High risk of stroke in ANCA-associated vasculitis—a population-based study

Abstract

Objective

To determine the incidence rate, predictors and outcome of stroke in a population-based cohort of individuals with ANCA-associated vasculitis (AAV).

Methods

The study included 325 patients diagnosed with AAV from 1997 through 2016 in a defined geographic area of Sweden. Patients who suffered a stroke were identified from Riksstroke, a national Swedish stroke register established in 1994, and the Skåne Healthcare Register (SHR), which includes data for all inhabitants of Skåne since 1998. Case record review was carried out to confirm the diagnosis of stroke in AAV patients identified in the SHR. The incidence rate of stroke was calculated per 1000 person-years of follow-up. Using data from the Swedish general population, the standardized incidence ratio (SIR) of stroke was estimated. Cox regression analysis was utilized to investigate survival and predictors of stroke.

Results

Twenty-five subjects (8%) suffered a stroke during 2206 person-years of follow-up. The incidence rate of stroke in AAV was 11.3/1000 person-years (95% CI 6.9, 15.8). Patients with AAV showed an increased risk of stroke compared with the general population [SIR 1.85 (95% CI 1.27, 2.59)], with a greater risk for those <65 years of age [SIR 3.19 (95% CI 1.53, 5.88)]. Higher platelet count at AAV diagnosis was an independent predictor of stroke [hazard ratio 1.14 (95% CI 1.00, 1.29)]. There were no differences in survival or other outcome measures between AAV patients with and without stroke.

Conclusions

The incidence rate of stroke in AAV is higher than in the general population. High platelet count at AAV diagnosis was associated with an increased risk of stroke.

Introduction

ANCA-associated vasculitis (AAV) is an umbrella term for a group of autoimmune diseases characterized by inflammation of the small blood vessels, including granulomatosis with polyangiitis (GPA), microscopic polyangiitis (MPA) and eosinophilic granulomatosis with polyangiitis (EGPA). AAV is characterized by systemic inflammation causing dysfunction of or damage to multiple organs including the kidneys, lungs and upper respiratory tract.

Since the introduction of AAV treatment with cytostatic drugs such as cyclophosphamide combined with high doses of glucocorticosteroids, the prognosis and survival has improved dramatically [1], transforming AAV from a disease with 80% mortality in the 12 months following diagnosis to a more stable and manageable condition with a 5 year survival rate of 72% [2]. However, patients with AAV exhibit a higher rate of comorbidities compared with the general population. This includes a 4-fold rate of venous thromboembolic disease and a 5-fold rate of severe infection and septicaemia [3], comorbidities with consequences to the individual and to the economic and societal burden of AAV [4].

Research has shown an increased risk of cardiovascular events in individuals with AAV [5–7]. A population-based study of a cohort of AAV patients in southern Sweden demonstrated an increased rate of ischaemic heart disease compared with the general population as well as a more frequent presence of traditional risk factors for cardiac disease, including hypertension and diabetes [3]. An increased rate of atherosclerosis in patients with chronic inflammatory disease, including AAV, has also been identified and is a possible risk factor contributing to cardiovascular events in AAV [8, 9].

While the risk of cardiovascular events in AAV has been well researched, only a few studies have investigated the incidence of stroke in patients with AAV [6, 7, 10–12], and results have been inconsistent with respect to stroke risk. In addition, most of the available studies originate from tertiary referral centres and, unlike population-based studies, can be associated with selection and referral bias.

In this population-based study from southern Sweden utilizing a large, validated cohort of patients with ANCA-associated vasculitis, we aimed to determine the incidence rate of stroke in AAV compared with the general population and explore possible predictors and outcomes of stroke in AAV.

Methods

Patients

A total of 325 patients diagnosed with AAV within the study area from 1 January 1997 through 31 December 2016 were included in the study. All patient information was retrieved from two healthcare districts in Skåne, the southernmost region of Sweden with a population of ≈700 000. The study area and population, case retrieval and confirmation of AAV diagnosis have previously been described [13]. Briefly, subjects included all those with clinical diagnoses of small vessel vasculitis confirmed by organ biopsy or, if not available, by surrogate markers for granulomatous and/or vasculitic disease [14]. Using the European Medicines Agency algorithm, all patients were classified with respect to AAV disease phenotype: GPA, MPA or EGPA [14]. The time of AAV diagnosis was defined as the date of a positive biopsy confirming small vessel vasculitis or, if no biopsy was performed, as the start of immunosuppressive therapy for treatment of AAV.

Data collection

Data recorded at the time of AAV diagnosis, including demographics, organ involvement, medical treatment and relevant laboratory tests, were collected. In this study, stroke events occurring in a period beginning 3 months prior to AAV diagnosis were included in the analysis and were considered related to AAV. Results of testing for ANCA at various times during follow-up included values of proteinase-3 (PR3)- and myeloperoxidase (MPO)-ANCA. Disease activity was calculated based on the BVAS version 3 at the time of AAV diagnosis and at 12 months post-diagnosis [15]. Irreversible organ damage was estimated using the Vasculitis Damage Index (VDI) [16] at 12 months. The occurrence of end-stage kidney disease (ESKD) was also recorded, defined as initiation of renal replacement therapy or successful renal transplant. Patient records were reviewed from the time of diagnosis of AAV until 31 December 2018 or death as revealed by Swedish census data [17].

Data sources for identifying individuals with stroke

Riksstroke

Riksstroke is a Swedish national quality register initiated in 1994 that records every patient diagnosed with ischaemic or haemorrhagic stroke and admitted to a stroke unit in any Swedish hospital [18]. The register contains data on patient condition before the onset of stroke, during the hospital stay and at discharge, as well as data on in-hospital treatment and intervention.

Skåne healthcare register

The Skåne Healthcare Register (SHR) is a central database that records data from all healthcare practitioners in both hospital and outpatient clinics in Skåne [19, 20]. Information from medical records and other administrative sources of all healthcare consultations with physicians, nurses and allied health professionals, public or private, in the Skåne region is regularly transferred to this register. The SHR entries contain data on personal identification numbers, age, sex, place of residence, dates of clinic visits, admission and discharge and healthcare providers, as well as up to eight diagnostic codes using the International Classification of Disease, 10th Revision (ICD-10) for every consultation. The register constitutes the basis for reimbursements and thus can be considered comprehensive [21].

Data linking

Individuals who had suffered a stroke were identified by linking the AAV cohort to the Riksstroke register and SHR using personal identification numbers. Riksstroke collects data only on patients diagnosed with stroke at a stroke unit. To identify those who suffered a stroke but were not included in Riksstroke, the SHR was searched using ICD codes I60–I68. All case records for patients assigned stroke codes at the SHR were reviewed to confirm the diagnosis using the World Health Organization (WHO) definition of stroke [22]. Date of stroke was defined as the date of discharge from the hospital, as it is on this date that the ICD code is assigned. Transient ischaemic attack and subarachnoid bleeding were excluded from the study.

Ethics

The study was conducted in compliance with the criteria of the Declaration of Helsinki and approved by the Swedish Ethical Review Authority (2010-517). Written consent from participants was not required according to the ethical approval.

Statistical analysis

Categorical variables are presented as frequencies and percentages. Differences among groups in categorical variables were calculated using a χ² test or Fisher’s exact test when appropriate. Data for normally distributed variables are presented as mean (s.d.) and for non-normally distributed variables as median and interquartile range (IQR). Differences among groups were calculated by Student’s t-test and Mann–Whitney U-test when appropriate. The incidence rate of stroke in AAV patients was calculated using the number of cases as the numerator and the sum of patient years of follow-up as the denominator. The follow-up time was defined as the time from the date of AAV diagnosis to stroke, death or termination of the study on 31 December 2018, whichever occurred first.

We estimated the standardized incidence ratio (SIR) as the ratio of observed stroke events among patients with AAV to the expected number in the general population. In this analysis, the observed number of strokes was identified from the diagnosis database in the SHR and was based on assigned ICD codes. Only ICD codes assigned after hospitalization were considered. To avoid any bias when comparing the SIR between the observed number of strokes (AAV cases) and the expected number of strokes (background population), all cases with the aforementioned ICD codes of stroke were included in the analysis. The expected number of stroke events in the general population was calculated by multiplying the 5 year age-, sex- and 1 year calendar period–specific number of person-years by sex, time period and age-specific incidence rates of stroke in Skåne, Sweden. Fisher’s exact test was used to calculate 95% CIs. The SIR was calculated for age ≤65 vs >65 years at AAV diagnosis, sex and GPA, MPA, PR3-ANCA and MPO-ANCA positivity. Predictors of stroke were studied using a Cox regression model. Age at AAV diagnosis, sex, ANCA serology, creatinine level, platelet count and BVAS at diagnosis were analysed in a univariable model and in a multivariable analysis. Estimation of patient survival was conducted using Kaplan–Meier survival analysis with a logrank test to study differences among groups. To account for immortal time bias, a time-dependent Cox model was fitted. P-values <0.05 were regarded as significant. Statistical analysis was performed using Statistical Package for the Social Sciences for Windows version 26 (IBM, Armonk, NY, USA).

Results

Characteristics of AAV patients with and without stroke

Twenty-five individuals (8%) suffered a stroke following diagnosis of AAV. Demographics and clinical/laboratory features of patients with and without a stroke after AAV diagnosis are summarized in Table 1. All the stroke events were ischaemic, apart from one patient who also had minor cerebral bleeding secondary to ischaemic changes. Patients with and without stroke were comparable in demographic characteristics, organ involvement and laboratory parameters at the time of AAV diagnosis. The frequency of MPO-ANCA was higher in patients with stroke but did not reach significance. Similarly, an increased frequency of mucocutaneous, ENT, upper respiratory tract and nervous system involvement was observed in patients with stroke compared with those without, as well as a slightly increased platelet count and CRP in those suffering a stroke, although these findings did not reach significance (Table 1). Those with stroke showed a trend of a longer median follow-up period compared with those without stroke [9.00 years (IQR 5.50, 12.00) vs 6.00 (2.00–10.00), P = 0.06].

Incidence of stroke in AAV

During a total follow-up of 2206 person-years, 25 subjects were diagnosed with stroke. The incidence rate in AAV was 11.3/1000 person-years (95% CI 7.6, 16.7) and was higher in the 3 months post-diagnosis of AAV than in the entire follow-up period, at 78.5/1000 person-years (95% CI 15.7, 141.4). Beyond the first 3 months after AAV diagnosis, incidence rates were consistent throughout the study period (Table 2 and Fig. 1A). The incidence of stroke was 12.4/1000 person-years (95% CI 5.9, 18.9) among males vs 10.2 (95% CI 4.2, 16.2) in females. Incidence increased with age at AAV diagnosis and was estimated at 5.2/1000 person-years (95% CI 0.0, 11.2) in the age group <50 years, 11.3/1000 person-years (95% CI 5.8, 16.8) in age group 50–79 years and 27.6/1000 person-years (95% CI 5.5, 49.6) in those ≥80 years (Table 3, Fig. 1B).

SIR

In the diagnosis database of the SHR in the study area, 33 AAV patients [19 male (58%)] were assigned an ICD code of stroke after hospitalization. The SIR of stroke in AAV patients was 1.85 (95% CI 1.27, 2.59): 1.84 (95% CI 1.11, 2.87) for males and 1.86 (95% CI 1.02, 3.12) for females. The highest SIR was estimated in those <65 years of age at AAV diagnosis [3.19 (95% CI 1.53, 5.88)] (Table 4).

Predictors of stroke in AAV

In a univariate analysis, age at AAV diagnosis was associated with an increased risk of stroke [hazard ratio (HR) 1.36 (95% CI 1.01, 1.85)] for each 10 year increase in age. Serum creatinine at diagnosis had a positive association with stroke [HR 1.14 (95% CI 1.00, 1.29)] for each 100 µmol/l increment. There was a positive association between platelet count and stroke. In the multivariable analysis, each increase in platelet count of 50 × 10⁹/l was associated with a 14% increase in risk of stroke. This was the only independent factor to show a significant association with stroke (Table 5).

Outcome of AAV patients with stroke

The VDI was available for 245 of those without stroke and 22 of those with stroke. There were no differences in the median VDI at 12 months post-diagnosis in patients with and without stroke (Table 1). However, nine patients without stroke (4%) presented a VDI score ≥5, a score not reached in any patient with stroke.

Five (20%) of those with stroke developed ESKD compared with 46 (15%) of those without stroke (P = 0.53). The median follow-up from diagnosis to death or end of study was 6 years (IQR 2–10) for all patients: 6 years (IQR 2–10) for individuals without stroke vs 9 years (IQR 5.5–12) for those with stroke. A total of 143 patients died during the follow-up period (Table 1). In the time-dependent Cox regression model, the presence of stroke after AAV diagnosis was associated with an increased risk of death [HR 2.6 (95% CI 2.16, 3.12)]. When the analysis was adjusted for stroke events prior to AAV diagnosis, the association remained unchanged [HR 2.32 (95% CI 1.28, 4.22)].

Discussion

In this population-based cohort study including 325 patients with AAV, we studied the occurrence of stroke and its predictors using validated sources of information. The incidence rate of stroke in AAV was higher than that of the general population in the studied region [23]. The incidence of stroke in the AAV cohort was greatest during the 3 months following AAV diagnosis and higher in patients >80 years of age at diagnosis than in younger patients. The risk of developing stroke among patients with AAV was three times that of the background population in persons <65 years of age at diagnosis. Among predictive factors for stroke, age at diagnosis, high creatinine and high platelet count were associated with an increased risk of stroke. However, in an adjusted multivariate model, only platelet count was associated with an increased risk of stroke: 14% for each increment of platelets of 50 × 10⁹/l.

The incidence of stroke in AAV was considerably higher than the 1.65/1000 person-years in a segment of the general population in the study area recently reported by Aked et al. [23] There are likely several factors that contribute to the higher rate of stroke in AAV. Atherosclerosis has been demonstrated to occur frequently in AAV [8, 9], possibly a direct result of local vascular inflammation caused by the vasculitis [8]. Long-term exposure to glucocorticoids may also accelerate atherosclerosis [24]. Increased endothelial dysfunction, which is proposed to increase the risk of thrombosis, has been observed in AAV [25]. A higher risk of stroke in AAV could be a result of systemic inflammation, but also of other consequences of high inflammation such as thrombocytosis. One of our main findings is that an increased platelet count was associated with an increased risk of stroke. Inflammation is known to drive the formation of thrombi, and platelets are a requisite for the formation of thrombi in blood vessels [26]. We recently demonstrated high rates of venous thromboembolic disease in patients with AAV [27], with both advanced age and high vasculitis activity as independent predictors of thromboembolic disease. Platelets in the setting of high inflammatory activity could be a key factor in the incidence of stroke events in AAV, and investigation of the prophylactic efficacy of anticoagulant drugs is needed. Finally, it cannot be excluded that cerebral vasculitis, a rare manifestation of AAV, could be a cause of stroke [28].

The observed incidence rate of stroke in AAV is in accordance with previous studies that estimated incidences of 8.9 [7] and 10.2/1000 person-years [10]. Other studies have reported a much lower incidence of stroke in AAV [11, 12]. The differences may be explained by the age of our patients (mean 66 years) compared with that of the patients in the cited studies (mean age 54 and 44, respectively). Our case identification, with a search of two registries including a national stroke register, together with case record review, may have contributed to the higher rate of stroke observed. We found older age at AAV diagnosis predictive of stroke and the highest incidence to be within the first months post-diagnosis. Our results confirm previous findings that showed the incidence of stroke, myocardial infarction and venous thromboembolic events to be highest early in the course of the AAV disease compared with the whole follow-up period [5, 10, 27]. This strengthens the hypothesis that inflammation is the driving factor of the increased incidence of stroke seen in AAV. Although we did not observe an association of baseline BVAS with stroke, such a relationship has been shown for cardiovascular risk. Cardiovascular disease (CVD) is the most common cause of death in patients with AAV, with MPO-ANCA positivity being associated with a higher risk of death from CVD than PR3-ANCA-positive patients [29]. However, because of the low number of events in our study, we could not demonstrate a correlation between baseline BVAS and a risk of stroke or differences in the incidence rate of stroke in patients with MPO-ANCA compared with PR3-ANCA.

We estimated the SIR of stroke in AAV, which is a robust method of comparing the incidence of a disease in a target cohort with that of the general population, taking into consideration factors affecting risk other than exposure, such as sex, age and year of diagnosis. An increased SIR of stroke was found for the entire cohort, especially in individuals <65 years of age at AAV diagnosis (SIR 3.2).

Our study has several strengths and certain limitations. We used a large well-established population-based cohort of AAV patients that included all three of the disease phenotypes of AAV with low or no selection bias. In addition to records review, the case identification of stroke was obtained from two validated registries, minimizing the risk of misclassification of stroke. By review of the registries and only including cases fulfilling the WHO definition of stroke, the validity of the stroke diagnosis was confirmed and the risk for detection bias was minimized.

A limitation of the study was the small number of events in our cohort, necessitating caution in drawing firm conclusions. The retrospective nature of the study also meant missing data regarding traditional risk factors for stroke, which could have led to BVAS estimates being skewed or incomplete. Another limitation is that when estimating the SIR of stroke, no matching was carried out for risk factors such as smoking and hypertension.

Our study revealed an increased incidence of stroke in AAV, especially during first few months following diagnosis, and platelet count was shown to be an independent predictor of stroke. We identified a higher risk of stroke in AAV patients compared with the general population in individuals <65 years of age at diagnosis. This study adds to the evidence that stroke is an important complication of AAV and emphasizes the need to investigate the prophylactic effectiveness of antithrombotic and anticoagulant drugs in AAV.

Data availability

Raw data are protected by confidentiality laws in Sweden and cannot be shared. All data relevant to the study are included in the article. Please contact the corresponding author.

Authors’ contributions

All authors were involved in the study conception and design, interpretation of data and drafting the article or revising it for intellectual content. All authors approved the final version to be published. D.T. and A.J.M. were responsible for data acquisition and analysis.

Funding

This study was supported by grants from the Swedish Research Council (Vetenskapsrådet: 2019-01655), Faculty of Medicine, Lund University (ALF-medel), Alfred Österlund’s Foundation, King Gustaf V’s 80-year Foundation, Anna Lisa and Sven-Eric Lundberg Foundation and STROKE-Riksförbundet (Swedish Stroke Association) (all to A.J.M.).

Disclosure statement: The authors have declared no conflicts of interest.

References