Introduction
Lupus nephritis (LN) is a significant contributor to the prevalence of kidney disease in South Africa (1) and an important predictor of morbidity and mortality in systemic lupus erythematosus (SLE).(2–5) Early diagnosis of LN is essential to impact patient and kidney outcomes meaningfully.(6)
Percutaneous kidney biopsy remains the gold standard for diagnosing LN and determining the risk of progression to kidney failure (KF),(7) and to facilitate early diagnosis, guidelines recommend biopsy for proteinuria above 500mg per day or the presence of red or white blood cell casts or acanthocytes.(8) Resource limitations in South Africa restrict native kidney biopsy to tertiary facilities with expertise in this procedure.(1) Clinical and biochemical markers may facilitate the identification of individuals with aggressive disease, potentially leading to earlier referral, diagnosis, and treatment.
Hypertension, hypocomplementemia, hypoalbuminemia, and anti-double stranded DNA (anti-dsDNA) are among the parameters associated with poorer renal outcomes in LN.(9–12) African ethnicity, age greater than 30 years, and elevated baseline serum creatinine levels may also confer a worse prognosis.(7,13) The utility of these parameters in the local context remains untested. Accordingly, this study was undertaken to analyze the patterns of presentation of LN in terms of clinical features, serology, and histopathology and to determine the effects of these parameters on renal outcomes.
Materials and methods
A retrospective single-center review was undertaken of all patients with biopsy-proven LN diagnosed from 1 January 2008 to 31 December 2017 at the Helen Joseph Hospital (HJH), a tertiary healthcare facility in Johannesburg, South Africa. Patients were identified by cross-referencing renal outpatient department files and biopsy registers at the HJH renal unit, the anatomical pathology department, and the National Health Laboratory Services (NHLS) electronic platform. The reviewed biopsies were performed based on Kidney Disease Improving Global Outcomes (KDIGO) criteria of suspected LN.(8)
A total of 84 patients were identified. Patients under the age of 18 at the time of diagnosis (n = 5), concomitant Human Immunodeficiency Virus (HIV) infection at the time of diagnosis (n = 9), active urinary tract infections at the time of biopsy (n = 0), inadequate biopsies (n = 3), or missing histopathology data (n = 20) were excluded from the analysis. Thus, 47 patients were included in the study.
Demographic data (sex, age, and self-reported race); total cholesterol; serum creatinine; estimated Glomerular Filtration Rate (eGFR) (as determined by the Modification of Diet in Renal Disease equation, then in use by the National Health Laboratory Service); urine cell counts (leukocyturia and haematuria per microbiology reports) and urine protein: creatinine ratio (UPCR); serum albumin; autoimmune serologies and titres (Anti-nuclear factor, Anti-dsDNA antibody, Anti-smith antibody, Anti-ro antibody, Anti-la antibody, Anti-RNP antibody); complement; presence of hypertension (defined as BP >140/90 mmHg); and WHO/ International Society of Nephrology (ISN) LN histological classification (established in 2003) were anonymously extracted for all included patients. Renal outcomes data was extracted for those patients with a minimum period of 12 months follow-up. Renal outcomes were categorized using KDIGO definitions, which included complete remission with stable kidney function, partial remission, ongoing treatment, and loss of kidney function, respectively. Complete remission was defined by stabilizing or improving kidney function, normalizing proteinuria to <300g per 24 hours, and resolving active urinary sediment. Partial remission was defined by a 50% improvement in proteinuria below the nephrotic range with inactive urinary sediment and stable kidney function. Loss of kidney function was described as an eGFR of <15ml/min/1.73m2 or the initiation of long-term kidney replacement therapy (KRT). Statistical analysis was performed using Statistica® (version 13.5,TIBCO, Pallo Alto, USA) for analysis.
The distribution of clinical and demographic data was determined using Shapiro-Wilk W testing and visual inspection of the histogram plot. Non-parametrically distributed continuous variables were analyzed using the Mann-Whitney U test. Categorical data was compared using Fischer Exact or Pearson's Chi-squared tests as appropriate. Statistically significant results were further tested using logistic regression. General discriminant analysis was applied to determine the level of the continuous variable with significant association. Survival curves of proliferative and non-proliferative groups were fitted to time to remission, time to loss of kidney function, and time to first relapse using the Kaplan-Meier technique. Survival outcomes were compared using the Cox Mantel F test. Multifactorial Cox regression and multifactorial linear regression were used to determine the effect of clinical and histological parameters on renal outcomes and kidney function at follow-up. Statistical significance was determined by a p-value <0.05.
The University of the Witwatersrand Human Research Ethics Committee (protocol number M200152) approved the study.
Results
Patients of Black African ethnicity comprised the largest demographic in the included cohort (32 patients, 72.7%). The majority (33 patients, 73.3%) of the cohort was female. The median age at diagnosis was 26.5 years old (IQR 24–36 years). Baseline clinical, biochemical, and autoimmune parameters are shown in Table 1.1. The median creatinine at diagnosis in this series was 74 μmol/L (IQR 60–117 μmol/L), equivalent to a median eGFR of 101.4 ml/min/1.73m2 (IQR 64–126.8 ml/min/1.73m2).
Presenting clinical parameters and histological patterns of LN in the sample population.
| Median [IQR] | Normal range | |
|---|---|---|
| Cholesterol | 4.57 [3.85–6.80] | <5.0 mmol/L |
| Creatinine | 74 [60–117] | Male: 62–115 umol/L Female: 53–97 umol/L |
| eGFR | 101.4 [64–126.8] | 90–120 ml/min/1.73m2 |
| UPCR | 0.343 [0.161–0.580] | <0.015g/mmol |
| UWCC | 6000 [3000–11000] | 4500–11000 wbc/ml |
| URCC | 3000 [0–36000] | <1000 rbc/ml |
| Active urine sediment (WCC and RCC > 5000) | Active: 13, 33.3% Bland: 26, 66.7% | <5000 cells/ml |
| Albumin | 25 [16–31] | 35–52 g/L |
| Anti dsDNA | 0 [0–160] | 0 |
| ANA | 320 [0–640] | 0–80 |
| Anti Sm | 0 [0–128] | 0 U/ml |
| Anti RNP | 50 [0–154] | 0 U/ml |
| Anti Ro | 0 [0–118] | 0 U/ml |
| Anti La | 0 [0–12] | 0 U/ml |
| C3 | 0.895 [0.550–1.300] | 0.90–1.80 g/L |
| C4 | 0.160 [0.100–0.290] | 0.10–0.40 g/L |
| Systolic BP | 150 [129–179] | <140 mmHg |
| Proliferate (III, IV+/+V) | 31 |
| Non-proliferatve (1, II, V) | 16 |
| WHO 1 | 1 |
| WHO II | 7 |
| WHO III | 12 |
| WHO IV | 10 |
| WHO V | 8 |
| WHO lll + V | 4 |
| WHO IV + V | 5 |
Proliferative lesions were more commonly diagnosed in this cohort (n = 31; 66%), with 12 cases (25.5%) of WHO/ISN pure class III, 10 cases (21.3%) of pure class IV, and 9 cases (19.1%) of mixed class III+V and IV+V (Table 1.1).
In the male cohort, most (92%) were diagnosed with proliferative lesions (p = 0.070) (Table 1.2). Kidney dysfunction was more prevalent in patients subsequently diagnosed with proliferative lesions, as evidenced by higher creatinine (p = 0.007) and lower eGFR (p = 0.035). Urine white cell count was higher in those with proliferative lesions (p = 0.023) (Table 1.2).
Comparison of Demographic and Clinical Parameters between Proliferative and Non-proliferative Lupus Nephritis.
| Proliferative | Non-proliferative | P | |||
|---|---|---|---|---|---|
| N, % | Median [IQR] | N, % | Median [IQR] | ||
| Gender a | 31 (66.0%) | Female: 20, 64.5% Male: 11, 35.4% | 14 (29.8%) | Female: 13, 92.9% Male: 1, 7.1% | 0.070 ** |
| Black / Non-black b | 29 (61.7%) | Black: 18, 62.1% Non-black: 11, 37.9% | 15 (31.9%) | Black: 14, 93.3% Non-black 1, 6.7% | 0.035 ** |
| Creatinine | 29 (61.7%) | 77(68–145] | 14 (29.8%) | 61.5 (59–70] | 0.007 * |
| eGFR | 29 (61.7%) | 92.9 [143–117] | 14 (29.8%) | 119.45 (100–134] | 0.035 * |
| UWCC | 28 (59.6%) | 7500 [5000–36500] | 13 (27.7%) | 2000 [1000–9000] | 0.023 * |
Gender and ethnicity data could not be retrospectively confirmed for 2 and 3 cases, respectively
Mann Whitey U
Fisher Exact
General discriminant analysis modeling found an eGFR below 90 mL/min/1.73 m2 to be associated with an increased probability of diagnosing a proliferative lesion. In a subsequent multivariate logistic regression analysis, eGFR below 90mL/min/1.73m2 was independently associated with increased odds of the diagnosis of proliferative LN (OR = 5.6; CI 1.06-29.59; p = 0.043). Male sex was associated with a trend towards increased odds of this diagnosis (OR = 8.31; CI 0.89-78.15; p = 0.064). Ethnicity was not included in the multivariate analysis due to the low numbers of non-Black patients with proliferative lesions.
Nine (19.1%) patients achieved partial remission, 10 (21.3%) had complete remission, and eight (17%) experienced a LN relapse after remission induction. Baseline parameters and demographic factors did not affect time to remission induction. The relapse rate was non-significantly higher in the non-proliferative group (n = 3; 23.1%) compared to the proliferative group (n = 5; 19.2%; p = 0.544). The median time to relapse in cases of non-proliferative LN was 28.9 months (IQR 18.63–39.07 months) compared to 59.3 months in proliferative LN (IQR 50.43-69.93 months) (p = 0.200).
Diagnosing a proliferative lesion was associated with a trend toward poorer kidney function survival compared to non-proliferative lesions (p = 0.057) (figure 1.1).
Stepwise multifactorial linear regression analysis found higher eGFR at baseline independently associated with preserved eGFR (p = 0.003) and creatinine (p = 0.009) at follow-up. Baseline urine WCC inversely correlated with eGFR and directly correlated with serum creatinine at follow-up (p = 0.041 and p = 0.001, respectively). The presence of a proliferative lesion was negatively associated with eGFR at follow-up (p = 0.002).
Discussion
The present study found baseline kidney dysfunction and leukocyturia to be important factors predictive of proliferative LN, which was also associated with poorer kidney function at follow-up. Additionally, males had higher odds of having proliferative LN, possibly indicating another group at risk of this aggressive lesion.
The overall prevalence of LN was most significant amongst females (73.3%), with African ethnicity being the predominant race affected. This is broadly consistent with current literature,(14–18) possibly reflecting the effects of underlying genetic predisposition to severe renal lesions such as Apolipoprotein L1 (APOL1) risk alleles.(19) Lupus nephritis predominantly affects younger patients.(9,14,15) The current study was similar in that the median age of diagnosis for proliferative LN was 27 (IQR: 24-37) years and 25 (IQR: 23-36) years for non-proliferative LN.
This cohort's proliferative lesions accounted for two-thirds of all biopsy diagnoses, most of which were WHO class III LN. This is consistent with both the local and international literature.(9,20) Our study found proliferative LN tended toward poorer kidney survival, with low baseline eGFR independently associated with this lesion. However, the wide confidence interval observed suggests a relatively weak association related to the small study sample.
Our finding that a high proportion of males present with proliferative LN is consistent with a general trend toward more severe disease in men.(9,10,18,21) Patients of Black ethnicity constituted most of the proliferative LN group, although an association was not demonstratable in this series. African ethnicity has been associated with aggressive LN histopathology in many other studies,(7,13,14) and failure to show similar trends in this study likely reflects sample bias. Baseline eGFR <90ml/min/1.73m2 was independently associated with proliferative LN (OR 5.6; CI1.06-29.59; p=0.043), consistent with other series.(6,12,16,18)
Previous reports from South Africa have noted a trend towards kidney dysfunction at presentation in patients with both proliferative and non-proliferative LN diagnoses.(9) In our study, kidney dysfunction was comparatively rare in both groups, suggesting differences in biopsy practices between units. The use of proteinuria as a primary indication for biopsy may have been selected to include patients early in their disease course with relatively preserved kidney function. Emphasis on proteinuria as a biopsy indication is partly evidenced by the median baseline urine PCR in this series of 0.343 g/mmol creatinine. Other studies have similarly reported nephrotic range proteinuria as a common feature of LN.(14,16,20)
Only 33% of this cohort had active urinary sediment, as evidenced by leukocyturia and haematuria, which may indicate glomerular inflammation.(19) Similar findings have been reported in a previous local LN study,(17) possibly reflecting reliance on proteinuria as a biopsy indication. Although leukocyturia was more significant in cases of proliferative LN in the present study, logistic regression suggested a lack of predictive ability for this marker in identifying such cases.
Hypocomplementemia has long been associated with active lupus, LN, and proliferative lesions.(7,21) In this study, though median C3 was decreased, no association with proliferative LN was found. Interestingly, previous local reports also failed to demonstrate a significant difference in C3 between membranous LN and proliferative LN.(4)
This study found that a lower baseline eGFR, elevated urine WCC (an indicator of an inflammatory response [19]), and proliferative lesions were independently associated with kidney dysfunction at the most recent follow-up. This is in keeping with several studies that also found proliferative LN predictive of worse outcomes.(7,9,14,16) The possible reasons for unfavourable outcomes in proliferative LN are multifactorial and include increased proteinuria, which in itself is tubulotoxic;(9) association with arterial hypertension;(11) the increased prevalence of crescents on kidney biopsy;(20) higher baseline serum creatinine indicating early kidney damage and poorer response to induction therapy.(12,16) The induction therapy included pulsed steroids and cytotoxic therapy with cyclophosphamide or mycophenolate mofetil (for class IV and III, respectively). Additionally, reflecting on the predominance of Black African patients within the proliferative LN group is the impact of genetic contributors, including APOL 1, on poorer renal outcomes.(19)
Surprisingly, we did not find a significant difference in the relapse rate across the different histological classes. The small number of relapses (n = 8) could reflect improving outcomes in class III/IV from modern cytotoxic regimens.(22) A similar lack of difference in relapse outcomes between histological lesions has been reported in other series.(4,12)
Limitations
The study had a small sample size related to this being a single center study, incomplete patient records, no established biopsy registry at HJH, inability to obtain retrospective pathology records predating 2012 due to changing of electronic data platform, lack of follow-up records, and other exclusion criteria such as HIV. Additionally, the same pathologist did not report histopathology results, and some information was lacking, including activity and chronicity indices. The WHO/ISN histological 2003 criteria excluded both the vascular and tubulointerstitial compartments, respectively: exclusion of this underestimates the severity of disease in LN. Furthermore, based on the criteria used, leukocyturia contributing to tubulointerstitial nephritis would be underrepresented in this cohort. The predominant African demographic also limits this study due to the population served by this public health facility.
Conclusion
LN is an important manifestation of SLE, with a higher prevalence in the younger African population, who seem to experience a poorer prognosis. Establishing predictors of severe disease is essential in facilitating early diagnosis and intervention. The present study found an association between low baseline eGFR and baseline leukocyturia with kidney dysfunction at follow-up. Additionally, males were found to have higher odds of proliferative LN, with proliferative LN associated with poorer kidney survival. Using these parameters in the clinical setting may add value in screening and monitoring such patients for aggressive disease.
Conflict of interest statement. None declared.