Objectives Short-term predictive endpoints of chronic kidney disease (CKD) are needed in lupus nephritis (LN). We tested response to therapy at 1 year.
Methods We considered patients with LN who underwent renal biopsy followed by induction therapy between January 1970 and December 2016. LN was assessed using the International Society of Nephrology/Renal Pathology Society (2003) criteria and the National Institute of Health (NIH) activity and chronicity index. The renal outcome was CKD. Response was defined according to EULAR/European League Against Rheumatism/European Renal Association-European Dialysis and Transplant Association (EULAR/ERA-EDTA) recommendations: complete: proteinuria <0.5 g/24 hours, (near) normal estimated glomerular filtration rate (eGFR); partial: ≥50% proteinuria reduction to subnephrotic levels, (near) normal eGFR; and no response: all the other cases. Logistic regression analysis was employed for 12-month response and Cox regression for CKD prediction.
Results We studied 381 patients (90.5% Caucasians). After 12-month therapy, 58%, 26% and 16% of patients achieved complete, partial and no response, respectively, according to EULAR/ERA-EDTA. During a median follow-up of 10.7 (IQR: 4.97–18.80) years, 53 patients developed CKD. At 15 years, CKD-free survival rate was 95.2%, 87.6% and 55.4% in patients with complete, partial and no response at 12 months, respectively (p<0.0001). CKD-free survival rates did not differ between complete and partial responders (p=0.067). Serum creatinine (HR: 1.485, 95% CI 1.276 to 1.625), eGFR (HR 0.967, 95% CI 0.957 to 0.977) and proteinuria at 12 months (HR 1.234, 95% CI 1.111 to 1.379) were associated with CKD, yet no reliable cut-offs were identified on the receiver operating characteristic curve. In multivariable analysis, no EULAR/ERA-EDTA response at 12 months (HR 5.165, 95% CI 2.770 to 7.628), low C4 (HR 1.053, 95% CI 1.019 to 1.089) and persistent arterial hypertension (HR 3.154, 95% CI 1.500 to 4.547) independently predicted CKD.
Conclusions Lack of EULAR/ERA-EDTA response at 12 months predicts CKD.
- arterial hypertension
- lupus nephritis
- systemic lupus erythematosus
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What is already known about this subject?
Preservation of long-term renal survival is the main goal of lupus nephritis treatment.
So far, tested composite and single short-term variables did not univocally function as long-term prognostic markers.
What does this study add?
This is the first study to evaluate the capability of an officially agreed definition of renal response to impact on long-term renal outcome in a large real-life setting.
How might this impact on clinical practice or future developments?
This study adds a piece of knowledge to the feasible and prognostically relevant targets that should be adopted in clinical practice in order to maximise the chances of renal survival, that is, early EULAR/ERA-EDTA (European League Against Rheumatism/European Renal Association-European Dialysis and Transplant Association) response and well-controlled blood pressure.
Patient and renal survival in lupus nephritis (LN) have progressively improved over the last 50 years.1–3 Nevertheless, renal damage continues to be the most important predictor of mortality within the damage index,4 and 10%–30% of patients with renal damage develop end-stage renal disease (ESRD) within 15 years from diagnosis.5 Preservation of renal function is a paramount treatment goal in LN. New drugs and therapeutic approaches are generally tested in clinical trials, whereby short-term predictive endpoints would be required to replace the hard kidney endpoints, namely chronic kidney disease (CKD) and death.
Short-term endpoints are also of utmost importance in clinical practice to help clinicians in guiding therapy particularly after the remission-induction phase. There is a general agreement that response to therapy can be one potential predictive endpoint for renal survival.6–11 However, no definition of renal response has unanimously been agreed and no time interval for assessment of renal response has clearly been established.8 12–19
The EULAR/ERA-EDTA (European League Against Rheumatism/European Renal Association-European Dialysis and Transplant Association) recommendations, developed by the review of available literature and expert opinion, have coined a shared definition of renal response to therapy (as categorised in the section Definition of renal variables).20 21
In this view, we tested the EULAR/ERA-EDTA definition of response20 21 in a large cohort of Italian patients with a long follow-up, with the aim of examining (1) the rate of renal response at 1 year from the initiation of therapy; (2) the ability of the EULAR/ERA-EDTA response to predict the long-term development of CKD, compared with the ability of some individual variables (renal function, proteinuria, haematuria); and (3) the predictors of no response to therapy among the clinical and histological features at the time of LN diagnosis.
Patients and methods
Patients older than 16 years followed at three Italian referral centres (Renal Divisions of Fondazione Ca’ Granda Ospedale Maggiore Milano and San Carlo Hospital Milano, and Rheumatology Unit of Padova University) and who had biopsy-proven LN with a follow-up longer than 1 year from the initiation of induction treatment were included in this retrospective analysis of prospectively collected data.
Inclusion criteria were (1) patients with systemic lupus erythematosus classified according to the American College of Rheumatology criteria22; (2) biopsy-proven LN performed between January 1970 and December 2016; and (3) at least three clinical and laboratory assessments during the first year from the initiation of induction therapy. Exclusion criterion was the need for renal replacement therapy at the time of renal biopsy.
The patients included in the study are part of a cohort of 499 patients described in a previous paper. Of these 499 patients, 118 were excluded due to incomplete data at the first year of the study.3 A comparison of clinical and histological characteristics of the included and excluded patients is reported in online supplementary table 1.
All patients signed an informed consent for the scientific use of their data, which were anonymised.
We considered as baseline of the study the initiation of induction therapy after the histological diagnosis of LN, which was classified according to the 2003 International Society of Nephrology/Renal Pathology Society criteria23 and assessed in terms of chronicity and activity indexes according to Austin et al.24 After renal biopsy all patients were followed by a dedicated team at each unit. They were evaluated 1 month after the diagnosis, then every 2–3 months, at 1 year, and every 3–6 months thereafter.
We set up a shared electronic database across the three participating centres. The following variables were recorded: baseline demographics, clinical and laboratory features (at baseline, at each clinical evaluation during the first year, at 12±2 months and at last observation), and induction and maintenance therapy. We considered the development of CKD as renal outcome.
Definition of renal variables
Normal renal function: serum creatinine ≤1 mg/dL and estimated glomerular filtration rate (eGFR) >60 mL/min/1.73 m2.
Acute renal dysfunction: acute nephritic syndrome (serum creatinine >1 mg/dL and eGFR ≤60 mL/min/1.73/m2, severe haematuria (urinary red blood cells >20/high power field) and/or erythrocyte casts, and proteinuria >0.5 g/day) or rapidly progressive renal insufficiency (rapid deterioration of renal function leading to CKD stage 3–5 within a few weeks, arterial hypertension and severe haematuria).
CKD: serum creatinine >1.0 mg/dL and eGFR <60 mL/min/1.73 m2 and inactive urinary sediment, confirmed by at least three determinations for at least 3 months (adapted from Levey et al 25).
Proteinuria: measured by benzethonium chloride on the urine collected over 24 hours, expressed as g/24 hours.
eGFR: evaluated with Cockcroft and Gault formula.
Arterial hypertension: the mean of three consecutive measurements of systolic blood pressure >140 mm Hg and/or diastolic blood pressure >90 mm Hg in a sitting position.
EULAR/ERA-EDTA response to therapy at 1 year20 was categorised retrospectively as follows:
Complete renal response: proteinuria <0.5 g/24 hours, normal or near-normal eGFR (within 10% of normal eGFR if previously abnormal).
Partial renal response: ≥50% reduction in proteinuria to subnephrotic levels (<3.5 g/24 hours), and normal or near-normal eGFR.
No renal response: all the other cases.
Descriptive statistics are reported as median and IQR for continuous variables, and as counts and percentages for dichotomous variables. The difference of the continuous variables between groups was tested with non-parametric Wilcoxon test for independent samples. Paired data tests were used to compare the values of clinical parameters at different timepoints. χ2 test was used to test the associations of qualitative or dichotomised variables among groups of patients.
Inferential analysis was performed using the Cox proportional hazard model. Both univariable and multivariable analyses were performed.
In the multivariable analysis, we tested all the variables with potential clinical significance.26
Schoenfeld residuals were computed to check the proportional hazards assumption (testing independence between residuals and time).
Kaplan-Meier estimate was used to draw survival curves, and log-rank test was used to test their difference. We used receiver operating characteristics (ROC) analysis to assess the ability of different variables to identify subjects who develop CKD. Threshold values have been chosen as those maximising the sum of sensitivity and specificity. Patients lost to follow-up were censored at last observation. The R statistical package was used for all analyses.27
Baseline patient features and outcome
The study included 381 patients from the three participating centres. The clinical and histological characteristics of patients and therapy at renal biopsy are reported in table 1. Of the patients, 86% were female, 345 (90.5%) were Caucasians and 36 (9.5%) belonged to different ethnicities. The median age at baseline was 30.4 years (IQR 23.9–40.4). Two hundred and ninety-one (76.4%) had normal renal function at baseline, while 90 (23.6%) had acute renal dysfunction (75 acute nephritic syndrome and 15 rapidly progressive renal insufficiency). No patients had CKD at diagnosis. Variations in the immunosuppressive therapy taken by the patients during follow-up are reported in online supplementary table 3.
No patients developed CKD during the first year of observation. At last observation, after a median follow-up of 10.7 years (IQR 4.97–18.80), 359 patients were alive, 44 (11.6%) developed CKD (of them 28 developed ESRD), and 315 had normal renal function. The remaining 22 (5.8%) patients died (of them 9 patients had CKD). Altogether 53 (13.9%) patients developed CKD during the follow-up. CKD occurred in 30 out of 90 (33.3%) patients with acute renal dysfunction at baseline vs 23 out of 291 (7.9%) with normal renal function. The baseline features of patients who developed and of those who did not develop CKD are reported in table 1.
Rates and prediction of response to therapy at 1 year
One year after the initiation of induction therapy, 61 (16%) patients were non-responders, 100 (26%) achieved partial renal response and 220 (58%) complete renal response according to EULAR/ERA-EDTA definition. As shown in figure 1, time to achieve complete response among complete responders by month 12 was weakly correlated with the level of baseline proteinuria (Spearman’s r=0.18, p=0.006). Baseline proteinuria and frequency of nephrotic syndrome were significantly lower in 72 patients who achieved complete response by 6 months in comparison with the 148 who required from 6 to 12 months to achieve the same target (median baseline proteinuria, 2.0 g/day, IQR 1.6–3.3, vs 3.2 g/day, IQR 1.9–5.5, p<0.01; frequency of nephrotic syndrome 22.2% vs 46.6%, p<0.0001) (online supplementary table 4).
The rate of complete response increased to 63% at 2 years and to 69% at 3 years. Altogether, 308 out of 381 patients (81%) of our cohort achieved complete renal response at some timepoint during the follow-up.
At last observation, 11 of 220 (5%) patients with complete renal response, 18 of 100 with partial renal response (18%) and 24 of 61 (39.3%) non-responders at 1 year developed CKD (p<0.0001).
At 5, 10 and 15 years, CKD-free survival was 99.5%, 96.5% and 95.2% in patients with complete renal response, 98.8%, 89.7% and 87.6% in those with partial renal response, and 81.3%, 65.3% and 55.4% in non-responders at 1 year (p<0.0001) (figure 2). The CKD-free survival curve of non-responders was significantly worse than that of complete responders and partial responders (p<0.0001, for both). CKD free survival rates did not differ between complete and partial responders (p=0.067).
Baseline predictors of no response at 1 year
Out of the 381 patients, 61 did not respond at 1 year. In multivariable analysis, arterial hypertension (OR. 2.567, 95% CI 1.199 to 5.884, p=0.013) and high chronicity index at the time of renal biopsy (OR: 1.221 for any unit increase, 95% CI 1.046 to 1.419, p=0.018) were independent predictors of no response at 1 year (table 2).
Clinical predictors at 1 year of CKD
Out of 381 patients, 53 developed CKD. The clinical characteristics of patients at 12 months from baseline were considered in this analysis. No response to therapy according to the EULAR/ERA-EDTA definition at 1 year predicted CKD (HR 5.604, p<0.0001). Among the clinical renal variables, serum creatinine (HR 1.485, p<0.0001), eGFR (HR 0.967, p<0.0001) and proteinuria (HR 1.234, p<0.0001) were associated with CKD. Other factors associated with CKD development were arterial hypertension (HR 3.522, p<0.0012) and low C4 (HR 1.049, p=0.004) at 1 year (table 3).
The number of urinary red blood cells/high power field and the differences between baseline and 1 year values of eGFR, proteinuria and haematuria did not predict CKD development.
In multivariable analysis, no response according to EULAR/ERA-EDTA definition (p<0.0001), arterial hypertension (p=0.013) and low C4 (p=0.0014) were independent predictors of CKD (table 3).
We performed ROC analysis in order to find a cut-off value for any clinical variable in our cohort, and we also used cut-off values found in the literature,17 but none of the cut-off values was able to accurately discriminate patients who did or did not develop CKD. The ROC analysis of our data showed an area under the curve (AUC) for serum creatinine of 0.759 (CI 0.680 to 0.837) and a cut-off of 1.195 mg/dL, while for proteinuria the AUC was 0.708 (CI 0.628 to 0.788) and the cut-off was 1.195 g/day. Sensitivity, specificity, and positive and negative predictive values are reported in online supplementary table 5.
In univariable and multivariable analyses, the predictors of ESRD were the same as those of CKD (online supplementary table 6).
In this large Italian LN cohort, we evaluated the 12-month response to induction therapy, as defined by EULAR/ERA-EDTA recommendations.20 21 We investigated which factors affect response to therapy and analysed the capability of renal response at 1 year to predict renal survival. We demonstrated that (1) complete and partial responses after 1 year of therapy were associated with a good long-term renal survival, while failure to achieve response predicts a worse renal survival; (2) high chronicity index at renal biopsy and arterial hypertension were independent predictors of no response to therapy; and (3) baseline urinary sediment did not predict renal survival, while serum creatinine, eGFR and proteinuria were associated with CKD. These observations are of clinical utility in a treat-to-target perspective, as consistent targets with defined prognostic value in the long term are still substantially lacking in LN.
Complete response was achieved by 56% of our patients at 1 year, 63% at 2 years, 69% at 3 years and 81% throughout the follow-up. Our results, in line with those of other studies,10 underline that the achievement of complete response after therapy is time-dependent, being influenced by several factors such as therapy, ethnicity, number of patients included in the study and the definition of response itself.12 16 28–34 In fact, the proportion of complete responders largely varied from study to study (from 10% to 95%) according to the different definitions used.12 16 28–34
The amount of baseline proteinuria was also crucial in predicting the time to response. Touma et al 35 found that patients with a higher levels of baseline proteinuria required longer time to improve, and Gomez Mendez et al,36 using data collected from two LN randomised trials, showed that nephrotic syndrome at baseline decreased the likelihood of renal response at 1 year. Our data were consistent with these observations as we showed that time to achieve complete response among complete responders by month 12 was correlated with both the amount of baseline proteinuria and the presence of nephrotic syndrome.
Few studies have evaluated the predictors of response to therapy.12 14 15 28 In our cohort, high chronicity index at renal biopsy and arterial hypertension at baseline were independent predictors of no response at 1 year, in keeping with previous data.8 28
Notably, patients who were non-responders to therapy at 1 year had the highest probability of developing CKD in the long term, while both complete and partial responses at 1 year were associated with good renal survival, consistent with other studies,13 28 thereby suggesting that any degree of response is always advisable.
The last issue addressed in this study was whether it is better to employ a preformed composite endpoint (such as response at 1 year) to predict long-term renal outcome, or to use a single variable. Some studies reported that proteinuria at 12 months was the single best predictor of long-term renal outcome in LN.17–19 Dall’Era et al 17 found that a proteinuria <0.8 g/day at 12 months predicted good renal function at 7 years with 81% sensitivity and 78% specificity in 76 patients of the Euro-Lupus Nephritis Trial. Adding serum creatinine and urinary sediment to this model did not improve the predictive power of proteinuria alone. The same results were obtained in 90 patients of another Lupus Nephritis Trial.18 In 101 patients from the Toronto cohort, proteinuria <0.6 g/day and serum creatinine <0.83 mg/dL at 1 year performed independently and moderately well in predicting good renal outcome.37 However, when the analysis was limited to patients with baseline proteinuria >1 g/day, the optimal cut-off raised to <0.95 g/day. The accuracy of any single variable in predicting renal outcome has not been confirmed by other studies.13 38 In 176 patients of the Hopkins Lupus Cohort, two composite remission scores that included both serum creatinine and proteinuria were tested. Both these scores seemed to perform better in predicting renal survival compared with proteinuria alone.13 These results were recently confirmed in 550 patients with LN where 12-month proteinuria and 12-month serum creatinine were able to predict the risk of developing CKD at 3 years.38
Similarly, in our cohort we were unable to identify a cut-off value of proteinuria, eGFR and serum creatinine which could accurately discriminate patients who did from those who did not develop CKD. Rather, we found that the composite variable ‘no EULAR/ERA-EDTA response’ at 1 year independently predicted CKD. It should be noted that arterial hypertension at 1 year was still an independent predictor of CKD, thus indicating that a better control of blood pressure may be helpful in preventing CKD.
This study has some strengths and limitations. Among the latter, it should be mentioned that since data were from a real-world LN cohort, treatment, time to data collection and duration of follow-up were not standardised; hence, a relevant proportion of patients were excluded due to incomplete records at 12 months. Similarly, other features at 12 months such as anti-DNA antibodies, serum albumin and haemoglobin were not evaluated due to inconsistency across laboratory techniques used for the detection of anti-DNA antibodies or due to incomplete records. The great majority of patients included in the study were Caucasians and the results cannot be extended to other ethnicities. Moreover, the study may have been underpowered to discriminate between complete and partial renal response to prevent CKD. Nevertheless, the number of patients included in the study and the duration of follow-up were robust enough to suggest that non-responders at 1 year according to EULAR/ERA-EDTA recommendations have a high probability of developing CKD.
In summary, according to our data, EULAR/ERA-EDTA response can be considered a good straightforward short-term endpoint for CKD prediction and a promising target to treat patients with LN. During treatment to achieve renal remission, careful attention should be paid to uncontrolled arterial hypertension.
We thank Professor Claudio Ponticelli for his enlightening suggestions.
RAS and AD are joint senior authors.
Handling editor Josef S Smolen
RAS and AD contributed equally.
Presented at The manuscript was based on a work presented at the EULAR conference in 2019 and was published as a conference abstract (number THU0685).
Contributors GM conceived and planned the study and took the lead in writing the manuscript. MG significantly contributed to drafting and revision of the paper. RAS and AD contributed substantially to data acquisition and interpretation, and to manuscript drafting. SQ and LS performed the statistical analysis. FT, GF, FS, FR, MG, BT, FA and VB helped in following patients and collected the data. PM, AV, AD and RAS critically revised the manuscript. All authors approved the final version.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The study was approved by the Ethics Committee of Fondazione Ca’ Granda IRCCS Ospedale Maggiore Policlinico di Milano, Italy (protocol number 505_2019bis) and of the other participating centres.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available for formal research purposes only upon request to the corresponding author.