Low- Versus High-dose Cyclophosphamide in Class III/IV Lupus Nephritis: A Retrospective Study from South Asia

Introduction

Lupus nephritis (LN) is one of the poorest prognostic factors in systemic lupus erythematosus (SLE).[1] Ethnic differences in prevalence and outcomes suggest that multiple biological and sociological factors contribute to the disease course. Asian SLE patients are known to have a higher incidence of renal involvement and more severe disease.[2] A combination of steroids and cyclophosphamide (CYC) is the gold-standard in treatment of LN with high disease activity.[3]

The Euro-Lupus Nephritis trial (ELNT), which was a European-based multicenter, prospective, randomized study, compared the then standard treatment of high-dose (HD) intravenous (IV) CYC and low-dose (LD) IV CYC as remission-inducing therapy in proliferative LN.[4,5] The study showed no difference in the rates of renal remission, treatment failure, or severe flares and a nonsignificant trend toward a reduction in adverse effects.[5] However, the patients included in the ELNT study were predominantly white Caucasians. The comparative efficacy of LD CYC in Asian populations has not been well investigated.[6] In particular, data from South Asian populations is scant. A recent systematic review from China suggests that both HD and LD regimens have similar efficacy outcomes.[7] However, data from a retrospective study done among Puerto Ricans and a prospective study among Indians showed poorer renal outcomes in patients treated with LD CYC.[8,11] While alternate treatment options such as mycophenolate and calcineurin inhibitors are becoming more popular, CYC is still commonly used, partly due to its lower cost.

The objective of our study was to compare the efficacy of LD- and HD-CYC in our population. The University Medical Unit, National Hospital of Sri Lanka, has been managing patients with LN since the 1980s. This cohort of patients includes those who received HD-CYC in accordance with the original National Institute of Health (NIH) trials and the evidence of Boumpas et al., [3,4] as well as those treated with LD CYC from 2013 onward, as in the ELNT.

Materials and Methods

The study was a single-center, retrospective, comparative study conducted in the University Medical Unit of the National Hospital of Sri Lanka. Patient data was collected between May 2017 and March 2020. The objective of the study was to compare the outcomes of LD-CYC and HD-CYC when used as induction therapy for class III and IV LN.

Results

A total of 74 patients with class III and IV LN were considered for the study. Seven were excluded due to inadequate data or failure to meet the inclusion and exclusion criteria. Thirty-four had received HD-CYC and 33 had LD-CYC. The baseline characteristics of the two groups have been presented in in Table 1. The HD-CYC group had received treatment between 2000 and 2013, and the LD regimen comprised patients initiated on treatment between 2013 and 2019.

Clinical characteristics on presentation were similar between the two groups. Only one of the LD- group had class III LN compared to 7/34 having in the HD- group. Activity index on renal biopsy was available for 43 of 66 patients and the overall median for the sample was 9/24. Median activity index was higher in the LD-CYC group. Within treatment groups, 18/23 treated with the LD-CYC regimen as opposed to 7/20 treated with the HD-CYC had an activity index of greater than 9 and this difference was statistically different (P < 0.005). This implies more histologically active lupus in the LD group. The details regarding treatment received and outcomes are summarized in Table 2.

There was no significant difference between the numbers of patients affected by treatment failure (21% in the HD-CYC group vs. 30% in the LD-CYC group). Though fewer patients appeared to achieve remission (complete or partial) in the LD-CYC group (73% vs. 82%), this difference was not statistically different. To further improve comparability with the primary outcome of the LD group, we looked at a composite end point of absence of primary response, occurrence of severe renal flares within the first 2 years, and doubling of S.Cr over the lowest value achieved. There was no difference between the LD-CYC group (10/33) and the HD-CYC group (8/34) (P = 0.3). None of the clinical criteria or histological criteria correlated with treatment failure [Table 3]. Within the subgroup of patients with an activity index >9, 1/7 in the HD-CYC group compared to 7/18 in the LD-CYC group had treatment failure, but this difference did not reach significance (P = 0.2), perhaps due to inadequate numbers [Table 2].

The HD group in the ELNT trial received eight doses of CYC, the last two being quarterly doses. In our cohort, only 11 of 34 treated with the HD-CYC had received eight or more cycles.[5] Within the HD-CYC group, none of those receiving eight or more doses of CYC, compared to seven of those receiving less than eight doses of CYC, had treatment failure (P = 0.045). The difference in treatment failure rates did not appear to be significantly different between the LD-CYC group when compared with either the HD group that received ≥8 doses of CYC or the group that received less than eight doses. However, as numbers in these subgroups are small, results will need to be interpreted with caution.

Only 20 of the HD-CYC group received mycophenolate mofetil (MMF) as the maintenance therapy, as opposed to all the patients in LD-CYC. This likely reflects the change in choice of maintenance therapy over time.

The adverse effects within each treatment group are shown in Table 4. There appeared to be no difference in the number of significant side effects in either group. Nausea and vomiting appeared to be more in the HD-CYC group. Infections were observed among 10 of the HD-CYC group, three of whom had more than one episode of infection. One patient had experienced severe pneumonia requiring intensive care. Other infections included cystitis, tinea corporis, onychomycosis, and multi-dermatomal shingles. Among the LD-CYC group, 7 experienced infections, with one experiencing two episodes of pyelonephritis. Infections among the others were pneumonia, shingles, and acute gastroenteritis.

Discussion

This study examined the hypothesis that HD-CYC and LD-CYC have similar outcomes in South Asian patients. Using retrospective data, it was suggested that both HD-CYC and LD-CYC therapies have comparable similar short-term outcomes in LN patients from a single center in Sri Lanka.

More than 62% of both groups in our cohort presented with nephrotic syndrome or nephrotic range proteinuria indicating severe disease, like the rates of nephrotic syndrome seen in the NIH cohorts.[4] Within our cohort, histological activity was more severe in the LD-CYC group and may reflect changes in CYC usage over time. Despite these differences in histological activity, there was no significant difference between the outcomes of the HD-CYC and LD-CYC groups. Notably, ELNT recruited patients with less-severe LN, with only 28% presenting with nephrotic syndrome.[5] The universal use of mycophenolate mofetil (MMF) maintenance in the LD-CYC group versus 20/34 in the HD-CYC group may have influenced the rates of relapses between the two arms. In the HD-CYC group, azathioprine doses were lower than those used in the ELNT and may have further increased risk of relapses.

The question of the efficacy of LD-CYC in non-European populations has been the subject of research worldwide. In the first phase of Abatacept and Cyclophosphamide combinations Efficacy and Safety Study (ACCESS), the placebo arm received a treatment course based on the ELNT and comprised six doses of LD-CYC followed by azathioprine.[10] The treatment group received abatacept in addition to the above. The primary outcome of CR at 24 weeks was 31% and 33%, respectively, in the treatment and placebo groups. Forty percent of the sample was African-American and 39% was Hispanic/Mestizo and, therefore, very different ethnically from the ELNT study population. The authors concluded that even in their ethnically diverse sample, the ELNT protocol appeared to have similar, if not superior, outcomes to HD-CYC and MMF-based regimens.

A systematic review comparing LD- and HD-CYC regimens among Chinese patients with LN described similar rates of PR and CR in both groups, with lower infection and menstrual disturbances in the LD group. This review studied only short-term effects but is reassuring in that early outcomes were similar.[7]

However, our study findings contrast with the only other study of its kind from South Asia. Mehra et al.[11] conducted a single-center, open-label, prospective, randomized controlled trial in India comparing remission rates achieved at 52 weeks with LD-CYC and HD-CYC in proliferative lupus. They defined CR as proteinuria <0.5 g, stable renal function, and inactive urinary sediments and PR as 50% reduction in proteinuria to subnephrotic range with stable renal function and inactive urinary sediments. In their study, the HD-CYC group received a median dose of 6 g (IQR 5.4–6.3 g), which was higher than that received in our HD-CYC group. Remission was significantly higher in the HD group (73%) compared to the LD (50%) group. During this 52-week follow-up, nine (24%) in the LD group compared to one (3%) in the HD group experienced relapse. While advising further studies, Mehra et al[11]. suggested that HD-CYC may be a more efficacious treatment approach among Indian patients.

While this cohort of patients appeared to have less clinically severe LN (only 24% and 11% of patients in the LD-CYC and HD-CYC groups, respectively, having a UPCR >3 in the Indian cohort), our response rates appeared to be higher overall with no between-group difference. This may partly be due to the more generous definition of CR in our study. The definitions of PR and CR are very variable, but current data supports that achieving proteinuria <1 g/d is associated with very good outcomes, making this a practical goal.[12]

Gonadotoxicity, assessed in terms of transient amenorrhoea and premature menopause, was similar between the two groups, despite clear differences in the duration of follow-up. However, our cohort of patients is quite young, and it will be worthwhile obtaining further data on fertility and pregnancy outcomes prospectively.

The main strength of the study is the inclusion of patients with pathologically severe LN. The main limitations of our study are that data was collected retrospectively, with risk of bias. We believe that data regarding response and renal/non-renal relapses were obtained with good accuracy as these were based on objective documentation. In addition, these were two distinct cohorts treated at two different periods in time. For our earlier patients, HD-CYC was followed by maintenance therapy with azathioprine, compared to our current practice of MMF. It is now standard practise for all our lupus patients to be on hydroxychloroquine, known to increase renal remission, prevent relapses, and prevent progression to end-stage kidney disease. Furthermore, better supportive care with antiproteinuric agents and blood pressure control is to be expected over time. However, we believe that the effect on the primary outcome of treatment failure at 6 months would not have been impacted greatly by these changes. A prospective trial comparing HD- and LD-CYC in this population will be unethical due to possible toxicity of HD regimens. Therefore, the data we have collected, when analyzed with full understanding of its limitations, is very valuable. It is reassuring to know that, when considering the efficacy of CYC, the South Asian (Sri Lankan) cohort fares similarly to Caucasian populations on whom the bulk of research has been conducted and on whom evidence-based treatment guidance has been formulated.

Conclusions

Our study suggests that an LD-CYC regimen is as effective as HD-CYC in those of South Asian ethnicity. Well-designed local multicentre prospective studies are needed, so that we can better understand our patient cohort and how to treat them.