Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Rheumatology 2001;40:492±498 Analysis of factors predictive of survival based on 49 patients with systemic Wegener's granulomatosis and prospective follow-up A. Mahr, T. Girard, R. Agher1 and L. Guillevin Service de MeÂdecine Interne, HoÃpital Avicenne, Universite Paris-Nord, 125, route de Stalingrad, 93009 Bobigny Cedex and 1Service des Maladies Infectieuses et Tropicales, HoÃpital PitieÂ-SalpeÂtrieÁre, 47±83, boulevard de l'HoÃpital, 75651 Paris Cedex 13, France Abstract Objectives. This prospective study attempted to determine factors predictive of survival in systemic Wegener's granulomatosis (WG) based on 49 patients. Patients and methods. All patients had previously untreated systemic WG. Treatment was with oral or pulse cyclophosphamide plus corticosteroids. Univariate and multivariate analyses of survival were performed using 13 parameters evaluated at diagnosis. Results. The mortality rate was 37% during a mean follow-up period of 1.9 yr. Among the 13 parameters evaluated, univariate analysis selected the following factors as predictors of a poor outcome: serum creatinine 018.1 mgudl, age 057 yr, and erythrocyte sedimentation rate (ESR) 090 mmu1st h. The absence of ear, nose and throat (ENT) involvement also tended to predict a greater risk of mortality. Glomerulonephritis, when present and regardless of creatininaemia, and pulmonary involvement had no signi®cant effect. Multivariate analysis retained serum creatinine 018.1 mgudl and age 057 yr as signi®cant predictors of poor prognosis. Conclusions. Our results suggest that impaired renal function and older age are independent factors predicting poor outcome in WG. ESR proved to be a good marker of disease severity. Conversely, univariate analysis indicated that patients with ENT involvement tended to have a better outcome, suggesting a more benign evolution of granulomatous disease compared with more aggressive vasculitis. KEY WORDS: Wegener's granulomatosis, Prognostic factors. Wegener's granulomatosis (WG) is a necrotizing vasculitis of small- and medium-sized vessels de®ned by granulomatous changes of the upper and lower airways, and is frequently associated with glomerulonephritis w1x. WG is further characterized by its strong association with antineutrophil cytoplasmic antibodies (ANCA), which generally have a cytoplasmic labelling pattern (c-ANCA) in immuno¯uorescence assays and proteinase 3 as their target w2, 3x. Although initially described as a triad combining ear, nose and throat (ENT), pulmonary and renal involvement w4±6x, WG is currently recognized as a more polymorphous disorder that comprises monovisceral forms as well as extensive multiorgan disease w7x. The spontaneous 2-yr mortality rate of 93% for WG w8x has declined dramatically since the introduction of immunosuppressive therapy. The gold standard Submitted 5 May 2000; revised version accepted 14 November 2000. Correspondence to: L. Guillevin, Service de MeÂdecine Interne, HoÃpital Avicenne, 125, route de Stalingrad, 93009 Bobigny Cedex, France. 492 for WG treatment combines corticosteroids (CS) and cyclophosphamide (CYC) w9, 10x, although other, less toxic, cytotoxic drugs, such as methotrexate w11x, have been investigated. More aggressive therapy with pooled immunoglobulins or plasma exchanges might be required for WG that is refractory to standard treatment w2x. Although remission is generally obtained with immunosuppressive therapy, the relapse rate of WG remains elevated and mortality high. Evaluation of the long-term outcome of WG patients showed a median survival time of 8.5 yr w12x, and the mortality rate was more than three times higher than that of the corresponding general population w13x. Even though overall survival was 87% in a series of 158 patients with a median follow-up of 8 yr w14x, 1-yr mortality rates of up to 56% have been reported by other authors w15x. Little is known about prognostic factors and disease activity assessment in WG. Several authors have proposed renal disease w16, 17x or impaired renal function w12, 18±21x as predictors of poor outcome. The poor prognostic value of pulmonary involvement has been ß 2001 British Society for Rheumatology Predictors of survival in Wegener's granulomatosis proposed w14, 22x but could not be demonstrated statistically w20x. Older age w18, 21x, initial white blood cell count 010 000uml w20x and anaemia w21x might also be predictive of increased risk of mortality. The aim of the present study was to identify factors predictive of survival based on 49 patients newly diagnosed with systemic WG and prospective follow-up. We also attempted to correlate patient survival with initial disease activity, as assessed by the erythrocyte sedimentation rate (ESR) and the Birmingham vasculitis activity score (BVAS) w22x. Patients and methods Patient selection The study included 49 patients with newly diagnosed and untreated systemic WG. These individuals were selected from among 50 subjects who had participated in a prospective therapeutic multicentre trial conducted in France between October 1990 and December 1995. The results of that trial, which compared the effectiveness and safety of two CYC regimens for systemic WG, have been published w24x. One patient who had been included in the original trial was not evaluated for the present study because certain data were missing. All the subjects were over 15 yr old and had systemic WG that met the American College of Rheumatology criteria w25x. Histopathological evidence of either necrotizing vasculitis with granulomatous in¯ammation or segmental necrotizing glomerulonephritis was required. The patients were considered to have systemic disease based on the presence of multiorgan involvement. However, subjects with monovisceral disease and severe life-threatening disease were also eligible. Treatment The ®rst-line therapy was administered according to the study protocol w24x. This regimen combined CYC and CS. Every patient received a pulse of CYC (0.7 gum2) and was then randomly assigned to receive either i.v. (0.7 gum2) or oral (2 mgukg) CYC. Intravenous CYC was administered at 3-week intervals for 1 yr and then at longer intervals. Oral CYC was taken daily; it was maintained at full dose for 1 yr and then tapered. CYC treatment could be discontinued after 2 yr when remission had been obtained. Initially, three consecutive daily pulses of methylprednisolone (15 mgukguday) were given before oral prednisone was started at a dose of 1 mgukguday. After 6 weeks, tapering of the CS dose could be started and, when complete remission (CR) was obtained, continued until total discontinuation 12±18 months later. Because of the high frequency of Pneumocystis carinii pneumonia (PCP) among the ®rst patients recruited, all patients were given co-trimoxazole (400 mguday) as PCP prophylaxis. When only partial remission (PR) was obtained, plasma exchanges were added to the assigned treatment. Minor relapses were managed by transient intensi®cation of the CS dose until the attenuation of the symptoms, whereas the CYC 493 dose was increased for severe relapses. Alternatively, patients who relapsed while receiving i.v. CYC were switched to oral CYC. In the case of complete treatment failure or the need for treatment after the completion of the study regimen, the therapeutic strategy was left to the discretion of the treating physician. Methods The medical charts of the 49 patients enrolled in the original study were reviewed in January 1998. All the patients had undergone an in-depth evaluation of their disease at the time of diagnosis and bene®ted from close monitoring during the therapeutic trial. After achievement of the trial end-points, further medical information about each patient's disease course was obtained by questioning the treating physicians, telephone contact with the patients or family members, or by consulting the local death registries. The following information was recorded for every patient at the time of diagnosis: clinical ENT and cutaneous involvement; ophthalmological involvement; peripheral neuropathy (based on electromyography results); glomerulonephritis (de®ned as abnormal urinary sediment anduor serum creatinine 015.8 mgudl anduor proteinuria of >0.5 guday); pulmonary involvement (de®ned as abnormal thoracic imaging in the absence of concomitant infectious pneumopathy). Furthermore, we assessed the following biological parameters at diagnosis: serum creatinine level, ESR (mmu1st h) and ANCA immuno¯uorescence assay results. Each patient's outcome was evaluated with regard to the response to treatment, relapses, severe infectious complications and death. CR was de®ned as the complete absence of disease activity and PR as the stabilization of disease activity. To be considered severe, an infectious disease necessitated hospitalization. Statistical analysis Patient survival was assessed by life-table analysis. Because all the patients had acute-onset systemic disease which resulted in the rapid initiation of treatment after diagnosis, we confounded the dates of diagnosis and treatment onset, and survival time was calculated from the date of inclusion in the therapeutic trial. The end-point of the present study was de®ned as the last time the patient's status could be determined or as the time of death. Survival curves were plotted using the Kaplan±Meier method w26x. The potential predictive value for survival of the following parameters assessed at diagnosis was investigated: age; sex; ENT involvement; pulmonary involvement; glomerulonephritis; peripheral neuropathy; ophthalmological involvement; cutaneous involvement; number of organs affected; serum creatinine level; ESR; the presence of ANCA; and BVAS. Univariate analysis was performed using log-rank tests w27x. To increase the power of univariate analysis, quantitative parameters (age, serum creatinine level, number of organs affected, ESR, BVAS) were also investigated as dichotomous qualitative parameters, 494 A. Mahr et al. using as the dividing threshold the median values observed for the whole population or, for the serum creatinine value, the median value observed for the patients with glomerulonephritis. The multivariate analysis used the Cox proportional hazards regression model w28x. The multivariate model included all the parameters that had a P value of <0.20 in univariate analysis. Then, a backward stepwise procedure was used to remove from the original model those parameters that did not signi®cantly explain patient survival according to a likelihood ratio test. To minimize a potential confounding bias in the survival analysis attributable to the difference in the initial CYC treatment modality (oral or i.v. administration), the mode of CYC administration was introduced as an adjusting parameter into the multivariate model. Statistical analysis was computed using the SAS statistical package, version 6.12 (SAS Institute, Cary, NC, USA). For all statistical analyses, P values of <0.05 were considered to be signi®cant. All con®dence intervals were calculated at the 95% level. Results Patient data All the patients had been diagnosed between October 1990 and March 1994. Principal data at diagnosis for the 49 patients are summarized in Table 1. Although TABLE 1. Principal demographic, clinical, biological and immunological data at diagnosis for 49 patients with systemic WG Characteristic Median age (yr) Sex Male Female ENT involvement Pulmonary involvement Glomerulonephritis ELK involvement w7x E L K EL EK LK ELK Ophthalmological involvement Peripheral neuropathy Cutaneous involvement Median serum creatinine level (mgudl) Patients (n = 47) Patients with glomerulonephritis (n=36) Median ESR (mmu1st h) (n = 44) Median BVAS ANCA positivity c-ANCA p-ANCA Undetermined Median number of organs affected Value 57 (range 23±78) 30 19 36 39 36 (61%) (39%) (73%) (80%) (73%) 2 4 3 7 5 6 22 16 (33%) 11 (22%) 18 (37%) 16.9 (range 7.82169.5) 18.1 (range 7.92169.5) 90 (range 52140) 18 (range 4±33) 43 (88%) 37 (76%) 4 (8%) 2 (4%) 4 (range 1±6) ELK, ear, lung and kidney; p-ANCA, ANCA giving a perinuclear labelling pattern in the immuno¯uorescence assay. monovisceral disease with ENT or pulmonary involvement were each diagnosed in one subject, both of these patients were considered to have systemic WG as they had severe general symptoms bearing a high risk of fatal outcome. The most prominent radiological ®ndings in the 39 patients with pulmonary involvement were nodules (n = 29), pneumonia-like condensations (n = 2), alveolar haemorrhage (n = 3), bilateral in®ltrates without evidence of alveolar haemorrhage (n = 3), and pleural effusion (n = 2). Glomerulonephritis was present in 36 patients and consisted of proteinuria >0.5 guday (n = 27), abnormal urinary sediment (n = 32) or serum creatinine 015.8 mgudl (n = 25); histological evidence of necrotizing glomerulonephritis was obtained for 26 subjects. Treatment Initially, all the patients received CYC and CS. CYC was administered i.v. to 28 patients and taken orally by 21 patients. After PR to the assigned treatment, four patients transiently underwent plasma exchanges and one patient received i.v. immunoglobulins. During the course of the study, 10u28 patients receiving i.v. CYC were switched to oral CYC because of unresponsiveness (n = 8) or side-effects (vomiting, n = 1; haemorrhagic cystitis, n = 1). Azathioprine was prescribed after a severe reaction to CYC (toxic epidermal necrolysis, n = 1), and for a relapse occurring after discontinuation of CYC (n = 1). Two patients received methotrexate, one in combination therapy with cyclosporin because of unresponsiveness to CYC and the other for a late relapse after discontinuation of CYC. For one patient, CYC was switched to i.v. immunoglobulins because of haematological toxicity. Eight patients required dialysis because of renal failure, and four patients died while on dialysis for acute renal failure. One patient was able to stop dialysis after recovery of renal function while three other patients required chronic dialysis for end-stage renal disease. Follow-up The mean follow-up was 1.9 yr (median 1.8 yr, range 2 days to 5.8 yr). Initial CR was achieved by 23 (47%) patients and PR by 20 (41%) patients; six (12%) patients had disease refractory to therapy. Twenty-one (43%) patients relapsed after a mean interval of 16 months (range 4±50 months). Nineteen of these patients experienced a single relapse; the remaining two each had two relapses. At the time of the last update, 21 (43%) patients were in CR and 10 (20%) patients in PR. Eighteen (37%) patients died during follow-up. The status of the 31 survivors was last determined within the last 12 months for six patients (19%), 24 months for two (6%), 36 months for two (6%), 36 months for 13 (42%), 48 months for 5 (16%), 60 months for 2 (6%), and 72 months for one (3%). Severe infectious complications occurred in 19 (39%) patients. The 31 severe infectious events recorded comprised 10 cases of PCP, three of bacterial pneumonia, ®ve of invasive aspergillosis, four of cytomegalovirus Predictors of survival in Wegener's granulomatosis 495 FIG. 1. (A) Overall survival curve for the 49 patients with systemic WG. (B±D) Survival rate as a function of (B) serum creatinine level (P = 0.03), (C) age (P = 0.04) and (D) ESR (P = 0.02) at diagnosis. pneumonia, one of cytomegalovirus retinitis, one of meningeal tuberculosis, one of pulmonary tuberculosis, two of herpes zoster, one of papovavirus leucoencephalitis, one of Serratia marcescens arthritis, one of Listeria monocytogenes septicaemia and one of septic shock of undetermined origin. The overall mortality was 37% (18u49 patients). The mean survival rates were 77.5 " 11.7% (95% con®dence interval) at 6 months and 67.5 " 13.7% at 2 yr of follow-up (Fig. 1A). The causes of death are detailed in Table 2, and were attributed to active vasculitis in six, severe infectious complication in seven and miscellaneous conditions in ®ve cases. Univariate analysis According to univariate analysis (Table 3), the presence of the following parameters at diagnosis was signi®cantly associated with shorter survival: serum creatinine level 018.1 mgudl (P = 0.03) (Fig. 1B), age 057 yr (P = 0.04) (Fig. 1C), and ESR 090 mmu1st h (P = 0.02) (Fig. 1D). Multivariate analysis The following parameters were selected for the multivariate analysis: age < or 057 yr, serum creatinine < or 018.1 mgudl, ENT involvement, cutaneous 496 A. Mahr et al. involvement and pulmonary involvement. All these variables except the latter had P values <0.20 in univariate analysis. Pulmonary involvement was, nevertheless, retained for the multivariate analysis because it had been reported previously to be a major prognostic factor in WG w14, 22x. Among the parameters selected for multivariate analysis, only serum creatinine 018.1 mgudl (P = 0.02) and age 057 yr (P = 0.02) were retained as independent factors signi®cantly predictive of poor prognosis (Table 4). Pulmonary involvement, ENT involvement and cutaneous involvement had been removed from the multivariate model using a backward stepwise procedure as their P values did not reach signi®cance. TABLE 2. Causes of death for 18 patients with systemic WG Patient Time after diagnosis 1 2 3 4 3 months 14 months 1 months 4 months 5 6 7 8 9 10 11 12 13 14 15 16 3 months 1 months 3 months 27 months 17 months 3 months 2 days 30 months 3 months 33 months 2 days 14 months 17 5 months 18 5 months Cause of death PCP Papovavirus multifocal leucoencephalitis Haemoptysis of unknown origin PCP and bacterial pneumonia, acute renal failure Cerebral haemorrhage Unknown Septic shock Unknown Alveolar haemorrhage PCP and CYC-induced pneumopathy Alveolar haemorrhage Pulmonary haemorrhage Suicide (chronic psychosis) Acute renal failure, bacterial pneumonia Unknown Cerebral haemorrhage, multifactorial thrombopenia PCP and bacterial pneumonia, invasive aspergillosis PCP and cytomegalovirus pneumonia, invasive aspergillosis TABLE 3. Univariate analysis of factors associated with survival of 49 patients with systemic WG Parameter Sex Age Age < or 057 yr ENT involvement Pulmonary involvement Glomerulonephritis Peripheral neuropathy Cutaneous involvement Ophthalmological involvement Serum creatinine Serum creatinine < or 018.1 mgudl Number of organs affected < or 04 One or more infectious complications ESR ESR < or 090 mmu1st h BVAS ( or >18 Values in bold are statistically signi®cant. Discussion According to the results of these univariate and multivariate analyses of factors predicting survival based on 49 patients with systemic WG, we conclude that the following parameters, when present at diagnosis, are independent predictors of poor outcome: impaired renal function, with a serum creatinine threshold value of 18.1 mgudl; and older age, with a threshold value of 57 yr. Furthermore, the signi®cant association between ESR and survival suggested that ESR could be a serological marker of prognosis in WG. No relationship could be established between any of the other demographic, clinical or biological parameters and patient survival. In particular, no signi®cant prognostic value could be attributed to pulmonary or renal disease when present and regardless of the serum creatinine level. Renal disease has been advocated as a predictor of poor outcome in WG since Carrington and Liebow w16x described a non-renal `limited' form of WG characterized by a better prognosis. The pejorative impact of kidney involvement on prognosis was con®rmed by Luqmani et al. w17x, who noted particularly poor outcomes for patients with serum creatinine >56.5 mgudl. In our analysis, only those patients whose renal involvement was associated with functional impairment were at greater risk of death. These ®ndings suggest that the prognosis might not depend on glomerulonephritis itself but on the deterioration of renal function. Similar ®ndings were reported by other investigators w18, 21x. The serum creatinine concentration associated with shortened survival in our cohort was 18.1 mgudl, and therefore lower than thresholds reported previously w17, 21x. An age of 057 yr at diagnosis was also a marker of poorer prognosis. This ®nding is in agreement with the results of previous univariate analyses which identi®ed ages of 060 w18, 29x and 50 yr w21x as predictors of poorer outcome. It is not yet clear whether the increased mortality of the older patients re¯ects more aggressive disease or whether it is the consequence of greater sensitivity to the toxicity of immunosuppressants in this category of patients. Krafcik et al. w29x demonstrated P 0.94 0.26 0.04 0.09 0.29 0.39 0.56 0.07 0.92 0.55 0.02 0.84 0.38 0.07 0.02 0.40 TABLE 4. Multivariate analysis, using Cox proportional hazards regression model, of factors associated with survival based on 49 patients with systemic WG Parameter Serum creatinine 018.1 mgudl Age 057 yr P Relative risk 95% con®dence interval 0.02 3.5 2.5±4.5 0.02 3.6 2.5±4.6 The model included the following variables determined at diagnosis: serum creatinine 018.1 mgudl; presence of pulmonary involvement; absence of ENT involvement; absence of cutaneous involvement; and age 057 yr. Variables that did not affect survival signi®cantly were removed by a backward stepwise procedure according to a likelihood ratio test. Results were adjusted according to the initial CYC regimen (i.v. or oral). Predictors of survival in Wegener's granulomatosis increased mortality attributable to infectious complications in the elderly, even though the overall incidence of infections was similar to that of younger patients. We also found a strong association between the ESR at diagnosis and patient survival. This ®nding suggests that ESR could be a good marker of disease severity in WG. On the other hand, we could not establish a relationship between the anatomical extent of the disease and survival, and there was, in particular, no evidence of shortened survival of patients with multiorgan involvement compared with those having less extensive disease. We therefore think that the intrinsic activity of WG might be re¯ected more accurately by markers such as the serum creatinine level and ESR rather than by more or less extensive disease. Similarly, disease activity assessment with BVAS could not be statistically correlated to survival. This is consistent with previous ®ndings, as BVAS relies primarily on the cumulative evaluation of the anatomical extent of the disease w23x. Therefore, on the basis of our results, disease activity scoring with the BVAS does not seem appropriate for the assessment of prognosis in WG. The overall mortality rate of 37% in our series is higher than that reported by other authors w15x. In particular, we noted a substantial early mortality rate, as 22.5% of the patients died within the ®rst 6 months of follow-up. Two deaths occurred only a few days after diagnosis and treatment onset. This higher mortality might be explained by the fact that we studied exclusively subjects with systemic WG. As discussed elsewhere w24x, the multicentre enrolment of the individuals, who were treated in primary, secondary and tertiary care centres, could have resulted in the inclusion of a higher number of critically ill patients in our cohort. It is noteworthy that, according to our univariate analysis, patients with ENT symptoms tended to have a better outcome than patients without such involvement. Even though this observation was not con®rmed by the multivariate analysis, this ®nding merits further examination. As mentioned above, two distinct pathological features are associated in WG: granulomatosis, predominantly involving the airways, and vasculitis, predominantly localized in the kidney. Although the precise signi®cance of the coexistence of these pathological ®ndings is unclear, the mechanisms leading to granuloma formation or vasculitis might be different. Indeed, it has been emphasized that the clinical heterogeneity of WG re¯ects a broad spectrum of disorders ranging from predominantly granulomatous disease at one end to predominantly vasculitic disease at the other w30x. It has also been suggested that granuloma formation occurs early during the course of WG, whereas the vasculitis might be a marker of later stages of the disease w31x. Furthermore, immunological studies suggest that these two pathological hallmarks of WG might correspond to different immune responses involving T cells w32, 33x. Consequently, our ®nding suggesting that ENT involvement might be a potential predictor of a better outcome could be explained by a possibly more benign course of granulomatous WG 497 than vasculitic WG. Further studies are needed to assess adequately the potential prognostic value of ENT involvement, which, in turn, might provide further insight into the pathogenesis of WG. The prognostic scoring of WG remains challenging. Certainly, the effectiveness of WG therapy could be improved by the choice of appropriate treatment guided by the severity of the disease. In this strategy, aggressive disease would be managed with intensive therapy while less toxic drugs could suf®ce for more benign disease. This therapeutic approach might enhance treatment ef®cacy in the severely ill patient and reduce therapy-associated toxicity in the milder forms. Further studies are needed to explore in greater detail potential prognostic factors in WG, and then to validate them. Acknowledgements The following individuals contributed to the study: P. Bielefeld, P. Bindi, O. BleÂtry, D. Boutin, M. Brouillard, J. Cadranel, P. Camus, P. Cohen, Y. Cohen, B. Combe, P. Congy, J. F. Cordier, A. Dallot, P. David, J. F. Desson, F. DeÂtreÂe, A. Dubois, D. Glotz, D. Grenet, E. Hachulla, D. Halloun, B. Hoen, D. Jacomy, B. Jarrousse, C. Jacquot, J. P. Jaulin, M. F. Kahn, A. Lassoued, D. Lauque, R. Leblay, P. Leclerc, P. Lesavre, E. Le Guen, H. Lena, F. Lhote, M. Longy-Boursier, D. MeÂchali, P. Miossec, E. MonchaÃtre, J. C. Piette, P. ReÂmy, J. Ribstein, J. Rossert, I. Royer, A. Sacchi, B. Saugier, C. Seigneuric and M. Stern. References 1. Jennette JC, Falk RJ, Andrassy K et al. Nomenclature of systemic vasculitides. Proposal of an international consensus conference. Arthritis Rheum 1994;37:187±92. 2. Savage COS, Harper L, Adu D. Primary systemic vasculitis. Lancet 1997;349:553±8. 3. Jennette JC, Falk RJ. Small-vessel vasculitis. N Engl J Med 1997;337:1512±23. È ber 4. Wegener F. U generalisierte, septische GefaÈsserkrankungen. Verh Dtsch Ges Pathol 1936;29: 202±10. È ber eine eigenartige rhinogene Granulo5. Wegener F. U matose mit besonderer Beteiligung des Arteriensystems und der Nieren. Beitr Pathol Anat Allg Pathol 1939; 36±68. 6. Godman GC, Churg J. Wegener's granulomatosis: pathology and review of the literature. Arch Pathol 1954; 58: 533±53. 7. De Remee RA, McDonald TJ, Harrison EG Jr, Coles DT. Wegener's granulomatosis: anatomic correlates, a proposed classi®cation. Mayo Clin Proc 1976;51:777±81. 8. Walton EW. Giant-cell granuloma of the respiratory tract (Wegener's granulomatosis). Br Med J 1958;2:265±70. 9. Novack SN, Pearson CM. Cyclophosphamide therapy in Wegener's granulomatosis. N Engl J Med 1971;284:938±42. 10. Fauci AS, Wolff SM. Wegener's granulomatosis: studies in 18 patients and a review of the literature. Medicine (Baltimore) 1973;52:535±61. 11. Sneller MC, Hoffman GS, Talar-Williams C, Kerr GS, Hallahan CW, Fauci AS. An analysis of 42 Wegener's 498 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. A. Mahr et al. granulomatosis patients treated with methotrexate and prednisone. Arthritis Rheum 1995;38:608±13. Anderson G, Coles ET, Crane M et al. Wegener's granuloma. A series of 265 British cases seen between 1975 and 1985. A report by a sub-committee of the British Thoracic Society Research Committee. Q J Med 1992;83:427±38. Matteson EL, Gold KN, Bloch DA, Hunder GG. Longterm survival of patients with Wegener's granulomatosis from the American College of Rheumatology Wegener's granulomatosis classi®cation criteria cohort. Am J Med 1996;101:129±34. Hoffman GS, Kerr GS, Leavitt RY et al. Wegener's granulomatosis: an analysis of 158 patients. Ann Intern Med 1992;116:488±98. Brandwein S, Esdaile J, Danoff D, Tannenbaum H. Wegener's granulomatosis. Clinical features and outcome in 13 patients. Arch Intern Med 1983;143:476±9. Carrington CB, Liebow AA. Limited forms of angiitis and granulomatosis of Wegener's type. Am J Med 1966; 41:497±527. Luqmani RA, Bacon PA, Beaman M et al. Classical versus non-renal Wegener's granulomatosis. Q J Med 1994;87:161±7. Vassallo M, Shepherd RJ, Iqbal P, Feehaly J. Agerelated variations in presentation and outcome in Wegener's granulomatosis. J R Coll Physicians Lond 1997;31: 396±400. Romas E, Murphy BF, D'Apice AJF, Kennedy JT, Niall JF. Wegener's granulomatosis: clinical features and prognosis in 37 patients. Aust NZ J Med 1993;23:168±75. Briedigkeit L, Kettritz R, GoÈbel U, Natusch R. Prognostic factors in Wegener's granulomatosis. Postgrad Med J 1993;69:856±61. Leà Thi Huong Du, Wechsler B, De Gennes C et al. Aspects eÂvolutifs et pronostiques de la granulomatose de Wegener. Rev Rhum 1989;56:583±8. 22. Fauci AS, Haynes BF, Katz P, Wolff SM. Wegener's granulomatosis: prospective clinical and therapeutic experience with 85 patients for 21 years. Ann Intern Med 1983;98:76±85. 23. Luqmani RA, Bacon PA, Moots RJ et al. Birmingham vasculitis activity score (BVAS) in systemic necrotizing vasculitis. Q J Med 1994;87:671±8. 24. Guillevin L, Cordier JF, Lhote F et al. A prospective, multicenter, randomized trial comparing steroids and pulse or oral cyclophosphamide in generalized Wegener's granulomatosis. Arthritis Rheum 1997;40:2187±98. 25. Leavitt RY, Fauci AS, Bloch DA et al. The American College of Rheumatology 1990 criteria of the classi®cation of Wegener's granulomatosis. Arthritis Rheum 1990; 33:101±7. 26. Kaplan E, Meier P. Non parametric estimation from incomplete observations. J Am Stat Assoc 1958;53: 457±81. 27. Peto R, Pike M, Armitage P. Design and analysis of randomized clinical trials requiring prolonged observations of each patient. Br J Cancer 1977;35:1±39. 28. Cox DR. Regression models and life tables. J R Stat Soc 1972;34:187±219. 29. Krafcik SS, Covin RB, Lynch JP 3d, Sitrin RG. Wegener's granulomatosis in the elderly. Chest 1996;109:430±7. 30. Bacon PA. Vasculitis from the rheumatologist's viewpoint. In: Gross WL, ed. ANCA-associated vasculitides: immunological and clinical aspects. New York: Plenum Press, 1993:185±95. 31. Wegener F. Wegener's granulomatosis. Eur Arch Otorhinolaryngol 1990;247:133±42. 32. Gross WL, Trabandt A, Csernok E. Pathogenesis of Wegener's granulomatosis. Ann MeÂd Interne (Paris) 1998; 149:280±6. 33. Csernok E, Trabandt A, Muller A et al. Cytokine pro®les in Wegener's granulomatosis: predominance of type 1 (Th1) in the granulomatous in¯ammation. Arthritis Rheum 1999;42:742±50.