To assess the epidemiology of the primary systemic vasculitides (PSV) in a well-defined population of southern Europe over a 14-year period using the Chapel Hill Consensus Conference (CHCC) definitions.
To assess the epidemiology of the primary systemic vasculitides (PSV) in a well-defined population of southern Europe over a 14-year period using the Chapel Hill Consensus Conference (CHCC) definitions.
The case records of all patients 15 years or older with vasculitis diagnosed between January 1988 and December 2001 at a single reference hospital in the Lugo region of northwest Spain were reviewed. Incidence rates were age- and sex-adjusted to the European standard population. Patients were classified as having PSV according to the CHCC definitions.
Fifty-four Lugo residents (29 men) fulfilled the CHCC definitions for PSV. The mean age was 60.7 ± 13.5 years (men: 61.0 ± 13.4; women: 60.4 ± 13.8 years). The overall annual incidence of PSV was 13.07/million (95% confidence interval [95% CI] 8.89–19.22). PSV were slightly more common in men. The age-specific incidence showed a clear increase with age. A peak in the 55–64 year age group for the whole group of patients with PSV was observed (34.9/million; 95% CI 28.6–42.6). Nonrandom periodical peaks of incidence every 3 years were only observed when the group of PSV was considered as a whole (P = 0.040). The annual incidence was 2.95/million (95% CI 1.44–6.05) for Wegener's granulomatosis (WG) and 7.91/million (95% CI 4.74–13.20) for microscopic polyangiitis (MPA) (P = 0.035). None of the patients with Churg Strauss syndrome (n = 4) lived in a rural area.
Our observations support an increasing incidence of PSV with age. In patients from northwest Spain defined by the CHCC definitions, MPA is more common than WG.
The primary systemic vasculitides (PSV) are a heterogeneous group of uncommon diseases characterized by blood vessel inflammation and necrosis (1). These conditions often have overlapping clinical and pathologic manifestations that sometimes make it difficult to reach a precise diagnosis (2). Pathogenic mechanisms of PSV are being investigated and the discovery of new autoantibodies, such as antineutrophil cytoplasmic antibodies (ANCA), has contributed to an increased awareness of these conditions (3, 4). Despite possible limitations (5, 6), the development of the classification criteria in 1990 by the American College of Rheumatology (ACR) (7) and in 1994 by the Chapel Hill Consensus Conference (CHCC) (8) has enabled new epidemiologic studies (9, 10). However, the incidence of PSV in southern Europe is still far from being well established. We have previously reported the epidemiology of vasculitides in northwest Spain over a 10-year period using the 1990 ACR classification criteria for vasculitides (11). In the present study we have applied, for the first time, the CHCC definitions for PSV to a well-defined population collected over a 14-year period. Due to this, we have assessed the incidence of classic polyarteritis nodosa (PAN), considered the paradigm of vasculitides involving medium arteries in adults, Wegener's granulomatosis (WG), Churg-Strauss syndrome (CSS), and microscopic polyangiitis (MPA) as the paradigm of vasculitides where small vessels and medium arteries are the dominant vessels involved (8). We examined trends of the PSV over the period of study, influence of sex, and differences between people living in urban and rural areas.
The case records of all patients diagnosed with vasculitis in the department of medicine of the Hospital Xeral-Calde (Lugo, Spain) from January 1988 through December 2001 were reviewed. Because children are seen in the department of pediatrics, only the medical records of those patients with vasculitis aged 15 years and older were reviewed.
Patients with a documented episode of vasculitis prior to 1988 were excluded. Also, only patients 15 years and older who fulfilled the CHCC definitions for WG, CSS, classic PAN, or MPA were included in this study. Thus, patients with vasculitis who fulfilled the CHCC definitions for other primary vasculitides, such as giant cell arteritis (GCA), cutaneous leukocytoclastic angiitis, essential cryoglobulinemic vasculitis, or Henoch-Schönlein purpura, were excluded. In addition, patients with vasculitides secondary to connective tissue disease, rheumatoid arthritis, malignancy, or infections were not included.
In all cases, a tissue biopsy sample showing vasculitis, necrotizing granulomatous inflammation, extravascular granulomas with eosinophils, or parenchymal infiltration with eosinophils or a kidney biopsy sample showing glomerulonephritis was required. As with a former report (10), patients without histologic evidence of granuloma or clinical evidence of upper respiratory tract involvement were considered to have MPA rather than WG using the CHCC definitions.
Complete ANCA data were not available because some patients were diagnosed prior to the widespread availability of ANCA testing. Until 1998, ANCA were detected by indirect immunofluorescence (IIF) on ethanol-fixed granulocytes. Since 1998, enzyme-linked immunosorbent assays (ELISAs) for proteinase 3 (PR3) ANCA or myeloperoxidase (MPO) ANCA have been used as the choice test if a systemic vasculitis is suspected.
Since 1988 (the rheumatology division was established in 1987), we have registered all patients with systemic vasculitis seen at the department of medicine of our hospital, which serves a well-defined, stable and ethnically homogenous population. The main characteristics of the Lugo population have previously been reported (11, 12). The Hospital Xeral-Calde is the single referral center for a mixed rural and urban white population of almost a quarter of a million people living in central Galicia (Spain) (13). In addition, our hospital often provides medical attention to patients from neighboring areas, but these patients are not included in this epidemiologic study.
In 1996, the population in the area of Lugo was 238,503. The adult (>15 years) population was 208,271 (men 101,375; women 106,896) and the population 65 years and older was 57,305 people (men 25,112; women 32,193). At that time, almost 48% lived in urban areas, most of them in the city of Lugo. This population is relatively static, and no important migration has occurred in the area during the last 2 decades. However, in Lugo the average age of the population is higher than in the rest of Spain (14). Due to this, the population older than 65 in Lugo is much higher than the national average for the Spanish population. The population in rural areas is older than in urban ones. The population structure is very similar for men and women. During the past 15 years, the population has fallen by 8,000.
Patients are sent to the hospital by general practitioners or they self refer to the emergency unit. All patients with vasculitis are diagnosed in the department of medicine of the hospital. In this regard, most of them are specifically diagnosed and treated at the rheumatology division. However, patients with vasculitis and renal and lung features are also examined by nephrology and respiratory staff physicians.
The target population was census based. The population distribution was provided by the Instituto Nacional de Estadística. In the Lugo region, people grouped by age and sex were estimated by exponential interpolation from census data. The population between censuses is established by extrapolation. Age-standardized incidence rates were calculated by the direct method using the European population as standard. The European standard was preferred to the world one because it better represents a developed country's population. Rates are reported as cases for 106 population 15 years or older. They were calculated using the number of new cases observed/the estimated population 15 years of age and older. Incidence rates were compared during the first (1988–1994) and second (1995–2001) periods of study.
Continuous data were expressed as mean ± standard deviation (SD). To calculate 95% confidence intervals (95% CIs) it was assumed that the number of cases followed the Poisson distribution. Statistical significance was defined as P ≤ 0.05.
To investigate if there were nonrandom peaks in the annual incidence, the Bartlett's periodogram test for white noise (variability at random) and the Q test from Box and Bierce for white noise were applied (15, 16).
All statistical analyses were performed with the software Stata Intercooled, release 6 (Stata Corporation, College Station, TX).
Between 1998 and 2001, 54 patients from Lugo with a new diagnosis of vasculitis fulfilled the CHCC definitions for PSV. Thirty-four of them fulfilled the CHCC definitions for MPA (8). However, only 16 of these 34 patients met classification criteria for PAN when the 1990 ACR classification criteria for PAN were applied (17). The 4 patients that met the CHCC definitions for classic PAN also fulfilled the 1990 ACR classification criteria for PAN (17). Due to the inclusion criteria used in this study, the 4 patients with CSS fulfilled both the CHCC definition and the 1990 ACR classification criteria for CSS (18). It was also the case for the 12 patients with WG (19).
Twenty-six patients were diagnosed before 1995. The number of new cases during the second 7-year period of study was 28. Twenty-nine of the 54 patients (54%) were men. The mean age ± SD for the whole group of patients was 60.7 ± 13.5 years (median 60.5). In men, mean age at diagnosis was 61.0 ± 13.4 years (median 60.0), whereas in women it was 60.4 ± 13.8 years (median 61.0) (P = 0.856).
The overall annual incidence of PSV was 13.07/million (95% CI 8.89–19.22). PSV was slightly more common in men (13.62/million; 95% CI 9.35–19.83) than in women (12.46/million; 95% CI 8.38–18.53; Table 1). The incidence in the period 1988–1994 was 13.01/million (95% CI 8.84–19.14) and in the period 1995–2001 it was 13.12/million (95% CI 8.93–19.27). Thus, no differences in incidence between the two 7-year periods of study were observed.
|All PSV||13.07 (8.89–19.22)||10.57 (6.83–16.37)||15.72 (11.15–22.16)||13.62 (9.35–19.83)||12.46 (8.38–18.53)|
|Wegener's granulomatosis||2.95 (1.44–6.05)†||2.67 (1.30–5.49)||3.33 (1.64–6.76)||2.68 (1.30–5.51)||3.35 (1.65–6.80)|
|Churg-Strauss syndrome||1.31 (0.87–1.96)||0.00‡||2.50 (1.22–5.13)‡||1.26 (0.88–1.81)||1.34 (0.87–2.06)|
|Classic polyarteritis||0.90 (0.72–1.13)||1.32 (0.87–1.99)||0.45 (0.01–3.49)||0.80 (0.46–1.38)||0.98 (0.94–1.02)|
|Microscopic polyangiitis||7.91 (4.74–13.20)†||6.58 (3.75–11.53)||9.44 (5.92–15.05)||8.88 (5.48–14.38)||6.79 (3.91–11.80)|
The age-specific incidence showed a clear increase with age (Figure 1). A peak in the 55–64 year age group for the whole group of patients with PSV was observed (34.9/million; 95% CI 28.6–42.6). This peak was mainly due to the men, because in this age group the incidence in men peaked at 48.7/million (95% CI 41.6–56.9) while for the same age group the incidence in women was 22.3/million (95% CI 17.0–29.3). In contrast, the peak incidence for women was observed at a younger age, in the 45–54 age group, with an incidence of 32.5/million (95% CI 26.3–40.1; Figure 1). The progressive increase of the incidence with age was also observed when patients were distributed according to rural or urban areas (Figure 2). The incidence peaked for patients living in urban areas at the 65–74 year age group (47.1/million; 95% CI 40.1–55.3), whereas the peak incidence for those from rural areas occurred at the 55–64 age group (32.1/million; 95% CI 26.0–39.7).
Annual adjusted rates for the whole group of patients, for men and women, and for population living in rural or urban areas are provided in Figures 3 and 4. In these figures, peaks of incidence for men appeared to occur every 3 years. However, when a statistical analysis on the peaks of incidence was assessed according to sex, no definitive conclusions were achieved. In this regard, based on the tests applied (15, 16), the presence of periodical peaks of incidence might occur at random. The only possibility for nonrandom periodical peaks of incidence every 3 years might be for the whole group of vasculitides when the Q test was applied (P = 0.040). However, it was not the case when PSV were stratified according to the site of residence (rural or urban).
The specific annual incidence rates for each one of the PSV differed from one vasculitis to another (Table 1). The incidence of WG using the CHHC definitions during the 14-year period was 2.95/million (95% CI 1.44–6.05). The incidence of MPA, in contrast, was much higher (7.91/million; 95% CI 4.74–13.20). The difference between the incidence of MPA and WG was statistically significant (P = 0.035). As described in Table 1, the incidence of CSS and classic PAN was very low.
When patients were stratified according to population living in urban or rural areas, the only statistically significant difference was observed in CSS. It was due to the absence of new cases of CSS during the period of study in rural areas (Table 1).
The age-specific incidence for each type of PSV according to CHCC definitions is shown in Figure 5. In both MPA and WG the incidence peaked at the 55–64 age group (MPA: 23.3/million; 95% CI 17.9–30.4; WG: 9.3/million; 95% CI 5.8–14.9).
Regarding ANCA results, 9 of the 10 patients with WG according to the CHCC definitions on whom ANCA tests were performed had positive ANCA test results (classic cytoplasmic ANCA pattern by IIF in all of them and PR3 ANCA in the 3 patients on whom an ELISA was performed). None of the 4 patients with classic PAN had positive results for ANCA. Thirteen of the 18 patients fulfilling definitions for MPA on whom ANCA tests were performed had positive results (most of them with perinuclear ANCA by IIF and/or anti-MPO specificity by ELISA and a few [n = 2] with cytoplasmic and/or anti-PR3 specificity). Finally, only 1 of the 3 patients with CSS on whom ANCA were tested had positive results (anti-PR3 ANCA).
Studies on the epidemiology of PSV have generally been of short duration and were frequently performed in tertiary referral centers. Thus, problems of referral and selection bias and uncertainty of denominator population in these centers might create difficulties when defining the actual incidence of PSV. This report constitutes the first attempt to examine the incidence of PSV in southern Europe using the CHCC definitions over a long period of time.
The particular characteristics of the Hospital Xeral-Calde, which is the only referral center for these conditions in a well-defined region of Spain, may help provide the information on the incidence of these poorly understood diseases in southern Europe. We understand that the CHCC definitions may be more restrictive than the ACR criteria for PSV and that the use of these definitions may underestimate the incidence of some PSV, in particular WG. However, sometimes patients fulfilling ACR criteria for PSV may not meet other definitions, in particular those proposed by the CHCC. Due to this, our purpose was to assess the implications of the CHCC definitions in the epidemiology of PSV in a defined population.
Scott et al (20) made the first study on the European annual incidence of PSV in the 1970s. In that study from Bath/Bristol (UK), the overall annual incidence of PSV was 10/million. More recently, in Norwich (UK) Watts et al (10) examined the epidemiology of PSV in patients aged >15 years between 1988 and 1997. Patients were classified according to the 1990 ACR criteria for WG, CSS, and PAN and the CHCC definitions for MPA. The overall annual incidence of PSV was 19.8/million. PSV was more common in men (23.5/million) than women (16.4/million). There was a trend toward an increase in annual incidence; the incidence in the first 5-year period was 16.5/million and in the second it was 22.4/million. These authors also observed an increase in the incidence with age in both men and women, with a peak in the 65–74 year age group of 60.1/million (10). Using the CHCC definitions only, the overall annual incidence of PSV in the population of Lugo aged 15 years and older over the period 1988–2001 was 13.07/million. PSV was slightly more common in men.
Peaks of incidence have been reported in patients with small vessel vasculitis (21). In the present study, nonrandom periodical peaks of incidence every 3 years were only observed when the group of PSV was considered as a whole. However, the implication of this finding needs further investigation.
In Germany, an increased prevalence of PSV was observed in urban compared with rural populations (22). This higher prevalence in urban populations was mainly due to GCA. In contrast, no significant differences in the incidence between rural and urban population were observed in Lugo. These contradictory observations may be explained by the fact that in our study, GCA was not included among the group of PSV and the main goal was to assess incidence.
The annual incidence of WG varies from one study to another. In 1984, Kurland et al (23) reported an annual incidence of 4/million in Rochester, Minnesota. In the UK, a progressive increase of this vasculitis has been observed. With respect to this, Scott et al (20) reported an annual incidence of 0.5/million in Bath/Bristol during the 1970s. In Leicester (UK), Andrews et al (24) reported an incidence of WG of 0.7/million from 1980 to 1986 and 2.8/million from 1987 to 1989. In Norwich, the annual incidences of WG and CSS using the 1990 ACR criteria during the period 1988–1997 were 9.7/million and 2.7/million, respectively (10). Also, based on the 1990 ACR classification criteria, in Tromsø (Norway) Koldingnsnes and Nossent (25) described a linear trend toward an increase in the annual incidence rate of WG, from 5.2/million in the period 1984–1988 to 12.0/million in the period 1994–1998. The adjusted annual incidence rates of WG and CSS in patients from Lugo defined by the CHCC were much lower (2.95/million for WG and 1.31/million for CSS). In addition, 4 cases in Lugo fulfilled the CHCC definitions for classic PAN, whereas no cases of classic PAN as defined by the CHCC were observed in Norwich (10). However, when the CHCC definitions were used, the annual incidence of MPA in Norwich was 8.0/million, which is similar to that found in northwest Spain. Of note, between the period 1988 and 1998 in Tromsø (Norway), MPA was less common (2.7/million) than in Norwich and Lugo. The annual incidence of WG (classified according to the 1990 ACR criteria), in contrast, was found to be much higher than in Lugo (10.6/million) (26).
In keeping with the data reported from Norwich (10), the mean age of the PSV patients from Lugo was higher than in the series from tertiary referral centers (27, 28). The Norwich data (10) showed an increase in the incidence with age, peaking in the 65–74 age group. This was also observed in Lugo with a peak between the 55 and 64 years. In both Norwich and Lugo, there is a higher number of patients who are aged >65 years compared with the national average for England and Spain, respectively. Also, in both regions the population is white (29). Thus, the apparent differences in the incidence of WG between both populations cannot be explained by differences in the age of the population between both regions.
The great disparity in data of incidence between populations of northern and southern Europe suggests that, besides the influence of different criteria for the diagnosis, genetic factors and possibly unknown environmental factors may account for such differences. In this regard, Duna et al (30) described higher rates of self reported environmental exposures to inhaled substances in North American patients with WG within 1 year before the onset of the vasculitis. However, there was no evidence of exposure to inhaled substances in most patients with PSV from Lugo (data not shown).
In Norwich, a predominance of men was observed in all types of PSV (10). In contrast, WG was slightly more common in North American women, with a peak of disease onset between 45 and 65 years (31). In Lugo, both WG and CSS were slightly more common in women. However, it was not the case for MPA. Of note, in Lugo, CSS was more common in people living in urban areas. This observation might also speak in favor of environmental factors influencing the development of the disease. Finally, the age of onset of MPA in Lugo was similar to that reported in France by Guillevin et al (32).
Another vasculitis often held to be more prominent in northern than southern Europe is GCA. During the period 1988–2001, the annual adjusted incidence rate of GCA in Lugo was 47.8/million for people 15 years and older and 98.1/million for population 50 years and older. As with the PSV examined in the present study, the incidence of GCA increased with age and peaked in the group of patients older than 75 years (202.2/million). In GCA, the incidence increases with latitude in the northern hemisphere, with more than a twofold increase in incidence in Scandinavian countries compared with Lugo in northwest Spain (1). It also seems to be the case for WG, as this vasculitis is more common in northern Europe than in the south (1). Thus, in both vasculitides, geographic variations and, specifically, the influence of latitude may account for the differences in the incidence.
In conclusion, our observations support an increasing incidence of PSV with age. Environmental factors in genetically predisposed populations may be responsible for the difference in the incidence of PSV in different populations.
The authors thank Drs. Javier Alvarez-Ferreira and Fernando Bal (pathology division) for their valuable help with performing this study. We thank Dr. Pablo Bouza (nephrology division) for his help in the study of patients with renal manifestations. We also thank Drs. Teresa Armada and Jaime Capella from the medical record department of the Hospital Xeral-Calde for their great help in the review of the medical records.