Abstract
Background and objectives ANCA-associated vasculitis is commonly found in elderly patients, but there are few data concerning outcome and treatment in the highest age groups.
Design, setting, participants, & measurements Consecutive patients (N=151) presenting between 1997 and 2009 were retrospectively included from local registries in six centers in Sweden, the United Kingdom, and the Czech Republic if diagnosed with microscopic polyangiitis or granulomatosis with polyangiitis at age ≥75 years during the study period. Patients were followed until 2 years from diagnosis or death. Data on survival and renal function were analyzed with respect to age, sex, ANCA specificity, renal function, C-reactive protein, comorbidities, and Birmingham Vasculitis Activity Score at diagnosis as well as treatment during the first month.
Results Median follow-up was 730 days (interquartile range, 244–730). Overall 1-year survival was 71.5% and 2-year survival was 64.6%. Older age, higher creatinine, and lower Birmingham Vasculitis Activity Score were associated with higher mortality in multivariable analysis. Patients who were not treated with standard immunosuppressive therapy had significantly worse survival. Renal survival was 74.8% at 1 year. No new cases of ESRD occurred during the second year. High creatinine at diagnosis was the only significant predictor of renal survival in multivariable analysis.
Conclusions ANCA-associated vasculitis is a disease with substantial mortality and morbidity among elderly patients. This study showed a better prognosis for those who received immunosuppressive treatment and those who were diagnosed before having developed advanced renal insufficiency.
Introduction
The ANCA-associated vasculitides (AAVs) comprise microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA; Wegener’s), and eosinophilic granulomatosis with polyangiitis (EGPA; Churg–Strauss) (1,2). AAVs are predominantly diseases of older patients, and the incidence increases with age (3,4). Several studies have shown that the risk of death and ESRD is higher in older patients, but their response to treatment differs between studies (5,6). Because of an increased risk of adverse events (7), dose adjustments of cyclophosphamide have been made for age in clinical trials (8) and are recommended in guidelines (9). Despite the fact that AAVs commonly affect older patients, the outcome of MPA and GPA in the elderly population is largely unknown (10). Many randomized trials exclude patients aged >75 years and there are few observational studies focusing on elderly patients (11,12). The aim of this study was to investigate demographic factors, treatment, and outcome in patients aged ≥75 years presenting with MPA or GPA.
Materials and Methods
Case Retrieval and Classification
Consecutive patients presenting at six centers in Sweden, the United Kingdom, and the Czech Republic between 1997 and 2009 were included. Inclusion criteria were age ≥75 years at diagnosis and a clinical diagnosis of MPA or GPA according to the European Medicines Agency algorithm during the study period (13). Patients classified as having eosinophilic granulomatosis with polyangiitis and polyarteritis nodosa were excluded, along with secondary vasculitis, drug-induced vasculitis, and anti–glomerular basement membrane disease in accordance with the exclusion criteria in the European Medicines Agency algorithm (13). Diagnosis was confirmed by review of patient charts. The project was approved by the Ethical Review Board in Lund, Sweden.
Participating Centers
The Swedish cohort was recruited from local vasculitis databases at the Nephrology and Rheumatology Departments at Linköping University Hospital (catchment area 430,000 inhabitants) (14), the Nephrology and Rheumatology Departments at Skåne University Hospital in Lund and Malmö (700,000 inhabitants) (3), and the Nephrology Department at Karolinska University Hospital (1.5 million inhabitants).
The English cohort was recruited from a local database at the Imperial College Renal and Transplant Centre (2 million inhabitants) and from vasculitis and renal pathology databases at the Nephrology Department at Royal Free Hospital (1.4 million inhabitants) in London.
The Czech cohort was recruited from the Department of Nephrology at General University Hospital in Prague (tertiary national referral center with catchment area for vasculitis about 5 million inhabitants), using a local database transferred into a nationwide registry in 2009.
Data Collection
The following data were collected retrospectively from the time of diagnosis: date of diagnosis, age, sex, diagnosis type (MPA/GPA), ANCA specificity, C-reactive protein, creatinine (at diagnosis or before start of dialysis in patients who were dialysis dependent at diagnosis), dialysis dependency, disease activity according to Birmingham Vasculitis Activity Score (BVAS) (15), and major comorbidities. Data on outcome up to 2 years from diagnosis included day of death and ESRD. Data on treatment during the first 3 months from diagnosis included cumulative dose of pulse steroids, oral steroids and cyclophosphamide, treatment with rituximab or other cytotoxic agents, and use of plasma exchange. Based on the treatment issued during the first month, patients surviving that point were divided into groups. Patients were assigned to the rituximab group if they were receiving any dose of rituximab. Patients were assigned to the oral cyclophosphamide group or the intravenous cyclophosphamide group if they were issued a minimum cumulative dose of 2000 mg oral or 1500 mg intravenous cyclophosphamide, respectively, during the first 3 months. Patients who were given less cyclophosphamide or treated with azathioprine, mycophenolate, or methotrexate were assigned to the “other regimens” group, whereas patients were assigned to the “steroids only” group if they were given a daily dosage of ≥30 mg of prednisolone. The remaining patients were placed in the “untreated” group. For each group, the proportion of patients given pulse steroids (cumulative dose ≥250 mg) or plasma exchange was recorded. The effect of treatment on survival and renal survival was estimated using an intention-to-treat approach starting on day 30.
Date of diagnosis was defined as follows: start of treatment with prednisolone ≥30 mg/d, plasma exchange, or cyclophosphamide; if never treated, the day of biopsy; and if no biopsy, the day of the first positive ANCA test result. A dipstick value of ≥2 was taken as representative of hematuria equal to 10 red blood cells per high-power field when assessing BVAS. Indirect immunofluorescence or antigen-specific ELISA was used to detect ANCA. ESRD was defined as need for dialysis for >90 days. Comorbidities were registered essentially as described by Davies et al. (16), with 1 point each given for malignancy, ischemic heart disease, peripheral vascular disease, heart failure, diabetes, systemic inflammatory disease (excluding AAV), pulmonary disease, and cirrhosis.
Statistical Analyses
Statistical analysis was performed using SPSS Statistics for Windows software (version 21.0; IBM Corp., Armonk, NY). P values <0.05 were considered significant. Double-positive patients were designated as having either myeloperoxidase-ANCA (MPO-ANCA)/perinuclear ANCA (P-ANCA) or proteinase-3–ANCA (PR3-ANCA)/cytoplasmic ANCA (C-ANCA) depending on the highest titer. Differences between groups were analyzed using the Mann–Whitney or Kruskal–Wallis test for nonparametric data and the chi-squared or Fisher’s exact test for categorical data. All analyses exclude missing data. The Kaplan–Meier method was used to estimate overall and renal survival, and the log-rank test was used to evaluate differences between groups. Censoring was performed at the day of loss to or completion of follow-up. Estimates of renal survival were censored for death. In analysis of treatment, patients who died within the first 30 days after diagnosis were excluded. Cox proportional hazards models were used to analyze variables at diagnosis and their association with overall survival and renal survival. Variables included were age, sex, ANCA specificity, BVAS, creatinine, C-reactive protein, and comorbidity index. In a second model, a treatment variable was included. Patients in the intravenous cyclophosphamide/oral cyclophosphamide and rituximab groups were considered to have received standard immunosuppressive therapy and were compared with those who had not received such treatment. The standardized mortality ratio (SMR) was calculated for the group aged 75–84 years comparing the observed death rates in the Swedish cohorts with expected death rates in the general Swedish population matched for age and calendar year. The death rate for the general population was calculated using life tables from 2004 provided by Statistics Sweden (http://www.scb.se).
Results
Baseline Patient Characteristics
A total of 151 patients were included in this study, and their characteristics at diagnosis are shown in Table 1. Median age at diagnosis was 79 years (interquartile range [IQR], 77–82). MPA was diagnosed in 70% of the patients and GPA in 30%. MPO-ANCA/P-ANCA positivity was seen in 60%, PR3-ANCA/C-ANCA positivity in 36%, double positivity in 1%, and ANCA negativity in 3%. Renal involvement (according to BVAS) was seen in 92% of the patients, more commonly in patients with MPA (P<0.001). Biopsy was performed in 113 patients (76.4%; data missing in three patients); of these, 96 (85.0%) were renal biopsies. Most parameters showed no significant differences between the three participating countries. However, dialysis dependency at diagnosis was seen in 16% of patients in the Swedish cohort compared with 44% in the English cohort and 60% in the Czech cohort (P<0.001). Comorbidities were more common in Czech patients; only 16% had no comorbidities compared with 47% and 40% of the Swedish and English patients, respectively (P<0.001). Median BVAS was 12 (IQR, 9–12) in patients with renal-limited disease and 16 (IQR, 13–19) in patients with more than one organ system involved (P<0.001).
Demographic factors at the time of diagnosis divided according to country and diagnosis type
Treatment
In total, seven patients died during the first month, and an additional 14 patients were excluded because of insufficient treatment data, leaving 130 patients for analysis of treatment. Characteristics for the different treatment groups are shown in Table 2. There were no significant differences between patients receiving standard immunosuppressive therapy (oral cyclophosphamide, intravenous cyclophosphamide, and rituximab groups) and those not receiving such therapy (other regimens, steroids only, or untreated groups) with regard to creatinine, BVAS, proportion of dialysis dependency, or proportion of patients with at least one comorbid condition at diagnosis. Likewise, there were no significant differences between the participating countries (Supplemental Table 1). The median cumulative dose of intravenous cyclophosphamide during the first 3 months in the standard treatment group was 3000 mg (IQR, 2100–4500) and the corresponding figure for oral cyclophosphamide was 5270 mg (IQR, 3100–6950).
Groups of patients based on the treatment given during the first month after diagnosis: Contribution to the treatment groups by the participating countries, baseline characteristics, and additional therapy
Survival
Patients were followed until 2 years after diagnosis or death. Six patients (all from England) were lost to follow-up, within a median time of 69 days (IQR, 34–140). One-year overall survival was 71.5% and 2-year survival was 64.6%. During the first 3 months after diagnosis, 22 patients (14.6%) died. Older age, higher creatinine, and lower BVAS at diagnosis were independent factors associated with higher mortality by univariable analysis (Figure 1, Supplemental Table 2) as well as multivariable analysis (Table 3). Two-year survival was 36.0% in the Czech Republic, 66.7% in Sweden, and 80.2% in the United Kingdom (P=0.001). Dialysis at diagnosis was not significantly associated with survival (hazard ratio [HR], 1.57; 95% confidence interval [95% CI], 0.90 to 2.74; P=0.12). Of the 37 patients who developed ESRD, 18 (48.6%) died during follow-up compared with 34 of the 114 patients (29.8%) who did not develop ESRD. The SMR in patients aged 75–84 years in the Swedish cohort was 3.69 (95% CI, 2.45 to 5.55) after 2 years. The SMR was 5.04 (95% CI, 3.13 to 8.11) in the first year and 1.86 (95% CI, 0.77 to 4.47) in the second year. Survival was significantly worse in patients not treated with standard immunosuppressive therapy compared with patients treated with either cyclophosphamide or rituximab (Figure 2). Two-year survival was 72.9% for patients treated with oral cyclophosphamide, 72.5% for intravenous cyclophosphamide, 81.3% for rituximab, and 45.0% for no/other treatment. When including treatment in the survival analysis, age, creatinine, and treatment remained significant predictors of mortality in both univariable and multivariable analyses, but BVAS lost its significance in the multivariable analysis (Table 4, Supplemental Table 3). The results remained essentially unchanged when including patients with lower cumulative doses of cyclophosphamide in the cyclophosphamide treatment group (HR, 0.30; 95% CI, 0.11 to 0.80; P=0.02).
Patient survival. Kaplan–Meier plot depicting overall 2-year survival divided according to age, creatinine level, and BVAS. (A) Age at diagnosis ≥80 years (n=69) versus <80 years (n=82). (B) Creatinine level at diagnosis equal to/above median (≥3.20 mg/dl) (n=71) versus creatinine below median (<3.20 mg/dl) (n=71). (C) BVAS at diagnosis ≥15 (n=74) versus BVAS <15 (n=73). BVAS, Birmingham Vasculitis Activity Score.
Multivariable Cox regression analysis of overall 2-year survival in the entire cohort
Treatment. Kaplan–Meier plot depicting overall 2-year survival divided according to the treatment given during the first month after diagnosis. CYC/RTX indicates rituximab or cyclophosphamide at minimum dose 2000 mg oral/1500 mg intravenous during the first 3 months (n=102), whereas no/other treatment indicates, other regimens, steroids only, or untreated (n=28). Number at risk starting on day 30.
Multivariable Cox regression analysis of overall 2-year survival in patients with complete treatment data available
Renal Function
Dialysis was needed in 45 patients (30.6%) at the time of diagnosis. Twelve of these patients (26.7%) recovered independent renal function during follow-up. By the end of the first year, a total of 37 patients (24.5%) had developed ESRD. No new cases of ESRD occurred during the second year. Renal survival censored for death was 74.8%. It was 35.2% for the Czech cohort, 88.2% for the Swedish cohort, and 71.1% for the English cohort (P<0.001). In multivariable analysis, creatinine level at diagnosis was the only significant predictor of renal survival (HR, 4.10; 95% CI, 2.25 to 7.49 per quartile increase in serum creatinine; P<0.001; Figure 3), whereas treatment was not significantly associated with renal outcome (HR, 1.88; 95% CI, 0.61 to 5.77; P=0.27). In the subgroup of patients who were dialysis dependent at diagnosis, nine of the 11 (81.8%) who recovered renal function were treated with plasma exchange compared with 14 of the 26 (53.8%) who did not recover renal function (P=0.15).
Renal survival. Kaplan–Meier plot depicting renal survival censored for death divided according to serum creatinine at diagnosis. First quartile, 0.69–1.73 mg/dl (n=37); second quartile, 1.74–3.34 mg/dl (n=38); third quartile, 3.35–5.67 mg/dl (n=37); and fourth quartile, 5.68 mg/dl-dialysis dependency (n=37).
Discussion
In this study, we show that AAV in older patients is a severe condition with high mortality and morbidity during the first year after diagnosis. Despite the significant burden of disease, we found a relatively favorable outcome for those who survive the first year. To our knowledge, this is the largest study of patients aged ≥75 years with AAV.
We found older age, high creatinine, and low BVAS to be predictors of worse survival. High age and impaired renal function have previously been shown to predict mortality in younger age groups as well (6,11,14,17–20). The inverse relationship between BVAS and survival is intriguing and opposite of some previous findings (11,18). Because the vast majority of patients in our study had renal involvement at diagnosis, patients with a high BVAS had both renal and extrarenal symptoms, whereas those with renal-limited disease fell in the low BVAS group. It is possible that patients with few extrarenal symptoms are diagnosed at a later stage with more irreversible renal damage and worse survival. However, this association needs to be confirmed in other studies and the cause remains to be elucidated.
Survival was significantly worse in those not treated with standard doses of cyclophosphamide or rituximab. There is a risk of confounding by indication when analyzing treatment data in retrospective studies (in this case, that frail patients receive less treatment), and we did not have data on the detailed clinical context including the goals and preferences of patients or the treating physician’s motive for therapy decisions. To reduce this effect, we did not include deaths occurring during the first month and used an intention-to-treat approach, reasoning that standard therapy given during the first month was intended to continue according to modern guidelines (9,21), whereas this was not the case when such therapy was not issued during the first month. It is, however, difficult to rule out the possibility that some patients who originally received immunosuppressive treatment developed complications and that therapy was stopped. The worse prognosis in these patients would thus be caused by early adverse events. However, also when considering all cyclophosphamide use as intention to treat with standard therapy, there was a clear survival benefit for standard therapy. The unfavorable prognosis in those treated less aggressively cannot readily be explained by more comorbidities, worse renal function, or differences in disease activity at diagnosis, because these parameters did not differ significantly between the groups. However, this does not exclude the possibility that some patients were still considered frail and not suitable for immunosuppressive therapy. The best survival was seen in patients treated with rituximab; however, the number of patients was rather small and no rituximab was given in the Czech cohort with the worst survival rate. Hence, this result should be interpreted with care.
The survival differences between countries may, at least partly, be explained by selection bias. The Department of Nephrology in Prague is a tertiary referral center receiving the most severe cases of AAV. In addition, the Czech patients had more comorbidities and more often had advanced renal insufficiency at diagnosis. Opposite to our results, others have reported dialysis at diagnosis to be an independent factor for survival (20), and it is possible that our analysis did not fully compensate for this factor. Comorbidities were not associated with mortality in the entire cohort but reached borderline significance in the analysis including treatment. In our cohort, 15% of the patients had a previous malignancy before the diagnosis of AAV. This is similar to other studies showing a frequency of malignancies of 8%–16% (17,22). The life expectancy at birth was 75.3 years in the Czech Republic in 2003, compared with 78.3 years in the United Kingdom and 80.2 years in Sweden (23). The effect of differences in life expectancies in the general population when comparing the survival of AAV is difficult to discern.
One-year survival was 72% and 2-year survival was 65% in this study compared with 82%–88% and 82%–85%, respectively, in previous studies (2,17,18). We also show that survival in the Swedish cohort was significantly worse compared with the background population during the first year after diagnosis but was not significantly worse during the second year. Together with the fact that no new cases of ESRD occurred during the second year, this implies a fairly good prognosis if patients survive the first year.
Importantly, we found that elderly patients usually present with significant renal involvement and a predominance of clinical MPA and MPO-ANCA/P-ANCA positivity. This is consistent with the finding that patients with MPA are older than patients with GPA and that patients with MPO/P-ANCA are older than patients with PR3/C-ANCA (5,6,11,18,24,25). Results showing that older patients with GPA have more renal insufficiency and less upper airway symptoms (26) add to this picture. We found a higher proportion of MPO/P-ANCA positive GPA compared with two large studies in Caucasian patients. In our cohort, 24% of patients with GPA were positive for MPO/P-ANCA compared with 9% and 10%, respectively, in a follow-up of patients from four European Vasculitis Study Group trials and a large genetic study in Europe (18,27).
High creatinine at presentation is a well known risk factor for impaired renal survival (2,12,17,28,29), and we confirmed this in older patients. High creatinine at diagnosis was the only significant predictor of renal survival in multivariable analysis, and treatment with immunosuppressive agents was not significantly associated with renal survival. Most patients reaching ESRD were dialysis dependent at presentation and conceivably already had irreversible renal damage at the time of diagnosis. Increased awareness of these conditions in elderly patients and early recognition are probably important for improved renal survival. Plasma exchange has been shown to reduce the risk of progression to ESRD in patients with severe renal impairment at diagnosis (30). In accordance with this, we found a tendency toward better renal survival in those treated with plasma exchange in the subgroup of patients who were dialysis dependent at diagnosis.
Our study has several limitations. The greatest limitation is the retrospective nature of the data and the risk of introducing bias related to this, as discussed above. A possible limitation is that the centers in Linköping and Lund received both rheumatology and nephrology patients, whereas London received relatively more nephrology patients and Stockholm and Prague almost exclusively received patients with renal involvement. However, referral bias cannot explain the large proportion of patients with MPA and renal involvement because this did not differ significantly between countries. Because of the heterogeneity of referral patterns in the participating centers, our results are difficult to generalize to sites with different patient characteristics. We did not have any data on renal histology, so we could not analyze whether this was predictive of outcome.
In conclusion, we demonstrate in a large cohort of elderly patients with AAV that survival is significantly better in patients who are treated with adequate doses of cyclophosphamide or rituximab at diagnosis compared with patients who receive less aggressive treatment.
Disclosures
None.
Acknowledgments
The authors thank Qianwen Ko for acquisition of data from Royal Free Hospital.
This study was supported by grants from the Ingrid Asp Foundation, the Swedish Renal Foundation, and the National Institute for Health Research Imperial Biomedical Research Centre.
Preliminary data were previously presented in abstract form (Weiner M et al., Nephrol Dial Transplant 27[Suppl 2]: 417, 2012; Weiner M et al., J Am Soc Nephrol 23: 709, 2012; Weiner M et al., Presse Med 42: 674, 2013).
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.00480115/-/DCSupplemental.
See related editorial, “Treating Elderly Patients with ANCA-Associated Vasculitis,” on pages 1110–1113.
- Received January 14, 2015.
- Accepted March 30, 2015.
- Copyright © 2015 by the American Society of Nephrology
References
In this issue
