria status (yes/no) at Month 24 was examined to investigate any association with renal survival.

Analysis

The primary study analysis was a comparison of the rates of renal survival between patients achieving PR or CR of LN at Month 24 and those who achieved no remission. Survival analysis (Kaplan–Meier plots with log-rank test and Cox proportional hazards regression) was used to compare event rates by remission status and assess the association between remission status at 24 months and renal death (ESRD or mortality), adjusting for potential confounding variables. Because this was an observational retrospective analysis, safety data such as adverse event reporting were not collected.

Multivariate analyses

Potential confounding variables considered included ISN class (III, IV, or V lupus glomerulonephritis at date of biopsy), early remission, age at date of biopsy, sex, race/ethnicity (black vs other), and SLICC Damage Index score (Day 0 window). A model-building approach was taken to produce final models according to the following rules: each variable was added in turn; those for which the remission category HR changed by > 10% were considered confounders and entered into the model. Owing to concerns about the low number of events observed and the high number of covariates, restricted models were constructed based on assessment of confounding. The number of covariates permitted in the model was set at a maximum of 2. Where there were > 2 confounders identified, the 2 confounders with the greatest percent change in HR were entered.