Primary systemic vasculitis: clinical features and mortality

Q J Med 2005; 98:97 111 doi:10.1093/qjmed/hci015 Advance Access publication 17 January 2005 Original papers Primary systemic vasculitis: clinical features and mortality S.E. LANE, R.A. WATTS, L. SHEPSTONE 1 and D.G.I. SCOTT 2 From the Department of Rheumatology, Ipswich Hospital, Ipswich, 1 School of Medicine, University of East Anglia, Norwich, and 2 Department of Rheumatology, Norfolk and Norwich University Hospital, Norwich, UK Received 2 February 2004 and in revised form 25 August 2004 Summary Background: Wegener s granulomatosis (WG), Churg Strauss syndrome (CSS) and microscopic polyangiitis (MPA) are primary systemic vasculitides (PSV), the clinical features of which have been described from tertiary centres. Aim: To provide the first clinical description of MPA from a general hospital and compare clinical features with WG and CSS. Design: Retrospective analysis of patient records. Methods: Records of 99 PSV patients attending a single hospital, from 1988 to 2000, were reviewed for: clinical features, date/age at diagnosis, sex, duration of illness, anti-neutrophil cytoplasmic antibodies (ANCA), treatment, comorbidity and deaths. Cases were classified using ACR, CHCC and Lanham criteria/definitions. Birmingham vasculitis activity scores (BVAS) and damage index (VDI) were calculated. Survival was assessed using Cox proportional hazards model and standardized mortality ratios (SMRs). Results: Compared to previous reports there was more ENT (29%) and respiratory (29%) but less renal (92%) involvement in MPA, and less ENT involvement in WG (81%). CSS showed high neurological (72%), cardiovascular (28%) and gastrointestinal (17%) involvement and the highest median (range) VDI (p ¼ 0.01 vs. WG; p ¼ 0.001 vs. MPA). BVAS1 was significantly lower in MPA than in WG [median (range) 15 (4 29) vs. 21 (6 39), (p ¼ 0.001)] but not in CSS [20 (7 28), p ¼ 0.08]. SMR (95%CI) for PSV was 4.8 (3.0 6.6); 5-year survival was 45.1% for MPA, 75.9% for WG and 68.1% for CSS. Age was a significant risk, but only to the same extent as in the reference population. When age was adjusted for, no other significant factor was found. Discussion: The clinical characteristics seen here are similar to those in previous series. There are difficulties in using the MPA CHCC definitions in classification. There is a high proportion of neurological involvement in CSS, causing permanent damage. MPA may have a poorer prognosis than WG or CSS. Introduction The primary systemic vasculitides are heterogeneous, multi-system disorders characterized by inflammation and necrosis of small and medium blood vessels. Their aetiology is unknown. Historically, the term polyarteritis nodosa was used collectively, but three distinct clinico-pathological syndromes, often associated with anti-neutrophil cytoplasmic antibodies (ANCA), have been Address correspondence to Dr S.E. Lane, Department of Rheumatology, Ipswich Hospital, Heath Rd, Ipswich IP4 5PD. e-mail: suzanne.lane@ipswichhospital.nhs.uk QJM vol. 98 no. 2! Association of Physicians 2005; all rights reserved.

98 S.E. Lane et al. identified: Wegener s granulomatosis (WG), Churg Strauss syndrome (CSS) and microscopic polyangiitis (MPA). 1 Although there is no ideal classification system for systemic vasculitis, the American College of Rheumatology (ACR) published classification criteria for WG, CSS and polyarteritis nodosa (PAN), but not MPA, and the Chapel Hill Consensus conference (CHCC) recommended definitions for WG, CSS and MPA. 2 5 The CHCC definitions were not intended for classification, but provide a method of describing MPA. Lanham et al. provided a clinical description for CSS in 1984. 6 The Birmingham Vasculitis Activity Score (BVAS) and Vasculitis Damage Index (VDI) are tools that have been developed to estimate the severity of disease. 7 Together, these criteria/definitions allow comparison of research findings between centres, although none incorporate ANCA. The use of cyclophosphamide and other immunosuppressive agents has transformed the prognosis of PSV, although there remains considerable morbidity associated with both disease and treatment. 8 Reports suggest that prognosis is worse for elderly patients, those with renal disease (especially high creatinine at presentation), pulmonary involvement, high ESR and a diagnosis of MPA. 9 14 Our aim was to study a series of patients with primary systemic vasculitis (PSV) in Norfolk, UK, using published criteria/definitions to provide the first description of MPA from a general hospital, and to study the clinical features of WG and CSS without the effect of tertiary referral bias inherent in previous reports. Methods Population information The Norfolk and Norwich University Hospital (NNUH) is the single, central referral centre for the former Norwich Health Authority (NHA) population previously used to study the epidemiology of PSV and rheumatoid arthritis. 15,16 The NHA provided administrative services to a group of general medical practices serving a population of 415 000 people. Despite the abolition of the NHA, details of the denominator population registered with these practices could be obtained from the successor organization, the East Norfolk Health Authority. In the UK, patients are referred by their family practitioner for secondary care to the nearest district general hospital. At the time of the study, the NNUH was a district general hospital providing services for the entire NHA population. Patient identification All patients who had been diagnosed with any type of vasculitis at the NNUH between May 1988 and July 2000 were registered at presentation. Patients were referred directly to the register by hospital consultants involved in their care. In addition, the computerized records of the histopathology department were searched for patients with a histological diagnosis of systemic vasculitis (renal and skin biopsy). Written records of renal biopsies and plasmapheresis episodes were also reviewed, and case notes identified were examined. The hospital discharge diagnostic index was searched for patients with a discharge diagnosis of PSV (WG, PAN, CSS) and renal biopsy using the International Classification of Diseases clinical modification codes (ICD 9 and 10) (previously published). 17 Case ascertainment was felt to be high, due to good communication between hospital specialities and between local primary and secondary care. Private health care in Norfolk is provided by the same consultants who work at NNUH. However, it is not possible to be entirely certain that no patient had all their care outside the NNUH, although this is thought unlikely, due to the distances involved. It was not possible to perform a capture recapture analysis to assess completeness of case identification for two reasons: firstly the sources of case identification were interdependent, and not independent as required for capture recapture analysis; secondly the source of the initial referral of each case to the register (i.e. consultant referral, histopathology records, hospital discharge data) was not reliably recorded for every patient. Case note review and disease classification A diagnosis of PSV was established in 166 patients following case-note review. Residence within the NHA was defined by the location of the general practice of each patient at the time of diagnosis, and the postcode of residence at the time of first symptom onset was also noted. In 18 cases, the diagnosis had been made prior to May 1988, and 49 lived outside the NHA (9 were tertiary referrals). In 16 cases, recorded details were insufficient for classification (all also excluded due to residence outside the NHA). For the 99 PSV cases resident in the NHA between May 1988 and July 2000, details of symptoms and organ involvement were recorded with a corresponding date using a BVAS form. 7 Results relevant to the classification of PSV

Primary systemic vasculitis 99 (radiology, histology, hepatitis B serology, urinalysis and eosinophil count) were also recorded. There are two main systems used to classify/ define PSV; the ACR classification criteria (1990) for WG, CSS and PAN and the CHCC definitions (1994) for WG, CSS, MPA and PAN. 2 5 MPA was not included in the ACR classification criteria. An additional classification system has been published by Lanham et al. for CSS. 6 We applied every classification criteria to each case. It should be noted that ANCA status is not part of any of these classifications/definitions. For the purposes of classification according to the ACR classification criteria, a person is said to have WG if at least two of the following four criteria are present: nasal/oral inflammation (painful or painless oral ulcers or purulent or bloody nasal discharge); abnormal chest X-ray (chest radiograph showing nodules, fixed infiltrates or cavities); urinary sediment; granulomatous inflammation on biopsy (histological changes showing granulomatous inflammation within the wall of an artery or in the perivascular or extravascular area (artery or arteriole). For CSS, four of six items must be present: asthma; eosinophilia (410% on white blood cell count); mononeuropathy or polyneuropathy; pulmonary infiltrates (non-fixed, i.e. migratory or transitory); paranasal sinus abnormality (pain, tenderness or radiographic opacification); extravascular eosinophils (biopsy including artery, arteriole or venule, showing accumulation of extravascular eosinophils). For PAN, three of ten criteria must be present: weight loss 44 kg; livedo reticularis; testicular pain/tenderness; myalgias; weakness or leg tenderness; mononeuropathy or polyneuropathy; diastolic blood pressure 499 mmhg; elevated urea and creatinine; hepatitis B virus, arteriographic abnormality; biopsy of small- or medium-sized artery containing polymorphonuclear neutrophils. The CHCC definitions were not designed to be used as classification criteria and their application can be difficult. 18 20 They are as follows: 5 Wegener s granulomatosis, Granulomatous inflammation involving the respiratory tract and necrotizing vasculitis affecting small to medium sized vessels (e.g. capillaries, venules, arterioles and arteries). Necrotizing glomerulonephritis is common) ; microscopic polyangiitis, Necrotizing vasculitis, with few or no immune deposits, affecting small vessels (i.e. capillaries, venules or arterioles). Necrotizing arteritis involving small and medium sized arteries may be present. Necrotizing glomerulonephritis is very common. Pulmonary capillaritis often occurs ; Churg Strauss syndrome, Eosinophil-rich and granulomatous inflammation involving the respiratory tract, necrotizing vasculitis affecting small to medium sized vessels, and associated with asthma ; classic polyarteritis nodosa, Necrotizing inflammation of medium sized or small arteries without glomerulonephritis or vasculitis in arterioles, capillaries, or venules. When we applied the CHCC definitions, we classified cases as CHCC WG only when there was histological evidence of granulomatous inflammation. Cases with histological evidence of necrotizing vasculitis and/or necrotizing glomerulonephritis in the absence of granulomatous change were classified as CHCC MPA. Patients sometimes fulfil more than one set of criteria; in particular, PAN and MPA commonly meet ACR criteria for WG and CSS (Figure 1). 1,18 For the purpose of the study, we needed to give cases a final diagnosis of WG, MPA or CSS. To achieve this, we first applied ACR criteria for WG and, CSS, which have been validated, and Lanham criteria for CSS. Patients who met one of these classifications were allocated the corresponding diagnosis. Where cases fulfilled criteria for both WG and CSS, the clinical records were reviewed and a final diagnosis was attributed. For those who did not fulfil these criteria, the CHCC definitions for WG and MPA were applied, and the respective diagnosis given as appropriate. Where cases failed to fulfil criteria/ definitions for WG, CSS or MPA, but met ACR criteria for PAN, clinical features were reviewed to allocate a diagnosis of WG, MPA or CSS (Figure 1b). Case notes were reviewed independently by two observers (SL and RW). RW, an external observer not involved in the patients clinical care, had the final decision. Patients who did not fulfil any of the classification criteria/definitions were excluded. The case notes were reviewed during 1999 2000. Clinical characteristics The following details were recorded: date and age at diagnosis, sex; hospital department of first referral; organ systems involved in vasculitis at presentation and throughout disease course (using BVAS form); permanent damage of organs (using Vasculitis Damage Index (VDI) form 7 ); duration of illness at case note review; and ANCA type (immunofluorescence (IIF) and/or proteinase 3 (PR3)/myeloperoxidase (MPO) specificity ELISA). Information about clinical features was supplemented by interviewing 67 surviving patients between 1998 and 2000 (one patient declined an interview). For analysis, organ involvement was defined as the presence of one or more items within each category of the BVAS form (systemic, cutaneous, mucous membranes, ENT, chest, cardiovascular, abdominal, nervous system). The exception is renal disease, which was defined

100 S.E. Lane et al. (a) 33 WG 21 ACR 11 ACR & CHCC 1 CHCC (15 also PAN) 22 WG*/MPA (19 also PAN) 23 MPA CHCC (14 also PAN) 0 1 WG*/CSS** (and PAN) 1 MPA/CSS** (and PAN) 17 CSS 1 ACR 3 Lanham 9 ACR & Lanham 4 ACR & Lanham & CHCC (8 also PAN) 2 PAN only Refer to Figure 1b 57 WG 33 WG 22 WG / MPA 1 WG/CSS 1 PAN 24 MPA (b) 55 ACR WG 18 CSS 17 CSS 1 CSS / MPA Cases with a clinical diagnosis of PSV 99 26 Do not fulfil ACR / Lanham WG or CSS 18 ACR and / Lanham CSS Fulfils any CHCC definitions? NO 2 ACR PAN YES 24 Clinical Review (1) (1) CHCC WG 1 CHCC MPA 23 Study Diagnosis WG 57 MPA 24 CSS 18 Figure 1. a Classification of patients. The Venn diagram illustrates all classification criteria fulfilled. The lower diagram shows the groups used in analysis. b Method used to obtain these groups. ACR, American College of Rheumatology 1990 criteria; CHCC, Chapel Hill Consensus definition; WG, Wegener s granulomatosis; MPA, microscopic polyangiitis; CSS, Churg Strauss syndrome; PAN, polyarteritis nodosas (ACR). *WG ACR; **CSS ACR & Lanham.

Primary systemic vasculitis 101 as the presence of haematuria, proteinuria, creatinine 4125 or rise in creatinine 430% or creatinine clearance fall 425% attributable only to vasculitis, as per the BVAS form, but hypertension was not included and recorded separately. In the analysis, the nervous system was sub-divided into central nervous system (organic confusion/dementia, seizures (not hypertensive), stroke, cord lesion and cranial nerve palsy) and peripheral nervous system (sensory peripheral neuropathy and motor mononeuritis multiplex). The number of deaths each year and the likely causes and comorbidity that may have contributed towards death (e.g. diabetes, ischaemic heart disease) was also recorded. Data analysis Patient characteristics The male: female ratio and mean, median and age range were calculated for total PSV, WG, MPA and CSS. Clinical features Clinical features were documented as a proportion of the total for PSV, WG, MPA and CSS for both the time of diagnosis and throughout the disease course. The categories (systemic, cutaneous, mucocutaneous, ENT, respiratory, cardiovascular, abdominal, renal and neurological) refer to those documented on the Birmingham Vasculitis Activity Score (BVAS) form, and involvement was defined as the presence of one or more clinical features. 7 The exception was in the renal category, where hypertension was excluded and documented separately. BVAS scores were calculated for each patient at diagnosis. The BVAS form was used in two ways. The first score (BVAS1) included symptoms that had developed or deteriorated within 1 month of diagnosis only. This excluded some grumbling disease that had been present since the Index Date. Therefore a second score (BVAS 1þ2) was calculated to include all clinical features attributable to vasculitis that had occurred prior to diagnosis. VDI scores were also established where symptoms attributable to vasculitis were present for at least 3 months. The mean, median and range of values were calculated for total PSV, WG, MPA and CSS for the following items: BVAS; VDI; number of organ systems (defined by the BVAS categories) involved at any time during the disease course; duration of the prediagnostic symptomatic period; and duration of disease at case note review. The first symptom attributable to vasculitis (index symptom), the hospital speciality of first referral, symptoms present at diagnosis and throughout disease course were calculated by organ system as a proportion of total PSV, WG, MPA and CSS. Comparison was made between disease types for each item analysed using the Mann-Whitney test. Comparison was made between surviving and deceased patients. Mortality Age- and gender-standardized mortality ratios (SMR, age-group size 5 years) were calculated for PSV patients diagnosed in the 10-year period between January 1989 and December 1998 compared to the NHA population, by Indirect Standardization, using the following equation. s ¼ r STUDY r REF P I i 1 ¼ n i P I i 1ð N im i =M i Þ where s ¼ standardized mortality ratio, r REF ¼ death rate expected if PSV mortality ¼ NHA mortality, n ¼ PSV deaths, N ¼ PSV patients, m ¼ NHA deaths, M ¼ NHA population, and i ¼ 5-year age group. The numbers of PSV cases (N i ) and PSV deaths (n i ) were recorded for each year (i ¼ 1989 1999). NHA Population estimates and mortality figures by agegroup for 1994 were obtained from the Norfolk City Council Demographics Unit and assumed to be stable throughout the study period. The denominator population (M i ) and number of expected NHA deaths (m i ) were taken to be the 1994 NHA population for each year. A Poisson distribution was assumed to obtain 95%CIs. Comparison was made between SMRs for cases 565 years of age at diagnosis, men and women, and disease classifications, by obtaining z values. BVAS and VDI scores were not included in the analyses, as they were calculated retrospectively, which may lead to bias. Kaplan-Meier curves were constructed for the 99 cases identified between May 1988 and July 2000, and the Cox proportional hazards model was used to compare survival by classification, sex, age, ANCA type, presence or absence of comorbidity, renal or respiratory involvement. Treatment Data were not analysed according to treatment received, because therapeutic regimens varied between individual patients and hospital specialities, according to disease severity and developments in treatment protocols over the study period. Three regimens were used: regimen 1, intravenous

102 S.E. Lane et al. (IV) cyclophosphamide (CYC), 15 mg/kg up to a maximum of 1 g, with IV methylprednisolone 1 g and mesna 400 mg (3 doses at 4-h intervals, commencing 2 h pre infusion), with a reducing dose of daily oral prednisolone; regimen 2, pulsed oral CYC (5 mg/kg) and prednisolone 100 mg on three consecutive days; regimen 3, continuous oral CYC 2 mg/kg with prednisolone reducing dose. Regimens 1 and 2 were both given at fortnightly intervals until remission induction (usually six pulses) then 3-, 4- and 6-weekly intervals for maintenance up to one year. The precise dose and duration of treatment was tailored individually according to clinical response. In some cases with renal involvement, methylprednisolone 1 g was administered on three consecutive days prior to commencing treatment. Maintenance therapy was with azathioprine or methotrexate. Additional plasmapheresis and/or intravenous immunoglobulin was administered in a minority of resistant cases during disease flares. There was no strict protocol for septrin prophylaxis for Pneumocystis carinii. Ethics Case-note review was undertaken as part of an ongoing research project registered with the Norfolk and Norwich Research and Development department, and ethical approval was confirmed by the Norwich District Ethics Committee. Informed consent was obtained for patient interviews approved separately by both committees. Results Classification Biopsies were available in 87 cases, and histological findings supported vasculitis in 78 cases, with 17 patients having more than one biopsy. Sites of biopsy were: 63 kidney (4 negative for vasculitis), 16 ENT (1 negative) 16 skin (2 negative), 5 lung, 3 nerve (2 negative) and 1 rectum. Figure 1 gives the results of applying all classification criteria/definitions to all cases and demonstrates the overlapping nature of classification systems. It also illustrates how cases were divided into the diagnoses of WG, CSS and MPA for the analyses. A detailed explanation follows. Polyarteritis nodosa Although 60 cases fulfilled ACR classification criteria for PAN, none met the CHCC definition for PAN. No patient was hepatitis-b-positive and, where results were available, there was no angiographic evidence of PAN. With the exception of two cases, all patients also fulfilled criteria for another diagnosis. The lack of specificity of the PAN criteria has previously been discussed, 1 and we did not analyse PAN as a separate group. One individual who could only be classified as PAN followed a clinical course consistent with WG. and was therefore included in the WG group. His clinical picture was of weight loss, bloody nasal discharge, opacification of the paranasal sinuses on X-ray, myalgia, hypertension and neuropathy. Serology was positive for canca. Subsequent to the completion of the study he developed focal segmental necrotizing glomerulonephritis. The second patient could not fulfil CHCC definitions because she was too unwell to have a biopsy. Her clinical course was of myalgias, arthralgias and malaise, then development of hypertension, renal failure and haemoptysis. Immunofluorescence revealed canca pattern but ELISA was not known. She was treated with aggressive immunosuppression but required dialysis, and subsequently went on to have pulmonary haemorrhage requiring plasmapheresis. In view of the absence of evidence for granulomatous vasculitis, she was considered to have MPA. Wegener s granulomatosis Overall, 57 patients were considered to have WG. One, already described, could be classified only as PAN, and one with limited WG with necrotizing granulomata formation on maxillary biopsy fulfilled only CHCC WG. The remaining 55 patients fulfilled ACR criteria for WG. Of these, 11 who had granulomatous inflammation on biopsy were also defined as WG CHCC, and 22 who had arteritis/venulitis and/or necrotizing glomerulonephritis on biopsy were defined as CHCC MPA. For the purposes of the study, the 22 patients who fulfilled ACR WG and CHCC MPA were included in the WG group. The remaining patient who fulfilled ACR criteria for WG also fulfilled both ACR and Lanham criteria for CSS. She initially presented with asthma, eosinophilia and neuropathy, but went on to develop renal failure, caused by necrotizing glomerulonephritis with vasculitis and eosinophilia, and a cavitating nodule in the lung. Granulomatous inflammation was also noted on histology. ELISA showed a raised PR3. She was included in the WG group because asthma was not a predominant feature, cavitation and PR3-ANCA were present, and she required aggressive immunosuppression. Churg Strauss syndrome A total of 18 patients were considered to have CSS; although 19 patients fulfilled ACR and/or Lanham criteria for CSS (4 also fulfilled the CHCC definition

Primary systemic vasculitis 103 for CSS), one, previously described, also fulfilled ACR criteria for WG and was assigned a diagnosis of WG for analyses. One patient, whose renal biopsy showed focal segmental proliferative glomerulonephritis with few eosinophils and low-grade arteriolitis, also met our definition for CHCC MPA, but as he also fulfilled both ACR and Lanham criteria for CSS and the clinical course included late-onset asthma, peripheral blood eosinophilia, pulmonary infiltrates and pericarditis, he was included in the CSS group for analyses. Microscopic polyangiitis We included 24 patients in the MPA group; 23 fulfilled the CHCC MPA definition with histological evidence of arteritis/venulitis or a necrotizing glomerulonephritis (14 of these cases also fulfilled ACR criteria for PAN but as discussed in the methods, MPA took precedent). One patient (already discussed) did not have a biopsy. Clinical features All patients were Caucasian, with a predominance of males for all diagnoses. Patient characteristics, duration of follow-up and time elapsed between the first symptom of vasculitis and diagnosis (prediagnostic period) are shown in Table 1. Figure 2 illustrates the first symptom attributable to vasculitis (index symptom) and the hospital speciality of first referral. The most common first symptom and hospital speciality to review cases of WG was ENT and MPA was renal. The first symptom of CSS was most commonly respiratory, and patients were commonly seen by the rheumatology team first, or another medical speciality. Table 2 shows organ system involvement at presentation and throughout disease course. There was significantly more mucocutaneous involvement in WG compared to MPA or CSS (p ¼ 0.0003 and 0.002, respectively), while ENT disease and respiratory vasculitis were significantly more common in WG than in MPA (p ¼ 0.0001), but not CSS Table 1 Demographics Total PSV WG MPA CSS n 99 57 24 18 Males 61 (61.1%) 33 (57.9%) 16 (66.7%) 12 (66.7%) Age (years) Mean 62.6 61.0 69.0 59.5 Median 65 62 70 62 Range 19 90 19 90 44 87 33 80 Follow-up (months) Mean 39.1 37.6 37.6 45.6 Median 26.0 24.0 25.0 41.0 Range 1 136 1 136 1-107 1 132 Prediagnostic period* Mean 8.7 10.3 6.7 6.5 Median 4 4.5 6 2 Range 0 61 0 59 0 24 0 61 *Not known in 9 cases (5 WG, 4 MPA). PSV, primary systemic vasculitis; WG, Wegener s granulomatosis; CSS, Churg Strauss syndrome; MPA, microscopic polyangiitis; ENT, ear, nose and throat. (a) ENT Systemic Respiratory Mucocutaneous Renal Abdominal Cutaneous WG MPA CSS Neurological Other Unknown 0 5 10 15 20 25 30 35 40 % Figure 2. a First symptom of vasculitis. b First hospital specialty. ENT, ear, nose and throat; WG, Wegener s granulomatosis; MPA, microscopic polyangiitis; CSS, Churg Strauss syndrome. Categories used refer to Birmingham Vasculitis Activity form groups, see text.

104 S.E. Lane et al. (b) ENT General medicine Respiratory Renal Gastroenterology Rheumatology Ophthalmology WG MPA CSS Neurology Other medical Surgical Figure 2. Continued. 0 5 10 15 20 25 30 35 % Table 2 Clinical features of primary systemic vasculitis Classification PSV Total WG MPA CSS n 99 57 24 18 Systemic 94/94 96/96 88/88 94/94 Cutaneous 38/41* 42/47* 25/25 44/44 Mucocutaneous All 32/37** 53/58** 4/13 11/11 Ophthalmic 24/28** 40/46** 4/8 6/6 ENT All 59/63 77/81 29/29 44/55 Nasal disease 49/52 67/70 21/21 33/39 Deafness/otitis 23/26 30/32 17/17 17/28 media Respiratory All 61/62* 63/67* 29/29 100 Haemoptysis 26/28 37/37 17/17 11/17 Nodules/cavitation 15/16 25/26 0/0 0/0 Infiltrates/effusion 21/22 37/56 13/13 17/22 Abdominal 7/9 5/7 8/8 11/17 Cardiovascular 5/7 2/2 0/4 22/28 Renal All 73/78 77/86 92/92 33/33 Hypertension 26/29* 23/28 38*/38 22/22 Neurological All 26/34 19/28 17/21 67/72 Central 8/9 5/5 4/8 22/22 Peripheral 20/30 12/25 17/17 56/67 Categories refer to groups used by Birmingham Vasculitis Activity Score form, see text. Data are percentages at presentation/percent throughout disease course. PSV, primary systemic vasculitis; WG, Wegener s granulomatosis; CSS, Churg Strauss syndrome; MPA, microscopic polyangiitis; ENT, ear, nose and throat. *In one case, data not available. **In two cases data not available. Table 3 Scoring for different forms of primary systemic vasculitis Score WG MPA CSS BVAS1 21 (6 39) 15 (4 29) 20 (7 28) BVAS1 þ 2 22 (6 39) 15 (4 29) 22 (7 34) VDI 2.0 (0 5) 1 (0 3) 3 (1 9) Systems affected 5 (1 7) 5 (3 8) 3 (2 7) Data are medians (range). WG, Wegener s granulomatosis; CSS, Churg Strauss syndrome; MPA, microscopic polyangiitis. (p ¼ 0.0033). In CSS, involvement of the cardiovascular system was more frequent than in WG (p ¼ 0.045) and neurological disease was significantly higher that either WG or MPA (p ¼ 0.0001 and p ¼ 0.0002, respectively). Renal vasculitis occurred more often in both MPA and WG than in CSS (p ¼ 0.03 and p ¼ 0.01, respectively). In addition, all CSS patients had asthma, 53.3% had rhinitis, 66.7% a personal history of allergy and 46.7% a family history of asthma (data not available for three patients). BVAS 1 was significantly lower in MPA than WG (p ¼ 0.001) but not CSS (p ¼ 0.08) Table 3). Both BVAS 1 þ 2 and VDI were significantly lower in MPA than WG (p ¼ 0.0001 and p ¼ 0.01, respectively) and CSS (p ¼ 0.007 and p ¼ 0.001, respectively) (Table 3). Significantly more systems were affected in WG and CSS than in MPA (p ¼ 0.0002 and p ¼ 0.0001, Table 3). Duration of follow-up was similar between groups (Table 1, WG vs. MPA, p ¼ 0.73; WG vs. CSS, p ¼ 0.18; MPA vs. CSS, p ¼ 0.47).

Primary systemic vasculitis 105 ANCA were positive for 73/94 PSV cases (77.7%), results being unavailable for five patients (3 WG, 1 MPA, 1 CSS). In WG, 88.9% of patients were ANCA-positive: 31 (57.4%) canca/pr3-anca, 9 (16.7%) panca/mpo-anca, 8 (14.8%) undetermined ANCA. In MPA, 73.9% of patients were ANCA-positive: 6 (26.1%) canca/pr3-anca, 7 (30.4%) panca/mpo-anca, 4 (17.3%) undetermined ANCA. In CSS, 47.1% of patients were ANCA-positive: 5 (29.4%) panca/mpo-anca, 3 (17.6%) undetermined ANCA. Mortality There were 31 deaths, of which 16 were associated with vasculitis (11 active vasculitis/progressive disease, 5 associated with previous vasculitis damage), 8 potentially associated with immunosuppression and 7 unrelated or unknown causes of death (Table 4). Of seven deaths associated with infection, only three were currently being treated with CYC (regimen 3): all others were receiving prednisolone alone, although they had received oral CYC between 6 months and 4 years earlier. Of 16 who died of vasculitis, only two were not currently being treated with CYC. In these cases, death was associated with the effects of permanent damage caused by vasculitis rather than active disease. The SMR (95%CI) for PSV patients compared to the NHA population was 4.8 (2.9 6.6), and was higher for men than women, at 5.9 (3.1 8.8) vs. 3.05 (1.2 4.9), although the difference was not significant (p ¼ 0.09). Survivors had significantly more cutaneous, mucocutaneous and nasal disease than deceased patients (Figure 3), both at presentation (p ¼ 0.03, 0.009 and 0.003, respectively) and throughout their illnesses (p ¼ 0.04, 0.006 and 0.003 respectively). There was significantly less abdominal involvement in survivors than in deceased patients (p ¼ 0.0019 and 0.017). There was no significant difference between BVAS or VDI scores between survivors and deceased cases, but survivors were followed up for a significantly longer duration (p ¼ 0.0005) and had more organ system involvement than survivors. There were no significant differences in SMRs between WG, MPA or CSS, or by age group (565 years vs. 465 years). Mean survival was 51.5 months for all PSV and 1 year survival was 84.8% overall, falling to 65.5% at 5 years. Survival was similar between WG, MPA and CSS at 1 year (85.5%, 82.7% and 83.2%, respectively) but was markedly reduced in MPA at 5 years (45.1%) compared to WG (75.9%) and CSS (68.1%). Figure 4 illustrates this difference. The Cox proportional hazards model demonstrated a significant increase in mortality with age, with a highly significant increase in risk comparing age groups over and under 65 years old [HR (95%CI) 9.9 (3.6 27.0), p50.001] and with increasing age [per year HR (95%CI) 1.1 (1.0 1.1), p50.001]. There were no significant differences between men and women (p ¼ 0.92) or diagnoses [WG vs. CSS, p ¼ 0.82; MPA vs. CSS, p ¼ 0.09]. There was no significant increase in risk for presence of renal disease or trend towards increasing risk with highest level of creatinine during the study period. Similarly, there was no significant risk for respiratory disease, panca/pr3-anca vs. panca/mpo-anca, or the presence or absence of comorbidity. After adjustment for age, the HR for comorbidity became 1.0 (95% CI 0.3 2.3, p ¼ 0.99). For total PSV patients there were significantly less deaths among cases without ENT disease than among those with ENT disease, log-rank p ¼ 0.003 (Figure 5). However comparing cases with abdominal vs. no abdominal involvement, there was no significant difference, log-rank p ¼ 0.66 (Figure 5). Discussion Classification Our PSV patients are well-defined and classification criteria were applied in a standardized manner. However, we experienced similar problems in classification to those reported in previous studies. In particular, we noted considerable overlap between the classification of MPA and WG. We have clearly described our methods to obtain diagnostic groups to allow comparison with previous studies. Patient characteristics The mean age of patients in Norfolk is higher than in other series for all diagnoses. One explanation is that our series, which excludes tertiary referrals, reflects a more representative picture of PSV in the general population than previous studies in tertiary referral centres. It is possible that these studies may be biased towards younger patients, and may exclude very ill patients who died rapidly after presentation. Norfolk has a similar percentage of elderly people to that in other regions of Great Britain (6.7% vs. 6.4% aged 475 years). Clinical features Microscopic polyangiitis Clinical features were generally similar to those reported by Savage and Westman et al., although a smaller proportion had renal involvement. 10,11

106 S.E. Lane et al. Table 4 Causes of death Cause of death CYC regimen* Survival (months) Comorbidity Deaths associated with active vasculitis or permanent damage Wegener s granulomatosis Active WG and pneumonia 1 2 HT, IHD Disease progression 3 100 DM Small-bowel perforation (WG) 1 3 Hypothyroidism Respiratory failure (WG) 2 32 Hyperlipidaemia Progression WG and CVA 3 1 Unknown, recent relapse 3 23 Sudden death, WG-related 1 1 Fibrosing alveolitis Disease progression (damage) Pred 57 Previous Ca Uterus Microscopic polyangiitis Active disease 1 14 General decline/poor renal function 3 46 Dementia, HT Cardiac arrest during haemodialysis 3 14 Previous Ca breast Active disease/chest infection** 3 2 Sepsis/renal failure** 3 13 PVD, DM Post operation for infected fistula Pred 68 HT, IHD, MI, hyperlipidaemia Churg Strauss syndrome Disease progression 1 4 Exacerbation of CSS 1 7 Deaths potentially associated with immunosuppression Wegener s granulomatosis Pneumocystis carinii pneumonia 3 2 Microscopic polyangiitis Septicaemia/bronchopneumonia 3 1 DM, IHD Pneumonia Pred 8 IHD, MI Streptococcal septicaemia Pred 45 HT Pneumonia/septicaemia Pred 61 LVF, PVD, Previous Ca rectum Pneumonia/respiratory failure 3 28 Pulmonary fibrosis Carcinoma of the oesophagus None 58 Depression Churg Strauss syndrome Ovarian cancer (never CYC) Pred 59 Other causes of death Wegener s granulomatosis Pulmonary embolism Pred 1 Unknown, general deterioration 3 9 Osteoarthritis Microscopic polyangiitis Unknown (moved out of area) NK 84 Unknown Pred 52 Unknown 3 61 Previous CVA Unknown Pred 3 Asthma Churg Strauss syndrome Suicide 1 3 Depression *Treatment regimen at time of death: CYC, cyclophosphamide; 1, intravenous; 2, pulse oral; 3, continuous oral; Pred, Prednisolone only. **Immunosuppression potentially contributory to death. HT, Hypertension; Ca, cancer; IHD, ischaemic heart disease; MI, myocardial infarction; PVD, peripheral vascular disease; DM, diabetes mellitus; CVA, cerebrovascular accident; NK, not known.

Primary systemic vasculitis 107 Abdominal CNS CVS PNS Ophthalmic Mucocutaneous Deceased Alive Cutaneous Respiratory ENT Renal 0 20 40 60 80 100 % Figure 3. Clinical features throughout disease course in survivors and non-survivors. CNS, central nervous system; CVS, cardiovascular system; PNS, peripheral nervous system; ENT, ear, nose and throat. Categories refer to Birmingham Vasculitis Activity form groups, see text. Kaplan-Meier Method Probability 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 10 20 30 40 50 60 70 Months Survival 80 WG CSS 90 MPA 100 Maximum follow-up (months) Log-rank p=0.02 Figure 4. Non-parametric survival plot comparing diagnoses. WG, Wegener s granulomatosis; MPA, microscopic polyangiitis; CSS, Churg Strauss syndrome. By contrast, there was less cutaneous, neurological, cardiovascular and abdominal features, compared to a French series. 21 A possible explanation for these differences is the exclusion of patients with ENT disease in that study. 10 This variation could be explained by disease differences between regions, patterns of referral, or may reflect the difficulty of applying the CHCC definitions in a standardized manner. Wegener s granulomatosis Although ENT disease was a prominent feature in WG and equivalent to previous studies at presentation, it was not as common throughout the course of disease in our series compared to some published studies. 2,13,23 25 This may be partly explained by a bias towards better recognition and tertiary referral of patients with classical symptoms of WG in previous studies. A higher proportion of surviving WG patients had ENT symptoms (72.1%) compared to deceased patients, which may reflect an underreporting of ENT symptoms in case-notes, as survivors were interviewed in addition to case-note review. Alternatively WG with ENT disease may have a better prognosis, as implied by Luqmani et al., who found a better prognosis in non-renal WG, 26 and suggested by Mahr et al. 13 Lower respiratory tract involvement was similar to previous

108 S.E. Lane et al. (a) 0.9 ENT Disease No ENT Disease 0.8 Probability 0.7 0.6 0.5 0.4 0.3 0 10 20 30 40 50 60 70 Months Survival (b) 1.0 0.8 Abdominal symptoms No Abdominal symptoms Probability 0.6 0.4 0.2 0.0 0 50 100 Months Survival Figure 5. Non-parametric survival plots comparing presence/absence of a ENT disease, b abdominal symptoms. reports at presentation, but lower throughout the course of disease. 10,13,22,25 27 A similar proportion had renal disease (abnormal urinalysis and/or creatinine) at presentation (78.6%) and throughout disease course (85.7%). We did not see the initial low percentage of renal disease reported by Hoffman et al. 22 Neurological and cardiovascular disease was relatively infrequent and ophthalmic, constitutional and cutaneous features were in keeping with previous estimates, although the proportion of patients with these manifestations varies widely between studies. 13 Churg Strauss syndrome Clinical features in CSS were similar to those of previously reported series, although Keough et al. described a higher percentage of ENT disease (74%). 6,28 30 Many of the cases were also included in an earlier study, and results were unsurprisingly similar. Fewer patients were included in this study, as the previous report included patients diagnosed prior to 1988 and living in Norfolk but outside the NHA. 31 All cases had respiratory disease defined by BVAS. 7 Unusually, one case had a nodular lesion on chest X-ray but also fulfilled classification criteria for WG. Fewer patients had rhinitis than previously reported, which may reflect bias in tertiary referrals. There was a high proportion of neurological involvement, which was responsible for much of the permanent damage in CSS, but gastrointestinal complications were relatively infrequent (15%). The number of ANCA-positive patients was similar to that in previous studies (41%), with the majority having panca/mpo-anca, although a higher

Primary systemic vasculitis 109 percentage (73%) was reported by Keogh et al. This may reflect the availability and increased usage of ANCA testing at the onset of disease in their Mayo clinic population, while many results available for our patients were later in the course of illness when improved ANCA assessment including ELISA became available. 30 BVAS and VDI BVAS and VDI scores were intended for prospective use in the monitoring of vasculitis, and their retrospective use may underestimate scores, due to the reliance upon record-keeping rather than direct patient interview. Estimates for surviving patients who were interviewed may be more accurate, which could potentially bias results. Mean BVAS1 was higher for WG than MPA, and was similar to figures reported in WG by Mahr et al. and in ANCA-positive vasculitis by Brijker et al. 13,32 The median BVAS1 at presentation (also calculated retrospectively) was 21 for WG and 16 for MPA. Similarly, in its original evaluation the median BVAS score for WG during periods of disease activity was 19.5 (11 25). In another study of outcome in WG, the median BVAS was slightly higher at the start of treatment, at 23 (4 46). These results suggest that our series is similar to those previously described in terms of initial disease activity in WG and MPA. However, the median VDI was lower than the Brijker study (5 for WG and 4 for MPA vs. 2 and 1, respectively, in our series). This may be attributable to data available in case notes, as features of permanent and grumbling disease, including treatment related side-effects, were generally less well recorded than active disease. The limitations of retrospective review mean that we cannot confidently compare VDI between series or comment on their relationship to outcome. This highlights the need for prospective recording of BVAS and VDI scores. Mortality Mortality increased significantly with age, which reproduces previous findings. It has been suggested that disease may be more severe and resistant to treatment in the elderly compared to younger patients. 9,14,33 However, there was no significant difference between SMRs for patients older and younger than 65 years. This suggests that the difference between age groups observed using Kaplan-Meier survival plots and the Cox proportional hazards model reflects the expected difference in mortality between age groups in the population rather than differences in disease severity. Comparison of BVAS1 at presentation between patients greater or less than 65 years old showed no significant differences. Mortality was higher for MPA compared to other diagnoses, and the ratios remained high when adjusted for age although not significantly so [HR 1.92 (0.66 5.57)]. Previous reports have also suggested that MPA has a poorer prognosis than other diagnoses, but the relatively preserved early survival compared to reduced late (5 year) survival has not previously been reported. 10,11 Overall, survival was slightly lower than many published figures (e.g. 1 year survivals of 93% in Norway and 99% in Germany in recent studies 24,33 ). Again, figures were from tertiary referral centres, so referral bias may contribute to the difference (early deaths may not be included and the case-mix may include patients fit to travel longer distances). Mahr et al. described an even lower survival for WG patients (77.5% at 6 months and 67.5% at 2 years) which they felt reflected the multicentre nature of their study, leading to the inclusion of critically ill WG patients with early deaths excluded by design in previous studies. 13 Figures were also higher than the earlier report by Guillevin et al., which may relate to a relatively high number of treatment side-effects in their series. 34 We did not reproduce an association between renal disease and poor prognosis that has previously been reported, and found no association with pulmonary disease. 10,26,35,36 This is likely to reflect the relatively broad definition of renal disease used in our study. We used the highest creatinine reached throughout disease course, rather than initial creatinine at presentation, to ensure consistency across patients, as initial blood results were not always available in some case records. This may mean that we have missed an association been higher mortality and severity of renal disease at diagnosis. Although the difference was not significant, there was an excess mortality in men compared to women, in keeping with findings by Mahr et al. and Slot et al. 13,14 Previous reports have, however, been conflicting, showing either no difference or increased mortality in females. 8,22 The excess of cutaneous, mucocutaneous and nasal disease in survivors may be explained by reporting bias between interviewed patients and case-note review alone. However, a number of other reports have also suggested that the presence of ENT disease may be associated with better outcome, and our results support this. 13,35 The relative excess of gastrointestinal features in non-survivors may indicate that this is a poor prognostic feature. Alternatively, abdominal disease may be less severe and under-recognized in survivors.

110 S.E. Lane et al. Conclusions Patient characteristics and disease activity in our series were similar to those in previous published series, although some differences were observed, including higher mean age. Differences in clinical features between this series and those from tertiary referral centres may be explained by referral bias. Alternatively, there may be some variation in disease expression between regions due to environmental or genetic factors. Mortality increased with age, but appears to reflect the expected increase in mortality associated with age rather than severity of disease. ENT disease may be associated with improved outcome, but possible reporting bias must be considered. MPA appears to have a poorer long-term outlook than other diagnoses; no other significant factors could be identified. References 1. Watts RA, Scott DGI. Overview of the inflammatory vasculitides. In: Hochberg MC, Silman A, Smolen JS, Weinblatt ME, Weissman MH, eds. Rheumatology, 3rd edn. 2003:1583 91. 2. Leavitt RY, Fauci AS, Bloch DA, et al. The American College of Rheumatology criteria for the classification of Wegener s Granulomatosis. Arthritis Rheum 1990; 33: 1101 7. 3. Masi AT, Hunder GG, Lie JT, et al. The American college of Rheumatology 1990 criteria for the classification of Churg- Strauss syndrome (Allergic granulomatosis and angiitis). Arthritis Rheum 1990; 33:1094 100. 4. Lightfoot RW, Michel BA, Bloch DA, et al. The American College of Rheumatology 1990 criteria for the classification of polyarteritis nodosa. Arthritis Rheum 1990; 33:1088 94. 5. Jennette JC, Falk RJ, Andrassy K, et al. Nomenclature of systemic vasculitides: proposal of an international consensus conference. Arthritis Rheum 1994; 37:187 92. 6. Lanham JG, Elkon KB, Pusey CD, Hughes GR. Systemic vasculitis with asthma and eosinophilia: A clinical approach to the Churg Strauss Syndrome. Medicine 1984; 63:65 81. 7. Luqmani RA, Exley AR, Kitas GD, Bacon PA. Disease assessment and management of the vasculitides. Baillieres Clin Rheumatol 1997; 11:423 46. 8. Watts RA, Lane SE, Scott DGI. Primary systemic vasculitis. In: Tugwell P, Shea B, Boers M, Brooks P, Simon L, Strand V, Wells G, eds. Evidence-Based Rheumatology. London, BMJ Books, 2004:495 528. 9. Vassallo M, Shepherd RJ, Iqbal P, Feehally J. Agerelated variations in presentation and outcome in Wegener s Granulomatosis. J R Coll Physicians 1997; 31: 396 400. 10. Westman K, Bygren P, Olsson H, Ranstam J, Weislander J. Relapse rate, renal survival and cancer morbidity in patients with Wegener s Granulomatosis or microscopic polyangiitis with renal involvement. J Am Soc Nephrol 1998; 9:842 52. 11. Savage CO, Winearls CG, Evans DJ, Rees AJ. Microscopic polyarteritis: presentation, pathology and prognosis. Q J Med 1985; 56:467 83. 12. Li PKT, Lui SF, Lai FM, Wang AYM, Leung CB, Lai KN. Microscopic polyarteritis has a poor prognosis in Chinese. J Rheumatol 1995; 22:1295 9. 13. Mahr A, Girard T, Agher R, Guillevin L. Analysis of factors predictive of survival based on 49 patients with systemic Wegener s Granulomatosis and prospective follow-up. Rheumatology 2001; 40:492 8. 14. Slot MC, Cohen Tervaert, Franssen CF. Renal Survival and prognostic factors in patients with PR3-ANCA associated vasculitis with renal involvement. Kidney Int 2003; 63: 670 7. 15. Symmons DP, Barrett EM, Bankhead CR, Scott DG, Silman AJ. The incidence of rheumatoid arthritis in the United Kingdom: results from the Norfolk Arthritis Register. Br J Rheumatol 1994; 33:735 9. 16. Watts RA, Lane SE, Bentham G, Scott DGI. Epidemiology of systemic vasculitis: a ten-year study in the United Kingdom. Arthritis Rheum 2000; 43:414 19. 17. Lane SE, Scott DGI, Heaton A, Watts RA. Primary renal vasculitis in Norfolk increasing incidence or increasing recognition? Nephrol Dial Transplant 2000; 15:23 7. 18. Lane SE, Watts RA, Barker T, Scott DGI. An evaluation of the sorenson classification criteria. Rheumatology 2002; 41: 1138 41. 19. Sorensen SF, Slot O, Tvede N, Petersen J. A prospective study of vasculitis patients collected in a five year period: evaluation of the Chapel Hill nomenclature. Ann Rheum Dis 2000; 59:478 82. 20. Hagen EC, Andrassy K, Csernok E, et al. Development and standardization of solid phase assays for the detection of antineutrophil cytoplasmic antibodies (ANCA): A report of the second phase of an international cooperative study on the standardisation of ANCA assays. J Immunol Methods 1996; 196:1 15. 21. Guillevin L, Durand-Gasselin B, Cevallos R, Gayraud M, Lhote F, Callard P, Amouroux J, Cassassus P, Jarrousse B. Microscopic polyangiitis. Arthritis Rheum 1999; 42: 421 30. 22. Hoffman GS, Kerr GS, Leavitt RY, Hallahan CW, Lebovics RS, Travis WD, Fauci AS. Wegener s An analysis of 158 patients. Ann Intern Med 1992; 116:488 98. 23. Cohen Tervaert JW, Huitema MG, Hene RJ, Sluiter WJ, The TH, van der Hem GK, Kallenberg CGM. Prevention of relapses in Wegener s Granulomatosis by treatment based on antineutrophil cytoplasmic antibody titre. Lancet 1990; 336:709 11. 24. Reinhold-Keller E, Zeidler A, Gutfleisch J, Peter HH, Raspe HH, Gross WL. Giant cell arteritis is more prevalent in urban than rural populations: results of an epidemiological study of primary systemic vascultiides in Germany. Rheumatology 2000; 39:1396 402. 25. Koldingsnes W, Gran JT, Omdal R, Husby G. Wegener s Granulomatosis: Long-term follow-up of patients treated with pulse cyclophosphamide. Br J Rheumatol 1998; 37:659 64.