Risk factors for infectious complications of ANCA-associated vasculitis: a cohort study

A total of 248 patients newly diagnosed with AAV and renal involvement who met the criteria of the Chapel Hill Consensus Conference [7] between January 1, 1998 and December 31, 2013 at the National Clinical Research Center of Kidney Diseases Jinling Hospital were included, among whom 194 patients had renal biopsies that showed pauci-immune necrotic and crescentic glomerulonephritis. All patients were ANCA-positive. Patients with secondary vasculitis, including Henoch-Schonlein purpura, allergy, autoimmune disease, tumour, cryoglobulinemia and infection, were not included. Patients with end-stage renal disease (ESRD) or who received only non-immunosuppressive treatment for infection at the time of diagnosis of AAV were excluded from the study. Ethical statement: This study was approved by the Institutional Review Board of our hospital and performed in accordance with the ethical standards laid down in appropriate version of the Declaration of Helsinki. All patients signed informed consent.

All clinical and laboratory data were collected retrospectively at diagnosis and during the follow-up period, including the patients’ age, gender, medical history, routine blood analysis, 24-h urine protein excretion, urinary sediment red blood cell count, serum albumin and serum creatinine (SCr), liver enzymes, immunoglobulin and T lymphocyte counts, serum ANCAs, lung involvement, Birmingham Vasculitis Activity Score (BVAS) [16], the usage of immunosuppressive agents, methlyprednisone pulse therapy, plasma exchange, and adverse events including major infection. Major infections were diagnosed according to common terminology criteria for adverse events (CTCAE) v4.0 in addition to clinical and radiological manifestations and microorganism cultures.

None of the patients had received any immunosuppressive therapy before diagnosis. Patients without contraindication initially received intravenous methylprednisolone pulse therapy (0.5 g, once daily, for 3 consecutive days) after diagnosis of AAV. Patients with severe manifestations of AAV underwent plasma exchange therapy. All patients received oral prednisone at a dose of 0.6–0.8 mg/kg/day for 4 weeks, which was then tapered by 5 mg each week to 10 mg/day. Induction immunosuppressive agents included MMF 1–1.5 g/day orally, monthly intravenous cyclophosphamide (IV-CYC) at 0.75–1.0 g/m² body surface area in monthly pulses, tripterygium glycosides (TW, extract from the traditional Chinese herb Tripterygium wilfordii, which mainly contains triptolide) and multi-target therapy (prednisone, mycophenolate mofetil and tacrolimus) [8]. Maintenance therapy included prednisone 5 mg/day combined with MMF and azathioprine. Prophylaxis of Pneumocystis Jirovecii pneumonia (PJP) with SMZ-CO (trimethoprim-sulfamethoxazole 400/80 0.48 g per day) was used in patients whose CD4⁺ T cell counts were less than 200/μl, and the doses were tapered in patients with renal dysfunction [17].

All immunosuppressive agents, except prednisone, were discontinued in patients with AAV who suffered from major infection during the follow-up period. Antimicrobial therapy was prescribed according to clinical and radiological manifestations and microbiological characteristics. Patients diagnosed with PJP were treated with SMZ-CO and echinocandin together.

Patients with weight loss were prescribed enteral nutrition. The patients with severe acute kidney injury or acute respiratory distress syndrome (ARDS) were treated with continuous blood purification.

A recorded severe infectious complication was defined as implying the administration of an antimicrobial medication for an observable clinical, microbiological and radiologic suspected infection requiring hospitalization. Immediate dialysis was defined as the clinical necessity of renal replacement therapy on admission. The first immunosuppressive agent used in addition to corticosteroids was termed induction therapy. The immunosuppressive regimen used during follow-up was termed the maintenance agent. The diagnosis criteria for deep fungal infection included clinical manifestations, such as fever, cough, diarrhoea or lower urinary tract symptoms, and the detection of fungi in sputum, urine, stool or tissue specimens. Cytomegalovirus (CMV) infection was diagnosed by CMV polymerase chain reaction (PCR). The range of quantification of this assay was 600–100,000 copies/ml for CMV. CMV pneumonia was defined as the detection of ground glass opacity by chest X-ray film or computed tomography, the detection of CMV in the bronchoalveolar lavage fluid or lung tissue samples, and clinical signs such as fever, cough, dyspnoea and hypoxemia. The diagnosis of PJP was made clinically or by the identification of Pneumocystis from sputum, bronchoalveolar fluid, tracheal secretions or lung tissue by special stains or a non-nested PCR, specifically designed to diagnose pneumonia rather than colonization [18]. ARDS was defined as the acute onset of hypoxemia (arterial partial pressure of oxygen to fraction of inspired oxygen [P_aO2/F_IO2] ≤ 200 mmHg) with bilateral infiltrates on chest radiographs, without left atrial hypertension. Multiple organ dysfunction syndrome (MODS) was defined as the simultaneous failure of at least two organs. ESRD was defined as eGFR < 15 ml/min per 1.73m² or requiring renal replacement treatment for > 3 months.

The follow-up endpoints included the final date of December 31, 2014, dropping out before the final date, reaching ESRD, or death.

Statistical analysis was performed with SPSS 20.0 for Windows (SPSS Inc., Chicago, IL, USA). Medians and ranges were reported for non-normally distributed data, and means ± standard deviations were reported for normal-distributed data. The Kruskal-Wallis test was applied for the comparison of non-normal distributed data. Differences between means were tested using the Student’s t-test. A Mann-Whitney U test was used for nonparametric distributions. Chi-squared tests were used for the comparison of categorical data. To address the independent predictive value of factors associated with the rate of infections, the variables with P values of less than 0.1 in univariate analysis as well as those reported in the literature were selected for multivariate analysis using the Cox regression model. The group with corticosteroids only was used as a reference group in multivariate analysis. Only the time to first severe infection was evaluated. Laboratory values and BVAS used for modelling were from the time of diagnosis. Receiver operating characteristic (ROC) curve analysis was performed to determine the cut-offs of SCr, haemoglobin, albumin, CD4+ T cells and BVAS. All tests were two-tailed, and P-values of < 0.05 were considered significant. Confidence intervals (CIs) were calculated at the 95% level.

This study identified 248 individuals with ages ranging from 14 to 78 years (median 55 years), including 214 cases diagnosed as MPA, 16 cases diagnosed as RLV, 10 cases diagnosed as GPA and 8 cases diagnosed as EGPA. Seventy-five patients (30.2%) showed lung involvement, 30 (12.1%) had alveolar haemorrhage, and 54 (21.8%) had sinus involvement. MPO-ANCA was more prevalent, and only 21 cases (8.5%) were PR3-ANCA-positive. Fifty-three patients started immediate dialysis. Initial immunosuppressive treatment consisted of pulse methylprednisolone (67.3%), plasma exchange (23.8%), IV-CYC (26.6%), MMF (31.0%) and TW (16.1%). Twenty-six percent of patients received only oral corticosteroids (Table 1). Forty-two patients (16.5%) received SMZ-CO to prevent PJP, and 29 of them were CYC users.

A total of 103 infectious episodes occurred in 86 patients (34.7%) during follow-up for 1~ 155 months (median 15 months). Fifteen cases experienced a second episode of infection, and one patient experienced a third episode. Seventy-six episodes (73.8%) of infection occurred during induction therapy (median 1.5 months). Twenty-seven episodes (26.2%) occurred during maintenance therapy (median 18 months), and six episodes (5.8%) occurred after 24 months. Pulmonary infections (71.8%, 74/103) were the most frequent type of infection, followed by skin (n = 7, 6.8%), digestive tract (n = 3, 2.9%), urinary tract (n = 2, 1.9%) and central nervous system (n = 1, 1.0%) infections. Six patients (5.8%) developed sepsis because of their reported infection.

The pathogens responsible for infection were confirmed in 87 episodes. The whole pathogen spectrum included bacteria, fungi and viruses. Bacterial infection was the most common (n = 57, 66%), especially Acinetobacter baumannii, followed by fungal (n = 21, 24%) and viral infections (n = 9, 10%) There were four CMV, seven PJP and 16 unspecified infections (Table 2).

The infectious rate of induction therapy with corticosteroids only was 38.5% (25/65), that for CS + IV-CYC was 39.0% (26/66), CS + MMF was 33.8% (26/77) and CS + TW was 22.5% (9/40). The incidence of smoking (36.0% vs. 17.3%, P = 0.000) and diabetes (8.1% vs. 2.5%, P = 0.032) was significantly higher among the infected patients. The cutoff level of SCr haemoglobin, albumin, CD4+ T cells, and BVAS were determined as 5.74 mg/dl, 7.75 g/dl, 33.95 g/l, 281/ul, and 25.5 respectively based on ROC curve analysis. Single factor analysis revealed that risk factors for complicated infection in patients with AAV included age, smoking, pulmonary involvement, hemoglobulin, albumin, SCr level, CD4 ⁺ T cell count, BVAS, and immunosuppressive therapy with MMF, CYC and TW. In adjusted models for the AAV cohort, increased risks of infection were observed in patients who were older at the time of diagnosis (HR = 1.003, 95% CI = 1.000–1.006), smoking (HR = 2.338, 95% CI = 1.236–4.424), with baseline Scr ≥5.74 mg/dl (HR = 2.153, 95% CI = 1.323–3.502), CD4⁺ T cell< 281 μl (HR = 1.813, 95% CI = 1.133–2.900), and users of intravenous cyclophosphamide regimen (HR = 1.951, 95% CI =1.520–2.740) (Table 3).

The exact pathogen was identified in 44 of 82 episodes of pneumonia. Bacteria were the most common pathogens (n = 27, 61.4%), especially Acinetobacter baumannii (n = 6, 13.6%), Staphylococcus aureus (n = 5, 18.5%) and Pseudomonas aeruginosa (n = 3, 6.0%). Thirteen cases were diagnosed as fungal infections, and most were caused by C. albicans (n = 8, 61.5%). CMV was identified in all four cases with viral pneumonia.

The main pulmonary radiologic findings included consolidation (n = 38, 51.4%), diffuse interstitial pneumonia (n = 21, 28.4%) and multiple nodules (n = 13, 17.6%). Bacterial pneumonia presented with consolidation (n = 24, 32.4%), nodules (n = 9, 12.2%) and a diffuse reticular pattern (n = 6, 8.1%). CMV pneumonia mainly presented with ground-glass opacities (4, 5.4%), diffuse reticular thickening (n = 3, 4.1%) and nodules (n = 1, 1.4%) on bilateral lungs. Fungal pneumonia was characterized by consolidation (n = 14, 18.9%), nodules (n = 4, 5.4%), halo (n = 4, 5.4%) and air crescent sign (n = 2, 2.7%). Thirteen cases were complicated by ARDS, and 10 were complicated by MODS. Nine patients required mechanical respiration (5 BiPAP and 4 endotracheal intubation).

All 103 episodes were treated with intravenous antibiotics. Twelve (11.7%) of 103 patients died and all due to severe pneumonia. The time to death was from one to sixteen months after the initiation of immunosuppressive therapy. None died due to AAV (Table 4).

There are some limitations in this retrospective study. First, the treatment protocols were not uniform and lack of data on Rituximab. Only a minority of patients were given SMZ-CO prophylaxis because of the insufficient awareness. None of these patients received prophylaxis for fungal infection. In addition, some cases with pulmonary or central nervous system infection failed to show a definitive pathogen. The frequency and severity of pneumonia should be lowered by prophylactic treatment and early diagnosis.