Long-term followup of polyarteritis nodosa, microscopic polyangiitis, and Churg-Strauss syndrome: Analysis of four prospective trials including 278 patients




To determine the long-term outcome of patients with polyarteritis nodosa (PAN), microscopic polyangiitis (MPA), and Churg-Strauss syndrome (CSS), to compare the long-term outcome with the overall French population, to evaluate the impact on outcome of the type of vasculitis, prognostic factors, and treatments administered at diagnosis, and to analyze treatment side effects and sequelae.


Data from PAN, MPA, and CSS patients (n = 278) who were enrolled between 1980 and 1993 were collected in 1996 and 1997 and analyzed. Two prognostic scoring systems, the Five-Factors Score (FFS) and the Birmingham Vasculitis Activity Score (BVAS), were used to evaluate all patients at the time of diagnosis.


The mean (±SD) followup of the entire population was 88.3 ± 51.9 months (range 3 days to 192 months). Of the 85 deaths recorded, at least 41 were due to progressive vasculitis or its consequences. Death rates reflected disease severity, as assessed by the FFS (P = 0.004) and the BVAS (P < 0.0002), and the 2 scores were correlated (r = 0.69). Relapses, rarer in hepatitis B virus (HBV)–related PAN (7.9%) than in MPA (34.5%) (P = 0.004), occurred in 56 patients (20.1%) and did not reflect disease severity. Survival curves were similar for the subpopulation of 215 patients with CSS, MPA, and non–HBV-related PAN who were given first-line corticosteroids (CS) with or without cyclophosphamide (CYC). However, CS with CYC therapy significantly prolonged survival for patients with FFS scores ≥2 (P = 0.041). Relapse rates were similar regardless of the treatment regimen; only patients treated with CS alone had uncontrolled disease. CYC was associated with a greater frequency of side effects (P < 0.00001).


Rates of mortality due to PAN (related or unrelated to HBV), MPA, and CSS reflected disease severity and were higher than the mortality rate in the general population (P < 0.0004). Rates of relapse, more common in MPA than HBV-related PAN patients, did not reflect disease severity. Survival rates were better among the more severely ill patients who had received first-line CYC. Based on these findings, we recommend that the intensity of the initial treatment be consistent with the severity of the disease. The use of the FFS and BVAS scores improved the ability to evaluate the therapeutic response.

At the beginning of the 1950s, the prognosis of systemic vasculitis dramatically improved with the introduction of corticosteroids (CS), increasing the 5-year survival rate from 13% without treatment to 48–57% (1). Patients with renal disease had a poorer prognosis (1–5). For vasculitides refractory to CS, Fauci et al (6) demonstrated that cyclophosphamide (CYC) further improved the prognosis. However, for polyarteritis nodosa (PAN), when CYC or other cytotoxic agents were given in combination with CS at the time of diagnosis, an advantage in terms of survival had not been demonstrated, except in 1 retrospective study (7). In a prospective study on the treatment of PAN, microscopic polyangiitis (MPA), and Churg-Strauss syndrome (CSS) conducted by the French Vasculitis Study Group (8), no survival advantage was demonstrated when CYC was given in combination with CS as first-line therapy.

The objectives of the present study were as follows: 1) to determine the long-term outcome of patients and to compare it with the overall French population; 2) to evaluate the impact, in terms of morbidity and mortality, of the type of vasculitis, the prognostic factors, and the treatments administered at the time of diagnosis; and 3) to analyze the number and severity of treatment side effects and sequelae.


Study population and study design

The 278 patients were enrolled from 1980 through 1993 in general or university hospitals participating in the French Vasculitis Study Group (see Acknowledgments for names of the investigators). Data concerning the patients were prospectively recorded in a computerized database. All patients, except 5, were randomized to 1 of 4 consecutive, prospective, multicenter therapeutic trials. Five patients had received treatment before a definitive diagnosis had been made; they were treated according to the protocol but were not randomized. These trials have been described in detail previously (8–12), and the protocols are briefly summarized below.

Protocol 1

For 68 patients who had PAN associated or not associated with hepatitis B virus (HBV) markers, MPA, or CSS (August 1980 through December 1983) (8), CS and plasma exchanges were prescribed for all patients, 31 of whom were subsequently randomized to receive CYC (2 mg/kg/day) for 1 year.

Protocol 2

For 71 patients who had PAN without HBV infection, MPA, or CSS (December 1983 through December 1988) (9), CS was prescribed for every patient and plasma exchanges were administered in a randomized manner (mean of 11 plasma exchanges for each of 32 patients).

Protocol 3

Patients who had PAN without HBV infection, MPA, or CSS (January 1989 through August 1993) were categorized into 2 subgroups: those with factors of poor prognosis (Five-Factors Score [FFS] ≥1; see explanation below) and those with factors of good prognosis. The 62 patients with poor prognostic factors were treated with CS and a monthly pulse of CYC for 1 year; after randomization, 34 received 9 plasma exchanges (3/week) (10). The 25 patients with good prognostic factors (FFS 0) were treated with CS and, after randomization, received oral CYC (n = 15) or monthly pulses of CYC (n = 10) for 1 year.

Protocol 4

For 47 patients with PAN with HBV markers (beginning in 1983) (11), treatment consisted of a combination of 2 weeks of CS followed by an antiviral agent, vidarabine (n = 39), until 1987, then interferon-alfa (n = 8) and plasma exchanges. In the case of relapse, treatment was adapted to the severity of the symptoms: CS, CYC, or other immunosuppressive agents.

Five patients were not randomized because they had been treated before a definitive diagnosis had been established. They all received CS, and 4 of them received oral CYC.

The CS regimen in protocols 1, 2, and 3 was the same. Prednisone was prescribed at a dosage of 1 mg/kg/day for 1 month, decreased by 2.5 mg every 10 days for 1 month, then decreased by 2.5 mg each week until a level equal to half the initial dose was reached. This dosage was maintained for 3 weeks, further decreased by 2.5 mg every week until patients were taking 20 mg/day, and then decreased by 1 mg every week.

Among the entire group of 278 patients, 122 received CYC as initial therapy: 50 orally and 72 via intravenous pulses. Ten patients (3 in protocol 1, 6 in protocol 2, 1 in protocol 4) received CYC therapy because of uncontrolled disease or relapse.

When our studies were initiated, the vasculitides were classified according to the Fauci classification system (13). Patients with vasculitis of the PAN group, which comprised PAN, CSS, and overlap vasculitis, were included. Diagnoses were based on histologic and/or angiographic features and, for 4 patients, on clinical criteria.

Patients have been reclassified according to the American College of Rheumatology criteria for CSS (14) and PAN (15). In addition, classic PAN and MPA were diagnosed according to the Chapel Hill Consensus Conference nomenclature (16) and our own recommendations for the diagnosis of MPA (17). Nevertheless, we cannot exclude the possibility that a few patients might have been misclassified between the PAN and MPA groups because of the limitations inherent in all classification systems (18).

Evaluation criteria

Two scores were applied to evaluate disease activity at the time of diagnosis. The 5 factors that comprise the FFS (19) are the following: classes of serum creatinine levels (≤1.58 mg/dl and >1.58 mg/dl) and proteinuria (≤1 gm/day and >1 gm/day), presence of severe gastrointestinal (GI) tract involvement, cardiomyopathy, and/or central nervous system involvement. The presence of each factor is accorded 1 point, and the score is defined as follows: 0 represents no factors present, 1 represents 1 factor present, 2 represents ≥2 factors present.

We also applied the Birmingham Vasculitis Activity Score (BVAS) (20) to test its ability to correlate with prognosis and to compare it with the FFS. The BVAS is a clinical index of disease activity based on symptoms and signs in 9 categories: systemic signs; skin; mucous membranes and eyes; ear, nose, and throat (ENT); chest; heart and vessels; GI tract; kidney; and nervous system. Maximum points were accorded to each category; the maximum score is 63. Disease features were scored only when they were attributable to active vasculitis. We used the BVAS in its first presentation form; modified forms of the BVAS have been published subsequently and can be used at successive steps of the disease (21). Our patients were evaluated according to 4 arbitrarily defined BVAS categories: scores of ≤10, 10 to ≤20, 20 to ≤30, and >30.

Uncontrolled vasculitis was defined as the occurrence of new manifestations or aggravation of manifestations already present despite treatment for the disease. Relapses were defined as recurrence of signs or new symptoms after an initial remission had been achieved; early relapses were those that occurred <12 months after the beginning of remission.

To gather information concerning the long-term survival of the patients, a questionnaire was sent to the hospital physician who had included the patients in the study and/or to their treating practitioner. The following questions were asked: 1) Is the patient dead or alive? 2) If the patient is dead, what was the cause of death? 3) Had the patient suffered a relapse(s) since inclusion in the study? If yes, what were the characteristics, date, and treatment of the relapse(s)? 4) Is the patient still receiving treatment for the vasculitis? 5) Did the patient experience treatment side effects? Answers were collected from January 1, 1996 to December 31, 1997. The end point was December 31, 1997. When the information was not obtained at the time of the inquiry or if the patient had not been seen by the physician who included the patient in the trial, the treating physician was contacted by mail or phone. When the patient was lost to followup, his or her demographic data were obtained from the French death registry, in which every French citizen's date and place of birth and of death are recorded; for ethical reasons, the cause of death is not reported.

Antineutrophil cytoplasmic antibodies (ANCA) had been sought in 118 of the 278 patients (42.4%). The technique that was used followed the recommendations of the European ANCA Assay Standardization Group (22).

Statistical analysis

Statistical analyses were carried out using SAS version 6.12 (SAS Institute, Cary, NC). The baseline characteristics of the study patients were expressed as percentages for categorical variables and as means for continuous variables. For univariate analysis, the chi-square test was used for categorical variables and Student's t-test for continuous variables. We used Spearman's correlation coefficient to determine the relationship between 2 continuous variables. Analysis of variance was performed for multiple comparisons of means and with the Bonferroni conversion for differences between means.

To compare our cohort with the general population, we used a model constructed according to the stochastic Monte Carlo method (23), which matched our control cohort of 278 persons according to year of study entry, age at study entry, sex, and duration of followup. A computerized program was designed with Visual Basic. Expected survival probabilities were calculated year-by-year on the basis of probabilities published by the Institut National de la Statistique et des Etudes Economiques (demographic data in France) (24). Survival curves were constructed according to the Kaplan-Meier method (25) using the log-rank test (26) or adjusted to prognostic factors and using the SAS Proc Lifetest, and using the log-rank test or the Cox model (27) adjusted to the prognostic factors using the SAS PROC PHREG. Survival curves were used to compare the clinical forms (CSS, MPA, PAN related or not related to HBV). These curves were plotted according to the 3 FFS classes and 4 BVAS classes described above. The survival curves evaluating the impact of CYC considered only 215 patients because those with HBV-related PAN were treated differently. The side effects were compared according to whether the patients received CYC or not. For all the statistical analyses, P < 0.05 was considered significant.


Demographics. Two hundred seventy-eight patients with a mean ± SD age of 53.8 ± 15.1 years at the time of diagnosis were included in the study; 54.3% were men and 45.7% were women. Regardless of the disease, no difference was noted between male and female patients in terms of clinical symptoms and causes of death. Of this total group, 64 had CSS, 58 had MPA, 93 had non–HBV-related PAN, and 63 had HBV-related PAN. Two hundred sixty-two of the patients (94.2%) were European Caucasians, 10 were born in North Africa, 3 were born in Asia, and 3 were black (2 Caribbean and 1 African). The patients were treated in different centers in France (see Acknowledgments). Nine patients were lost to followup after a mean of 82.3 months. The mean ± SD followup of the entire series of patients was 88.3 ± 51.9 months (range 3 days to 192 months).

The sex ratio, mean age, mean FFS and BVAS scores, deaths, and relapses for the different clinical forms of vasculitis are shown in Table 1. For the 64 patients with CSS, the ENT and chest manifestations accounted, respectively, for 3.52 ± 1.6 and 2.6 ± 1.1 BVAS points (mean ± SD). The mean BVAS, excluding ENT and chest signs (with the exception of hemorrhagic alveolitis), was 16 ± 6.4.

Table 1. Demographics and vasculitis scores of the 278 patients with CSS, MPA, and PAN
ParameterCSS (n = 64)MPA (n = 58)Non–HBV-related PAN (n = 93)HBV-related PAN (n = 63)
  • *

    Differences in the Five-Factors Score (FFS) between the group with microscopic polyangiitis (MPA) and the other 3 groups were significant (P < 0.001). No significant differences were noted between the groups with Churg-Strauss syndrome (CSS), non–hepatitis B virus (HBV)–related polyarteritis nodosa (PAN), and HBV-related PAN.

  • Differences in the Birmingham Vasculitis Activity Score (BVAS) between the group with non–HBV-related PAN and the other 3 groups were significant (P < 0.001). No significant differences were noted between the groups with CSS, MPA, and HBV-related PAN.

  • Death rates were comparable among the 4 groups.

  • §

    Relapse rates differed significantly among the 4 groups (P = 0.004).

Sex, no. females/no. males35/2928/3046/4722/41
Mean age, years51.159.553.451.9
FFS, mean ± SD*0.75 ± 0.91.86 ± 1.30.72 ± 10.95 ± 1
BVAS, mean ± SD22.1 ± 6.323.4 ± 8.415.1 ± 6.418.6 ± 7.3
Deaths, no. (%)20 (31.3)22 (37.9)22 (23.7)21 (33.3)
Deaths due to vasculitis, no.4756
Patients with relapses, no. (%)§13 (20.3)20 (34.5)18 (19.4)5 (7.9)

ANCA were sought in 118 of the 278 patients and were positive by immunofluorescence assay in 40 (14 of 30 CSS patients, 19 of 30 MPA, 7 of 35 non–HBV-related PAN, and 0 of 23 in HBV-related PAN).

Outcome. As shown by the Kaplan-Meier survival curves (Figure 1), the overall death rate of the study patients exceeded that of the general population of France, regardless of their age (P < 0.0004). A sharp drop in survival was observed within the first 18 months of followup. Afterwards, the curves tended to be parallel.

Figure 1.

Survival of vasculitis patients (Churg-Strauss syndrome, microscopic polyangiitis, and polyarteritis nodosa with and without hepatitis B virus markers; n = 278) as compared with the general population of France (P < 0.0004).

The mean ± SD FFS score for the entire cohort was 1 ± 1.13. The numbers of patients in each FFS group were 126 for an FFS of 0, 60 for an FFS of 1, 63 for an FFS of 2, 20 for an FFS of 3, 7 for an FFS of 4, and 2 for an FFS of 5. The mean ± SD BVAS score for the entire group was 19.1 ± 7.8, which was distributed as follows: 32 for a BVAS of ≤10, 130 for a BVAS of 10 to ≤20, 91 for a BVAS of 20 to ≤30, and 25 for a BVAS of >30.

Mortality was significantly associated with disease severity, as assessed by the FFS (P = 0.004) (Figure 2A) and the BVAS (P < 0.0002) (Figure 2B). Despite a tendency toward increased mortality in the MPA patients, no statistically significant difference was observed among the 4 clinical forms of vasculitis (Table 1 and Figure 3). Nevertheless, we observed more deaths during the followup of the MPA patients. Patients in this group were older at the time of diagnosis, 2 of them experienced severe relapses (108 and 120 months, respectively, after the diagnosis of vasculitis), and 2 CS-dependent patients developed sepsis due to severe immunosuppression. A correlation was established between the FFS and BVAS scores (r = 0.69, P < 0.01) (Figure 4).

Figure 2.

Survival of vasculitis patients (Churg-Strauss syndrome, microscopic polyangiitis, and polyarteritis nodosa with and without hepatitis B virus markers; n = 278) as a function of A, their Five-Factors Scores (FFS) (P = 0.004) and B, their Birmingham Vasculitis Activity Scores (BVAS) (P < 0.0002).

Figure 3.

Survival of vasculitis patients (Churg-Strauss syndrome [CSS], microscopic polyangiitis [MPA], and polyarteritis nodosa [PAN] with and without hepatitis B virus [HBV] markers; n = 278) as a function of the type of vasculitis, adjusted to the Five-Factors Score (P = 0.53) and to the Birmingham Vasculitis Activity Score (P = 0.79).

Figure 4.

Correlation between the Five-Factors Score (FFS) and the Birmingham Vasculitis Activity Score (BVAS) in 278 patients with vasculitis (Churg-Strauss syndrome, microscopic polyangiitis, and polyarteritis nodosa with and without hepatitis B virus markers) (r = 0.69, P < 0.01).

Relapses. Fifty-six patients (20.1%) experienced 84 relapses. No relationship was observed between the initial disease severity and the occurrence of relapse (mean ± SD FFS 0.94 ± 1.1 and BVAS 18.8 ± 8 in the group with relapses and FFS 1.05 ± 1.1 and BVAS 19 ± 7.8 in the group without). The first relapse occurred a mean ± SD of 31.6 ± 26.9 months after the initiation of first-line therapy. Treatment with cytotoxic agents did not prevent relapses: among the 215 patients treated with CS and CYC for CSS, MPA, or non–HBV-related PAN, 23.4% experienced a relapse, compared with 24.2% among those treated with CS alone. The time to relapse for the group treated with CYC and CS was 33.6 ± 20.8 months, and the time to relapse or uncontrolled disease for those treated with CS alone was 26.9 ± 24 months. It should be noted that the 6 patients whose disease became uncontrollable before the end of this initial treatment, between the sixth and eighth months, had received CS alone. No patient receiving CS and CYC had uncontrolled disease during the treatment period.

The number of relapses differed according to the type of vasculitis: 5 (7.9%) of the 63 patients with HBV-related PAN, 18 (19.4%) of the 93 patients with non–HBV-related PAN, 13 (20.3%) of the 64 CSS patients, and 20 (34.5%) of the 58 MPA patients experienced at least 1 relapse (P = 0.004 for the comparison of relapses versus no relapses among the 4 groups, by chi-square test with 3 degrees of freedom). The time to the first relapse was comparable for all the vasculitides: 24.6 ± 18 months in CSS, 37 ± 26 months in MPA, 29.4 ± 20.5 months in non–HBV-related PAN, and 36.6 ± 60.1 months in HBV-related PAN.

Deaths. The causes of the 85 deaths that were recorded are shown in Table 2. Four sudden deaths of unknown cause were noted during the first 6 months after the start of treatment and were considered to be related to the disease. Three patients died of heart-related conditions (sequelae of specific cardiomyopathy in 2, hypertensive heart disease in 1). Two occurrences of pulmonary emboli were observed during the first 4 months of treatment. Thirteen patients died of cancers that developed in the following sites: lung (n = 3), throat (n = 2), GI tract (2 colon, 1 stomach), liver (n = 1), prostate (n = 1), lymphatic system (n = 2), unknown (n = 1). These patients were older than the rest of the study patients at the time of study inclusion (60.3 ± 10 years); their mean ± SD duration of followup was 71.9 ± 57.6 months. One patient developed breast carcinoma but died of cardiac failure secondary to cardiomyopathy that was specifically associated with CSS.

Table 2. Causes of the 85 deaths of the vasculitis patients
Cause of deathNo. (%) of patients
Progressive vasculitis22 (25.9)
 Bowel infarction10
 Cardiac failure5
 Multivisceral involvement3
 Renal failure3
Infectious side effects of treatment11 (12.9)
 Bacterial pneumonia5
Sudden deaths9 (10.6)
Heart disease8 (9.4)
Cancer13 (15.3)
Pulmonary embolism3 (3.5)
Chronic respiratory disease3 (3.5)
Fulminant viral hepatitis2 (2.4)
Miscellaneous14 (16.5)

Two patients were treated for malignancies during followup. One had bladder cancer secondary to CYC therapy, and the other had uterine cancer considered to be independent of the vasculitis and its treatment. Both are still alive and in remission of both the vasculitis and the malignancy. The patient who had uterine cancer also developed thrombotic thrombocytopenic purpura 16 years after HBV-related PAN; the condition was attributed to the radiotherapy that had been used to treat the cancer.

Among the total of 15 patients who developed cancers, 7 received CS plus CYC, and 8 received CS alone. One of the 2 lymphoma patients received CYC.

Forty-one patients died as a consequence of their vasculitis or its treatment: 22 died of uncontrolled vasculitis, 11 of treatment side effects (sepsis), 6 of early sudden death, and 2 of sequelae of cardiomyopathy, representing 14.7% of the total population and 48.2% of all the deaths (Table 3). Eighteen of the 22 patients who died of uncontrolled vasculitis had poor prognostic factors. The mean followup of this group was 39 ± 17.4 months. Thirty (73.2%) of the 41 deaths due to the disease or its consequences occurred during the first 18 months after starting treatment.

Table 3. Patient characteristics and vasculitis scores, according to outcome*
ParameterStudy population (n = 278)Survivors (n = 193)Deaths
Overall (n = 85)Uncontrolled vasculitis (n = 22)Sepsis (n = 11)
  • *

    The patients' age, Five-Factors Score (FFS), and Birmingham Vasculities Activity Score (BVAS) were recorded at diagnosis. Values are the mean ± SD.

  • The SD is not reported because 9 of the 11 deaths occurred before 4 months and the other 2 occurred at 10 years and 11 years, respectively.

Age, years53.6 ± 15.551.3 ± 15.358.9 ± 13.652.1 ± 16.263.5 ± 9.2
FFS1 ± 1.10.9 ± 11.3 ± 1.22.3 ± 1.31.5 ± 1.1
BVAS19.1 ± 7.717.9 ± 7.521.9 ± 7.625.2 ± 9.222.1 ± 6.3
Followup, months88.5 ± 51.7106 ± 42.549 ± 42.526.6 ± 40.417.9

Influence of treatment. Survival was comparable for the 215 patients receiving either CS or CS plus CYC (Figure 5). When these patients were stratified according to FFS, those with scores ≥2 who received CYC (n = 91) benefited from significantly prolonged survival (P = 0.044). No difference was observed between the overall population and those with an FFS score of 0 or 1 (P = 0.46 and P = 0.31, respectively). Stratification according to the BVAS failed to reveal a difference in survival. Nevertheless, among the 25 patients for whom the BVAS was >30, a tendency toward significance was observed in favor of CYC (P = 0.069). Fifteen of the 22 patients who died of severe vasculitis had been treated with CS alone.

Figure 5.

Survival of patients with vasculitis (Churg-Strauss syndrome, microscopic polyangiitis, and polyarteritis nodosa without hepatitis B virus markers; n = 215) as a function of treatment with corticosteroids (CS) versus CS plus cyclophosphamide (CYC) as first-line therapy (P not significant).

Treatment side effects. Only moderate and severe side effects were recorded; mild side effects were not taken into account because they were not always reported. One hundred seventy-five treatment side effects (Table 4) were reported in 93 of the 278 patients; 79 of the 93 patients (84.9%) had received CYC. The distribution of side effects among the protocols and forms of diseases were as follows: 41% in protocol 1, 9.8% in protocol 2, 64.2% in protocol 3, and none in protocol 4 (only mild effects due to antiviral agents); 37.5% of the CSS patients, 58.6% of MPA, 29% of non–HBV-related PAN, and 7.9% of HBV-related PAN. Among the 185 patients who did not experience side effects, only 53 (28.6%) had received CYC (P < 0.0001). The mean ± SD ages of the patients with and without side effects were, respectively, 56.5 ± 12.7 years and 52.4 ± 15.5 years.

Table 4. Distribution of side effects as a function of treatment with CS with and without CYC*
Side effectNo. of events (n = 175)No. of patients with side effects
Total (n = 93)No. taking CYC (n = 79)
  • *

    Some patients experienced more than 1 side effect; therefore, the sum of each of the 2 last columns is higher than the number of patients indicated at the top of the column. CS = corticosteroids; CYC = cyclophosphamide; GI = gastrointestinal.

 Bacterial pneumonia8
 Acute bronchitis6
 Pneumocystis carinii pneumonia1
 Urinary tract infections2
 Herpes zoster infections5
 Osteomyelitis, arthritis2
 Aspergillosis, candidiasis, cryptococcosis5
Osteoporotic fractures262414
Hemorrhagic cystitis777
Cushing's disease242419
Weight gain993
GI symptoms555
Bladder cancer111

Infectious diseases and osteoporotic fractures were the 2 major events, respectively, in 13.3% and 8.6% of the entire study population. In protocol 3, in which all patients received CS and CYC, infections occurred in 28.6% (mean ± SD age 61.8 ± 10 years), and osteoporotic fractures occurred in 19% of the whole group and in 35% of the women in the group. The 11 patients who died of sepsis had taken CYC (6 oral doses and 5 intravenous pulse doses) and were older than the other patients in the series (63.5 years versus 53.6 years). Side effects were observed in 59% of patients receiving oral CYC and in 62% of those receiving pulse doses.

Sequelae were noted in 58 patients. The main sequelae were end-stage renal failure necessitating hemodialysis (n = 8; 2 of whom subsequently received renal allografts), moderate renal insufficiency (n = 7), peripheral neuropathy (n = 10), congestive heart failure (n = 5), hypertension (n = 4), chronic respiratory disease (n = 5), hemiplegia (n = 1), visual impairment (n = 1), amputation (n = 2), marked weight gain (n = 2), pain from vertebral fractures (n = 11), and inability to get out of bed (n = 2).


Several studies (6–12) have demonstrated an overall improvement in the prognosis of vasculitis and, recently, scores predicting the prognosis have been devised (19). However, to date, the impact of therapeutic strategies on subgroups of vasculitides according to the prognostic factors and the occurrence of long-term side effects has not been evaluated. We demonstrated in the present study that the overall mortality was significantly higher than that observed in the general population of France, despite the improvement in treatments. Deaths usually occurred early, within the first 18 months of followup. Later, survival curves tended to be parallel despite the occurrence of late relapses. Survival reflected the initial disease severity (Figures 2A and B), and in our opinion prognostic scores should be used in clinical trials to guide the initial treatment choice as a function of disease severity in the individual patient.

We observed a few discrepancies between the BVAS and the FFS. Although the scores were correlated (r = 0.69), for CSS the BVAS (20) overestimated the disease severity (Table 1) as a consequence of the points attributed to ENT involvement and chest involvement, despite the fact that neither maxillary sinusitis nor asthma has ever been demonstrated to be a factor of poor prognosis in CSS (19). Based on these data, the FFS was easier to use and better adapted for comparing the initial severity of the different types of vasculitides. However, the BVAS is a more precise measurement of the dissemination of the disease and could be used to evaluate followup (21), which is not the case for the FFS. The 2 scores are therefore complementary and are useful for decision-making and treatment followup.

Figure 6.

Survival of patients with vasculitis (Churg-Strauss syndrome, microscopic polyangiitis, and polyarteritis nodosa without hepatitis B virus markers and Five-Factors Scores [FFS] of ≥2; n = 91) as a function of treatment with corticosteroids (CS) versus CS plus cyclophosphamide (CYC) as first-line therapy (P = 0.044).

In our opinion, deaths due to vasculitis or its complications did not reflect the type of vasculitis but, rather, the extent of organ involvement, e.g., kidney, GI tract, or heart, or the patient's age at the time of diagnosis (in terms of treatment-related sepsis). The increased, but not significant, number of late deaths of MPA patients was partly due to more severe disease (with kidney involvement in nearly all patients), to their tendency to relapse more frequently (and thus require more treatments, which are potentially responsible for a higher number of side effects), and to their older age at study entry. Considering the entire cohort of patients, we did not observe that patients receiving CS plus CYC had better outcomes. Careful analysis of survival stratified according to prognostic factors showed that CS plus CYC significantly prolonged the survival of the more severely ill patients, who had FFS scores ≥2 (P = 0.044). A tendency toward prolonged survival was also observed for patients who received CYC and had BVAS scores >30.

Also, 15 patients taking CS alone died of uncontrolled disease. We found that CYC in combination with CS did not prevent the occurrence of the first relapse, which seemed more often to be related to the type of vasculitis and its spontaneous tendency to relapse than to the first-line therapy prescribed. Nevertheless, only patients who were being treated with CS alone had uncontrolled disease.

The causes of death in a previous French cohort of 149 patients (for the years 1958–1983) (4) were nearly identical to those seen in the present group, except for sepsis, which was lower in the present study (26.5% versus 12.9%, respectively), and can be attributed to the improved management of infections.

Side effects were frequent, being reported in at least 33.5% of the patients. Infectious events were strongly associated with the combination of CS and CYC, were more frequent in patients older than 65 years, and were responsible for 11 deaths. The increased frequency of infections in patients taking the combination of CS and CYC has previously been reported (2, 28). In a few of our patients, the vasculitis was controlled by this combination, but the treatment was responsible for infection-related deaths. Although no difference was observed between oral and pulse administration of CYC, hematuria was predominantly observed with the oral form (6 of 7 cases).

Despite recommendations to prescribe calcium and vitamin D supplements to all patients, we observed a high frequency of osteoporotic fractures, higher than that which would be expected in the general population of France (annual incidence of femoral fractures 170/100,000 females, 62/100,000 males; no data concerning vertebral fractures are available) (29). Bone demineralization can be explained by the prolonged and high-dose CS regimen given in these prospective trials, the mean age of the patients at the time of diagnosis, and the high number of menopausal women without hormone replacement therapy and/or women with CYC-induced amenorrhea. Indeed, a lower CS dose, close to that used in National Institutes of Health and European trials (30), is prescribed in the ongoing trials in an attempt to lower the frequency of osteoporosis. We also recommend preventing osteoporosis in different manners: hormone replacement therapy for menopausal women, and/or calcium, vitamin D, and bisphosphonates.

In our patients, malignancies were not clearly attributable to CYC, except for 1 case of bladder cancer. Those who developed cancer were older at the time of the diagnosis of vasculitis, the mean interval between the two diagnoses was rather long, and 5 of them had risk factors (smokers and/or heavy drinkers). Comparison of the frequencies of cancers in this series with those of the general population is difficult because there is no exhaustive registry of cancers in France. They are recorded in some districts of France and extrapolated to the entire population (31). Nevertheless, the incidence of lymphomas was higher than that of the general population.

It has been reported that, in Wegener's granulomatosis, long-term treatment with CYC was the cause of a dramatically increased number of bladder cancers, with a long latency period between the first exposure and the diagnosis of cancer (32–35). The presence of microscopic hematuria is an independent risk factor for the development of cancer. Although the total CYC dose and the duration of therapy were not clearly associated with the development of cancers, most of them occurred in patients who had received a cumulative dose that exceeded 100 gm (32). In our patients, the duration of CYC treatment did not exceed 12 months, and, since 1987, pulses have been predominantly prescribed, thereby lowering the total dose, with the systematic use of hydration and mesna. The relatively short duration of CYC therapy compared with that in Wegener's granulomatosis and the use of pulses to lower the total CYC dose could explain the low number of bladder cancers observed in our patients.

Based on the results of these studies, our group has undertaken new trials that aim to specify the indications for CYC and other immunosuppressive agents and to lower the CYC dose in patients with severe vasculitis. We also intend to establish a strategy adapted to the type of vasculitis, its severity, and its predicted (according to the FFS) outcome.


This research was conducted with the help of The Société Nationale Française de Médecine Interne (SNFMI). The following investigators participated in the study: M. Alcalay and P. Roblot (Poitiers); P. Amarenco, J. Cabane, P. Cacoub, J. Cadranel, B. Christoforov, P. Deblois, G. Deray, P. Druet, P. Even, M. Gayraud, P. Godeau, M. Krulik, F. Lioté, D. Molle, P. Pasquier, J. C. Piette, J. M. Visy, B. Wechsler, and J. M. Ziza (Paris); K. S. Ang, G. Dien, and P. Simon (St. Brieuc); P. Arlet, P. Bayle, P. Carles, O. Pourrat, and S. Ollier (Toulouse); P. Babinet and F. Lhote (St. Denis); J. Baillet, J. P. Ducroix, A. Fournier, R. Madkassi and S. Smaïl (Amiens); G. Baralis (Arles); J. Barrier and J. Y. Grolleau (Nantes); P. Bironne and J. Glowinski (Gonesse); J. P. Bouchon (Ivry); E. Bercoff, O. Deshayes, B. Desrumeaux, X. Le Loët, and B. Pinel (Rouen); P. Canton and M. Maignan (Nancy); J. P. Cassuto, P. Dujardin, J. G. Fuzibet and J. F. Quaranta (Nice); B. Coevoet (St. Quentin); P. Casassus, P. Cohen, D. Daupleix, P. Dreyfus, L. Guillevin, B. Jarrousse, P. Le Toumelin, and J. N. Hugues (Bobigny); G. Desproges-Gotteron, F. Liozon, R. Treves, and E. Vidal (Limoges); J. F. Desson (Caen); P. Doumovo (Eaubonne); Y. Echard (Montfermeil); A. Fischer (Evry); J. L. Debru, A. Franco, B. Imbert, and P. Micoud (Grenoble); B. Godeau, A. Schaeffer, J. C. Roujeau, and A. Sobel (Créteil); G. Janin (Mâcon); J. Jouquan (Brest); M. Kerjean (St. Malo); M. Imler (Strasbourg); S. Lassoued (Cahors); J. Mallecourt (Dreux); J. Maupetit (Libourne); O. Maurisset (La Roche-sur-Yon); P. Mornet (St. Germain-en-Laye); M. Mougeot Martin (Creil); J. Ninet, J. Pasquier, C. Trépo, and D. Vital-Durand (Lyon); O. Patey (Villeneuve St. Georges); A. M. Piette (Suresnes); M. Pointud (Montluçon); F. Rossi (Vendôme); H. Rousset (St. Etienne); M. Ruel (Senlis); M. Sigal Nahum (Argenteuil); and P. Veyssier (Compiègne). The coordinating center is at the Department of Internal Medicine, Hôpital Avicenne, Assistance Publique des Hôpitaux de Paris, Université Paris-Nord, Bobigny, France.