http://www.kidney-international.org & 2013 International Society of Nephrology Long-term of patients with severe ANCAassociated vasculitis comparing plasma exchange to intravenous methylprednisolone treatment is unclear Michael Walsh 1, Alina Casian 2, Oliver Flossmann 3, Kerstin Westman 4, Peter Höglund 5, Charles Pusey 6 and David R.W. Jayne 2 on behalf of the European Vasculitis Study Group (EUVAS) 1 Departments of Medicine (Nephrology), and Clinical Epidemiology and Biostatistics, St Joseph s Hospital, McMaster University, Hamilton, Ontario, Canada; 2 Lupus and Vasculitis Clinic, Addenbrooke s Hospital, Cambridge, UK; 3 Renal Department, Royal Berkshire Hospital, Reading, UK; 4 Department of Nephrology, University Hospital Malmo, Malmo, Sweden; 5 Competence Centre for Clinical Research, Skane University Hospital, Lund, Sweden and 6 Department of Medicine, Imperial College London, Hammersmith Hospital, London, UK Patients with antineutrophil cytoplasm antibody associated vasculitis (AAV) requiring dialysis at diagnosis are at risk for developing end-stage renal disease (ESRD) or dying. Shortterm results of a trial comparing plasma exchange () to intravenous methylprednisolone () suggested improved renal recovery. Here we conducted long-term to see if this trend persisted. A total of 137 patients with newly diagnosed AAV and a serum creatinine over 500 lmol/l or requiring dialysis were randomized such that 69 received and 68 received in addition to cyclophosphamide and oral glucocorticoids. The patients were followed for a median of 3.95 years. In each group there were 35 deaths, while 23 and 33 patients developed ESRD. The hazard ratio for compared to IV MeP for the primary composite outcome of death or ESRD was 0.81 (95% confidence interval 0.53 1.23). The hazard ratio for all-cause death was 1.08 with a subhazard ratio for ESRD of 0.64 (95% confidence interval 0.40 1.05). Thus, although short-term results with are encouraging, the long-term benefits remain unclear. Further research is required to determine the role of in AAV. Given the poor outcomes of patients with severe AAV, improved treatment is urgently needed. Kidney International (2013) 84, 397 402; doi:10.1038/ki.2013.131; published online 24 April 2013 KEYWORDS: plasma exchange; randomized controlled trial; vasculitis Correspondence: Michael Walsh, Departments of Medicine (Nephrology), and Clinical Epidemiology and Biostatistics, St Joseph s Hospital, McMaster University, Marian Wing, Hamilton, Ontario, Canada L8N 4A6. E-mail: lastwalsh1975@gmail.com Received 15 June 2012; revised 22 January 2013; accepted 31 January 2013; published online 24 April 2013 Granulomatosis polyangiitis (formerly Wegener s granulomatosis) and microscopic polyangiitis are associated with antineutrophil cytoplasm antibodies and are collectively referred to as the antineutrophil cytoplasm antibody associated vasculitides (AAV). Although treatment with immunosuppressive drugs and glucocorticoids has improved outcomes for patients with AAV, those with severely reduced kidney function at diagnosis are at high risk of death and end-stage renal disease (ESRD). 1 Rapid and complete initial disease control may improve the clinical outcomes of patients with severe AAV by reducing the burden of organ scarring that occurs early in the course of disease. However, given that patients with severe AAV are also at high risk of serious infections, adjuvant therapies must be minimally toxic. 2 Both plasma exchange () and pulse intravenous methylprednisolone () are potential treatments that may enhance remission induction without serious toxicity. The MEPEX trial compared with as adjuvants to cyclophosphamide and oral prednisolone in patients with a new diagnosis of AAV and a serum creatinine 4500 mmol/l or requiring dialysis. 3 The primary outcome for MEPEX was renal recovery defined as a creatinine o500 mmol/l or dialysis independence within 12 months of randomization. MEPEX demonstrated a 50% relative reduction in the need for dialysis at 12 months for participants that received compared with those who received IV methylprednisolone. 3 We hypothesized that such a substantial reduction in early dialysis dependency represented dramatic improvement in disease control and likely arrested organ damage, which would result in significant improvement in medium and long-term prognosis. However, the MEPEX trial had a higher-than-anticipated mortality rate and recent metaanalyses call into question whether is truly effective at reducing the clinically important endpoints of death and ESRD for patients with severe AAV. 4 We conducted a study of Kidney International (2013) 84, 397 402 397
the long-term outcomes of patients from the MEPEX trial, to determine whether the benefits of on renal function in the short term resulted in a long-term lower risk of death or ESRD (ISRCTN52952998 and NCT01408836). RESULTS One hundred and thirty-seven patients were enrolled and followed for a median of 3.95 years for a total of 515 patientyears at risk. Patients enrolled in the MEPEX trial have been previously described (Table 1). Patients with granulomatosis polyangiitis and those with alveolar hemorrhage were slightly more common in the group than in the group. Table 1 Baseline characteristics of patients enrolled in the MEPEX trial (n ¼ 68) (n ¼ 69) Age, median (IQR) 66 (59 71) 67 (60 70) Female, n (%) 24(35) 28(49) Granulomatosis with polyangiitis, n (%) 25(37) 17(25) Anti-PR3 positive, n (%) 31(46) 27(39) Creatinine at entry, median (IQR) 732 (611 918) 701 (583 896) BVAS at entry, median (IQR) 19.5 (14.5 27.0) 19 (15 23) Alveolar hemorrhage present, n (%) 18(26.5) 13(18.8) Abbreviations: BVAS, Birmingham Vasculitis Activity Score; IQR, 25th to 75th percentile interquartile range;, intravenous methylprednisolone;, plasma exchange; PR3, proteinase 3. 151 Screened 137 Randomized 69 68 50 Included in original trial 12-month dialysis independence outcome 45 Completed long-term 19 Deaths before 12-month 5 Lost after 12-month 52 Included in original trial 12-month dialysis independence outcome 41 Completed long-term 69 Included for analysis 68 Included for analysis 16 Deaths before 12-month 10 Lost after 12-month Figure 1 Flow diagram. Flow diagram of patients., intravenous methylprednisolone;, plasma exchange. One hundred and twenty-two patients (90%) had their vital status ascertained in long-term. Fifteen (5 and 10 ) patients did not complete the long-term (Figure 1). Three patients from the group developed ESRD and one from the same group relapsed before being lost to long-term. No patients from the group developed ESRD, died, or relapsed before being lost to long-term. Patients did not differ significantly with respect to therapy used after the end of the original study period (Table 2). Outcomes Table 3 summarizes the clincial outcomes. Of the 70 patients (51%) that died, 35 (51%) were in the group and 35 (51%) were in the group. Fifty-six patients (41%) developed ESRD (23 vs. 33 ) and 86 (63%) developed the composite endpoint of death or ESRD (40 vs. 46 ). Three in the group compared with 1 in the group died of pulmonary hemorrhage, whereas 9 in the group compared with 15 in the group died of infection. Infections were the most common cause of death (24 of 70 deaths (34%)) and deaths attributed to active vasculitis were uncommon (2 of 70 deaths (3%)). Overall, there was no statistically significant difference in the distribution of causes of death (P ¼ 0.80), although infectious causes were more common in the group (15 (43%)) than in the group (9 (26%); Table 3). Only 26 patients (10 vs. 16 ) suffered a relapse of vasculitis. We did not find an effect of on the composite endpoint (hazard ratio (HR), 0.81; 95% confidence interval (CI), 0.53 1.23; P ¼ 0.32) or on the all-cause death (HR, 1.08; 95% CI, 0.67 1.73; P ¼ 0.75; Figures 2 and 3). Similarly, in competing-risk models in which death was the competing risk, we did not find an effect of on ESRD (sub-hr, 0.64; 95% CI, 0.40 1.05; P ¼ 0.08; Figure 4). Twenty-six patients had a relapse, 10 (14.4%) in the group and 16 (20.5%) in the group (sub-hr, 0.56; 95% CI, 0.26 1.21; P ¼ 0.14). Similarly, there was no demonstrable difference between groups with an sub-hr for renal relapse of 1.14 (95% CI, 0.41 3.14; P ¼ 0.80) for compared with (Table 4). Sensitivity analyses that adjusted for other prognostic factors did not materially alter the assessment of compared with, with respect to the primary composite outcome or the outcomes of ESRD, death, or relapse (i.e., in no analyses was one treatment superior to the other). At 5 years, the median creatinine in the 37 patients alive and not on dialysis with a measure was 167 mmol/l (130 159) and was not different between patients treated with (176 (142 292)) compared with those treated with (158 (123 238)). No difference between groups was detected when either comparing the 5-year keratinize levels (P ¼ 0.43 by Kreskas Wallis test) or after adjusting for baseline creatinine with analysis of covariance (mean difference, 398 Kidney International (2013) 84, 397 402
Table 2 Therapy after the end of the orignal trial period for patients in the MEPEX trial Cyclophosphamide, n (%) Azathioprine, n (%) Prednisone, n (%) Months after randomization P-value P-value P-value 13 18 3 (7) 5 (12) 0.43 28 (64) 23 (55) 0.4 36 (82) 32 (76) 0.52 19 24 2 (5) 1 (3) 0.65 24 (59) 20 (51) 0.51 30 (73) 26 (67) 0.53 25 36 3 (8) 2 (5) 0.67 17 (44) 14 (37) 0.55 21 (54) 25 (66) 0.29 37 48 3 (8) 5 (14) 0.41 14 (37) 13 (36) 0.95 20 (53) 20 (56) 0.8 49 60 2 (6) 4 (12) 0.37 14 (41) 12 (36) 0.67 16 (47) 15 (45) 0.9, intravenous methylprednisolone;, plasma exchange. P-values obtained from two-tailed w 2 -test. Table 3 Long-term primary and secondary outcomes by treatment group Outcome, n ¼ 68 (%), n ¼ 69 (%) HR (95% CI) P-value Death or ESRD 46 (68) 40 (58) 0.81 (0.53 1.23) 0.32 Death 35 (51) 35 (51) 1.08 (0.67 1.73) 0.75 ESRD a 33 (49) 23 (33) 0.64 (0.40 1.05) 0.08 Relapse a 16 (21) 10 (14) 0.56 (0.26 1.21) 0.14 Abbreviations: ESRD, end-stage renal disease; HR, hazard ratio;, intravenous methylprednisolone;, plasma exchange group. a Subhazard ratio presented. P-value based on Cox regression or competing risk regression model. Table 4 Causes of death over the course of long-term followup Primary cause of death, n (%) (n ¼ 35) (n ¼ 35) Vasculitis (%) 1 (2.9) 1 (2.9) Pulmonary hemorrhage (%) 3 (8.6) 1 (2.9) Infection (%) 9 (25.7) 15 (42.9) Cardiovascular (%) 6 (17.1) 4 (11.4) ESRD (%) 1 (2.9) 1 (2.9) Cancer (%) 3 (8.6) 2 (5.7) Other 12 (34.3%) 11 (31.4%) Proportion free of ESRD or death 1 0.75 0.50 0.25 0 Number at risk 0 2 4 6 8 10 Time (years) 68 25 23 15 7 1 69 32 26 13 6 0 Figure 2 Kaplan Meier curve of composite outcome. Kaplan Meier curve of the composite outcome of end-stage renal disease (ESRD) or death according to randomization to treatment with intravenous methylprednisolone () or plasma exchange (). 1 Abbreviations: ESRD, end-stage renal disease;, intravenous methylprednisolone;, plasma exchange. There was no difference in the overall distribution of cause of death (P ¼ 0.80). Proportion dead 0.75 0.50 29 mmol/l; 95% CI, 44 to 101 mmol/l; P ¼ 0.43). Similarly, there were no significant differences between groups when comparing the last measured keratinize values of 70 patients who were alive at the end of the first year and were not on dialysis by the study end (median 221 (150 335) in the group compared with 192 (141 268) in the group; P ¼ 0.41). 0.25 0 Number at risk 0 2 4 6 8 10 Time (years) 68 38 33 21 13 2 69 39 34 17 7 0 DISCUSSION We assessed the long-term outcomes of patients with severe AAV that were treated with either or as initial adjuvant therapy. After a median of almost 4 years, there was no evidence of a net difference between and in clinical outcomes. Furthermore, over half of patients with severe AAV at diagnosis died, and almost twothirds had either developed ESRD or died. The outcomes for patients with severe AAV continue to be very poor. Figure 3 Kaplan Meier curve of deaths. Kaplan Meier curve of deaths in patients with severe antineutrophil cytoplasm antibody associated vasculitis randomized to intravenous methylprednisolone () or plasma exchange (). The MEPEX trial was designed to assess renal recovery and was found to be superior to at 12 months. 3 In our extended study, we focused on ESRD and survival. If truly reduces ESRD, it seems logical that longer-term mortality should also be reduced given the Kidney International (2013) 84, 397 402 399
Cumulative incidence of ESRD 0.5 0.4 0.3 0.2 0 2 4 6 8 10 Time (years) Figure 4 Cumulative incidence curve. Cumulative incidence curve of end-stage renal disease (ESRD) in patients with severe antineutrophil cytoplasm antibody associated vasculitis randomized to intravenous methylprednisolone () or plasma exchange () when death is treated as a competing risk. strong association between ESRD and death. The fact that this was not seen could be due to several reasons. First, it is possible that ESRD in severe AAV does not share the same relationship with death as seen in other conditions (i.e., ESRD caused by AAV is not strongly associated with death). This is possible as patients with severe, acute AAV do not accumulate chronic morbidity before initiating ESRD as, e.g., patients with chronic diabetes mellitus do. 5 Second, it is possible that the effect of early in AAV results in a true reduction in ESRD, but that this effect is obscured over time by other non-aav morbidities causing death in both groups. This results in a low signal-to-noise ratio in a small trial. Given the advanced age of the population at the start of the trial, this is certainly possible. This finding could also be explained if reduces ESRD but also increases mortality via a mechanism independent of ESRD (e.g., infection) or if there were a bias towards initiating dialysis in the group. Finally, it must be considered that the early benefits observed were random error. Other studies have suggested that 1500 randomized patients are optimally required to reliably demonstrate whether is effective in reducing death and ESRD in patients with AAV. 4 is frequently advocated as the standard of care for patients with pulmonary hemorrhage or advanced renal failure secondary to AAV. 6,7 Our study demonstrated that death from pulmonary hemorrhage was infrequent in both and -treated patients despite similar occurrence at trial entry in both groups. Further, although the results are encouraging regarding the prevention of ESRD, we did not demonstrate that the reduction in ESRD lead to a reduction in mortality. These findings raise uncertainty regarding the net benefit of as a first-line treatment in patients with severe manifestations of AAV, and this uncertainty is compounded when the relative paucity of information regarding clinical outcomes is considered. Previous studies have suggested that even when we consider all randomized trials of in AAV, there is insufficient evidence to make reliable conclusions. 4 Perhaps the most important finding of our study is not that did not materially improve the incidence of ESRD or death compared with for patients with severe AAV. The most important finding is that half the study cohort died within 5 years from the time of diagnosis, with many of the survivors requiring dialysis. Further, the most frequent cause of death was infectious in nature, whereas very few were due to active vasculitis. Pateints with severe AAV have several important risk factors for infections including: the use of chronic immunosuppression, the placement of central intravenous catheters, uremia/dialysis dependency, and gammaglobulin depletion from. These complicated risk factors result both from the treatment for AAV being toxic (i.e., immunosuppressive) and being inadequately effective (i.e., need for ongoing immunosuppression and failure to prevent ESRD). As such, research to reduce treatment toxicity while improving disease control both in the short- and long term is urgently needed. Our findings must be considered in the context of the study s limitations. Although this is the largest trial in severe AAV, it has limited statistical power to reliably conclude whether there is a difference between and and is therefore at risk of failing to identify a clinically important benefit. This risk is increased by the loss of 15 patients to long-term with more patients in the group lost. This differential loss of may have introduced an unpredictable bias in the results. We believe this bias is, however, small in magnitude and unlikely to alter our conclusions. We cannot, however, conclude that no benefit exists. Further, is frequently considered in addition to rather than as an alternative to. This trial was also restricted to patients with a creatinine 4500 or on dialysis. If is effective, it is likely that it would also be effective in patients with rapidly progressive glomerulonephritis but a creatinine o500 mmol/l. This is also suggested by the finding that the patients with the least amount of chronic scarring on renal biopsy were the most likely to recover and from a meta-analysis of trials. 4,8 Finally, we are unable to inform the potential use of in relapses as this information was not collected. continues to be a promising therapy for AAV, but further trials are required to inform the potential for an increased risk of infections and death. This is particularly relevant given the invasiveness and cost of. The ongoing 500 patient PEXIVAS trial with up to 7 years of planned (ISRCTN07757494; clinicaltrials.gov NCT00987389) will improve our knowledge of the effect of on AAV. Even if is demonstrated an effective therapy for AAV, we will require further improvements in the care of patients with severe manifestations to reduce the current unacceptably high mortality and ESRD rates. Given the relative frequency of infectious causes of death, reducing this risk must be a priority in future studies. 400 Kidney International (2013) 84, 397 402
MATERIALS AND METHODS Study design The trial was conducted in accordance with the Declaration of Helsinki Principles, and the protocol is previously described and available online (www.vasculitis.org). 3 Briefly, MEPEX was a two parallel limb, randomized control trial conducted in 28 hospitals in nine European countries. Informed consent was obtained for all patients and they were treated with the same regimen of oral cyclophosphamide for 6 months and with a tapering regimen of oral prednisolone. At trial entry, patients were randomly assigned to receive either or in addition to cyclophosphamide and prednisolone. was prescribed as seven exchanges of 5% albumin (in the absence of a risk for bleeding in which case plasma could be used) over 14 days at a volume of 60 ml/kg per exchange with an upper limit of 4.5 l per exchange. was prescribed as three doses of 1 g/day. Patients were maintained on azathioprine after 6 months of treatment. Patients Eligible patients had a diagnosis of granulomatosis polyangiitis or microscopic polyangiitis using criteria adapted from the Chapel Hill consensus conference, 9 biopsy evidence of pauci-immune, necrotizing and/or crescentic glomerulonephritis, and a serum creatinine 4500 mmol/l or required dialysis. Follow-up Patients were recruited between March 1995 and October 2001. The original trial lasted 12 months. Extended was conducted between October 2004 and November 2006, with clinical data abstracted to standardized data collection forms by local investigators. Outcomes The primary outcome for this analysis was the composite of ESRD or death from any cause. ESRD was defined as the need for permanent renal replacement therapy as determined by permanent dialysis dependence or receipt of a renal transplant. The composite was chosen, because death acts as a competing event for ESRD. Competing events are important to consider when the occurrence of one event that occurs frequently precludes the occurrence of the outcome of interest and, particularly, when the effect of treatment may be different on the competing event than the primary event of interest. 10 In this case, early deaths preclude the occurrence of ESRD (i.e., death competes for ESRD). We also assessed the occurrence of ESRD and death separately. Relapses of AAV were considered a secondary outcome. A relapse was defined as any new or worsened state of disease activity that required an escalation in treatment (immunosupressives and/or glucocorticoids). Renal function: Serum creatinine at 5 years after onset of disease was recorded in survivors not on dialysis. Statistics Descriptive data is presented as mean (s.d.) or median (25th to 75th percentile) for continuous variables and as frequency (%) for categorical data. The outcomes were assessed using time-to-event analyses. We assessed the primary composite outcome of ESRD or death with a standard Cox proportional hazard model. The Cox proportional hazard model was also used to assess the effect of on all-cause death. Analysis of ESRD incorporated the competing risk of death by calculating the subhazard for ESRD as a function of model covariates that were jointly fit to the hazard of death. 11 In both models, all-cause death was considered the competing risk, as patients that die before ESRD or relapse are unable to experience the event of interest (ESRD). Relapses were also assessed using the competing risk framework in which death was considered a competing risk. In addition, renal relapses were assessed using both death and ESRD as a competing risk. Sensitivity analyses adjusted for previously described risk factors for death and relapse: age, sex, diagnosis of granulomatosis with polyangiitis versus microscopic polyangiitis, and cytoplasmic antineutrophil cytoplasm antibody immunofluorescence/anti-pr3 antibodies (Supplementary Table S1 online). We did not adjust for serum creatinine given its uncertain prognostic value across the range included in this trial (i.e., all 4500 mmol/l). In models in which ESRD was an outcome, patients were right censored on the first day of dialysis without subsequent recovery. Renal function at 5 years was compared using analysis of covariance to account for the correlation between pretreatment and posttreatment correlations. Estimates of the effect of, their 95% CI, and associated P-values were calculated for each outcome. A two-tailed P-value o0.05 was considered significant for all analyses, with no adjustments made for multiple comparisons. All analyses were performed with Stata 11 MP (Stata, College Station, TX). DISCLOSURE MW and DRWJ are principle investigators of a trial that has in-kind support from Caridian BCT and Fresenius Australia for plasma exchange disposables. All the other authors declared no competing interests. ACKNOWLEDGMENTS This trial was supported by grants from the European League Against Rheumatism and by the European Union (European Community Systemic Vasculitis Trial project (BMH1-CT93-1078 and CIDP-CT94-0307) and Associated Vasculitis European Randomized Trial project (BMH4-CT97-2328 and IC20-CT97-0019)). MW is supported by a Kidney Research Scientist Core Education National Training (KRESCENT) New Investigator Award from the Canadian Institutes of Health Research, Kidney Foundation of Canada, and the Canadian Society of Nephrology. DRWJ is supported by the Cambridge Biomedical Research Centre. CP is supported by the Imperial Biomedical Research Centre. SUPPLEMENTARY MATERIAL Table S1. Results of senstivity analyses adjusted for age, sex, disease activity, and antibody specificity. Supplementary Material is linked to the online version of the paper at http://www.nature.com/ki REFERENCES 1. Flossmann O, Berden A, De GK et al. Long-term patient survival in ANCAassociated vasculitis. Ann Rheum Dis 2011; 70: 488 494. 2. Little MA, Nightingale P, Verburgh CA et al. Early mortality in systemic vasculitis: relative contribution of adverse events and active vasculitis. Ann Rheum Dis 2010; 69: 1036 1043. 3. Jayne DR, Gaskin G, Rasmussen N et al. Randomized trial of plasma exchange or high-dosage methylprednisolone as adjunctive therapy for severe renal vasculitis. J Am Soc Nephrol 2007; 18: 2180 2188. Kidney International (2013) 84, 397 402 401
4. Walsh M, Catapano F, Szpirt W et al. Plasma exchange for renal vasculitis and idiopathic rapidly progressive glomerulonephritis: a meta-analysis. Am J Kidney Dis 2011; 57: 566 574. 5. Keith DS, Nichols GA, Gullion CM et al. Longitudinal and outcomes among a population with chronic kidney disease in a large managed care organization. Arch Intern Med 2004; 164: 659 663. 6. Menahem S, Hiremagalur B, Mudge D et al. The CARI guidelines. Induction and maintenance therapy in ANCA-associated systemic vasculitis. Nephrology (Carlton) 2008; 13(Suppl 2): S24 S36. 7. Szczepiorkowski ZM, Bandarenko N, Kim HC et al. Guidelines on the use of therapeutic apheresis in clinical practice: evidence-based approach from the Apheresis Applications Committee of the American Society for Apheresis. J Clin Apher 2007; 22: 106 175. 8. de Lind van Wijngaarden RA, Hauer HA, Wolterbeek R et al. Clinical and histologic determinants of renal outcome in ANCA-associated vasculitis: A prospective analysis of 100 patients with severe renal involvement. J Am Soc Nephrol 2006; 17: 2264 2274. 9. Jennette JC, Falk RJ, Andrassy K et al. Nomenclature of systemic vasculitides. Proposal of an international consensus conference. Arthritis Rheum 1994; 37: 187 192. 10. Ferreira-Gonzalez I, Permanyer-Miralda G, Busse JW et al. Methodologic discussions for using and interpreting composite endpoints are limited, but still identify major concerns. J Clin Epidemiol 2007; 60: 651 657. 11. Fine JP. Regression modeling of competing crude failure probabilities. Biostatistics 2001; 2: 85 97. 402 Kidney International (2013) 84, 397 402