Data collection: demographic features, clinical, laboratory and angiographic findings

Baseline demographic data including age at symptom onset and at diagnosis, gender, ethnicity and past medical history were recorded. Clinical characteristics included constitutional symptoms such as weight loss, malaise and fever, as well as signs and symptoms of organ involvement. Arterial hypertension was defined as blood pressure > 95th percentile for age. Blood pressure discrepancy was defined as a difference of ≥ 10 mmHg between limbs. Laboratory results included erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complete blood cell count, creatinine, von Willebrand factor antigen, antinuclear and antineutrophil cytoplasmic antibodies, proteinuria and hematuria. Tuberculosis testing results were recorded.

Vascular imaging studies included conventional angiography, magnetic resonance angiography (MRA), computed tomography angiography (CTA) and echocardiography studies. Location, type and characteristics of lesions such as stenosis (<50% lumen), narrowing, aneurysm, dilatation, dissection, vessel wall thickening and post-contrast enhancement were captured. Vascular imaging was documented when performed within 1 month of the predefined data collection time points.

Treatment and complications

Current medication, vascular and surgical interventions were recorded. Disease and treatment-related complications included arterial dissection, treatment-related side effects and severe infections necessitating hospitalization.

Disease activity and damage

Disease activity was assessed at each visit using the validated Pediatric Vasculitis Activity Score (PVAS) [21] and the Indian Takayasu Arteritis Activity Score (ITAS2010), which was recently validated in adult Indian TAK patients [22]. Disease damage was assessed using the Pediatric Vasculitis Damage Index (PVDI), a damage assessment tool modified from the adult Vasculitis Damage Index [23]. The PVAS and PVDI were systematically completed at each clinic visit as part of routine clinical practice at our institution since 2009. For patient visits before 2009, the PVAS and PVDI were retrospectively completed by two independent reviewers (FAA, SS); discrepancies were resolved by a third reviewer (SMB). The ITAS2010 was retrospectively assessed for all visits. All three scores assess the multisystem involvement in childhood TAK. PVAS and ITAS2010 measure disease activity including symptoms that newly occurred or worsened over the past 4 weeks or are persistent for less than 3 months. The PVAS is currently the only validated disease activity measurement tool in childhood vasculitis and routinely used for clinical research [21]; however, it may not be optimal for assessment of disease activity in pediatric large vessel vasculitis. The ITAS2010 has been specifically developed for assessment of disease activity in TAK patients [22], but has not been validated in children and is therefore not routinely used in childhood TAK. Given these specific features, both, the PVAS and ITAS2010 were used for assessment of disease activity. In contrast to the PVAS and ITAS2010, the PVDI assesses irreversible cumulative disease-related damage and only includes symptoms present for more than 3 months.

Definitions

Active disease was defined as PVAS ≥ 1 and/or raised inflammatory markers and/or imaging evidence of active disease including new angiographic lesions, vessel wall inflammation by post-contrast enhancement and thickening. Inactive disease was defined by the concurrence of all three, a PVAS of zero and normal inflammatory markers and inactive disease on imaging. Disease flare was defined as occurrence of new TAK symptoms and/or increase of inflammatory markers necessitating altered treatment and/or new or worsening angiographic lesions. The duration of stable disease prior to disease flare was not further defined. Inactive disease was defined by the treating expert clinician.

Treatment response was defined as a ≥ 50% reduction in the PVAS at 6 months after treatment start.

Outcome

Primary outcome was disease flare or death. Secondary outcomes included disease activity as defined above and disease damage as measured by PVDI at last follow-up.

Analysis

Patients were analyzed in groups depending on type of induction therapy. Flare-free survival of biologic or non-biologic treatments was calculated using Kaplan-Meier methods. As each child could experience multiple flares, a time-to-event analysis was used. Therefore, the time origin for each treatment was considered the time when the child was started on treatment. Logistic regression was used to identify prognostic factors of adverse outcome. The following variables were studied as potential predictors: age, gender, PVAS, ITAS2010, ESR, platelet count and treatment category (biologic versus non-biologic treatment). The analysis was adjusted for the number of treatment episodes per patient. Variables with more than 20% missing values were excluded. Categorical variables were compared by Fisher’s exact test, odds ratios (OR) were calculated including the 95% confidence intervals (95%-CI). Continuous variables were reported as medians with interquartile range (IQR), categorical variables as percentages with frequencies. Analyses were conducted in Prism (GraphPad 6.0 g, San Diego, CA, USA) and R 3.12 (Vienna, Austria).

Demographic features, clinical and laboratory findings

Three patients (11%) were diagnosed with tuberculosis at presentation of childhood TAK. Two children had been diagnosed with inflammatory bowel disease based on classic histopathology findings prior to the onset of TAK. At time of diagnosis, inflammatory markers were elevated in 21 children (78%). ESR was increased in 20/25 (80%), CRP in 14/19 children (74%). Laboratory results are presented in Table 2. The median PVAS at diagnosis was 10 (IQR 7–18), the median ITAS2010 14 (IQR 10–18).

Angiographic features

MRA alone (11/27, 41%) or in combination with CTA or conventional angiography (3/27, 11%) was most commonly performed at diagnosis. Conventional angiography was the initial imaging modality in 9/27 (30%) children, and CTA in 3/27 (11%). The most frequently involved vessels were the abdominal aorta (89%), the renal (67%) and carotid arteries (56%). Figure 1 depicts the frequencies of the involved vessels.

Treatment

Twenty-two of 27 children (81%) received immunosuppressive therapy (Figs. 2 and 3). Five (19%) were considered to have inactive disease at diagnosis and therefore did not receive immunosuppressive agents.

Induction treatment

Patients with active disease at diagnosis (22) received different treatment regimens. Initially, 4/27 children (15%) received high-dose corticosteroids only, and 18 (67%) received a combination of corticosteroids plus another immunosuppressive agent. These immunosuppressive agents prescribed in combination with corticosteroids included MTX in ten (37%), cyclophosphamide in five (19%), and MTX plus a biologic agent in three (11%) children. The latter three children received tumor necrosis factor alpha (TNF-α) inhibitors, two infliximab and one adalimumab. Two of them had been already on TNF-α inhibitors for pre-existing inflammatory bowel disease, when diagnosed with childhood TAK. At diagnosis of childhood TAK, both were started on high-dose corticosteroids and MTX; in addition, the TNF-α inhibitor dosing was increased. Therapy was chosen at the discretion of the treating physician and the availability of the drug.

PVAS and ITAS2010 did not significantly differ between the four induction treatment groups at diagnosis. At 6 months 20/22 treated children survived and 18/20 surviving children (90%) had responded to the treatment (Fig. 3).

Maintenance treatment

At 6-month follow-up, all 20 surviving children remained on corticosteroids at a median dose 0.4 mg/kg/day prednisone equivalent (IQR 0.4–0.7 mg/kg/day). Maintenance treatment varied and included MTX, azathioprine, MMF and leflunomide or one of them combined with biologics. An overview of the treatment regimens is shown in Figs. 2 and 3.

Co-therapies

Most children (18/27, 67%) received antihypertensive drugs (median 1, 0–3). Low-dose acetylsalicylic acid was prescribed to 15/27 (56%) and anticoagulation to 6/27 (22%) children. A 6 to 9-month course of antituberculosis treatment was initiated simultaneously with immunosuppressive therapy in the three children diagnosed concomitantly with tuberculosis.

Surgical and endovascular interventions

Eight of 27 patients (30%) required vascular surgery or intervention after diagnosis of childhood TAK. Eighteen procedures were performed: renal artery angioplasty (eight procedures in four patients), balloon dilatation of the aorta (four procedures in two patients) and axillary-femoral bypass, coronary bypass and embolectomy of the femoral artery in one patient each. Another child required a splenorenal shunt, local thrombolysis due to thrombosis and eventual nephrectomy. Two patients received surgical treatment prior to childhood TAK diagnosis (unilateral nephrectomy with subsequent renal revascularisation procedure and balloon dilatation of the aorta).

Complications

Complications were divided into disease- and treatment-related complications. Arterial dissection was noted in three children (11%) at diagnosis; no new dissections were seen during follow-up. Cerebral infarction with consecutive craniectomy (due to increased intracranial pressure) and intestinal ischemia requiring intestinal resection were documented in one child each. Relevant treatment-associated side effects included Cushing syndrome in 21/22 children on corticosteroids, transaminitis defined as transaminase levels ≥ twice the upper limit of normal in four children and vertebral fractures in one child. Four severe infections in four children required hospitalization including one child treated with corticosteroids and MTX who developed candidemia. Three children treated with corticosteroids and cyclophosphamide were hospitalized for severe Epstein-Barr virus (EBV) infection, varicella zoster infection and bacterial sepsis following bowel ischemia.

Primary outcome

A total of 14/27 (52%) patients experienced adverse outcomes; 13 (48%) developed a disease flare and two children died (7%), one during a flare. The first flare occurred at a median of 9.0 months after diagnosis (IQR 3.9–12.1). Twenty-one flares were observed during the study period: 16 (76%) were diagnosed by new angiographic lesions, three (14%) by both new clinical symptoms and increase of acute phase reactants and two (10%) by increase of acute phase reactants only necessitating treatment alteration (start of infliximab in one, dose increase of corticosteroids and methotrexate in the other patient). The number of flares varied between one and three per patient. All but one child flared while receiving immunosuppressive therapy. Flares were treated by intensification of therapeutic regimens such as increase or restart of corticosteroids and/or start of an additional immunosuppressive drug such as cyclophosphamide, azathioprine or a biologic agent. Figure 2 depicts flares and individual treatment regimens.

When analyzing treatment episodes separately, 19 flares occurred during 44 non-biologic treatment episodes (43%) compared to only two flares during 12 biologic treatment episodes (17%) (p = 0.18; OR 3.80, 95% CI 0.81–18.59). Median corticosteroid dose at time of flare was 0.4 mg/kg/day (IQR 0.2–0.9) prednisone equivalent. None of the patients with inactive disease at time of diagnosis experienced an adverse outcome.

The 2-year flare-free survival was 80% with biologic treatments compared to 43% in non-biologic treatments when adjusted for the number of treatment episodes per patient (p = 0.03) (Fig. 4). Biologic agents included infliximab (5 mg/kg at 0, 2 and then every 4 weeks) in nine treatment episodes, adalimumab (40 mg every 2 weeks) in one and tocilizumab (8 mg/kg monthly) in two.

Two children died from TAK, both within 6 months of diagnosis. A four-year-old boy presented with a 3-month history of arterial hypertension, blood pressure discrepancy between limbs and absent pedal pulses. Conventional angiography revealed involvement of the abdominal aorta, the superior mesenteric, celiac and bilateral renal arteries. High-dose corticosteroids, anticoagulation and quintuple antihypertensive therapy were started. Cyclophosphamide was added 1 month after diagnosis due to progressive disease. He died 4 months after diagnosis following disease flare with uncontrollable hypertension and massive intestinal ischemia with sepsis. A 13-year-old girl presented with a 3-week history of constitutional symptoms, cough, chest pain, headaches and arterial hypertension. MRA showed extensive involvement of the abdominal aorta, celiac, superior mesenteric, pulmonary, renal and iliac arteries. She received high-dose corticosteroids, cyclophosphamide, anticoagulation and quadruple antihypertensive therapy. She died of uncontrollable disease and massive left middle cerebral arterial infarction following an embolism of a left ventricular thrombus and multi-organ failure 12 days after diagnosis. The treatment regimens of both children are shown in Fig. 2.

Prediction

No additional prognostic factors of adverse outcome were identified using logistic regression. None of the putative predictive variables were independently associated with an increased risk for adverse outcome; therefore, none were able to identify high-risk children early on.

Secondary outcomes

At last follow-up at a median of 2.1 years, eight of 20 surviving patients (40%) with active disease at diagnosis had continued active disease, while 12 of 20 (60%) were considered inactive. All 20 surviving patients remained on immunosuppressive treatment at last follow-up visit. Median PVAS was 0 (IQR 0–0) and median ITAS2010 0 (IQR 0–1) at last follow-up visit. Of the eight children with ongoing disease activity, two received biologic agents and six non-biologic immunosuppressive therapies. Most of the patients on non-biologic drugs were diagnosed earlier in the study period, prior to first available publications of successful treatment of childhood TAK with biologic agents [10] (Fig. 2). Importantly, 9/12 children with inactive disease on treatment had received biologics; three were treated with non-biologic therapies including corticosteroids, MTX and azathioprine (p = 0.02, OR 21.00, 95% CI 1.78–248.30). The five children with biochemically inactive disease at diagnosis maintained inactive disease without treatment throughout the study period until last follow-up.

The most commonly accrued disease damage at last follow-up were vessel stenosis (100% of surviving children), absent pulses (70%), claudication of the extremities (33%), cerebrovascular accidents (26%) and seizures (11%). Median PVDI was 4 (IQR 3–6) and did not significantly differ in the four treatment groups.