Introduction

In the evolution of classification schemes for vasculitis, distinct clinical entities have been classified according to the predominant size of the inflamed blood vessels (large, medium, or small), histopathology (eg, granulomatous versus nongranulomatous inflammation), and more recently according to the association with antineutrophil cytoplasmic antibodies (ANCAs). Within the context of these schema, granulomatosis with polyangiitis (GPA) is a chronic vasculitis involving small- to medium-sized arteries characterized by granulomatous inflammation and an association with ANCAs [1, 2]. GPA was previously known as Wegener’s granulomatosis (WG). In April 2011, a report of a consensus of the American College of Rheumatology (ACR), American Society of Nephrology, and the European League Against Rheumatism (EULAR) recommended a nomenclature change such that WG now be described as GPA [3••]. We generally use this term in this article.

GPA is characterized by pauci-immune necrotizing crescentic glomerulonephritis and granulomatous inflammation primarily involving the lungs and upper respiratory tract; however, the vasculitis frequently may involve other organs or organ systems. Although childhood GPA is rare, the burden of disease to affected children, their caregivers, and the health care system is high. Acutely, children with GPA account for an appreciable number of rheumatology intensive care admissions and deaths [4]; chronically, the disease is organ and/or life threatening [5, 6, 7•], and the large majority of patients require cytotoxic or other immunosuppressive therapy [8••]. Anecdotally, affected children have continuing therapy requirements, with sustained disease activity and damage from disease and treatment that extend long into adulthood. Because of disease rarity, most data on GPA are derived from the adult literature, although evidence for their applicability to the growing child is lacking. The need for international pediatric collaborations including registries to study sufficient patient numbers has been recognized [9, 10•], and recent efforts to this end have begun to provide more substantial pediatric data. In this review of GPA, the focus is on childhood disease where data are available, more recent literature relating to children and adults, and also includes some preliminary pediatric data that have only been presented in abstract form.

Epidemiology

The incidence of GPA has risen from 0.2 to 1.2 per 100,000 persons per year over the period from the 1970s to the 1990s, as reported in multiple European consecutive cohort studies [11–15]. In the general population, males outnumber females 1.6 to 1 [16], and the peak incidence of GPA is the fourth and sixth decade of life [12, 17, 18], with a higher incidence in higher latitudes [13, 19]. In the pediatric population, the disease occurs in the second decade of life, with a female preponderance [5, 6, 8••, 20]. A recent Canadian study demonstrated an increase in the incidence of childhood GPA in the past 5 years from 0.28 to 0.64 per 100,000 per year in southern Alberta [21].

Classification

The 1990 ACR criteria for classifying vasculitis [22], including GPA [23] and derived primarily from adult–patient data, are suboptimal in classifying children with vasculitis. As such, a pediatric adaptation of ACR criteria taking into account common pediatric manifestations and the presence of ANCAs was proposed by a consensus of pediatric experts and reported in 2006 [24]. These proposed criteria were subsequently tested and improved upon using a cohort of pediatric patients. The final validated criteria had improved sensitivity when compared with the ACR criteria and were subsequently published in 2010, having been endorsed by EULAR, the Pediatric Rheumatology European Society, and Pediatric Rheumatology International Trial Organization (EULAR/PRINTO/PReS) [25••]. Criteria for classifying GPA according to ACR or the EULAR/PRINTO/PReS criteria are compared in Table 1. Unfortunately, the new pediatric criteria retain some inherent limitations of the ACR criteria that also have been recognized in the adult literature [26•, 27]. For example, because there are no criteria for microscopic polyarteritis (MPA), pediatric patients classified as GPA by ACR criteria may be concurrently described as having MPA using the Chapel Hill consensus conference definition [8••]. In a cohort of 155 pediatric patients with ANCA-associated vasculitides (AAV) and unclassified vasculitis in which one sixth of the patients were diagnosed with MPA, both the ACR and the new EULAR/PRINTO/PReS criteria had poor sensitivity and specificity to diagnose GPA when compared with sensitivity described in the classification validation exercises in which the patient cohort included extremely few patients diagnosed with MPA [28•]. This conundrum that also exists for adult patients has led on the one hand to the convenient grouping of GPA and MPA in clinical trials under the rubric of AAV. On the other hand, complex algorithms and systems have been proposed or required to uniquely discriminate patients with the various types of AAV and polyarteritis nodosa for study [29–31]. Because of this and other classification inadequacies, the whole system for classifying vasculitis is currently under scrutiny [27].

Table 1 Comparison of the ACR and EULAR/PRINTO/PReS criteria for classification of GPA
Full size table

Pathogenesis

The cause of GPA is unknown, but ANCAs likely have a role in the pathogenesis and may also determine the extent or severity of disease manifestations, being more strongly associated with generalized versus limited disease [32]. ANCAs associated with GPA can have a cytoplasmic (c) or perinuclear (p) immunofluorescence pattern, with their primary antigenic targets being proteinase 3 (PR3) or myeloperoxidase (MPO), respectively [33]. The cANCA pattern and PR3 specificity are the more frequent ANCAs in children [5, 8••] and adults [16, 34] with GPA. A direct pathogenic role for MPO-ANCA in lesion development is supported by in vivo experimental studies in mice and rats [35, 36]; however, this may be more relevant to MPA. Among the AAV, MPO-ANCA is strongly associated with MPA, whereas PR3-ANCAs are associated with GPA [37••]. Attempts to develop a PR3-ANCA animal model corresponding to the MPO-ANCA model have been unsuccessful.

The pathogenic mechanisms of GPA seem to be more complex than those of MPA, and knowledge about the relative role of ANCAs and other elements of the humoral, cellular, and innate immune system continues to accumulate. The presence of ANCAs implicates neutrophils as key effector cells. ANCAs promote degranulation of neutrophils and monocytes, leading to endothelial damage and activation of complement pathway [38], with the oxidative burst of neutrophils being modulated by the sialylation levels of anti-PR3 antibodies. Bronchiolar lavage fluid from GPA patients compared with controls had MPO, higher neutrophil count, and a high level of neutrophil chemotaxis mediated through interleukin-1β that persists even in clinical remission [39]. Another recent discovery is that chromatin fibers, so-called neutrophil extracellular traps (NETs), are released by ANCA-stimulated neutrophils and may be deposited in the kidney and may trigger vasculitis [40•]. NET formation is a mechanism of host defense that allows efficient microbial clearance. NET formation also increases the interferon-γ level, which sustains inflammation [37••].

Genetic factors have been implicated in the etiology and pathogenesis, but the relative late mean age at onset, the infrequent familial occurrence, and the environmental geographical differences suggest that nongenetic factors make a major contribution. A link between the DRB1*15 allele and the risk of PR3-ANCA disease in African Americans has been demonstrated [41], but these findings have yet to be replicated. Many reports suggest links among bacterial infections, autoimmunity, and development of GPA, and a recent review described several potential mechanisms, including a role for molecular mimicry and Toll-like receptor signaling [42•]. The complete role of ANCAs in the pathogenesis of GPA is continuing to unravel.

Disease Presentation in Children

The triad of upper and lower respiratory tract inflammation and renal disease is the hallmark of GPA, and manifestations in these organs characterized childhood GPA in the four largest pediatric series describing 17 to 65 patients [5, 8••, 20]. The largest cohort describes 65 patients collected within ARChiVe (A Registry for Children with Vasculitis: e-entry) [8••], a Canadian and US collaboration since 2007. At disease onset, the most common features were constitutional symptoms, including fatigue, weight loss, fever, and malaise (89%), and the frequencies of ear, nose, and throat (ENT); pulmonary; and renal manifestations were 80%, 75%, and 80%, respectively [8••]. Less frequently involved organ systems were musculoskeletal (57%), gastrointestinal (42%), eyes (37%), skin (35%), and nervous system (25%) [8••]. Notably, no cardiovascular features were reported, unlike in the series reported from the Hospital for Sick Children in Toronto, in which 12% of 25 patients were found to have venous thromboses at diagnosis [5]. Although two small pediatric studies suggest that childhood GPA more frequently presents as a localized or limited disease than adult GPA [14, 43], in the larger series [5, 6, 8••], the large majority of children with GPA presented with multiple organ involvement. Among patients in ARChiVe, at disease onset, specific pulmonary features included shortness of breath (52.5%), chronic cough (52.3%), hemoptysis/alveolar hemorrhage (44.6%), and nodules (42.5%); renal features included abnormal urinalysis (75.4%), biopsy-proven glomerulonephritis (52.3%), and elevated serum creatinine (41.5%); and ENT features were nasal involvement (64.6%), sinusitis (60%), and subglottic involvement (13.8%) [8••]. Rottem and colleagues [6] reported that subglottic stenosis during the course of the disease is five times more common in childhood- versus adult-onset disease, and this led to the inclusion of this symptom to be proposed as one of the pediatric classification criteria [24]. Although subglottic disease may be no more frequent than other upper respiratory features of GPA at diagnosis [8••], its ultimate inclusion as a validation criterion was likely because of its specificity (Table 1) [25••].

Clinical Course and Outcome/Prognosis in Pediatrics

The few reports on the clinical course and outcome for pediatric GPA patients include series from 1993, 2007, and 2011, respectively, reporting 23, 25, and 8 patients observed for a mean of 8.7, 2.7, and 19 years [5, 6, 7•]. The last report describes seven GPA patients and one patient with MPA who had pediatric-onset disease and were observed for as long as 27 years in an adult institution [7•]. Similar to the report from 1993 [6], the outcomes from the 2011 report do not reflect contemporary treatment practices, and because all three series are from specialized centers, they may represent a concentration of patients with more severe disease. Notwithstanding, important outcomes from the 2011 report inherent to the much longer follow-up are that four of eight patients became infertile, two had severe skeletal complications of osteoporosis or avascular necrosis of the femoral head, one female patient developed breast cancer at 30 years of age, and one patient died 25 years from initial presentation; no patients were described with cardiac disease [7•]. Similar to the other two series [5, 6], most patients achieved remission, but all had relapsing disease and infections were common. In the 1993 and 2007 reports [5, 6], no patients developed malignancy, and in the earlier of the reports, cyclophosphamide treatment was associated with cystitis in 50% and infertility in 28%, and two patients died, one from severe lung disease and cor pulmonale and one from sepsis.

In all the pediatric follow-up studies, severe ENT disease accumulated after diagnosis. In the 2011 series, half of the patients suffered hearing impairment and nasal septal or upper airway deformities [7•]. In the 1993 series, 4% of patients had ENT involvement at presentation, but 48% developed this problem during follow-up, and half of these required tracheostomy. An additional report from 2011 focused primarily on ENT manifestations of GPA in children [44]. From a retrospective cohort of 28 patients, it reported on seven children having airway lesions, three of whom had isolated and limited lesions and four of whom had multilevel disease. These latter four patients needed a multidisciplinary treatment approach involving ENT, pulmonology, and rheumatology, as they required several surgical interventions and biological therapy [44].

Important Considerations from Recent Adult Studies

An adult cohort including 445 patients with GPA showed an improved outcome over the past four decades [45•]. During the follow-up period, the interval between first symptom and diagnosis reduced from 8 to 4 months. Eighty percent of the patients are still initially treated with cyclophosphamide, but the cumulative dose has been significantly reduced over time, as have the mortality and relapse rates. Young men were more inclined to develop renal manifestations and had a considerably higher mortality rate than young females. No increased risk in malignancies was reported over four decades. Another report describes long-term survival in 535 patients with AAV who had been recruited to 4 different clinical trials [46]. They reported an increased risk of death in patients compared with the age- and sex-matched population, with mortality within the first year of disease being due to infection or active vasculitis, whereas later in the disease course, mortality was related to cardiovascular disease, malignancy, and infection [46]. The European Vasculitis Study Group Clinical Trials Group also reported a small increased incidence of malignancies in patients with AAV compared with the general population, primarily due to an increase in nonmelanoma skin cancer [47•]. They speculated that the lower cancer risk in this more contemporary cohort compared with previous reports may have been due to less extensive use of cyclophosphamide in current treatment regimens. A pilot study described an increased frequency of fibromyalgia, depression, and sleep disorders in adult patients with GPA [48]. Another study reported that health-related quality of life in patients with newly diagnosed AAV seemed to be impaired in complex ways and variably in different domains as measured in the Medical Outcomes Survey Short Form 36 (SF-36) [49].

Several recent case series and adult cohort studies in GPA have focused on specific organ systems. One retrospective study described patients presenting, often insidiously, with otological symptoms, including otitis media, facial palsy, and sensorineural hypoacusis [50]. The authors commented on the challenge of considering GPA within the differential diagnosis and the young age of the patients, ranging from 32 to 46 years. They noted a high frequency of mortality and renal failure in patients who had early involvement of multiple organs, and for them, they recommended early, aggressive systemic therapy [50]. In another study, adult GPA patients were found on screening to have impaired smell and taste compared with age-matched controls, and they recommended that the assessment of these senses should be part of ongoing evaluations, with subsequent counseling as necessary [51]. Although cardiac disease has not been described for pediatric patients with GPA, in the adult population, recent reports describe early, disease-related cardiac involvement [52] and longer-term risks of cardiovascular events [53•]. In the first study, in a selected group of nine adult patients with GPA who were resistant to 6 months of induction therapy, seven patients (82%) had cardiac involvement identified as gadolinium-enhancing lesions on imaging or pericardial effusion or thickening using transthoracic echocardiography [52]. In the second study, 75 (13.8%) of 535 patients with AAV who were involved in European Vasculitis Study Group trials and observed for more than 5 years were found to have had at least 1 cardiovascular event (defined as cardiovascular death, stroke, myocardial infarction, or coronary artery disease requiring bypass grafting or percutaneous intervention) [53•]. They developed and validated a model to predict cardiovascular events in adult patients with AAV and no previous cardiovascular disease, and to quantify the risk based on age, diastolic hypertension, and ANCA status. PR3-ANCA reduced the risk of a cardiovascular event compared with MPO-ANCA or ANCA-negative patients [53•].

Treatment

Prior to the aggressive use of glucocorticoids and cyclophosphamide for the treatment of GPA [54], the disease was fatal for most children [55] and also most adults, in whom 1-year mortality was in the order of 80% with death from renal or pulmonary failure or from infection [56]. Glucocorticoids and oral cyclophosphamide originally prescribed for periods of 1 to 2 years induced remission in more than 90% of adult patients [57, 58]. As eloquently described in a recent review [59••], although cyclophosphamide has been life saving, it has not prevented relapses from occurring in 50% of patients within 5 years and has been associated with significant short- and long-term toxicity risks, particularly an increased incidence of life-threatening infections, malignancy, and infertility [58, 60]. Thus, the clinical challenge and aim of ongoing clinical trials is to balance the risks associated with current or emerging therapies against the damage associated with under-treating aggressive disease [61] or over-treating less aggressive disease [58, 62–65]. Strategies to reduce cyclophosphamide burden, including intermittent administration with different intravenous regimens, the concept of induction-maintenance regimens, and the complete avoidance of cyclophosphamide for less severe disease, were reviewed [59••]. Finally, in the same review, the place of cyclophosphamide in the treatment of AAV since the introduction of rituximab was questioned [59••]. A summary of the conclusions of several representative clinical trials in adults addressing the above issues follows.

For patients who went into remission within 6 months following standard glucocorticoid and cyclophosphamide therapy for AAV, remission maintenance using either continuing oral cyclophosphamide or azathioprine was equally effective with respect to relapse rates (CYCAZAREM trial) [66]. However, mycophenolate mofetil was found not to be as effective as azathioprine in maintaining remission [67]. In a randomized trial, pulse cyclophosphamide was found to be to be as effective as oral daily cyclophosphamide in inducing remission in patients with AAV, with a lower cumulative dose and less leukopenia [68•]. On the other hand, for patients who fail to achieve remission on an intravenous cyclophosphamide regimen, subsequently switching to oral cyclophosphamide can be an effective rescue treatment (WEGENT trial) [69].

For patients with non–life-threatening or non–organ-threatening disease and minimal to no renal disease, remission rates for methotrexate and cyclophosphamide were comparable [70]. For patients with more extensive disease, remission induction using methotrexate was not as effective as cyclophosphamide, and more relapses were seen [71]. In one small, unblinded trial, mycophenolate mofetil for inducing remission and improving renal function seemed to have some role when compared with cyclophosphamide, but the conclusions could not be considered definitive [72].

Plasma exchange was recommended as induction therapy for patients with moderate to severe renal disease in the conclusions of one recent clinical trial [73]. Another review suggested that there remains some uncertainty as to what severity level of renal failure plasma exchange might be beneficial, particularly because of its expense and the frequency of adverse events [74].

Rituximab, a monoclonal antibody that depletes B cells by binding to CD20 on activated B cells, is challenging the role of cyclophosphamide in treatment of AAV [75••, 76••, 77, 78, 79••]. In the first of only two randomized controlled trials of rituximab, patients newly diagnosed with AAV were treated with intravenous cyclophosphamide as standard induction therapy followed by azathioprine, or four consecutive weekly doses of rituximab accompanied by two intravenous doses of cyclophosphamide [75••]. By 12 months, high remission rates were achieved in both groups, the rituximab-based treatment was found not to be superior, and there was no reduction in early severe adverse events associated with rituximab [80]. In the second trial, newly diagnosed or relapsing patients with AAV were treated with daily oral cyclophosphamide or four consecutive weekly doses of rituximab [76••]. By 6 months, rituximab was found to be noninferior to daily cyclophosphamide for remission induction, rituximab may have been superior in relapsing disease, and there was no difference in frequency of adverse effects [76••].

Adequate long-term efficacy and safety trials for rituximab are lacking. There are a few small series—for example, including 5 [78] and 11 [77] patients with GPA refractory to or intolerant of conventional treatment. They describe efficacy in inducing remission, in some patients in the long term and for other patients repeatedly with recurrent relapses. Patients were observed for up to 52 months, and infections were described as rare, but generalizable conclusions about long-term safety and outcome could not be drawn. Following a systematic review of rituximab use with AAV, several recommendations were provided by a group of rheumatology and nephrology experts [79••]. Despite the acknowledged low levels of evidence, the clearest recommendation was for use of rituximab for GPA patients with refractory disease, and as an alternative to cyclophosphamide for previously untreated patients. However, it was also clearly stated that the long-term efficacy and safety of rituximab remain to be determined [79••].

In 2008, the EULAR recommendations for management of primary small- and medium-sized vessel vasculitis were published [81•]. In life-threatening and organ-threatening generalized, primary, small and medium vessel vasculitis, induction consists of treatment with cyclophosphamide, 15 mg/kg (maximum, 1.2 g), every 2 weeks for the first three pulses, followed by infusions every 3 weeks for the next three to six pulses; glucocorticoids, 1 mg/kg per day; and plasma exchange (in case of severe renal disease in order to improve renal survival). In non–organ-threatening or non–life-threatening disease, induction with high-dose glucocorticoids and methotrexate is suggested. Remission/maintenance therapy includes azathioprine, leflunomide, or methotrexate in combination with low-dose glucocorticoids. These recommendations were based on trials and observational adult data in the literature and preceded the results of recent clinical trials using rituximab.

No clinical trials have been conducted in pediatric GPA. Currently, as described in ARChiVe, pediatric patients in the United States and Canada are given remission-induction treatments with corticosteroid therapy (pulse methylprednisolone for 3–5 days, followed by oral prednisone) and cyclophosphamide (orally or with one of two intravenous regimens), and this is followed by maintenance therapy, most frequently with methotrexate [8••]. Additional treatments used in children with GPA have been reported in small series and include plasma exchange (especially in pulmonary capillaritis or rapidly progressive glomerulonephritis), antiplatelet treatment (aspirin) [10•], and rituximab for refractory disease [82]. There is also a case report of one child developing pneumocystis jiroveci following rituximab use [83]. In the above-described systematic review [79••], in which there was a 1 pediatrician represented among 12 adult physician experts, it was recommended based on a low evidence level that rituximab should be considered for the treatment of children with AAV who fail to respond to conventional induction therapy with glucocorticoids and cyclophosphamide, and also for patients with relapsing disease in whom there is particular concern regarding cumulative glucocorticoid and/or cyclophosphamide toxicity [79••].

Disease Measurement Tools

The ability to conduct clinical trials in pediatric GPA and all forms of chronic vasculitis in children has been hampered not only by disease rarity but also by the lack of available “scoring” tools to measure disease activity, damage, and outcome before, during, and after treatment. Various tools have been developed and used with adult patients in clinical practice and clinical trials. Two such tools, the Birmingham Vasculitis Activity Score (BVAS) and the Vasculitis Damage Index (VDI) [84–87], have become widely established in adult AAV and are accepted by the Outcome Measures in Rheumatology Clinical Trials (OMERACT) initiative [88]. The tools are an integral part of clinical trials and studies because they allow the more formal and reproducible definition of disease states while taking into consideration the variable spectrum of affected organs that are involved to varying degrees. However, the adult tools may not be fully applicable to childhood chronic vasculitis, and when the performance of BVAS was examined in pediatric AAV patients in ARChiVe, there was a weak correlation with the physician global assessment of disease activity and only a moderate correlation with erythrocyte sedimentation rate and treatment decision [89•]. Currently, there are no available validated tools to measure disease activity or disease damage in pediatric vasculitis, although initiatives to develop pediatric-specific modifications of adult tools are under way.

Conclusions

The pathogenesis of AAV remains unclear but is unraveling, with both humoral and cellular immunity playing a role. PR3-ANCA is associated with GPA and relapsing disease. Cyclophosphamide with glucocorticoids is the induction treatment of choice in pediatric GPA, but rituximab is increasingly recommended in adult GPA for induction treatment or in refractory disease. Long-term outcomes have improved significantly in adult GPA. In pediatric GPA, relapses are seen in most patients, and long-term renal damage seems to be relatively mild, but larger cohorts and longer follow-up are necessary to better characterize the outcomes of all systems. Validated disease activity and damage tools are necessary in pediatric vasculitis for disease monitoring and assessment of treatment response. International multicenter studies are needed to better define the long-term outcome of pediatric GPA and to determine the efficacy, outcome, and adverse events of existing and emerging treatments.