Prevalence of locoregional and distant lymph node metastases in children and adolescents/young adults with soft tissue sarcomas: a Bayesian meta-analysis of proportions

Abstract

Background: The presence of both regional and distant lymph node metastases (LNM) in paediatric and adolescent/young adult (AYA) patients with soft tissue sarcomas (STS) significantly impacts clinical outcomes. However, reported rates of LNM vary widely across the literature and are often accompanied by substantial uncertainty. We aimed to quantitatively synthesise global proportions of LNM across different histological subtypes and tumour sites in this population.

Methods: In this meta-analysis, we systematically searched MEDLINE, Scopus, and Web of Science from inception until May 1, 2024 (updated on June 1, 2025) for studies published in English that reported LNM rates in patients with STS aged 0-21 years. Eligible study designs included cohort studies, case-control studies, case series, and randomised controlled trials. Patient-level data were not requested from study authors. LNM had to be confirmed clinically, by imaging, or histologically. We excluded reviews, editorials, and case reports with fewer than three patients. We conducted a random-effects Bayesian meta-analysis using a logit transformation to synthesise LNM proportions. The posterior distributions of LNM prevalence were summarised by the posterior mean and 95% credible intervals (CrIs). Study quality was assessed across seven domains: confounding, participant selection, exposure classification, deviations from intended exposures, missing data, outcome measurement, and selective reporting.

Findings: Of 3969 records screened, 263 articles were included in the data synthesis. These comprised 147 studies on rhabdomyosarcoma (RMS), 106 on non-rhabdomyosarcoma soft tissue sarcoma (NRSTS), and 10 on mixed RMS/NRSTS cohorts, representing 53,093 patients with STS. The pooled posterior mean proportion of LNM in patients with RMS (n = 41,547) was 0.228 (95% CrI: 0.202-0.255), with the highest rates observed in patients with alveolar RMS (posterior mean proportion: 0.370; 95% CrI: 0.276-0.473). Subgroup analysis by RMS primary site revealed the highest LNM rates in the perianal/perineal region (0.466; 95% CrI: 0.397-0.537), extremity (0.281; 0.210-0.363), and non-parameningeal head/neck region (0.259; 0.167-0.376). Among patients with NRSTS (n = 11,546), the pooled posterior mean proportion of LNM was 0.111 (95% CrI: 0.092-0.133), with desmoplastic small round cell tumour (0.440; 0.335-0.552), clear cell sarcoma (0.212; 0.163-0.275), and malignant rhabdoid tumour (0.199; 0.141-0.273) showing the highest rates. Most analyses had moderate-to-high heterogeneity (95% CrI for tau: 0.7443-1.1139).

Interpretation: Our Bayesian meta-analysis synthesises global evidence on the prevalence of LNM in paediatric and AYA patients with STS, highlighting the significant heterogeneity in LNM rates by histological subtype, particularly in NRSTS, and by tumour location, especially in RMS. Future studies should aim to standardise lymph node staging protocols and reporting practices to improve classification accuracy and enhance comparability across studies.

Funding: None.

Keywords: Bayesian meta-analysis; Lymph node metastases; Rhabdomyosarcoma; Soft tissue sarcomas; ‘Non-rhabdomyosarcoma’ soft tissue sarcoma.

Fig. 1

PRISMA flow diagram of study selection. This diagram illustrates the study selection process according to the PRISMA guidelines. RMS = rhabdomyosarcoma; NRSTS = non-rhabdomyosarcoma soft tissue sarcoma.

Fig. 2

Risk of bias assessments. The figure displays the distribution of risk of bias ratings (low, moderate, serious, unclear) across individual domains (D1–D7) and overall for all included studies. Colour intensity corresponds to the number of studies, as indicated by the legend. D1 = confounding, D2 = selection of participants, D3 = classification of exposures, D4 = deviations from intended exposures, D5 = missing data, D6 = measurement of outcomes, D7 = selection of the reported result.

Fig. 3

Pooled LNM prevalences in patients with RMS, by tumour location. Each row shows the posterior mean proportion of lymph node metastases (LNM) with 95% credible intervals (CrI), derived from Bayesian meta-analysis using logit-transformed proportions. Black dots represent the posterior mean; horizontal lines indicate the 95% CrI. The density curves illustrate the posterior distributions. RMS = rhabdomyosarcoma; LNM = lymph node metastases; CrI = credible interval; GU = genitourinary; NS = not specified; BP = bladder-prostate.

Fig. 4

Pooled LNM prevalences in patients with NRSTS, by tumour histology. Each row shows the posterior mean proportion of lymph node metastases (LNM) with 95% credible intervals (CrI), derived from Bayesian meta-analysis using logit-transformed proportions. Black dots represent the posterior mean; horizontal lines indicate the 95% CrI. The density curves illustrate the posterior distributions. NRSTS = non-rhabdomyosarcoma soft tissue sarcoma; DSRCT = desmoplastic small round cell tumour; PPNT = peripheral primitive neuroectodermal tumour; MPNST = malignant peripheral nerve sheath tumour; CrI = credible interval.

Supplementary Fig. S1

Supplementary Fig. S2

Supplementary Fig. S7