Treatment outcomes in refractory lupus nephritis: Data from an observational study

Objectives Despite current advances in treatment, refractory lupus nephritis (RLN) continues to pose a challenge. The present paper studies the clinical profile and treatment outcomes in patients with RLN. Methods This observational, bidirectional study enrolled consecutive lupus nephritis (LN) patients from August 2018 to January 2019, who either failed to improve within three months, did not achieve partial renal response (PR) at six months, or did not achieve complete renal response (CR) after two years of treatment. Patients were followed every three months; treatment details and outcomes [CR, PR, no renal response (NR)], doubling serum creatinine, and death were recorded. Group comparisons were made using ANOVA and chi-square test. Factors affecting renal response were studied using linear regression. Results Forty-five of forty-eight enrolled patients completed at least nine months of follow-up and were included in outcome analysis. The median (IQR) SLE duration was three years (2-6 years). The majority of patients (n = 25) had proliferative LN (ISN/RPS class III/IV), with nine patients having pure membranous LN (class V). The mean activity and chronicity indices were 8 and 0. Over a median (IQR) follow-up period of 15 (12-27) months, 28 had CR, 9 had PR, and 8 showed no response to a switch in an immunosuppressive (IS) agent. Repeat renal biopsy (n = 8) with a mean (±SD) biopsy interval of 2 (±1) years showed histological class transformation in more than half of the patients. There was no significant difference in treatment outcome and time to attain response based on individual IS agent or sequence of IS agents used. None of the variables (duration of SLE or nephritis, baseline SLEDAI, leukopenia, hypertension, elevated anti-dsDNA, low complements, serum albumin, 24-hour urinary protein, biopsy class) predicted renal response on univariate analysis. No patient had a doubling of serum creatinine or progression to end-stage renal disease. There were three deaths, all related to infection. Conclusion A change in immunosuppression produces response in most RLN patients while a fifth of them showed no response to therapy. No predictor of renal response was identified. Histologic class switch was frequent. Renal function did not decline over a year of follow-up.


Introduction
Lupus nephritis (LN) is present in about one-third of systemic lupus erythematosus (SLE) patients at diagnosis and up to two-thirds during the disease course [1][2][3] Progression to end-stage renal disease (ESRD) is a dreaded outcome. Despite the use of newer agents in the treatment, 20%-70% of patients with LN, do not respond to initial immunosuppressive (IS) therapy [4][5][6] Lupus 30 (11) Non-responders have poor survival, including renal survival at ten years compared to complete or partial responders. 7 In epidemiologic studies, poor prognostic factors reported for LN include hypertension, hypocomplementemia, anti-dsDNA antibodies, high serum creatinine and nephrotic syndrome at initial presentation, as well as failure to achieve clinical remission within the first year of treatment. 8 Higher NIH activity index (AI) and chronicity index (CI) scores on renal histology are associated with doubling serum creatinine, end-stage renal disease, or death. 9 Treatment non-compliance, class VI LN, thrombotic microangiopathy, antiphospholipid syndrome, or lupus podocytopathy are other causes of persistent renal dysfunction to be considered before labeling as refractory LN (RLN). 10 ACR defines an inadequate response to the first-line treatment in LN as disease worsening at three months or treatment failure by six months. 11 EULAR/EDTA labels patients who fail to improve within 3-4 months, or do not achieve partial response after 6-12 months, or complete response after two years of treatment, as having refractory disease. 12 The ACR definition leads to an earlier declaration of RLN since assessment is done at 3-6 months, while in EULAR-EDTA criteria, it could be done up to 12 months.
There is paucity of literature capturing outcomes after change in therapy in non-responders to first line therapy for lupus nephritis. Since it is not easy to conduct a randomized controlled trial in this group of patients, we believe that this study presenting such data in 48 patients followed prospectively with well documented details of therapy and outcome, will contribute to evidence which will guide therapy in initial non-responders. This study aims to assess the clinical profile and treatment outcomes in patients with LN resistant to first-line drugs. Documenting the response to a switch to an alternate immunosuppressive (IS) agent will shed light on whether refractoriness in LN is due to the patient or drug-related factors.

Methods
This was an observational, longitudinal, bidirectional study. Consecutive patients fulfilling the classification criteria of SLE based on Systemic lupus international collaborating clinic criteria (SLICC) 2012, 13 attending the Rheumatology Department at Nizam's Institute of Medical Sciences, Hyderabad, and satisfying the EULAR/EDTA definition of RLN, were enrolled in the study between August 2018 and January 2019. Patients with overlap connective tissue diseases, irregular follow-up, or non-compliance to therapy were excluded. Each patient was followed for at least nine months after enrollment.
Lupus nephritis was defined as per ACR criteria. 14 Complete renal response (CR) was defined as spot UPCR <50 mg/mmol (equivalent to proteinuria <0.5 g/24 h) and normal or near-normal (within 10% of normal GFR, if previously abnormal) GFR. Partial renal response (PR) was a � 50% reduction in proteinuria to sub-nephrotic levels and normal or near-normal GFR. Lack of improvement, i.e., no reduction in proteinuria or deterioration of GFR, was defined as no renal response (NR). RLN was defined according to EULAR/ERA-EDTA recommendations as NR at three months, failure to achieve either PR at six months or CR after two years of treatment induction. 12 Demographic, clinical, and laboratory parameters (complete blood count, anti-dsDNA antibody and complement levels, serum creatinine, urine routine examination, 24-hour urinary protein or spot UPCR, and renal biopsy findings) were noted. The following details of treatment received prior to enrolment were noted: use of intravenous pulse steroid before starting oral prednisolone, choice of induction immunosuppressive agent [cyclophosphamide (CYC) as per the Euro-Lupus Nephritis Trial regimen (ELNT-CYC), CYC as per the NIH protocol (NIH-CYC), mycophenolic acid (MPA), tacrolimus (Tac) or rituximab (RTX)], use of hydroxychloroquine and angiotensin-convertingenzyme (ACE) inhibitors.

Treatment protocol
Once RLN was considered, the treatment strategies employed included a switch to another first-line drug, for example, MPA if the patient had received CYC and vice versa. Alternative treatment options considered during a switch were tacrolimus, multitarget therapy (MPAþTac), or RTX. This decision was based on the treating physician's discretion in consultation with the patient. Mycophenolate mofetil or enteric coated-MPA dose was 2 g and 1.44 g respectively, adjusted to the maximum tolerable dose. CYC was given monthly in the dose range of 500-1000 mg/m 2 (NIH-CYC). Tacrolimus was given in a dose of 0.025-0.1 mg/kg with the blood levels titrated between 4-10 ng/ml in those showing inadequate response. Rituximab was given as two 1 g intravenous infusions, 15 days apart. All patients received a maintenance dose (<0.15 mg/kg/ d) of oral corticosteroids, hydroxychloroquine, calcium/vitamin D, and ACE inhibitors. During switching, no increase in oral steroid dose was done. As in the case of the first switch, subsequent switches were made if there was no improvement at 3 months, no partial response by 6 months or no complete response at 2 years.
Following were recorded during a flare and every three months: Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), response to treatment, and adverse events. The study complies with the Declaration of Helsinki and approved by the institutional ethics committee of Nizam's Institute of Medical Sciences, Hyderabad, India (EC/NIMS/2209/2018). Written informed consent was obtained from all the patients (or their legally authorized representative).

Statistical analysis
Baseline characteristics of patients were presented as mean (SD) or median (IQR), as appropriate. Comparisons among CR, PR, and NR groups were made using ANOVA and chi-square tests. Predictors of renal outcome were assessed using Cox regression analysis. Statistical analysis was carried out with SPSS software (version 21). P-value < 0.05 was considered to represent a statistically significant difference.

Results
Out of 48 patients enrolled for the study, 12 were those who had no response by three months, 29 had not achieved PR at six months, and 7 had not attained CR after two years of starting induction treatment. Two patients were lost to follow-up after six months, and one died within two months of enrolment. Fortyfive patients completed nine months of follow-up and were considered for further analysis. The baseline disease characteristics are summarized in Table 1.
Renal biopsy was available in 43 patients (Table 2). Five patients either refused consent or had inadequate tissue yield. In 8 patients, a repeat renal biopsy was available. The mean interval between biopsies was 2 (�1) years. A histological class transformation was seen in 6 patients (Table 3). In 3 patients, the renal biopsy was repeated for the third time. Renal biopsy could not be performed in the rest of the patients due to contraindications (thrombocytopenia, ascites, anasarca, orthopnoea, active infection) or unwillingness for the procedure. They were not different from those who were biopsied in terms of urinary and serological abnormalities.
All the patients included in the study were refractory to one or more or combinations of IS agents. The average number of IS agents used in each patient in the past was 3 (�1). At the time of enrollment, the disease was refractory to CYC in 38 patients (NIH-CYC in 33 and ELNT-CYC in 5); MPA in 17; Tac, RTX, and MPAþTac in one patient each. After enrollment, the  choice of IS agent(s) was guided by prior IS used, the severity of nephritis, involvement of other organ systems, comorbid conditions, and purchasing capacity of the patient. Thus MPA was used as the next IS agent in 20, MPAþTac in 9, NIH-CYC in 8, tacrolimus in 3, rituximab as monotherapy in 1, and in combination with tacrolimus and MPA in 2 patients each. In 18 patients, a second switch (MPAþTac in 9, rituximab as monotherapy in 3 and in combination with MPA in 2, tacrolimus alone in 2, NIH-CYC and MPA in 1 patient each) was needed due to persistent refractoriness to treatment. Seven patients required a third switch (MPAþTac in 2, RTXþMPA in 2 and, RTX monotherapy, CYCþTac, CYC-NIH in one each). One further non-responder was switched to a combination of MPAþTac. The most common sequence of switch among IS agents was CYC to MPA in 13 patients ( Figure 1) and there was no significant difference in treatment outcome based on the sequence of IS agents followed.
At the end of a median follow-up of 15 (12-27) months, 28 of 45 had a CR, 9 had a PR, and 8 showed no response to therapy. Univariate analysis of the following factors: age, sex, duration of SLE or nephritis, number of organ systems involved, baseline SLEDAI, leukopenia, hypertension, presence of anti-dsDNA, low C3 and / or C4, serum albumin, 24-hour proteinuria, spot UPCR, biopsy class or IS therapy, revealed that none predicted renal response. There was no significant difference in treatment outcome among individual IS agents (Figure 2; Table 4). No patient had either doubling of baseline creatinine or ESRD. There were three deaths.
Four patients had severe adverse effects. One patient developed tubercular monoarthritis of the right knee at 12 months. Another developed CYC induced severe leukopenia (TLC ¼ 500/mm 3 ). The same patient later developed an epidural abscess and succumbed to sepsis with multi-organ dysfunction. Two other deaths due to sepsis occurred, one within three months of starting RTX and another in a non-responder to RTX and MPA. None of these patients had rising creatinine or ESRD before death.

Discussion
This study shows that in refractory lupus nephritis (defined, as suggested by the EULAR/EDTA group), a change in immunosuppressive agent leads to an overall response (CRþPR) in four-fifth of the patients. The  1  III  IV  IV  1  5  Yes  2  V  V  V  4  3  No  3  II  IVþV  IV  2  4  Yes  4  V  V  -1  -No  5  III  IIIþV  -2  -Yes  6  III  IV  -2  -Yes  7  IV  IVþV  -2  -Yes  8 IV V -1 -Yes Figure 1. Most common sequence of switches made among IS agents. combination of tacrolimus and mycophenolate appears to be most effective in producing a response. Renal failure is rare during short-term follow-up and death is related to infection. In our study, a combination of MPA and tacrolimus, also termed multitarget therapy (MTT), produced a high global response rate. Response rates (complete and partial) ranging from 57-95% have been reported in previous studies (RCT 15 or observational studies [16][17][18] ) which included 64 patients receiving this combination.
The response to MPA as a single agent, in almost half the patients in our study was much lower than the 87% response seen in Hispanic patients with relapsing or refractory LN. 19 While their population had a higher incidence of proliferative nephritis, it did include patients with relapsed disease which may explain the lower response rate in our study. However racial differences in treatment response to MPA have also been suggested in literature. 20 With NIH-CYC, half the patients had CR. One patient who was refractory to multiple IS agents, including RTX had a complete response on receiving a combination of NIH-CYC and tacrolimus. While EULAR/EDTA recommends either MPA or CYC as first-line therapy in LN and a switch to either agent if the first fails, there is no data available on cyclophosphamide as the next remission inducing agent, if the primary regimen fails. 12 In the present study, regimes that included rituximab produced a response rate of 54%. Many studies which have used RTX, alone or in combination, in RLN have been included in a systematic review 21 and a meta-analysis, 22 reporting global response rates of 74% and 70-78%, respectively. There have been limiting factors with these studies that include marked heterogeneity in the definition of refractoriness to therapy and variation in the RTX doses and regimens used.
Tacrolimus as monotherapy showed no response in any of our patients at a mean dose of 4 mg at the last follow-up. Two studies on tacrolimus in 35 LN patients not responding to sufficient cyclophosphamide report very high response rates. 23,24 These differences may be explained by the rapid escalation of dose on follow-up in those studies.
Within the median duration of follow-up of 15 months, no patient in our study had a doubling of serum creatinine or progression to ESRD. Deaths in RLN patients are often due to infection 25 but may also be due to disease activity. 19 All deaths in our study followed severe infection.
In a small subset where repeat renal biopsies were available, frequent class transformation was noted. There was accrual of membranous lupus nephritis in four patients with the consequent increase in proteinuria pushing these patients to fulfil the definition of RLN. A change from Class II or III to IV could mean non-response to the therapy used. In studies reporting repeat renal biopsies in patients with ongoing or relapsed LN, it is shown that patients with proliferative LN continue to show the same findings with a transformation from class 3 to 4, while those with non-proliferative LN (Class 2 or Class 5) may accrue proliferative changes. 26,27 Though our sample size is not representative, switching from membranous to proliferative LN was not noted in our study.
The present study is, to our knowledge, the first prospective study from India and among the largest reported in the literature assessing treatment outcomes in patients of RLN in a real-world scenario over one year. Numerically best response was seen using MPA and tacrolimus combination while switching IS agents Lupus 30 (11) in any sequence proved effective in four-fifth of the patients. Limitations of sample size and problems in using randomization in this situation will continue to thwart efforts to understand the best way forward in patients with refractory lupus nephritis.

Conclusion
In the present study on patients with lupus nephritis who worsened within three months of induction therapy or did not achieve at least a partial response at six months or complete response at 24 months, slightly less than two-thirds of the patients responded completely to a change in immunosuppression while a significant minority still showed no response to therapy. During the follow-up period of more than a year, none of these patients had a worsening renal function. However, three deaths were noted, all in the rituximab recipients, mainly due to infection.

Key messages
• A complete response following a switch in immunosuppression occurs in a significant proportion of patients with RLN. • No response after switching the IS agent is seen in a significant minority. • Histologic class switch often occurs over time.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

Contributorship
SG, PKD and LR analyzed and interpreted the patient data and were major contributors in writing the manuscript. MU performed the renal histological examination. All authors read and approved the final manuscript.

Data availability statement
The data underlying this article will be shared on reasonable request to the corresponding author.