From January 2017 to September 2020, 368 newly diagnosed pediatric LCH patients were identified through the electronic medical records of Beijing Children's Hospital [18]. Among them, 330 patients met the inclusion criteria: confirmed LCH diagnosis, age < 18 years, complete baseline peripheral lymphocyte subsets and cytokines testing, and standardized treatment with follow-up. Thirty-eight were excluded due to missing data or loss to follow-up (Supplementary Figure S1).

The Ethics Committee of Beijing Children’s Hospital approved this study. Written informed consent was exempted due to the retrospective design, with all personal identifiers securely protected.

Peripheral blood for lymphocyte subset profiling and cytokine measurement was collected at diagnosis, before the start of any systemic therapy. The samples were immediately processed as fresh specimens. None of the patients had received systemic corticosteroids before the baseline blood draw. Upon admission, all patients underwent standardized tests for pathogens, and no evidence of concurrent infection was identified.

Cells (stimulated or unstimulated) were stained for surface antigens using fluorochrome-conjugated monoclonal antibodies, then fixed and permeabilized for intracellular staining. Isotype controls with irrelevant specificity were run in parallel. Samples were acquired on a BD FACSCanto II, with lymphocytes defined by low side scatter and bright CD45 expression. CD3+T, CD19+B, and CD16+/CD56+ natural killer (NK) cells were subsequently assessed from total lymphocytes. The percentages of CD4+ and CD8+T cells were derived from the CD3+ population. T helper 1 (Th1, defined as CD3+ CD8- IFN-γ+), Th2 (CD3+ CD8- IL-4+) and Th17 (CD3+ CD8- IL-17+) subsets were calculated from CD3+ CD8- cells. Tregs (CD3+ CD4+ CD25+ FoxP3+) were derived from CD3+ CD4+ cells. Ratios (CD4/CD8, Th1/Th2, and Treg/Th17) were calculated. Antibodies used are listed in Supplementary Table S1, and the gating strategy is depicted in Supplementary Figure S2.

Serum cytokines (IL-2, IL-4, IL-6, IL-10, TNF-α, and IFN-γ) were quantified by cytometric bead array (CBA, BD Human Th1/Th2 Cytokine Kit II, 551,809). Standards were prepared by serial dilution. Capture beads were incubated with serum samples, standards, and kit controls for three hours at room temperature, washed twice, and acquired by flow cytometry (FC, BD FACSCanto II). Data were analyzed with FCAP Array v3.0. Potential outliers — values ± 3 standard deviations (SD) from group means—were re-analyzed on raw fluorescence histograms; exclusions were limited to biologically implausible signals consistent with technical artifact (e.g., instrument noise). Dynamic range was supported by antibody-titration optimization and per-run internal controls. Protocols followed peer-reviewed standards.

Patients were stratified by disease extent: SS, MS RO-, and MS RO+LCH. RO refers to the liver, spleen, or hematologic system. Treatment followed the BCH-LCH 2014 protocol [6]. First-line therapy consisted of six or twelve weeks of induction with vindesine and prednisone, followed by maintenance that included vindesine and prednisone, with or without 6-mercaptopurine. Second-line chemotherapy consisted of an intensified regimen of cytarabine (Cytarabine), vindesine, and dexamethasone, with or without cladribine (Cladribine). Maintenance treatment then continued with vindesine, prednisone, and 6-mercaptopurine. Response was defined as non-active disease or better [19]. Disease progression was defined as clinical worsening or new lesions while on therapy. Relapse was defined as recurrence of active disease after complete remission or more than three months post-maintenance [20].

Categorical variables were compared using Fisher's exact test, while continuous variables were analyzed with the Mann-Whitney U or Kruskal-Wallis test. Associations were assessed by Spearman correlation. Overall survival (OS) was defined as the time interval from diagnosis to death, and progression-free survival (PFS) was defined as the time from treatment initiation to disease progression, relapse, or death, whichever occurred first. The performance of binary predictions was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC); optimal cut-offs were determined by Youden's index. Variables with ROC P < 0.05 were included in the univariate analyses (Kaplan-Meier with log-rank test). Those with P < 0.10 progressed to the multivariable Cox model to identify independent prognostic factors. A prognostic nomogram was built with the “rms” R package (v6.3–0). The model’s ability to discriminate between outcomes was assessed using the concordance index (C-index). The calibration was evaluated through bootstrap-corrected calibration curves, utilizing 200 resamples. A composite risk score was calculated from the independent prognostic factors. Continuous indicators, specifically IL-6, IL-10, and the Th1/Th2 ratio, were processed via the “survival” R package (v3.3.1). Patients were categorized into high- and low-risk groups based on the median risk score. Analysis was performed in SPSS 25.0 (IBM) and R (v4.2.1); Visualizations, including donut plots, bubble heatmaps, correlation pie charts, and risk score plots, were generated with the ggplot2 package (v3.4.4). Unless otherwise specified, tests were two-sided, and P < 0.05 was considered statistically significant.

Baseline demographics of 330 enrolled patients are presented in Supplementary Figure S3A. The median diagnostic age was 2.8 years (0.5–16.1). Disease extent distribution included 158 (47.9 %) SS-LCH, 115 (34.8 %) MS RO-, and 57 (17.3 %) MS RO+-LCH. Involved organs included bone, skin, lung, liver, pituitary, lymph nodes, spleen, hematologic system, ear, eye, and, rarely, the thymus. Of 233 genotyped, BRAF-V600E was detected in 136 (58.4 %), MAP2K1 variants in 21 (9.0 %), and other BRAF alterations in 17 (7.3 %). Baseline features did not differ between enrolled and excluded cohorts (Supplementary Table S2).

Initial management comprised first-line therapy (n = 294), upfront second-line chemotherapy for pituitary involvement (n = 10), targeted therapy with dabrafenib/trametinib for age or vascular constraints (n = 10), and observation for non-central nervous system (CNS)-risk single-site lesion (n = 16). Over a median follow-up of 73.0 months, three (0.9 %) deaths occurred (5-year OS, 99.1 % ± 0.5 %). Fifty-three patients (16.1 %) progressed and 97 (29.4 %) relapsed, yielding a 5-year PFS of 54.7 % ± 2.7 %. Permanent consequences (PCs) developed in 16 patients (five neurodegeneration, four liver cirrhosis, four anterior pituitary hormone deficiency, three diabetes insipidus) with a 5-year cumulative incidence of 3.1 % ± 1.0 % (Supplementary Figure S3B). Seventeen patients had diabetes insipidus at baseline.

Across LCH subtypes (Figure 1), the percentages of T cells were lowest in MS RO+-LCH (vs. SS-LCH, P < 0.001; vs. MS RO--LCH, P = 0.018, respectively), whereas B cells were lowest in SS-LCH (vs. MS RO--LCH, P = 0.002; vs. MS RO+-LCH, P < 0.001, respectively). MS RO+ LCH showed higher CD4+T cells (P = 0.014) but lower Th1, Tregs, and Th17 cells than SS-LCH (P < 0.001, P = 0.045, and P = 0.004, respectively), and higher IL-6, IL-10, and IFN-γ (P = 0.034, P < 0.001, and P = 0.010, respectively).

Figure 1.

Box-and-whisker plots showing lymphocyte subset proportions and cytokine levels across the three clinical subtypes of LCH. Abbreviations: LCH, Langerhans cell histiocytosis; RO, risk organ; SS, single-system; MS, multisystem.

Patients with progression/relapse had higher B and CD4+T cells (P = 0.048, and P = 0.014, respectively) but lower total T cells, Th1 cells, and Th1/Th2 ratios (P = 0.033, 0.003, and 0.048, respectively), with elevated IL-6, IL-10, and IFN-γ (P = 0.013, 0.006, and 0.036, respectively); IL-10 was particularly higher in non-survivors (P = 0.020). PCs showed no association with lymphocyte subsets/cytokines (Figure 2). Within MS-LCH, lower total T and Th1 cell proportions and elevated IL-10 levels were associated with progression, relapse, or death (Supplementary Figure S4). No significant difference was observed in SS-LCH (Supplementary Figure S5).

Figure 2.

Box-and-whisker plots comparing lymphocyte subset proportions and cytokine levels by event status. Events included death, progression/relapse (P/R), and permanent consequences (PC).

At baseline (Supplementary Figure S6), age < 2 years associated with lower total T, CD8+T, Th1, Th2, Th17, Tregs, and Th1/Th2 ratios, but higher B, CD4+T cells, CD4/CD8 and Treg/Th17 ratios. BRAF-V600E carriers had fewer T cells and more B cells. By organ, skin involvement was associated with lower total T and Th1 cells and higher total B and CD4+T cells; splenic or hematologic involvement correlated with lower Th17 cells and higher Treg/Th17 rations; pulmonary involvement was associated with lower Tregs. IL-6 and IL-10 levels were higher in younger age, BRAF-V600E positivity, or cutaneous disease; IFN-γ was higher with splenic or hematologic involvement (all P < 0.05).

Patients presenting with fever at diagnosis had higher IL-10 and IFN-γ at baseline compared to those without fever (P < 0.001 and P = 0.014, respectively). Among 20 patients with secondary Hemophagocytic lymphohistiocytosis (sHLH), levels of IL-10 and TNF-α were significantly elevated compared to other MS cases without sHLH (P < 0.001 and P = 0.025, respectively; Supplementary Figure S7). Additionally, Inter-marker correlations were generally weak, with most correlation coefficients below 0.2 (Supplementary Figure S8).

Age, sex, and mutation-related trends remained consistent when the cohort was divided into MS-LCH and SS-LCH (Supplementary Figure S9). The organ-specific patterns observed in MS-LCH were similar to those found in the overall cohort. In SS-LCH, immune profiles differed by site of involvement. Specifically, compared with cutaneous disease, bone involvement displayed higher proportions of Th1, CD8+T, and NK cells and higher Th1/Th2 ratios, but lower proportions of CD4+T and B cells and lower CD4/CD8 ratios (all P < 0.05).

In 294 patients receiving first-line chemotherapy, the proportions of T, B, CD4+T, CD8+T, and Th1 cells, ratios of CD4/CD8 and Th1/Th2, and levels of IL-6, IL-10 predicted progression/relapse (ROC, Figure 3A, Supplementary Table S3). Youden-derived cut-offs dichotomized with distinct PFS (Figure 3B). The Prognostic significance of clinico-biological characteristics, lymphocyte subsets, and cytokines was assessed by univariate analysis (Supplementary Table S4). On multivariable Cox, IL-6, IL-10, Th1/Th2 ratio, RO status, and week-6 treatment response were independent prognostic factors (Figure 4A), informing a nomogram model with good predictive capability (C-index 0.715, 95 % CI 0.692–0.738, Figure 4B). Furthermore, the calibration plots indicated a good agreement between the predicted PFS rates and the actual observations (Figure 4C).

Figure 3.

Prognostic significance of lymphocyte subsets and cytokines in pediatric Langerhans cell histiocytosis. (A) ROC curves for variables with significant predictive value (P < 0.05); full results are provided in Supplementary Table S3. (B) Kaplan-Meier survival curves comparing groups stratified by the cut-off values for lymphocyte subsets and cytokines, determined by Youden's index from the ROC analysis.

Figure 4.

Multivariate analysis of risk factors for progression-free survival (PFS) in patients receiving first-line therapy. (A) Forest plot of the multivariate Cox regression. (B) Nomogram for PFS prediction. (C) Calibration curves at 1, 3, and 5 years. (D) Risk-score distribution, survival status, and heatmaps of IL-6, IL-10, and the Th1/Th2 ratio. (E) Kaplan-Meier survival curves for high- vs. low-risk score groups.

Additionally, the authors performed a risk score analysis to evaluate the predictive value of IL-6, IL-10, and Th1/Th2 ratio, treating them as continuous variables (Figure 4D). The authors then classified the patients into high-risk and low-risk score groups using the median risk score as a cut-off. The survival analysis revealed inferior PFS in the high-risk group compared to the low-risk group (P < 0.001; Figure 4E).

Across the three clinical subtypes of LCH, most of the three indicators (IL-6, IL-10, and Th1/Th2 ratio) showed significant or trending PFS differences (Supplementary Figure S10). Higher IL-6 tended to be associated with progression/relapse after treatment with BRAF/MEK inhibitors (P = 0.068). However, the authors did not observe significant differences among other treatment groups (Supplementary Figure S11). In MS-LCH, total T cells, Th1 cells, and IL-10 levels retained prognostic value by ROC analysis, and IL-10 remained independently prognostic in multivariable Cox models (Supplementary Table S5-S6 and Figure S12). In contrast, these immune indicators showed no significant prognostic relevance in SS-LCH (Supplementary Table S7).

An independent validation cohort of 103 pediatric LCH patients diagnosed between October 2020 and June 2022 was included. There were 60 boys (58.3 %) and 43 girls (41.7 %), with a median age of 4.6 years (range 1.9–7.7). Disease subtypes included SS-LCH (50 cases, 48.5 %), MS RO--LCH (34 cases, 33.0 %), and MS RO+-LCH (19 cases, 18.4 %). With a median follow-up of 45.0 months, the 3-year PFS rate was 61.2 % ± 4.8 %. Using the discovery cut-offs for Th1/Th2, IL-6, and IL-10, PFS differed significantly between groups (Supplementary Figure S13A). The prognostic nomogram model showed good discrimination in the validation cohort, achieving a C-index of 0.800 (95 % CI: 0.767–0.833) (Supplementary Figure S13B). Calibration plots showed strong agreement between predicted and observed PFS (Supplementary Figure S13C).