Article Text

Extended report
Long-term patient survival in ANCA-associated vasculitis
  1. Oliver Flossmann1,
  2. Annelies Berden2,
  3. Kirsten de Groot3,
  4. Chris Hagen4,
  5. Lorraine Harper5,
  6. Caroline Heijl6,
  7. Peter Höglund6,
  8. David Jayne7,
  9. Raashid Luqmani8,
  10. Alfred Mahr9,
  11. Chetan Mukhtyar10,
  12. Charles Pusey11,
  13. Niels Rasmussen12,
  14. Coen Stegeman13,
  15. Michael Walsh14,
  16. Kerstin Westman6,
  17. for the European Vasculitis Study Group
  1. 1Renal Unit, Royal Berkshire Hospital, Reading, UK
  2. 2Leiden University Medical Center, Leiden, The Netherlands
  3. 3Klinikum Offenbach, Offenbach, Germany
  4. 4Meander Medical Center, Amersfoort, Netherlands
  5. 5College of Immunity and Infection, University of Birmingham, Birmingham, UK
  6. 6University of Lund, Skane University Hospital, Sweden
  7. 7Addenbrooke's Hospital, Cambridge, UK
  8. 8NIHR Musculoskeletal Biomedical Research Unit, Nuffield Orthopaedic Centre, University of Oxford, Oxford, UK
  9. 9Hôpital Cochin, Paris, France
  10. 10Norfolk and Norwich University Hospital, Norwich, UK
  11. 11Imperial College, London, UK
  12. 12Rigshospitalet, Copenhagen, Denmark
  13. 13University Hospital Groningen, Netherlands
  14. 14Department of Clinical Epidemiology and Biostatistics, McMaster University Hamilton, Canada
  1. Correspondence to Dr Oliver Flossmann, Royal Berkshire Hospital, Renal Unit, London Road, Reading, Berkshire RG1 5AN, UK; oflossmann{at}doctors.org.uk

Abstract

Background Wegener's granulomatosis and microscopic polyangiitis are antineutrophil cytoplasm antibodies (ANCA)-associated vasculitides with significant morbidity and mortality. The long-term survival of patients with ANCA associated vasculitis treated with current regimens is uncertain.

Objective To describe the long-term patient survival and possible prognostic factors at presentation in an international, multicentre, prospectively recruited representative patient cohort who were treated according to strictly defined protocols at presentation and included the full spectrum of ANCA-associated vasculitis disease.

Methods Outcome data were collected for 535 patients who had been recruited at the time of diagnosis to four randomised controlled trials between 1995 and 2002. Trial eligibility was defined by disease severity and extent, covered the spectrum of severity of ANCA-associated vasculitis and used consistent diagnostic criteria. Demographic, clinical and laboratory parameters at trial entry were tested as potential prognostic factors in multivariable models.

Results The median duration of follow-up was 5.2 years and 133 (25%) deaths were recorded. Compared with an age- and sex-matched general population there was a mortality ratio of 2.6 (95% CI 2.2 to 3.1). Main causes of death within the first year were infection (48%) and active vasculitis (19%). After the first year the major causes of death were cardiovascular disease (26%), malignancy (22%) and infection (20%). Multivariable analysis showed an estimated glomerular filtration rate <15 ml/min, advancing age, higher Birmingham Vasculitis Activity Score, lower haemoglobin and higher white cell count were significant negative prognostic factors for patient survival.

Conclusion Patients with ANCA-associated vasculitis treated with conventional regimens are at increased risk of death compared with an age- and sex-matched population.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Wegener's granulomatosis (WG) and microscopic polyangiitis (MPA) are primary vasculitides of unknown cause characterised by necrotising inflammation involving predominantly small blood vessels and the presence of circulating antineutrophil cytoplasm antibodies (ANCA) in the majority of patients at diagnosis. They are often grouped as ‘ANCA-associated vasculitis’ because of striking similarities in histology, absence of immune deposits, response to treatment and the likely contribution of ANCA to the pathogenesis.

The introduction of immunosuppression with cyclophosphamide and corticosteroids in the 1960s led to a dramatic improvement in prognosis, with over 90% of patients achieving remission compared with a mortality of 80% at 1 year in untreated patients.1 2 The addition of plasma exchange to immunosuppression showed promise in improving renal recovery in early small studies.3 However, these advances were achieved at the cost of treatment-related toxicity such as haemorrhagic cystitis, bladder cancer, lymphoproliferative disease, myelodysplasia and infertility due to the cumulative exposure to cyclophosphamide.4 5

The European Vasculitis Study Group (EUVAS) standardised disease definitions, diagnostic criteria and disease assessment and defined four disease stages according to severity and extent at presentation.6 Four randomised controlled trials were launched to study treatment regimens tailored to different disease stages in patients with newly diagnosed ANCA-associated vasculitis.7,,10

Despite advances in the treatment of ANCA-associated vasculitis, retrospective studies limited to patients with WG suggest that there still is an excess mortality compared with the general population.11 12

This study aimed to describe the long-term patient survival and possible prognostic factors at presentation in an international, multicentre, prospectively recruited representative patient cohort who were treated according to strictly defined protocols at presentation and included the full spectrum of ANCA-associated vasculitis disease severity.

Patients and methods

Study population

The patients were recruited into four randomised therapeutic multicentre trials organised by the EUVAS in 70 general and university hospitals in 15 countries between 1995 and 2002.7,,10 All studies were approved by the local ethics committees and all patients gave written informed consent. The trials were conducted according to the 1964 Declaration of Helsinki and subsequent amendments.

Patients were eligible if they had a new diagnosis of WG or MPA according to criteria adapted from the 1994 Chapel Hill disease definitions.13 The diagnosis was based on a clinical presentation compatible with ANCA-associated vasculitis and substantiated by a positive ANCA serology and/or histology.14 Patients were excluded if they had a coexistent multisystem autoimmune condition, concurrent malignancy, active infection, pregnancy or were aged <18 or >80 years.

Entry into the different trials was dependent on the disease stage, as defined by severity subgroup.14 Patients with life-threatening pulmonary haemorrhage within 24 h of presentation were excluded. The protocols for the different trials are summarised in table 1.

Table 1

Summary of trials

Baseline evaluation

Disease activity was assessed with the Birmingham Vasculitis Activity Score (BVAS).15 Laboratory investigations included serum creatinine, full blood count, C-reactive protein and ANCA. Type of ANCA was categorised as proteinase 3 ANCA (PR3-ANCA) if there was positivity by PR3-ANCA ELISA or a cytoplasmic ANCA (cANCA) pattern by indirect immunofluorescence microscopy; as myeloperoxidase ANCA (MPO-ANCA) if there was positivity by MPO-ANCA ELISA or a perinuclear ANCA (pANCA) pattern on immunofluorescence microscopy, or as double positive if both PR3-ANCA and MPO-ANCA were detected. These data were recorded as baseline data at trial entry.

The estimated glomerular filtration rate (eGFR) was calculated using the four-factor Modification of Diet in Renal Disease study equation (MDRD) formula.16

Questionnaire

This long-term follow-up was performed in accordance with the principles laid down in the 1964 Declaration of Helsinki and subsequent amendments, and ethical approval was obtained by national and local ethics committees in accordance with national legislation. In order to determine long-term survival a questionnaire was sent to participating doctors. Data collected were date of last doctor's visit, survival status and date of death. In cases of deaths the local investigators were asked to provide information about the cause of death and contributing factors according to death certificate if available and in their own opinion. They were also asked to rate separately the likelihood that active vasculitis, immunosuppression or sepsis contributed to the death on a scale from 0 (not related) to 5 (definite).

Two authors (KW, OF) independently categorised the cause of death as being directly attributable to or contributed to by vasculitis, infection, malignancy, cardiovascular, miscellaneous and unknown. In cases of disagreement a third author (DJ) adjudicated.

Replies were collected from September 2004 until January 2007.

Statistical analysis

Single imputation was done to capture missing values.17 Continuous variables are expressed as median with 25th and 75th centiles. Categorical variables are presented as percentages and frequencies. Survival was calculated using the Kaplan–Meier method18 and presented graphically for the whole cohort. The log-rank test was used to compare the overall patient mortality with that of a control cohort matched for age, sex, year and country calculated by the Hakulinen method.19 Sources for these controls were official vital statistics for participating countries: Belgium, Czech Republic, Germany, Denmark, Spain, Finland, France, England and Wales, Italy, the Netherlands, Poland, Sweden and Scotland. For a few countries from which we could not obtain information, data were retrieved from a neighbouring country. For each year and country we obtained the average number of inhabitants and death rates by age groups and gender for these groups. The following predetermined parameters were included in a multivariable Cox regression: age at entry into clinical trial, sex, clinical diagnosis (WG, MPA), ANCA-type (PR3-ANCA, MPO-ANCA, double positive ANCA, negative), eGFR (MDRD4), BVAS, haemoglobin (Hb), white cell count (WBC), platelets and C-reactive protein. Patients' eGFR was classified into stages 1–5 with the same cut-off points as chronic kidney disease stages.20 For the analysis, these were trichotomised into stages 1–2, 3–4 and 5, respectively. The proportional hazards assumption was tested by the weighted scaled Schoenfeld residuals test. We included interaction terms between haemoglobin, WBC and platelets and log time (subsequently referred to as ‘time-dependent covariates’) in order to satisfy the proportional hazards assumption. This model was applied to the patient material both without and with subtraction of general mortality in a control cohort matched for age, sex, year and country using the Hakulinen method.

Data were analysed using SAS (SAS Institute, Cary, North Carolina, USA) version 8.2 and the R software (R foundation for Statistical Computing, Vienna, Austria) version 2.6.2. A two-tailed p value of ≤0.05 was considered statistically significant.

Results

Demography

Five hundred and thirty-five patients with a median (25th–75th centile) age of 61 (49–69) at diagnosis participated in the trials; 54% of whom were male. Of these patients 281 (53%) were diagnosed with WG and 254 (47%) with MPA. The duration of follow-up ranged from 1 day to 11.46 years, with a median length of follow-up of 5.16 years for all patients and a median of 5.95 years for patients alive at the last visit/end of follow-up.

Further demographic data are summarised in table 2.

Table 2

Descriptive statistics (median, lower and upper quartile) at baseline by dichotomised entry diagnosis microscopic polyangiitis (MPA) and Wegener's granulomatosis (WG)

Survival

Cumulative survival at 1, 2 and 5 years was 88% (95% CI 86% to 91%), 85% (95% CI 82% to 88%) and 78% (95% CI 75% to 82%), respectively. The calculated survival in the matched population was 98%, 97% and 92%, respectively. The overall mortality ratio among the patients compared with the controls was 2.6 (95% CI 2.2 to 3.1) and was significant by log-rank test (p<0.0001). Although the mortality ratio was highest in the first year it was still significantly raised for patients with ANCA-associated vasculitis in subsequent years with a ratio of 1.3 (95% CI 1.04 to 1.65) (figure 1).

Figure 1

Patient survival overall (solid line) with 95% CI (dashed lines) compared with a matched general population (dot/dash line).

Cause of death

A total of 133 (25%) deaths occurred during the observation period. The causes of death are summarised in table 3. Fifty-nine of the 133 deaths (44%) occurred within 1 year of enrolment. The cause of death differed between patients who died within the first year after enrolment and patients who died later (see table 3). Patients in the first year were most likely to die either of infection (48%) or active vasculitis (19%), whereas patients after 1 year died mostly of cardiovascular disease (26%), malignancy (22%) or infection (20%).

Table 3

Causes of death within and after the first year of follow-up, respectively

Predictors of death: multivariable analysis

In the multivariable Cox regression the following entry parameters were significant predictors of mortality: advancing age, an eGFR <15 ml/min, higher BVAS, lower haemoglobin and the time interaction for a higher WBC (figure 2).

Figure 2

Multivariable analysis (Forrester plot). ANCA, antineutrophil cytoplasmic antibodies; BVAS, Birmingham Vasculitis Activity Score; CKD, chronic kidney disease; CRP, C-reactive protein; MPO, myeloperoxidase; PR3, proteinase 3; WBC, white blood cell count.

In the multivariable Cox regression where the mortality of a control cohort matched for age, sex, year and country was used as comparator group the following entry parameters were significant predictors of reduced survival: younger age, an eGFR <15 ml/min, higher BVAS, MPO-ANCA, and lower haemoglobin, higher WBC and their respective interaction terms for log time (see online supplementary figure S1).

Patient survival according to ANCA specificity, age, renal function as chronic kidney disease stage and BVAS is presented in figure 3A–D as adjusted Cox curves.

Figure 3

Calculated patient survival probability from the multivariable Cox regression. (A) Type of antineutrophil cytoplasmic antibody (ANCA); proteinase 3 (PR3)-ANCA, myeloperoxidase (MPO)-ANCA, MPO- and PR3-ANCA (double positive); (B) renal function estimated as chronic kidney disease (CKD) stage; stage 1+2, stage 3+4 and stage 5; (C) patient age at entry, age 60 years (reference) and 45 and 75 years; (D) Birmingham Vasculitis Activity Score (BVAS) which was set at 5, 15 (reference) and 25.

Kaplan–Meier curves (see online supplementary figure S2A–E), univariate analyses and Cox regressions without time-dependent covariates (see online supplementary tables S1–2) are given in a online supplement to this paper.

Discussion

The past 20 years has seen the advent of randomised therapeutic trials in ANCA-associated vasculitis, which have substantially contributed to establishing modern treatment paradigms in these diseases.21 This study assessed the long-term survival and factors predicting mortality in patients with ANCA-associated vasculitis who participated in four randomised trials organised by the EUVAS capturing almost the full range of disease severity.

The 1-, 2- and 5-year survival was 88%, 85% and 78%, respectively. The mortality ratio was 2.6 compared with the general population. The main predictors of poor outcome at presentation were advanced renal failure, increasing age, a high BVAS and WBC and a low haemoglobin. Relative mortality was higher in younger patients than in a matched general population.

This study shows that patients with ANCA-associated vasculitis continue to have a substantially higher mortality than a matched general population despite advances in diagnosis and treatment in recent years. Mortality is high in the first year after diagnosis, particularly for patients with severe renal impairment and advanced age. Disease and treatment-related complications, in particular infections and active vasculitis, account for the majority of deaths within the first year. It emphasises the need to strike the right balance between gaining rapid control of life-threatening disease manifestations without exposing the patients to undue risk of heavy immunosuppression. Improving the safety while at least maintaining current efficacy needs to be the focus for the further development of remission induction protocols.

The excess risk of mortality persists after the first year. The main causes of death, infections, cardiovascular causes and malignancies, are similar to the pattern seen in the general population. It is noteworthy, however, that death due to infection remains prevalent. So far it is unclear whether there is an excess of cardiovascular and malignant death. It is interesting that in this study most of the excess mortality compared with the general population occurred in the younger age group. This is an important finding, and has not been documented before and highlights the importance of comparing patient survival with the general population.

In our cohort increasing age, advanced renal impairment, higher disease activity and low haemoglobin were predictors of mortality. Advanced renal failure and higher age are independent risk factors for treatment-related toxicity as previously shown by our group.22 Although current treatment guidelines already make recommendations to modify cyclophosphamide dosing according to age and renal function,21 further efforts are required to tailor treatment to individual patients taking into account the specific risk from disease manifestations and treatment toxicity. A timely diagnosis might help to prevent irreversible loss of kidney function but this is hampered by the often silent nature of kidney disease. The identified adverse factors will be useful in the design of new therapeutic studies for risk stratification to deal with these problems.

Patient survival in our study was comparable to smaller previous studies, which have reported a survival of between 82% and 97% at 1 year, 45% and 91% at 5 years and 75% and 88% at 10 years for patients with WG and MPA.23 Several previous studies identified advancing age and various markers of impaired kidney function as predictors of early death5 24,,27 in keeping with our findings. Disease activity as measured by the BVAS had a small but statistically significant effect on survival in the multivariable model. This is in keeping with some previous reports.28 A low haemoglobin and high WBC imparted an increased risk of death, possibly as markers of the severity of the systemic inflammation. In addition, the importance, particularly of neutrophils, in the pathogenesis of ANCA-associated vasculitis is well recognised.29 A possible bias is, however, that patients with more severe disease may have been more likely to have received corticosteroids before enrolment which would have increased their neutrophil count.

In contrast with several previous studies,27 30 31 we found no difference in survival in patients with either WG or MPA or patients with different ANCA specificities. The reported differences in previous studies are probably due to either reporting unadjusted mortality without correcting for the older age and more severe renal involvement of patients with MPA and anti-MPO-ANCA, underpowered multivariable analysis and/or inadequate case mix.

The strengths of this study are that it is to our knowledge the largest study investigating survival in ANCA-associated vasculitis. The patients had a broad range of disease manifestations and were followed up for a median of 6 years for surviving patients.

However, there are some limitations to this study. This cohort was derived from four randomised controlled trials rather than a typical inception cohort. Although the proportions of patients with each severity of ANCA-associated vasculitis (limited, generalised, or severe) in our cohort may differ from an inception cohort, their characteristics are representative of the broader population of patients with ANCA-associated vasculitis. The relative magnitude of effect of each parameter in our multivariable model is therefore likely to be valid and representative of the broader population. Furthermore, most inception cohorts in ANCA-associated vasculitis are derived from a limited number of centres and geographical areas while ours covers a large part of Europe, further enhancing its generalisability. The ratio between patients with WG and those with MPA varies geographically; the ratio in our study was typical for ratios found at middle European latitudes.32 The overall mortality of ANCA-associated vasculitis may have been underestimated as our studies also excluded very elderly patients and those with lung haemorrhage at the time of diagnosis. In addition, by necessity, patients who died before they could be recruited are also not captured. Patients with localised disease only may be under-represented as researchers might have been reluctant to enrol them into studies with possible exposure to cytotoxic drugs. Thus it is difficult to generalise our results to those specific populations.

Despite recent advances in treatment, patients with ANCA-associated vasculitis continue to have an excess mortality. This is particularly true for younger patients and patients with severe renal involvement. This increased risk persists after the initial acute presentation. The development of newer treatment strategies has the potential to reduce early deaths due to vasculitis and deaths consequent on the toxicity of current agents. It is unclear to what extent improved treatments will have an impact on later mortality.

Acknowledgments

We would like to thank those who returned the questionnaires and the investigators of the included trials.

References

Supplementary materials

Footnotes

  • Funding This study was supported by a grant from the European League Against Rheumatism and grants from Region Skåne, Sweden. OF and DJ were supported by the Cambridge Biomedical Research Centre. MW was supported by the Kidney Research Scientist Core Education and National Training (KRESCENT) Program and the Alberta Heritage Foundation for Medical Research.

    The funding sources had no involvement in the study's design, conduct or reporting.

  • Conflict of interest None.

  • Ethics approval This study was conducted with the approval of the West Midlands Multi-centre Research Ethics Committee (ref: MREC/98/7/37).

  • Provenance and peer review Not commissioned; externally peer reviewed