Does Takayasu arteritis differ between children and adolescents? Insights from a Russian multicenter cohort

Abstract

BACKGROUND

Takayasu arteritis (TA) is a chronic large-vessel vasculitis with significant age-related heterogeneity. Global data highlight differences between adult and pediatric TA, as well as among different pediatric age groups. However, the epidemiology and clinical features of childhood-onset TA in the Russian population are largely unknown.

AIM

To characterize the clinical and laboratory features, age-related differences, and long-term outcomes of TA in a Russian pediatric cohort.

METHODS

A retrospective multicenter cohort study included 79 patients aged under 18 years with TA, diagnosed according to EULAR/PRINTO/PRES criteria, from four centers in the Russian Federation. Demographic, clinical, laboratory, and instrumental data were collected. Patients were stratified into two age groups according to WHO definitions: Children (aged less than 10 years, n = 25) and adolescents (aged 10 years to 17 years inclusive, n = 54). Disease activity was assessed using ITAS2010 and ITAS.A. Statistical analysis was performed using StatTech v.4.12.2.

RESULTS

The median age at onset was 11.0 (8.0-14.0) years, and the time before diagnosis was 8.0 (2.0-17.5) months; females were 78.5%. The most common clinical manifestations were malaise (84.8%), fever (54.4%), and vascular bruits (54.4%). Children under 10 years had a lower proportion of girls (P = 0.016), lower erythrocyte sedimentation rate, C-reactive protein, and platelet levels (P < 0.05), ITAS.A scores (8.9 ± 4.3 vs 12.0 ± 4.4, P = 0.005), required more surgical interventions at diagnosis (32% vs 7.4%, P = 0.008) compared to adolescents. At follow-up, 34.4% of patients had persistent vascular imaging activity despite normalized laboratory parameters. Biological drugs were used in 50.6% of patients. Three patients died, all due to postoperative complications.

CONCLUSION

This pioneer Russian cohort study of childhood-onset TA reveals a unique clinical phenotype. Younger children show lower levels of inflammation than adolescents, yet they still require more surgical intervention.

Key Words: Takayasu arteritis; Children; Adolescents; Vasculitis; Age; Surgery; Disease-modifying anti-rheumatic drugs; Biological drugs; Tocilizumab; Russia

Core Tip: In this first Russian multicenter cohort study, childhood-onset Takayasu arteritis (TA) presents with distinct age-specific clinical phenotypes. Younger children (< 10 years) exhibit lower inflammatory activity but paradoxically require more frequent surgical interventions, suggesting a “burnt-out” presentation at diagnosis due to diagnostic delays. Vascular damage progresses in one-third of patients despite normalized laboratory parameters, confirming the dissociation between systemic inflammation and vessel wall injury. Tocilizumab was the predominant biologic choice with favorable outcomes. These findings emphasize that conventional inflammatory markers are insufficient for monitoring disease activity and that age-specific management strategies are essential for improving long-term outcomes in pediatric TA.

INTRODUCTION

Takayasu arteritis (TA) is a chronic granulomatous large-vessel vasculitis. It mainly affects the aorta and its branches[1]. The disease shows strong age-related features. In adults, TA typically involves women, the brachiocephalic vessels, and active inflammation. In children, sex differences are less pronounced, and the disease is more aggressive. Children often have aorta and renal artery involvement. A long asymptomatic period often delays diagnosis[2,3].

TA heterogeneity is not just about ‘adults vs children’. Recent studies show that the clinical phenotype also varies within the pediatric group, depending on age of onset[4,5]. Clemente et al[6] divided children into two groups: Those under 10 years old and those 10 years or older. Younger patients had more laboratory abnormalities (anemia, thrombocytosis) and experienced longer diagnostic delays. Xu et al[7] used more detailed age groups. In their study, infants often showed fever, high acute-phase reactants, and frequent coronary artery involvement. Adolescents often had neurological symptoms (headache, dizziness) and greater involvement of the visceral branches of the aorta.

Ethnicity is a key driver of TA variability. This points to the importance of genetic predisposition. Asia reports the highest incidence rates, while other regions report much lower rates[8-11]. Genetic studies support these patterns. International genome-wide association studies have identified key genes for TA. The greatest risk is linked to HLA-B*52. Other genes involve the IL-12/IL-23 immune pathway (IL12B), interleukin-6 (IL6), NK cell activation (LILRA3/LILRB3), and TGF-β-mediated fibrosis (PCSK5)[12-15]. HLA-B*52 is particularly associated with childhood and adolescent-onset, severe course, and therapy resistance[16]. Newer data suggest that childhood-onset TA may have an oligogenic basis involving complement and collagen genes[17]. The latest view sees TA as a form of ‘MHC-I-opathy’ and supports targeted therapy[18]. However, almost no data exist on these genes in Russian populations, especially in children, or on their effects on disease course and response.

International data link the age of onset, ethnicity, and TA’s clinical course. However, features of childhood TA in Russia are largely unknown. Russia’s genetic landscape is unique, so findings from Asia or Western Europe may not apply. There are no national registries or systematized data. This prevents accurate assessment of prevalence, description of age-based clinical variants, and identification of prognostic factors.

Therefore, this study aimed to characterize the clinical and laboratory features of childhood-onset TA in a Russian pediatric cohort, with a focus on age-related differences and long-term outcomes.

MATERIALS AND METHODS

Study design and participants

A retrospective multicenter cohort study was conducted involving four clinical centers in the Russian Federation: The Clinic of Children’s Diseases at Sechenov First Moscow State Medical University (Sechenov University), Russian Children’s Clinical Hospital of Pirogov Russian National Research Medical University, Morozov Children’s Moscow City Clinical Hospital, and the University Clinic of Saint Petersburg State Pediatric Medical University. The study included patients treated or followed at these centers between January 2007 and November 2025. Data were extracted from medical records across all centers between April 2025 and January 2026 by two independent investigators using a standardized case report form with predefined variable definitions. Discrepancies were resolved through consensus discussion. Imaging data were collected from the original radiology reports at each center, as central reassessment was not feasible.

A total of 90 patients with a confirmed diagnosis of TA were initially identified from the medical records. Inclusion criteria were: (1) Age under 18 years at the time of diagnosis; (2) Fulfillment of the EULAR/PRINTO/PRES classification criteria for TA in children[19]; and (3) Availability of complete medical documentation sufficient for the planned analyses. After applying these criteria, 79 patients were included in the study. Due to the rarity of the disease, a formal sample size calculation was not performed; all available eligible patients were included in the analysis.

For the analysis of long-term outcomes, we further restricted the cohort to patients with at least 6 months of follow-up. Accordingly, an additional 17 patients with a follow-up duration of less than 6 months were excluded from the long-term outcome analysis. These 17 patients were analyzed separately to characterize their baseline clinical features and compare them with the group included in the long-term outcome analysis.

The study protocol was approved by the Interuniversity Ethics Committee of Sechenov’s First Moscow State Medical University (Protocol No. 08-25 dated April 10, 2025). Informed consent was obtained from all participants and/or their legal representatives.

Data collection and age stratification

Demographic, epidemiological, clinical, laboratory, and instrumental parameters were assessed retrospectively by analyzing medical records. The analyzed parameters included: Sex, age at disease onset, disease duration, time from the appearance of the first typical symptoms (or laboratory/instrumental changes suggestive of vasculitis) to diagnosis verification (diagnostic delay), and follow-up duration. Laboratory evaluation included a complete blood count with assessment of the erythrocyte sedimentation rate (ESR), determination of C-reactive protein (CRP) levels, and biochemical and immunological parameters. Instrumental diagnostics were performed using Doppler ultrasound of the aorta and its branches, computed tomography angiography, and catheter angiography. Persistent imaging activity was defined based on original radiology reports as documentation of at least one of the following features on any imaging modality: Wall thickening with contrast enhancement, wall edema, or new vascular changes. The choice of methods varied for each patient depending on clinical necessity and the stage of follow-up. TA activity was assessed using the Indian Takayasu Clinical Activity Score (ITAS2010) and its modification with acute-phase reactants, ITAS.A[20]. Although ITAS2010 and ITAS. These indices were originally developed for adult populations; they have been previously used in pediatric TA cohorts and demonstrate comparable sensitivity to the pediatric vasculitis activity score[21-24]. A dedicated pediatric tool for isolated large-vessel vasculitis is currently lacking.

The TA type was determined according to the angiographic classification by Hata et al[25]. The following types were distinguished: Type I - involvement of the branches of the aortic arch; type IIa - involvement of the ascending aorta, aortic arch, and its branches; type IIb - involvement of the ascending aorta, aortic arch with branches, and the descending thoracic aorta; type III - involvement of the descending thoracic and abdominal aorta and/or renal arteries; type IV - isolated involvement of the abdominal aorta and/or renal arteries; type V - combined involvement corresponding to types IIb and IV.

In accordance with the study’s objective - to assess age-related features of TA - all patients were divided into two groups according to the World Health Organization (WHO) recommendations. WHO defines adolescence as the period from 10 years to 19 years old[26]. However, due to the specifics of healthcare delivery in the Russian Federation, where patients aged 18 and older are managed in the adult healthcare network, the upper limit of the adolescent group in this study was set at 17 years inclusive. Thus, Group 1 (prepubertal/early pubertal period) included patients aged 0 to 9 years; group 2 (adolescent period) included patients aged 10 years to 17 years.

Statistical analysis

Statistical data processing was performed using StatTech v. 4.12.2 software (Stattech LLC, Russia). Quantitative variables were tested for normality using the Shapiro-Wilk test (for n < 50) or the Kolmogorov-Smirnov test (for n ≥ 50). For normally distributed data, results were described using the arithmetic mean ± SD, along with the 95% confidence interval (95%CI). For non-normally distributed data, results were presented as the median (Me) and interquartile range (Q₁-Q₃). Categorical data were described using absolute and relative (%) frequencies; 95%CIs for proportions were calculated using the Clopper-Pearson method.

Comparisons between two groups for quantitative traits were performed using the Student’s t-test or Welch’s t-test (for normal distribution with equal or unequal variances, respectively), the Mann-Whitney U test (for non-normal distribution with equal variances), or the Brunner-Munzel test (for non-normal distribution with unequal variances). For comparing categorical variables, Pearson’s χ² test (for expected frequencies > 10) or Fisher’s exact test (for expected frequencies < 10) was used. For the analysis of contingency tables with multiple rows/columns, Pearson’s χ² test was applied. The odds ratio (OR) with 95%CI was calculated as a measure of effect for relative outcomes. In cases of zero values in contingency table cells, the Haldane-Anscombe correction was applied.

To build a predictive model of outcome probability, logistic regression was used, and the proportion of explained variance was assessed using Nagelkerke’s R². Differences were considered statistically significant at P < 0.05.

RESULTS

General cohort characteristics and clinical-laboratory findings at disease onset

In the studied cohort, the median age at TA onset was 11.0 (8.0-14.0) years, with girls comprising 78.5% of patients. The median diagnostic delay was 8.0 (2.0-17.5) months, and the mean follow-up duration was 28.0 ± 22.0 months (Table 1).

Table 1 Demographic and baseline characteristics of the total cohort and stratified by age group (< 10 years vs ≥ 10 years), n (%).

Characteristic	Total cohort (n = 79)	< 10 years (n = 25)	≥ 10 years (n = 54)	P value
Age at onset, years, median (IQR)	11.0 (8.0-14.0)	7.0 (5.0-8.5)	13.0 (11.0-14.0)	< 0.001
Female sex	62 (78.5)	15 (60.0)	47 (87.0)	0.016
Diagnostic delay, months, median (IQR)	8.0 (2.0-17.5)	10.0 (2,0-24.0)	7.0 (2.0-12.0)	0.209
Follow-up duration, months, mean ± SD	28.0 ± 22.0	35.8 ± 28.4	24.6 ± 19.6	0.088
Surgical intervention at diagnosis	12 (15.2)	8 (32.0)	4 (7.4)	0.008
ITAS.A at onset, mean ± SD	11.0 ± 4.6	8.9 ± 4.3	12.0 ± 4.4	0.005
ITAS 2010 at onset, median (IQR)	7.0 (5.0-12.5)	5.0 (3.0-8.0)	7.0 (5.0-9.0)	0.061
ESR at onset, mm/hour, mean ± SD	47.8 ± 24.9	37.5 ± 19.5	52.7 ± 25.8	0.012
CRP at onset, mg/L, median (IQR)	35.5 (11.8-90.2)	17.0 (5.0-36.0)	47.0 (15.0-100.2)	0.003

IQR: Interquartile range; ESR: Erythrocyte sedimentation rate; CRP: C-reactive protein.

A high frequency of constitutional and vascular symptoms characterized the clinical picture at onset. Most commonly reported were malaise (84.8%), fever (54.4%), and vascular bruits (54.4%). Other frequent manifestations included headache (46.8%) and diminished or absent pulse (31.6%). Symptoms of critical ischemia were less common but still occurred in a significant proportion of cases: Arterial hypertension (35.4%), limb claudication (26.6%), limb numbness (16.5%), and carotidynia (13.9%) (Table 2).

Table 2 Clinical manifestations at disease onset in the total cohort and stratified by age group (< 10 years vs ≥ 10 years), n (%).

Parameter	Total cohort (n = 79)	Total cohort (n = 79), %, 95%CI	< 10 years	≥ 10 years	P value
Malaise	67	84.8, 75.0-91.9	18 (72.0)	49 (90.7)	0.044
Fever	43	54.4, 42.8-65.7	10 (40.0)	33 (61.1)	0.080
Headache	37	46.8, 35.5-58.4	11 (44.0)	26 (48.1)	0.731
Arthralgia	21	26.6, 17.3-37.7	4 (16.0)	17 (31.5)	0.179
Arthritis	5	6.3, 2.1-14.2	0 (0.0)	5 (9.3)	0.173
Weight loss	20	25.3, 16.2-36.4	4 (16.0)	16 (29/6)	0.269
Myalgia	14	17.7, 10.0-27.9	4 (16.0)	10 (18.5)	1.000
Abdominal pain	14	17.7, 10.0-27.9	6 (24.0)	8 (14.8)	0.353
Limb claudication	21	26.6, 17.3-37.7	10 (40.0)	11 (20.4)	0.099
Limb numbness	13	16.5, 9.1-26.5	4 (16.0)	9 (16.7)	1.000
Diminished/absent pulse	25	31.6, 21.6-43.1	8 (32.0)	17 (31.5)	1.000
Vascular bruits	43	54.4, 42.8-65.7	11 (44.0)	32 (59.3)	0.205
Blood pressure difference between extremeties ≥ 10 mmHg	22	27.8, 18.3-39.1	7 (28.0)	15 (27.8)	1.000
Carotidynia	11	13.9, 7.2-23.5	2 (8.0)	9 (16.7)	0.487
Arterial hypertension	28	35.4, 25.0-47.0	9 (36.0)	19 (35.2)	1.000
Vascular pain	21	26.6, 17.3-37.7	5 (20.0)	16 (29.6)	0.424
Epistaxis	4	5.1, 1.4-12.5	3 (12.0)	1 (1.9)	0.091
Lymphadenopathy	7	8.9, 3.6-17.4	1 (4.0)	6 (11.1)	0.422
Hearing loss	1	1.3, 0.0-6.9	0 (0.0)	1 (1.9)	1.000
Nausea	4	5.1, 1.4-12.5	0 (0.0)	3 (5.6)	1.000
Syncope	8	10.1, 4.5-19.0	3 (12.0)	5 (9.3)	0.703
Dizziness	8	10.1, 4.5-19.0	1 (4.0)	7 (13.0)	0.424
Visual impairment	4	5.1, 1.4-12.5	0 (0.0)	4 (7.4)	0.302

Diagnosis was primarily based on imaging modalities. Doppler ultrasound of the aorta and its branches was performed in 92.4% of patients, computed tomography angiography in 78.5%, and conventional catheter angiography in 39.2%. According to instrumental examination, type V aortic involvement according to the Numano classification was the most frequent (58.2%), followed by type IV involving the abdominal aorta and its branches (15.2%); type III was the rarest (3.8%) (Figure 1). The median number of affected vessels at diagnosis was 4.0 (3.0-7.0). The most common sites of involvement were the left common carotid artery (51.9%) and the abdominal aorta (51.9%), followed by the left subclavian artery (41.8%), descending aorta (36.7%), right common carotid artery (36.7%), and celiac trunk (32.9%). Involvement of the coronary (2.5%), splenic (2.5%), and pulmonary (3.8%) arteries was extremely rare (Table 3).

Figure 1 Gender distribution in the younger (< 10 years) and older (≥ 10 years) age groups. P = 0.016 (Fisher’s exact test).

Table 3 Frequency of vascular involvement at disease onset in the total cohort and stratified by age group (< 10 years vs ≥ 10 years), n (%).

Vessel	Total cohort (n = 79)	Total cohort (n = 79), %, 95%CI	< 10 years (n = 25)	≥ 10 years (n = 54)	P value
Brachiocephalic trunk	20	25.3, 16.2-36.4	6 (24.0)	14 (25.9)	1.000
Left common carotid artery	41	51.9, 40.4-63.3	11 (44.0)	30 (55.6)	0.339
Right common carotid artery	29	36.7, 26.1-48.3	7 (28.0)	22 (40.7)	0.323
Left subclavian artery	33	41.8, 30.8-53.4	9 (36.0)	24 (44.4)	0.479
Right subclavian artery	13	16.5, 9.1-26.5	3 (12.0)	10 (18.5)	0.536
Axillary artery	1	1.3, 0.0-6.9	0 (0.0)	1 (1.9)	1.000
Left vertebral artery	14	17.7, 10.0-27.9	4 (16.0)	10 (18.5)	1.000
Right vertebral artery	4	5.1, 1.4-12.5	2 (8.0)	2 (3.7)	0.587
Ascending aorta	17	21.5, 1.4-12.5	6 (24.0)	11 (20.4)	0.772
Aortic arch	15	19.0, 11.0-29.4	7 (28.0)	8 (14.8)	0.219
Descending aorta	29	36.7, 26.1-48.3	11 (44.0)	18 (33.3)	0.453
Abdominal aorta	41	51.9, 40.4-63.3	15 (60.0)	26 (48.1)	0.327
Celiac trunk	26	32.9, 22.7-44.4	8 (32.0)	18 (33.3)	1.000
Superior mesenteric artery	21	26.6, 17.3-37.7	7 (28.0)	14 (25.9)	1.000
Left renal artery	24	30.4, 20.5-41.8	9 (36.0)	15 (27.8)	0.600
Right renal artery	20	25.3, 16.2-36.4	8 (32.0)	12 (22.2)	0.409
Splenic artery	2	2.5, 0.3-8.8	1 (4.0)	1 (1.9)	0.536
Iliac arteries	8	10.1, 4.5-19.0	3 (12.0)	5 (9.3)	0.703
Coronary arteries	2	2.5, 0.3-8.8	1 (4.0)	1 (1.9)	0.536
Pulmonary trunk	3	3.8, 0.8-10.7	1 (4.0)	2 (3.7)	1.000
Femoral arteries	5	6.3, 2.1-14.2	2 (8.0)	3 (5.6)	0.849

Laboratory parameters reflected high systemic inflammatory activity: Mean ESR was 47.8 ± 24.9 mm/hour (95%CI: 42.1-53.6), median CRP level was 35.5 (11.8-90.2) mg/L. Characteristic hematological abnormalities included anemia [mean hemoglobin 106.4 ± 14.1 g/L (95%CI: 103.1-109.8)] and thrombocytosis [mean platelet count 439.9 ± 147.1 × 10⁹/L (95%CI: 404.7-475.3)]. The median ITAS2010 activity index was 6.0 (4.0-9.0) points, and the mean ITAS.A was 11.0 ± 4.6 (95%CI: 9.9-12.1) points.

All patients received glucocorticoids (GC) at disease onset at a dose of 1.0 (0.5-1.0) mg/kg/day (prednisolone equivalent). Methotrexate was the most frequently used first conventional disease-modifying antirheumatic drug (cDMARD) (71.8%), followed less often by cyclophosphamide (4.2%) and mycophenolate mofetil (1.4%). Twelve patients (24.5%) were initiated on biologic disease-modifying antirheumatic drugs (bDMARDs) immediately after diagnosis verification. Overall, 40 children (50.6%) received bDMARDs during the entire follow-up period, with the follow-up duration significantly longer in this group than in patients without bDMARDs: 30.5 (30.8-53.0) months vs 15.0 (2.0-33.8) months (P < 0.001). Among bDMARDs, tocilizumab was the first-line agent (70%), followed by infliximab (12.5%), etanercept (10%), and adalimumab (7.5%).

Age-related features of the clinical course of Takayasu arteritis

To analyze age-related features, the cohort was divided into two groups according to WHO recommendations: Children under 10 years (n = 25) and adolescents aged 10 years and older (n = 54). The mean follow-up duration was 35.8 ± 28.4 months (95%CI: 23.8-47.8) in the younger group and 24.6 ± 19.6 months (95%CI: 19.3-29.9) in the older group (P = 0.088).

The most striking difference was in gender distribution: Among children under 10 years, boys constituted 40%, whereas among adolescents, only 13% were male (P = 0.016). The odds of being female in the younger group were 4.5 times lower (OR = 0.223; 95%CI: 0.072-0.690) (Figure 1).

Notably, patients in the younger age group were significantly more likely to require surgical treatment for the underlying disease at the time of diagnosis: 32% vs 7.4% (P = 0.008). Furthermore, we identified a statistically significant, moderate negative correlation (ρ = -0.316; P = 0.012): The younger the child at the time of TA manifestation, the higher the probability of subsequent surgical intervention. Regression analysis confirmed this trend: An increase of one surgical procedure was associated with a decrease in the mean age of onset by 1.3 years (P < 0.001) (Table 4).

Table 4 Association between age of Takayasu arteritis onset and surgical treatment.

Parameter	Value	95%CI	P value
Spearman’s correlation (age of onset and number of surgeries)	ρ = -0.316	-	0.012
Linear regression: Change in age of onset per one surgery	β = -1.3 years	-2.1 to -0.5	0.002
Linear regression model fit (R2)	0.24	-	0.002

ρ: Spearman’s correlation coefficient; β: Linear regression coefficient; CI: Confidence interval. All shown associations are statistically significant (P < 0.05).

To evaluate the discriminatory ability of the number of surgical interventions in distinguishing between patients with early and late disease onset, receiver operating characteristic (ROC) curve analysis was performed (Figure 2). The area under the curve (AUC) was 0.877 (95%CI: 0.738-1.000; P < 0.001), indicating good discriminative performance. At the optimal cut-off value of 0.673 (determined by the Youden index), the model demonstrated a sensitivity of 81.8% and a specificity of 91.2% for predicting early TA onset.

Figure 2 Receiver operating characteristic curve for the number of surgical interventions as a predictor of early-onset Takayasu arteritis (≤ 10 years). Area under the curve = 0.877 (95%CI: 0.738-1.000).

Parents of children under 10 years reported malaise less frequently in their child (72% vs 90.7% in the older group, P = 0.044). The frequency of fever was also lower in the younger group (40% vs 61.1%), although this difference did not reach statistical significance (P = 0.080).

In our cohort, laboratory markers of inflammation at diagnosis were lower in the younger age group: ESR levels in the < 10 years group were significantly lower [37.5 ± 19.5 mm/hour (95%CI: 29.3-45.7) vs 52.7 ± 25.8 mm/hour (95%CI: 45.5-59.9); P = 0.012], as were CRP levels [median 17.0 (5.0-36.0) mg/L vs 47.0 (15.0-100.2) mg/L; P = 0.003]. Platelet counts in children under 10 years also remained within the normal range and were significantly lower than in adolescents [331.0 (286.0-406.0) × 10⁹/L vs 454.0 (363.8-588.0) × 10⁹/L; P = 0.004]. These differences were correspondingly reflected in the integrated activity indices: The ITAS.A index in the younger group was 8.9 ± 4.3 points (95%CI: 7.1-10.7) vs 12.0 ± 4.4 points (95%CI: 10.8-13.2) in the older group (P = 0.005).

Analysis of angiographic involvement types revealed no statistically significant differences between the groups, although type IIb (12.0% vs 5.6%) and type III (8.0% vs 1.9%) variants were somewhat more frequent in younger children (Figure 3). The pattern of involvement in individual vascular territories also showed no significant intergroup differences.

Figure 3  Distribution of angiographic types (Hata classification[25]) in patients with childhood-onset Takayasu arteritis, stratified by age group.

Regarding therapeutic approaches, patients in both groups did not differ. The proportion of patients receiving bDMARDs during the entire follow-up was 48.1% among those aged 10 years or older, compared with 48% among those aged 10 years or younger.

Outcomes and disease dynamics at the end of the follow-up

Of the 79 patients, 17 (21.5%) were followed for less than six months, which is insufficient for assessing long-term treatment outcomes; therefore, outcome analysis was performed for 62 (78.5%) children. Comparison of baseline characteristics between the excluded patients and the analyzed cohort (n = 62) revealed no significant differences in age at onset, sex distribution, or vascular involvement pattern. However, excluded patients had lower baseline inflammatory markers (CRP, ESR) and lower ITAS.A scores, and lower initial glucocorticoid doses, suggesting a milder disease course at presentation (Table 5).

Table 5 Baseline characteristics of patients included in and excluded from outcome analysis.

Parameter	Outcome analysis (n = 62)	Excluded (n = 17)	P value
Disease duration, months, median	49.00 [26.00; 65.00]	9.50 [3.25; 15.25]	< 0.001
Time from first symptoms to diagnosis, months, median	7.00 [2.00; 20.00]	9.50 [2.25; 14.25]	0.930
Follow-up duration after diagnosis, months, median	30.00 [19.50; 44.25]	1.00 [1.00; 2.00]	< 0.001
Age at onset, years, median	11.00 [9.00; 14.00]	11.00 [8.00; 13.00]	0.647
ITAS2010 at onset, median	7.00 [5.00; 9.00]	5.00 [4.00; 7.00]	0.173
ITAS.A at onset, median	11.00 [9.00; 14.00]	7.00 [5.00; 11.00]	0.015
CRP at onset, mg/L, median	40.00 [15.73; 99.25]	4.90 [1.23; 36.75]	0.001
ESR at onset, mm/hour	50.90 (22.06)	36.56 (31.49)	0.040
bDMARD use	38 (61.3)	2 (11.8)	0.001
Female sex	51 (82.3)	11 (64.7)	0.197
Surgical intervention at diagnosis	7 (11.3)	4 (23.5)	0.131

Data are presented as median (interquartile range), mean (SD), or n (%). IQR: Interquartile range; bDMARD: Biologic disease-modifying antirheumatic drugs; ESR: Erythrocyte sedimentation rate; CRP: C-reactive protein.

The mean follow-up duration in this group was 35.5 ± 20.7 months (95%CI: 30.2-40.8). At the end of follow-up, signs of disease activity on instrumental imaging persisted in 34.4% of patients. However, laboratory markers of inflammation normalized in most children: Median ESR was 8.0 (4.0-17.0) mm/hour, median CRP was 1.0 (0.1-5.9) mg/L. Activity indices also reached minimal values: ITAS2010 - 1.0 (0.0-2.0), ITAS.A - 1.0 (0.0-3.5). The median number of affected vessels increased during follow-up to 6.0 (4.0-8.0) arteries.

The median maintenance dose of glucocorticoids at the end of follow-up was 0.19 (0.12-0.28) mg/kg/day. Complete GC discontinuation was achieved in 11 patients (6.8%). Among cDMARDs, methotrexate was most commonly used (n = 46; 74.2%), followed less frequently by mycophenolate mofetil (n = 7; 11.3%). Nine children were able to discontinue csDMARD therapy completely.

Forty children (50.6%) received bDMARD therapy during the entire follow-up period. Tocilizumab was the most frequently used first-line bDMARD (n = 28; 70%), followed by infliximab (n = 5; 12.5%), etanercept (n = 4; 10%), and adalimumab (n = 3; 7.5%). Biologic disease-modifying antirheumatic drugs were discontinued in two patients: One due to sustained remission and the other due to recurrent infections despite remission of the underlying disease. Notably, patients who received bDMARDs had a significantly longer follow-up duration compared to those who did not: Median 30.0 (19.0-50.0) months vs 15.0 (2.0-33.8) months, respectively (P = 0.002). To further investigate this association, a logistic regression analysis was performed to evaluate the link between follow-up duration and bDMARD prescription. The analysis identified follow-up duration as a statistically significant predictor of biologic therapy (P = 0.006). An increase in follow-up duration by 1 month increased the odds of bDMARD prescription by 1.03 (95%CI: 1.007-1.052; P = 0.010). However, the constructed model explained only 12.3% of the variance in the dependent variable (Nagelkerke R²), indicating the influence of other unaccounted factors (Table 6).

Table 6 Logistic regression model for the prediction of biologic disease-modifying antirheumatic drugs prescription.

Parameter	Value
Model characteristics	Model characteristics
Number of patients (n)	78
Nagelkerke R2, %	12.3
Model P value	0.006
Predictor: Follow-up duration (per 1 month)	Predictor: Follow-up duration (per 1 month)
Odds ratio	1.030
95%CI	1.007-1.052
P value	0.010
Equation	Equation
Logit function (Z)	Z = -0.802 + 0.029 × follow-up (months)
Predicted probability (P)	P = 1/(1 + e-z)
Discriminatory performance	Discriminatory performance
Area under the curve	0.704
95%CI	0.589-0.820
P value	0.002
Optimal cut-off (Youden index)	0.396
Sensitivity at cut-off, %	89.7
Specificity at cut-off, %	46.2
Additional predictors assessed	Additional predictors assessed
Other baseline clinical/laboratory parameters	No statistically significant associations identified

CI: Confidence interval.

Subsequently, the model’s discriminatory ability was evaluated using ROC analysis. The AUC was 0.704 (95%CI: 0.589-0.820; P = 0.002), indicating acceptable model quality. Furthermore, the cut-off value of predicted probability P ≥ 0.396 demonstrated high sensitivity (89.7%) but low specificity (46.2%) in predicting bDMARD prescription (Figure 4).

Figure 4 Receiver operating characteristic curve for the logistic regression model predicting biologic disease-modifying antirheumatic drugs prescription. Area under the curve = 0.704 (95%CI: 0.589-0.820; P = 0.002). The optimal cut-off value (Youden index) was 0.396, corresponding to a sensitivity of 89.7% and a specificity of 46.2%.

Surgery for TA

Surgical intervention for TA-related vascular complications was undergone in 21 (26.6%) patients. Three fatal outcomes (3.8% of the total cohort) were recorded in this subgroup (Table 7). All three patients underwent surgery for bilateral critical renal artery stenosis, and two remaining boys died in the early postoperative period (within 1-2 months) from multi-organ failure secondary to septic complications. The third case involved a girl who successfully underwent simultaneous aortic and renal artery replacement. However, in the long term (three years later), she developed an aneurysm at the aortic anastomosis proximal to the graft site, the rupture of which led to death. No other deaths were recorded in the study cohort. Non-fatal postoperative complications occurred in one additional patient who developed cervical phlegmon following right common carotid artery replacement; this was successfully managed with antibiotic therapy, and the patient recovered without further sequelae.

Table 7 Clinical characteristics of three deceased patients.

Parameter	Patient 1	Patient 2	Patient 3
Sex	Male	Male	Female
Age at onset, years	5	11	8
TA type (Hata et al[25])	IIb	IV	V
ITAS2010/ITAS.A at diagnosis	11/14	4/6	6/12
Follow-up, months	57	4	88
Time from onset to surgery, months	57	3	12
Surgery reason	Bilateral critical renal artery stenosis	Bilateral critical renal artery stenosis	Bilateral critical renal artery stenosis, abdominal aorta stenosis
Cause of death	Multi-organ failure, sepsis	Multi-organ failure, sepsis	Rupture of aortic anastomotic aneurysm
bDMARDs	Yes	No	Yes

TA: Takayasu arteritis; bDMARDs: Biologic disease-modifying antirheumatic drugs.

DISCUSSION

Childhood-onset TA is a rare disease with a difficult-to-predict course and outcomes. The non-specific clinical phenotype in the early stages of the disease leads to diagnostic delays and, consequently, to the development of irreversible vascular lesions. The increased interest in pediatric TA over the last decade has enabled the identification of several features distinguishing it from the adult form of the disease[5,27,28]. We present the results from the first Russian registry of childhood-onset TA. For this analysis, younger patients were defined as those diagnosed before age 10, while older patients were those diagnosed at age 10 or above. Age-specific differences were observed: Younger patients demonstrated lower systemic inflammatory activity yet required surgical treatment more frequently. Additionally, the likelihood of bDMARD initiation in children with TA was found to correlate with follow-up duration.

In our study, the leading clinical manifestations at disease onset were constitutional symptoms: Malaise (84.8%), fever (54.4%), and vascular bruits (54.4%). These data differ from the results of a large study by Danda et al[3] in India and China, which included 119 children with TA. In that study, the most frequent findings were arterial hypertension (66.4%), pulse asymmetry (61.3%), and vascular bruits (46.2%), whereas fever and malaise were less common, occurring in 29.4% and 33.6% of cases, respectively, although fever was significantly more frequent in children than in adults (P < 0.01). The male-to-female ratio in our cohort was 1:3.7, consistent with the literature, which indicates a less pronounced gender imbalance in the pediatric population than in adults.

Regarding the angiographic pattern, our results are generally consistent with global observations: A predominance of type V aortic involvement according to the Hata classification and frequent involvement of the abdominal aorta[3,27].

It is noteworthy that, despite the relatively low prevalence of critical ischemia symptoms in the overall cohort, 15.2% of children already required surgical intervention at the time of diagnosis. Overall, 21 patients (26.6%) required surgical treatment during follow-up. The median time from first symptoms to diagnosis verification in operated patients was 11.5-month, almost twice that in the non-operated group (6.5 months). Although these differences did not reach statistical significance (P = 0.142), likely due to the limited sample size, the observed trend underscores the importance of timely diagnosis in preventing irreversible vascular complications.

Our analysis of age-related features yielded results that initially appear to contradict data from international colleagues. In our cohort, laboratory inflammatory markers (ESR, CRP, platelet count) were lower in children under 10 years than in adolescents. In contrast, studies by Clemente et al[6] and Xu et al[7] observed the opposite trend. Possible explanations for this finding include the inherent difficulty in accurately determining the true disease onset in younger children, as non-specific constitutional symptoms may precede overt vascular manifestations for prolonged periods. In such cases, acute-phase inflammatory markers may have already subsided by the time of diagnosis. At the same time, irreversible ischemic complications have already developed - a pattern sometimes referred to as a “burnt-out” phenotype. However, this interpretation remains speculative. Alternative explanations include heterogeneity of the study population (e.g., unaccounted-for genetic or environmental factors) and the small number of patients with infantile-onset TA (under 1 year) in our sample, a subgroup known to exhibit higher inflammatory activity. According to current data, the infantile form of TA is characterized by the highest inflammatory activity[4,28], and the underrepresentation of this age group may have influenced the overall results.

The 17 patients excluded from the outcome analysis due to short follow-up (< 6 months) had lower baseline inflammatory activity. This likely reflects the referral pattern: Three of the four participating centers are federal referral centers that primarily manage severe cases. Patients with milder disease may have been followed locally rather than at these specialized centers, leading to shorter documented follow-up. Thus, the analyzed cohort likely represents a subgroup with more active disease, which should be considered when interpreting the long-term outcomes.

Analysis of long-term outcomes demonstrated that, despite combined immunosuppressive therapy, including bDMARDs, a significant proportion of children (34.4%) had persistent vasculitis activity according to instrumental imaging. Meanwhile, laboratory markers of inflammation and activity indices (ITAS2010, ITAS.A) were within normal ranges for most patients. This observation confirms that standard clinical and laboratory parameters are non-specific for assessing true TA activity and highlights the need to identify new biomarkers that reflect persistent vascular inflammation.

The increase in the median number of affected vessels by the end of follow-up is likely related to a refractory or relapsing disease course. Consequently, the search for predictors of an aggressive TA course at the diagnostic stage is particularly relevant. For instance, in our previous work, we proposed that an ITAS.A level above 12 points at onset is a risk factor for a relapsing course[29]. Deniz et al[30] demonstrated the prognostic value of the Combined Arteritis Damage Score for the need for bDMARD therapy. Further studies like these are needed to develop effective routine clinical tools.

In our study, half of the patients (50.6%) ultimately received bDMARDs, with tocilizumab being the first-line agent in 70% of cases. The observed frequency of tocilizumab use aligns with its established efficacy and favorable therapeutic survival profile reported in systematic reviews and meta-analyses[31-33]. Nevertheless, the lack of a comparative group in our study precludes direct assessment of treatment outcomes. It is plausible that the use of tocilizumab, which blocks interleukin-6 and consequently CRP production, may have contributed to the normalization of laboratory parameters despite persistent vascular inflammation. The median glucocorticoid dose at the end of follow-up was reduced to a maintenance level (0.19 mg/kg/day), and 11 patients were able to discontinue steroids completely.

Furthermore, our study demonstrated a correlation between bDMARD prescription and follow-up duration, which appears clinically logical. This correlation aligns with the progressive nature of TA, a vasculitis marked by frequent relapses. International studies evaluating 5-year relapse-free survival have reported rates of 62.3% by Fan et al[34] and 58.6% by Comarmond et al[35], confirming the aggressive nature of the disease. As a result, clinicians increasingly encounter relapses and must escalate treatment over time. The latest international guidelines recommend bDMARDs as the agents of choice for managing TA exacerbations[36,37]. In our cohort, logistic regression analysis identified follow-up duration as a significant predictor of bDMARD prescription. However, the model explained only 12.3% of the variance (Nagelkerke R²), indicating that other unmeasured factors - such as disease severity, refractory course, and physician treatment preferences - likely play a more important role in treatment escalation.

All three fatal outcomes in our cohort were associated with surgical treatment of the underlying disease and subsequent postoperative complications (sepsis, aortic aneurysm rupture). These cases highlight the severity of the condition in patients requiring surgery and the complexity of clinical decision-making in children with critical vascular lesions. The data obtained emphasize the need for careful risk-benefit assessment when planning surgical interventions in this patient category[36,38].

Limitations

Several limitations should be acknowledged. The retrospective design may have affected the completeness and uniformity of the data. Although data were extracted by two investigators using a standardized case report form, they were not blinded to each other's assessments, which may have introduced measurement bias. The lack of centralized imaging review and reliance on local radiology reports from different centers with varying equipment and interpretation practices may have introduced variability in the assessment of vascular lesions. The absence of a standardized treatment protocol and inconsistent access to biologic agents over time may have influenced treatment-related outcomes.

Additionally, 17 patients (21.5%) with follow-up duration less than six months were excluded from the outcome analysis. While these patients did not differ from the analyzed cohort in key demographic or angiographic characteristics, they had lower baseline inflammatory activity, which may have introduced a selection bias toward including more severe cases in the long-term outcome analysis.

The relatively small sample size (n = 79) limited our ability to perform multivariable analyses adjusting for potential confounders (e.g., diagnostic delay, disease severity, treatment exposure). Therefore, our findings should be interpreted as exploratory. The ITAS2010 and ITAS.A indices, while previously used in pediatric TA studies, are not formally validated in children. Finally, the study did not investigate genetic markers that could explain the identified phenotypic features and disease heterogeneity in the Russian population.

CONCLUSION

This pioneer Russian cohort study of childhood-onset TA reveals a unique clinical phenotype. Younger children show lower inflammatory activity than adolescents, but they still require more surgical intervention. Laboratory normalization and persistent vascular damage often do not match, highlighting the need for better biomarkers. Establishing a national registry for pediatric TA is urgently needed to study the genetic and environmental factors underlying TA heterogeneity in Russia and to improve diagnosis and tailored therapy.