Treatment of polyarteritis nodosa and microscopic polyangiitis with poor prognosis factors: A prospective trial comparing glucocorticoids and six or twelve cyclophosphamide pulses in sixty-five patients
Hôpital Avicenne, Assistance Publique-Hôpitaux de Paris, Université Paris-Nord, Bobigny, France
Department of Internal Medicine, Hôpital Avicenne, Assistance Publique-Hôpitaux de Paris, Université Paris-Nord, UPRES–EA 3409 Recherche Clinique et Thérapeutique, 125, rue de Stalingrad, 93009 Bobigny Cedex, France
Because the optimal cyclophosphamide (CY) treatment duration for severe polyarteritis nodosa (PAN) without virus infection and microscopic polyangiitis (MPA) has not been established, we conducted a trial to compare the effectiveness of 6 versus 12 CY pulses given in combination with corticosteroids (CS).
Sixty-five (18 PAN, 47 MPA) previously untreated patients were randomized to receive 12 (n = 34) or 6 (n = 31) CY pulses combined with CS. PAN and MPA were histologically proven or met ACR criteria. All patients presented ≥1 factor of severity according to the five factor score (FFS). CY pulses were administered every 2 weeks for 1 month, then every 4 weeks. The end point of the study was the number of events (relapses and/or deaths) occurring in each group, analyzed according to an intention-to-treat strategy. The outcome was evaluated by Cox proportional hazards analysis.
The baseline characteristics were similar for both groups. The mean (± SD) followup was 32 ± 21 months. Survival analysis showed a significantly lower relapse probability (P = 0.02; hazards ratio [HR] = 0.34) and higher event-free survival (P = 0.02, HR = 0.44) for the 12 CY-pulse group while the mortality rates were not significantly different (P = 0.47).
These results suggest that 6 CY pulses are less effective than 12 CY pulses to treat severe PAN and MPA, particularly with respect to the risk of relapses.
Polyarteritis nodosa (PAN) and microscopic polyangiitis (MPA) are primary systemic necrotizing vasculitides (1, 2). Formerly considered as a single entity (3), PAN and MPA were subsequently discriminated by the Chapel Hill Consensus Conference (CHCC) nomenclature, in particular with regard to differences in the size of the predominantly involved vessels and clinical presentation (1). Furthermore, the underlying pathogenic mechanisms appear to be different, involving immune complexes in PAN and antineutrophil cytoplasm antibodies (ANCA) in MPA (1, 2).
The standard regimen for PAN, not related to hepatitis B virus (HBV) infection, and MPA is based on a combination of corticosteroids (CS) and cyclophosphamide (CY). According to a recently published retrospective analysis, it would appear that–for either disease–the addition of CY to CS benefits particularly patients presenting with factors of poor prognosis (4) and not for those without such signs (5). There is an ongoing controversy in the field of vasculitides whether CY should be administered orally (6, 7) or intravenously (8, 9). For PAN and MPA, prior studies suggested that an intermittent pulse-therapy may be as efficacious as oral CY at inducing remission (8, 10) while generating fewer side effects (11).
Independently of the choice of the route of administration, the optimal duration of CY treatment for PAN and MPA has not yet been codified. A treatment duration of at least 12 months has been empirically recommended, but no study has directly addressed this question, and some arguments would support shorter treatment courses. In HBV-induced PAN, e.g., patients recover in less than 3 months when the appropriate treatment, combining antiviral drug(s) and plasma exchanges, is prescribed (12).
We therefore carried out the present randomized trial that aimed to evaluate the effectiveness of a shortened 6 CY-pulse treatment compared to the standard 12-pulse regimen—both in combination with the same CS regimen—for patients with severe PAN and MPA.
PATIENTS AND METHODS
Sixty-five patients were included in a prospective, multicenter trial, organized by the French Vasculitis Study Group (FVSG). None of them had received previous treatment for vasculitis. PAN and MPA were histologically proven or met the American College of Rheumatology (ACR) classification criteria (3) and the CHCC nomenclature for MPA (1). Once the diagnosis was made, the patients were included in the trial if they presented at least one factor of poor prognosis, according to the five factor score (FFS) (4). The FFS comprises the following items: classes of serum creatinine (≤ and > 1.58 mg/dl) and proteinuria (≤ and > 1 gm/day), presence of severe gastrointestinal tract involvement, cardiomyopathy and/or central nervous system involvement; the presence of each factor is accorded one point. We also applied the Birmingham vasculitis activity score (BVAS) in its first presentation (13) to test its ability to correlate with prognosis and compare it with FFS. This score was not used to select the patients for the study but was applied retrospectively. Disease features were only taken into consideration when they were attributable to active vasculitis.
Randomization was centralized at the coordinating center and made by phone, fax, or E-mail. Criteria for inclusion were checked by the coordinating center.
Abdominal and renal angiographies were performed only in selected patients suffering from abdominal pain and/or presenting with renal insufficiency or other renal symptoms for which the diagnosis of nephropathy was not clearly established and a renal biopsy was required.
Every patient's serum was tested for the presence of ANCA by immunofluorescence according to the recommendations of the European Vasculitis Study Group (14) using sera diluted 1/16 and ethanol-fixed neutrophils. The immunofluorescence assay was completed by enzyme linked immunosorbent assay (ELISA) for every patient with ANCA detected by the former. ELISA was recommended but not compulsory when the fluorescence assay was negative.
CS were administered as follows: a daily pulse of 15 mg/kg for 3 days, then 1 mg/kg/day orally for 3 weeks. CS were then tapered by 5 mg every 10 days until half the dose was obtained; then diminished by 2.5 mg every 10 days until a dosage of 15 mg/day was obtained. When the dosages were under 10 mg, CS were reduced by 1 mg every 10 days until being definitively stopped. CY pulses were administered every 2 weeks for 1 month, then every 4 weeks. Patients were randomly assigned to receive 6 or 12 pulses. No maintenance treatment was given after stopping CY. The end point of the study was the number of events (relapses and/or deaths) occurring in each group, according to intention to treat.
Adjuvant treatments were compulsory: cotrimoxazole for prophylaxis of Pneumocystis carinii pneumonia, potassium, calcium, and vitamin D3. Since 1998, bisphosphonates have been recommended to treat or prevent CS-induced osteoporosis. In the case of relapse, the attending physician was free to modify treatment and to choose other immunosuppressants or different administration routes.
This interim analysis was performed 7 years after the start of inclusion. Quantitative variables, expressed as means ± SD values, were compared using Student's t-test or nonparametric tests. Categorical variables were compared with chi-square or, when appropriate, Fisher's exact test. The relapse, mortality, and event-free survival (EFS) probabilities were assessed by life-table analysis. All the data were analyzed according to an intention-to-treat strategy. The time to relapse and/or death was calculated relative to the time of diagnosis. Survival curves were plotted using the Kaplan-Meier method (15). The effects of the treatment protocol were evaluated by Cox proportional hazards analysis (16). The risks of relapses and death were also evaluated as a function of the initial presence of ANCA, FFS and BVAS. Continuous variables (FFS, BVAS) and their impact were rendered dichotomous according to the median values of their distribution. P values less than 0.05 were considered statistically significant. All statistical tests were 2-tailed with confidence intervals calculated at the 95% level (95% CI). Analyses were conducted using SAS Statistical Package, version 8.12 (SAS Institute Inc., Cary, NC).
Patient data and therapy.
Patients were enrolled between January 1994 and April 2000. Among the 65 patients included (43 males, 22 females), 18 were diagnosed with PAN and 47 with MPA. Among the clinical symptoms, we noted that 3 patients had paranasal sinusitis and 2 conjunctivitis. Sinusitis was present before the occurrence of vasculitis and no arguments supporting a diagnosis of Wegener's granulomatosis were found. Conjunctivitis resolved quickly with symptomatic treatment. An additional 5 patients (6-pulse group, n = 3; 12-pulse group, n = 2) had symptoms usually found in both PAN and MPA. This overlap was a matter of discussion but they were finally classified as having MPA in accordance with the CHCC nomenclature. Mean FFS at entry was 1.8 ± 0.8 (median 2.0). Mean BVAS was 21.8 ± 7.7 (median 22). The main clinical and biologic manifestations at the time of diagnosis are summarized in Table 1. Randomization assigned 31 and 34 patients, respectively, to receive either 6 or 12 CY pulses. No statistically significant between-group differences were noted with respect to the principal clinical, biologic, and immunologic parameters (Table 1). Creatininemia tended to be higher in the 12-pulse group (2.89 ± 2.95) than in the 6 pulse group (1.87 ± 1.33 mg/dl; P = 0.08), but the median values were quite similar (1.71 versus 1.70 mg/dl).
Table 1. Baseline characteristics observed in the entire study population*
Total (n = 65)
6-pulse CY (n = 31)
12-pulse CY (n = 34)
Values are numbers (%) unless otherwise specified. CY = cyclophosphamide; PAN = polyarteritis nodosa; MPA = microscopic polyangiitis; CNS = central nervous system; ESR = erythrocyte sedimentation rate; ANCA = antineutrophil cytoplasm antibodies.
Age, mean ± SD years
55 ± 16
58 ± 16
53 ± 16
Renal and/or urogenital involvement
Proteinuria and/or hematuria
Creatininemia, mean ± SD, mg/dl
2.41 ± 2.36
1.87 ± 1.33
2.89 ± 2.95
Cranial nerve involvement
Gastrointestinal tract involvement
Pneumonitis and/or infiltrates
ESR >30 mm/1st hour
The mean followup was 32 ± 21 months (Table 2). Complete remission was obtained in 56 patients (86%). All the patients not achieving remission with the assigned treatment died of active vasculitis or infection within a few days or weeks after diagnosis. Among the 56 patients who had achieved a complete remission, 20 relapsed (Table 2); 4 of the 20 relapses were diagnosed in patients still receiving therapy with CS and/or CY. Most of these patients (18/20) reentered remission after intensifying or reinitiating therapy with CS and immunosuppressants or, alternatively, intravenous immunoglobulins (n = 1). During the 12-month period following the end of CY-pulse therapy, 2 of 7 relapses of the 12-pulse group and 6 of 13 relapses of the 6-pulse group occurred.
Table 2. Outcome of 65 patients with PAN and MPA and poor prognostic factors according to the treatment regimen*
12 CY pulses
6 CY pulses
MPA (n = 28)
PAN (n = 6)
Total (n = 34)
MPA (n = 19)
PAN (n = 12)
Total (n = 31)
Values are number (%). See Table 1 for definitions.
The causes and times of deaths are listed in Table 3. The 14 patients who died were older than 60 years and 2 of them were older than 85 years. Uncontrolled vasculitis was responsible for 5 deaths, and infections for 3 deaths. The deaths of 10 patients occurred within the first 6 months and 5 patients died during the first 3 weeks after treatment onset. The other deaths occurred later than 6 months and were not obviously related to vasculitis. Two patients died of malignancies; both had received 6 CY pulses.
Table 3. Causes and times post-diagnosis of deaths*
When more than one entity are present, the one considered the principal cause of death is given first.
6 CY pulses
Sudden death occurring at home, probably septic shock
Small bowel perforation
Uncontrolled vasculitis, fungal infection
12 CY pulses
Pneumonia, mechanical ventilated at the time of inclusion, uncontrolled vasculitis
Alveolar hemorrhage, uncontrolled vasculitis
Uncontrolled vasculitis, acute GI manifestations
Uncontrolled vasculitis, acute GI and cardiac manifestations
Aortic aneurysm rupture (not attributed to vasculitis)
Adverse effects are summarized in Table 4. Nausea and vomiting due to CY were not considered in the table because they occurred in almost all patients but could be successfully treated or prevented with adequate treatments. Thirty-two (49%) patients suffered no side effects. Most adverse effects were attributed to CS treatment. Infections were favored by treatments but were transient in most cases. Four patients developed herpes zoster infections at the end of or following immunosuppressive treatment.
Table 4. Side effects observed during treatment of patients
Time of occurrence, months (outcome)
6 CY pulses
12 CY pulses
Prescribed for relapses. IV = intravenous. See Table 1 for other definitions.
The mean relapse rates for 12- and 6-pulse groups, respectively, were 22 ± 16% (95% CI) and 66 ± 27% (95% CI) at 3 years of followup (P = 0.02, HR = 0.34) (Figure 1A). The 3-year survival rates were 85 ± 12% (95% CI) for the 12-pulse group versus 74 ± 17% (95% CI) for the 6-pulse group (P = 0.47) (Figure 1B). When considering the occurrence of events (relapses and/or deaths), EFS was significantly higher for the 12-pulse group (68 ± 17% [95% CI]) than in the 6-pulse group (19 ± 22% [95% CI]) (P = 0.02, HR = 0.41) (Figure 1C).
With regard to the responses to therapy of PAN and MPA patients, the 3-year EFS rates were 80 ± 35% (95% CI) for PAN patients receiving 12 pulses, 71 ± 20% (95% CI) for PAN patients treated with 6 CY pulses, 54 ± 34% (95% CI) for MPA patients given 12 CY pulses and 55 ± 27% (95% CI) for MPA patients receiving 6 CY pulses (Figure 2). The EFS was significantly different for PAN patients treated with 12 pulses compared with MPA patients receiving 6 pulses (P = 0.04, HR = 0.47).
We also evaluated the potential predictive value for relapse or death of ANCA, FFS (≤ or > 2) and BVAS (≤ or > 22), as assessed at diagnosis and relapse. ANCA positivity did not predict the occurrence of relapses (P = 0.81) but was significantly associated with mortality (3-year survival rate of 70 ± 17% [95% CI] for ANCA-positive patients versus 93 ± 10% [95% CI] for ANCA-negative patients) (P = 0.04; HR = 4.8). Neither FFS (P = 0.17) nor BVAS (P = 0.19) was associated with a risk of relapse. With regard to mortality, we found no association with BVAS (P = 0.12) but noted a significantly lower survival rate at 3 years for patients with an initial FFS > 2 (48 ± 34% [95% CI]) than those with an FFS ≤ 2 (87 ± 9% [95% CI]) (P = 0.03, HR = 3.5).
This randomized, prospective multicenter trial was designed to determine whether a shorter course of CY would be able to treat successfully PAN or MPA with factors of poor prognosis. The 6 and 12 CY-pulse regimens were delivered over 4 and 10 months, respectively, and were not followed by a systematic prescription of a maintenance treatment.
Regarding the baseline characteristics, the 2 groups were comparable. However, the mean creatininemia was higher in the 12-pulse group. Although that difference was not statistically significant, this finding was a matter of concern because of the known pejorative prognostic value on outcome of renal dysfunction in PAN and MPA (17). Given the very similar median creatininemia values for the two groups, this difference probably reflects the inclusion of 2 anuric patients in the 12-pulse group. Moreover, the higher serum creatinine level concerned the group of patients that had the better outcome, and thus it is unlikely that this finding might have substantially biased the results of the trial.
Our results provide evidence that, for patients with severe PAN or MPA, the shorter 6 pulse-course of CY is significantly less effective at preventing relapses and/or deaths than the standard 12-pulse course. In particular, the difference between the two regimens concerned the relapse risk that was significantly higher for the 6 CY-pulse group. We do not think that this difference might be simply due to a postponement of the relapses in the patients that were kept on immunosuppressive therapy for 6 more months. Indeed, the widely divergent relapse-probability curves would rather suggest that the shorter immunosuppressant regimen resulted in a persistently higher relapse risk. This divergence could point to a sustained effect of prolonged initial immunosuppression on the subsequent relapse risk. Consequently, these findings would indicate that the initial CY therapy has to be maintained for more than 4 months in order to obtain sustained disease remission and, eventually, cure of the disease.
Some concern about the validity of this trial may arise from the fact that it combines PAN and MPA patients, in particular with respect to the higher propensity for relapses in MPA (5). Indeed, this study was set up prior to the publication of the CHCC nomenclature that now clearly separates these 2 disorders (1). There is nonetheless some evidence coming from previous trials that both diseases respond to the same treatment, at least to induce the remission (8, 10, 17).
We consequently attempted to evaluate the long-term effects of therapy separately for PAN and MPA. Although clear conclusions are difficult to draw because of the low numbers of patients in each subgroup, our data might give some indications on the optimal treatment of PAN and MPA. The 3-year EFS rate for patients treated with 12 CY pulses was 80% for PAN patients and 71% for MPA patients, which might indicate that after administration of 12 CY pulses it is not mandatory to pursue a maintenance treatment for either vasculitis. It now remains to be elucidated whether similar results can be obtained with a strategy dividing the time on immunosuppressants into 2 phases: induction therapy with CY pulses until remission and subsequent maintenance using another and less toxic immunosuppressant. This question will be addressed in a forthcoming trial.
When occurring, relapses could usually be well controlled by treatment. Some were minor and required only an increase of the CS dose, whereas the more severe relapses required a switch from pulse to oral CY, the prescription of alternative immunosuppressants or, in 1 case, intravenous immunoglobulins. Neither ANCA positivity nor BVAS or FFS at the time of diagnosis predicted the occurrence of relapses. Conversely, ANCA positivity predicted a higher risk of death, which might indicate that ANCA is a predictor of poor survival.
Although several authors reported high rates of malignancies associated with CY (6, 18), we observed a very low malignancy rate, including 1 prostate cancer that was probably not attributable to CY. No patient developed cystitis and none developed myelodysplasia. One patient had moderate thrombocytopenia that was attributed to acute alcoholism. Although these findings have to be confirmed by longer followup, the low malignancy rate might highlight the lower toxicity of CY when administered by pulses. In this trial, the total CY dose was < 9 g in one group (6 pulses) and 18 g in the other (12 pulses). Additional CY was given to some of the patients who relapsed, but none of them developed malignancies. In fact, the most frequent side effects were attributable to CS, e.g., fractures and osteonecrosis necessitating surgery were observed. Compared to previous studies (10, 17), even though we reduced the total CS dose in this protocol by maintaining an initial dose of 1 mg/kg/day of prednisolone for only 3 weeks, side effects did not disappear. We now recommend the systematic prescription of bisphosphonates and increased calcium and vitamin D intake from the time of the first CS dose and we hope to reduce the number of osteoporotic complications. Therapeutic strategies aiming at further lowering CS doses are currently under discussion.
The 3-year mortality rate for this series of patients with factors of poor prognosis was 18% for the 12-pulse group and 26% for the 6-pulse group. The number of deaths might be explained by our policy of inclusion, as we enrolled critically ill patients, even those requiring intubation and ventilatory support at the time of randomization. In fact, 5 of 14 deaths occurred within 3 weeks following randomization. Some deaths were not related to vasculitis but attributed to age. The others were due to the activity of the vasculitis and these patients should probably be treated in another manner, and might benefit from new therapeutic strategies. Promising results have been reported for anti-tumor necrosis factor antibodies in a limited number of patients (19, 20) or for CY dose intensification with or without autologous bone marrow transplant (21). The benefit/risk ratios of these strategies should nevertheless be weighed very carefully.
We conclude that treating PAN or MPA patients with factors of poor prognosis with 12 rather than 6 CY pulses significantly lowered the relapse rate and significantly increased the probability of EFS. Our findings further support that this regimen might not require subsequent maintenance therapy although this possibility needs to be confirmed by longer followup.
Information on the protocol was also disseminated by the European Vasculitis Study Group (EUVAS).
The following institutions and principal investigators comprise the French Vasculitis Study Group and contributed to this study: L. Agron (Caen), M. Alcalay (Poitiers), S. Arfi (Fort-de-France), P. Babinet (Saint-Denis), F. Bani Sadr (Garches), M. Bardet (Coulommiers), J. Barrier (Nantes), D. Blockmans (Leuven), G. Bolla (Cannes), A. Bruet (Poissy), P. Chevalet (Nantes), P. Cohen (Bobigny), D. Coronel (Lyon), I. Delacroix-Szmania (Créteil), P. Dournovo (Eaubonne), A. Dubois (Montpellier), D. El Kouri (Nantes), M. Ergan (Caen), H. Esvant (Clamart), R. Fior (Clamart), S. François (Caen), P. Galanaud (Clamart), B. Gilson (Verdun), B. Godeau (Créteil), M. Hamidou (Nantes), F. Houssiau (Bruxelles), B. Hurault de Ligny (Caen), M. Karmochkine (Paris), J.M. Kerleau (Rouen), X. Kyndt (Valenciennes), P. Lanoux (Charlesville-Mézières), C. Larroche (Suresnes), D. Lauque (Toulouse), S. Le Moal (Saint-Brieuc), P. Letellier (Caen), F. Lhote (Saint-Denis), F. Lioté (Paris), E. Liozon (Limoges), J. Liwartovski (Clamart), T. Lobbedez (Caen), D. Merrien (Nantes), P. Mornet (St-Germain-en-Laye), E. Palazzo (Paris), B. Patri (Paris), J. L. Pennaforte (Reims), C. Perret-Guillaume (Nancy), E. Pertuiset (Pontoise), M. Peureux (Le Havre), P. Philippe (Clermont-Ferrrand), T. Ponge (Nantes), M.A. Pottier (Nantes), X. Puechal (Le Mans), O. Puyoo (Toulouse), I. Quéré (Montpellier), J. Rossert (Paris), S. Roux (Paris), M. Ruivard (Clermont-Ferrand), J.L. Saraux (Eaubonne), P. Soria (Limoges), J. C. Stoléar (Tournai, Belgique), A. Tazi (Bobigny), D. Tirot (Le Mans), J. C. Valcke (Paris), P. Vanhille (Valenciennes), J.F. Viallard (Bordeaux), E. Vidal (Limoges), D. Wahl (Nancy), M. Wissing (Bruxelles), M. Zaegel (Chartres).