Abstract
Objective Lupus nephritis (LN), a common manifestation of systemic lupus erythematosus, is associated with a higher risk of kidney failure and death. The renal pathology of LN helps elucidate the severity of inflammation and the extent of irreversible damage. We aimed to identify histologic variables that correlate with risks of kidney failure and mortality.
Methods Between 2006 and 2019, a total of 526 patients with LN were enrolled. Renal pathology was classified according to the International Society of Nephrology/Renal Pathology Society classification. Components of activity and chronicity indices were analyzed to determine which variables correlated with an increased risk of kidney failure and death, with the adjustment of potential confounders.
Results During the follow-up period (median 7.5, IQR 3.5-10.7 years), 58 patients progressed to kidney failure and 64 died. In the multivariate Cox regression analysis, tubular atrophy (hazard ratio [HR] 2.28, 95% CI 1.66-3.14) and tubulointerstitial inflammation (HR 3.13, 95% CI 1.34-7.33) predicted kidney failure. The renal outcome was even worse if tubular atrophy and tubulointerstitial inflammation coexisted (10-year kidney survival rate: 63.22%). The presence of cellular crescents was associated with an increased risk of death in male patients with LN (HR 1.91, 95% CI 1.02-3.57), whereas the presence of fibrous crescents predicted death in female patients with LN (HR 5.70, 95% CI 1.61-20.25).
Conclusion Histologic variables of renal biopsy in LN could be regarded as prognostic indicators for kidney failure and mortality.
Lupus nephritis (LN), a common manifestation of systemic lupus erythematosus (SLE), affects 20% to 60% of patients with SLE.1-3 Patients with LN are at higher risk for renal failure and mortality compared to patients without LN.4 Unless a contraindication presents, current guidelines recommend that all patients with SLE with clinical evidence of sufficient renal involvement (eg, clinically significant proteinuria) undergo kidney biopsy.5,6 LN is classified into 6 subtypes by the International Society of Nephrology (ISN)/Renal Pathology Society (RPS) classification7,8; proliferative LN—classes III and IV—is considered to be more inflammatory than other subtypes, with worse renal outcomes and long-term survival in SLE. However, whether histologic indices of activity and chronicity indices correlate with long-term renal survival and overall mortality has yet to be fully elucidated.
Prior studies have demonstrated that proliferative LN is associated with increased risks of kidney failure. High LN activity and chronicity indices are predictors of kidney failure.9 Both glomerular pathology and the presence of tubulointerstitial damage have been noted to correlate with kidney failure.10-13 One study examining a cohort with a median follow-up duration of 9.9 years found that fibrinoid necrosis, fibrous crescents, and interstitial fibrosis (IF)/tubular atrophy (TA) were histopathologic variables in predicting kidney failure.14 However, studies focusing on long-term renal outcomes in terms of different histopathologic variables remain scarce.
Class IV LN is known to be associated with mortality in SLE.15 In addition, histopathologic worsening on repeat biopsy was related to an increased risk of death.16 Prior studies have shown that IF/TA is a significant prognosticator of death.17 To date, there is a paucity of data on the effect of pathologic indices of LN on mortality, and previous studies’ sample sizes were too small. Whether a composite of activity and chronicity indices can be a better predictor of mortality needs to be elucidated.
Our study aims to investigate whether any histopathologic variables are associated with increased risks of kidney failure and mortality in our cohort.
METHODS
Study cohort. A retrospective cohort was created by enrolling 537 patients with SLE who had a renal biopsy performed from January 2006 to August 2019 at Taichung Veterans General Hospital, Taiwan. The diagnosis of SLE met the 1997 American College of Rheumatology classification criteria for SLE.18 Patients with SLE were recognized using the International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) code 710.0 and the International Classification of Diseases, 10th revision, Clinical Modification (ICD-10-CM) codes M32.0 to M32.9. The diagnosis of SLE was verified by confirming the certificate of “catastrophic illness” for SLE issued by the Bureau of National Health Insurance, Taiwan, which necessitates a review and verification of the diagnosis of SLE by another 2 rheumatologists. Exclusion criteria were as follows: inadequate renal biopsy specimen (< 10 glomeruli) and causes of glomerulopathies other than SLE. In total, 9 patients were excluded: 4 had inadequate renal tissue and 5 showed histologic patterns not compatible with lupus glomerulonephritis. Further, 2 patients with class VI LN were excluded from the following analysis.
Data source. In this study, we extracted data from the patients’ electronic health records using the Hyperion Enterprise Performance Management System (Oracle). To protect patients’ privacy, all outpatient and inpatient data obtained from the data warehouse were anonymized in advance of analysis.
Ethics approval. This study was approved by the Institutional Review Board of Taichung Veterans General Hospital (CE20338B) and was conducted according to the Declaration of Helsinki. Since all data were deidentified prior to analysis, the requirement to obtain informed consent was waived.
Renal histopathology. Renal biopsy specimens underwent light microscopy, immunofluorescence, and electron microscopy examination. Renal biopsy specimens were fixed in 10% buffered formaldehyde for light microscopy. Consecutive serial 3-μm sections were used for histological staining, including hematoxylin and eosin, periodic acid silver methenamine stain, and Masson trichrome. Slides were reviewed by an experienced nephropathologist (MCW). All specimens contained > 10 glomeruli and were each evaluated using the 2003 ISN/RPS classification and, subsequently, the 2018 revised ISN/RPS classification.7,8 The National Institutes of Health activity and chronicity indices were measured as described in previous reports.8,19
Clinical assessment, treatment, and comorbidity. Follow-ups were scheduled every 1-3 months depending on the treating physician and based on each patient’s disease activity. The estimated glomerular filtration rate (eGFR) for each participant was determined using the 4-variable Modification of Diet in Renal Disease equation.20 Estimated daily urine protein was obtained by spot urine protein-to-creatinine ratio (UPCR). Anti-dsDNA antibody detection was performed using the ELISA-based QUANTA Lite dsDNA assay (Inova Diagnostics, Inc.). Complement levels of C3 and C4 were determined by polyethylene glycol–enhanced immunoturbidimetric assay (Siemens Healthineers), and SLE disease activity was assessed using the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI).21 The presence of antiphospholipid antibodies in all LN subgroups was also extracted from the patients’ electronic health records. Lupus anticoagulant assays were performed using the dilute Russell viper venom time test (HemosIL, Werfen). Anticardiolipin antibodies and anti–ß2-glycoprotein I antibodies were measured by QUANTA Lite ELISA (Inova Diagnostics).
The presence of comorbidities was confirmed by checking the patients’ electronic medical records within a 3-month period before the kidney biopsy. Comorbidities of diabetes mellitus (ICD-9-CM codes 250.00-250.92; ICD-10-CM codes E08.00-E13.9), hypertension (ICD-9-CM codes 401.00-404.93; ICD-10-CM code I10-I13.2), hyperlipidemia (ICD-9-CM codes 272.0-272.4; ICD-10-CM codes E78.0-E78.5), and coronary atherosclerosis (ICD-9-CM codes 414.00-414.07, 414.4, 414.8, and 414.9; ICD-10-CM codes I24 and I25) were categorized in accordance with the adequate ICD-9-CM or ICD-10-CM codes.
The electronic medical records of the enrolled patients were reviewed to identify the medications for LN during the follow-up period, including glucocorticoids (oral or parenteral), hydroxychloroquine, cyclophosphamide (CYC), mycophenolate, azathioprine (AZA), and cyclosporine. For each LN class, the prescription of induction therapy (ie, CYC, mycophenolate, and AZA) and maintenance therapy (ie, mycophenolate and AZA) were calculated. The proportion of LN flares was analyzed. Nephrotic flare was defined as nephrotic-range proteinuria (UPCR > 3.5 mg/g), and nephritic flare was defined as an increase of serum creatinine concentration of ≥ 30% above the baseline value.22
Kidney failure and mortality. The primary outcomes were kidney failure and mortality. Kidney failure was defined by an eGFR < 15 mL/min/1.73 m2 requirement for maintenance dialysis or kidney transplantation. We determined the date of kidney failure onset as the earliest date of regular hemodialysis, peritoneal dialysis, or renal transplant. The occurrence of mortality and causes of death were obtained from the electronic cause-of-death database operated by the Taiwan Ministry of Health and Welfare. Survival time was defined as the time from baseline (ie, kidney biopsy) until the date of either mortality or last follow-up.
Statistical methods. Continuous variables are reported as mean (SD), and categorical variables are reported as the number of patients and the corresponding percentage. We used the chi-square test, Kruskal-Wallis test, and Mann-Whitney U test to examine the unadjusted comparison among different LN classes and histopathology components. Correlations between clinical and histologic variables were assessed using the Spearman test. The predictive factors associated with renal survival and mortality were determined using Cox proportional hazard regression. Covariates with P values < 0.05 in the nonadjusted analysis were included in the subsequent multivariate analysis using the forward selection method. Potential confounders for kidney failure and death were assessed, including age; sex; SLE disease activity, as measured by the SLEDAI, anti-dsDNA antibodies, hypocomplementemia, and proteinuria; and renal histopathology. Given that kidney failure may be associated with presenting renal function, eGFR was excluded from the Cox regression analysis for kidney failure. The significance of variates was determined by calculating hazard ratios (HRs) with 95% CI.
Kaplan-Meier estimates and a log-rank test were performed to calculate 10-year renal survival and mortality. All data analyses were carried out using SPSS, version 22.0 (IBM Corp.). All P values < 0.05 were considered statistically significant.
RESULTS
Patient characteristics and renal histopathology. A total of 526 patients with LN were analyzed. The number of patients within each LN histological class was as follows: 20 patients in class I + II, 68 patients in class III, 43 patients in class III + V, 43 patients in class IV + V, 267 patients in class IV, and 85 patients in class V. The median postbiopsy follow-up time was 7.5 (IQR 3.5-10.7) years. Table 1 displays the patient baseline demographics according to renal histopathology. There was a female predilection among all classes and no statistical differences in age distribution. Although the total leukocyte count did not differ, patients with class IV LN tended to have a lower lymphocyte count (P = 0.02) and lower hemoglobin (P < 0.001). Patients with class IV + V exhibited greater proteinuria, followed by classes IV and V (P < 0.001). Declines in renal function tests were more significant in patients with class III or IV LN than in those with class V LN (P < 0.001). A higher SLEDAI, increased anti-dsDNA antibody levels, and hypocomplementemia were observed in class IV + V LN (P < 0.001); however, the positivity of antiphospholipid antibodies in each class was similar (Supplementary Table S1, available with the online version of this article).
Patient characteristics by ISN/RPS classification of lupus nephritis (N = 526).
The total activity index score was higher in class IV than in class IV + V (median 8, IQR 5-11 vs median 6, IQR 4-9; P < 0.001). Further, when comparing class IV with class IV + V LN, cellular crescents, subendothelial hyaline deposits, leukocyte infiltration, and interstitial inflammation were more frequent in class IV LN (P < 0.001). Cellular crescents, fibrinoid necrosis, and interstitial inflammation were more common in class III than in class III + V LN (P < 0.001). In our cohort, patients with class IV LN had higher total chronicity index scores than other classes (median 1, IQR 0-4; P < 0.001). In terms of nonglomerular pathology, the presence of tubulointerstitial inflammation was more frequent in patients with class IV LN (P < 0.001), but the presence of thrombotic microangiopathy did not differ significantly among all subgroups (P = 0.54). Patients with class V LN were more likely to have diabetes mellitus compared to their counterparts. The distribution of coronary atherosclerosis, hyperlipidemia, and hypertension did not differ significantly among all LN classes. Among the induction therapies for LN, CYC was used more frequently in class IV ± V LN (P < 0.001). Further, patients with class IV ± V LN were more commonly treated with mycophenolate as the maintenance therapy (P = 0.03). In contrast, cyclosporine was most frequently prescribed in class V LN (P < 0.001). Nephrotic flares were most frequently observed in patients with class IV + V LN, followed by class IV and class III + V (P < 0.001).
Correlations between clinical manifestations and activity and chronicity indices. Supplementary Table S2 (available with the online version of this article) demonstrates the association of clinical features with components of the activity and chronicity indices. Lower hemoglobin, higher proteinuria, impaired kidney function at the time of the initial kidney biopsy, increased anti-dsDNA levels, and hypocomplementemia were associated with a higher total activity index (> 6) and the presence of components of the activity index. Further, a higher total chronicity index (> 2) and the presence of components of the chronicity index correlated with declines in eGFR as well. In contrast, the presence of components of the chronicity index did not lead to an association with increased proteinuria.
Predictors for kidney failure. Of the 526 patients, 58 progressed to kidney failure during the follow-up period. Table 2 details the Cox regression analysis for renal survival. In multivariate analysis 1, the presence of TA (HR 2.28, 95% CI 1.66-3.14; P < 0.001), tubulointerstitial inflammation (HR 3.13, 95% CI 1.34-7.33; P = 0.01), and nephrotic flares (HR 2.55, 95% CI 1.39-4.65; P = 0.002) were associated with kidney failure. Similarly, in multivariate analysis 2, TA (HR 1.75, 95% CI 1.26-2.44; P = 0.001), tubulointerstitial inflammation (HR 4.55, 95% CI 1.88-11.00; P = 0.001), and nephritic flares (HR 9.39, 95% CI 2.88-30.60; P < 0.001) were found to be significant prognostic factors of renal survival. To determine whether TA and tubulointerstitial damage were both predictive of renal outcomes, Kaplan-Meier survival curves were delineated according to variable combination (Figure 1). The worst long-term renal outcome was observed in the subgroup that showed the coexistence of TA and tubulointerstitial inflammation, compared with its counterparts (P < 0.01).
Predictors for kidney failure by Cox regression analysis.
Kaplan-Meier renal survival curves according to the combination of TA and tubulointerstitial inflammation. A: absence of TA and tubulointerstitial inflammation; B: presence of TA alone; C: presence of tubulointerstitial inflammation alone; D: coexistence of TA and tubulointerstitial inflammation. TA: tubular atrophy.
Predictors and causes for mortality. Out of 526 patients, 64 (12.2%) died during follow-up; of these, 9 (14.1%) developed kidney failure before death. Table 3 displays the Cox regression analysis for mortality. Since sex is a significant predictor of mortality, we further conducted multivariate regression analyses by sex. Among male patients, the presence of cellular crescent (HR 1.91, 95% CI 1.02-3.57; P = 0.04) predicted mortality; among female patients, fibrous crescent was related to an increased risk in mortality (HR 5.70, 95% CI 1.61-20.25; P = 0.01). Figure 2 shows the Kaplan-Meier survival curves for differences in survival. Female patients had favorable long-term survival (Figure 2A; P = 0.01) compared to male patients. The existence of cellular crescent in male patients with LN was associated with the worst outcome (Figure 2B; P = 0.02). In the female subgroup, the presence of fibrous crescent predicted an increased risk of mortality (Figure 2C; P = 0.001). The main cause of death was active disease status (n = 47), followed by infection (n = 10), malignancy (n = 4), and cardiovascular/cerebrovascular disease (n = 3).
Predictors of mortality in male and female patients with lupus nephritis by Cox regression analysis.
Kaplan-Meier survival curves for 10-year risk of mortality according to (A) sex, (B) the presence of cellular crescent in male patients with LN, and (C) the presence of fibrous crescent in female patients with LN. LN: lupus nephritis.
DISCUSSION
Kidney failure in LN is partly attributable to the cumulative immutable damage to the affected nephrons. Clinical variables are associated with kidney failure and death in patients with SLE; renal histopathologic variables are also associated with an increased risk of poor renal outcome or death.23-25 In our study, we showed that the histologic variables of both activity and chronicity indices were prognostic factors of kidney failure and death, as well as kidney flares. Patients with poor prognostic histologic characteristics should have a comprehensive management plan developed to preserve renal survival and avoid deleterious outcomes.
Our study showed that TA and tubulointerstitial inflammation independently predicted renal survival. Histopathologic variables predictive of ominous renal outcomes varied in different studies.10-14 Interestingly, both glomerular pathology and tubulointerstitial inflammation were predictive factors of renal outcome in previous reports.12,14 Cai et al12 reported that glomerular sclerosis, cellular crescents, and fibrocellular crescents were independent risk factors for kidney failure in 231 Chinese patients with proliferative LN. In that study, severe TA, rather than chronic glomerular lesions (ie, glomerulosclerosis and fibrous crescent), predicted kidney failure12; our study exhibited similar findings. Broder et al11 analyzed 131 patients with LN and elucidated an association between kidney failure progression and nonglomerular pathology. Their study cohort mainly comprised Black and Hispanic patients and found that moderate to severe tubulointerstitial damage, defined by moderate to severe TA and/or IF, rather than tubulointerstitial inflammation, was a compelling predictor of kidney failure.11 In a retrospective study of 105 patients with LN by Rijnink et al,14 kidney failure was predicted by the following variables: fibrinoid necrosis, fibrous crescents, and IF/TA of 25% or greater. Further, progression of IF and TA may potentiate the development of kidney failure.26-28 These findings support our result that chronic tubulointerstitial damage, as the sequelae of precedent inflammation, was associated with poor renal outcome.
Previous studies by Cai at al12 and Rijnink et al14 have shown that active glomerular lesions, especially cellular crescents and fibrinoid necrosis, are predictors of kidney failure. In contrast, our study showed that tubulointerstitial inflammation predicted kidney failure. One potential explanation is that most acute glomerular lesions of the index kidney biopsy might have been effectively ameliorated by induction therapy, whereas tubulointerstitial inflammation was not. Gomes at al29 analyzed 166 patients with LN and showed that moderate to severe tubulointerstitial inflammation was associated with poor renal survival in LN. Taken together, tubulointerstitial inflammation could be a new pathologic biomarker to elucidate the inflammation state of LN, predict renal outcome, and guide future management.25,30
In line with the observations of Wilson et al,13 our data suggested that both tubulointerstitial inflammation and IF/TA predicted worse renal survival. Tubulointerstitial inflammation contributes significantly to the pathogenesis of fibrosis,31 but whether the coexistence of IF/TA has a significant effect on renal survival requires investigation. Further, the severity of tubulointerstitial inflammation may be independent of glomerular pathology.10 Although there have been no prediction models of kidney failure using composite scores of tubulointerstitial damages, chronicity scores of renal histopathology have been added to stratify and predict LN.32 Our results indicate that the composite of active and chronic tubulointerstitial lesions could be a key predictor of renal survival. Clinicians should identify patients with LN who have coexisting IF/TA lesions and manage to prevent kidney failure.
LN is associated with an increased risk of death, but long-term data on the mortality rate in terms of different histopathologic patterns remain lacking. In a longitudinal cohort study by Mok et al4 that enrolled Chinese patients with SLE, the presence of proliferative LN had a significant effect on mortality. In our study, the presence of cellular crescents was associated with mortality in male patients with LN in the multivariate regression analysis. Further, the presence of fibrous crescents was correlated with an increased risk of death in female patients with LN. Surprisingly, tubulointerstitial injuries were not associated with mortality in the multivariate analysis. In contrast, Leatherwood et al17 discovered that IF/TA was a strong predictive factor of death, independent of LN class. There is a paucity of studies that determine the glomerular pathologies associated with death. Our study demonstrated a sex difference in the prognostic factors for mortality. One possible explanation is that LN among male patients with cellular crescents is associated with worse renal function and requires extensive immunosuppressive therapy, which may contribute to an increased risk of infection and mortality. Further studies are required to confirm our findings.
Previous studies have focused on the risk factors that predict composite renal outcomes (ie, death, kidney failure, or a 30% reduction of eGFR from baseline).33,34 The strength of the present study was that we attempted to examine the prognostic factors for kidney failure and death independently among patients with LN; however, there are several limitations. First, because of the retrospective study design, the effect of delay in LN diagnosis and nonadherence of induction/maintenance therapy on the outcomes remained unknown. Moreover, the study enrollment period was prolonged. Within the long follow-up period, the management of LN had substantially evolved. Second, our study only enrolled Taiwanese patients with LN. However, CYC and AZA comprised the majority of induction therapies for LN before the reimbursement of mycophenolate for LN by the Taiwanese Bureau of National Health Insurance in 2017. Though it is possible that our results may not be generalizable to other ethnicities or to patients using contemporary therapy, our results are largely consistent with the published literature. Third, some patients were referred to our hospital for kidney biopsy and it was not feasible to detect the first appearance of urinary abnormalities. In addition, we only analyzed the data from the first renal biopsy, so the potential effect of histological changes following treatment on kidney failure and mortality remains unknown. The number of outcomes (ie, kidney failure and death) might not be large enough for proper analysis. Fourth, we used Cox regression analysis to identify variables predicting clinical outcomes in our study. However, some clinically relevant variables, such as age, sex, and immunosuppressive agents, were not included in the multivariate analysis, which may have underestimated their importance. Last, since all data were deidentified prior to analysis, it would not be feasible to divide clinical SLEDAI scores into renal and nonrenal domains or to elucidate the correlation between causes of death and kidney failure. Future research is needed to investigate the effects of renal and nonrenal domains of SLEDAI scores on the outcomes of patients with SLE.
In conclusion, the histopathology of LN could be regarded as a potential determinant of kidney failure and mortality. Physicians treating patients with LN should be aware of these outcome predictors and provide a stringent management strategy to prevent foreseeable deleterious events.
ACKNOWLEDGMENT
The authors would like to thank the staff of the Biostatistics Task Force of Taichung Veterans General Hospital for their assistance in performing the statistical analyses.
Footnotes
This study was funded by Taichung Veterans General Hospital, Taiwan (TCVGH-1117306C, TCVGH-1117302D).
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication March 3, 2023.
- Copyright © 2023 by the Journal of Rheumatology
