Langerhans cell histiocytosis (LCH) is a disorder characterized by an abnormal accumulation of CD207+ and CD1a+ cells in almost any tissue. Currently, there is a lack of prognostic markers to follow up patients and track disease reactivation or treatment response. Putative myeloid precursors CD207+ and CD1a+ cells were previously identified circulating in the blood. Therefore, we aim to develop a sensitive tracing method to monitor circulating CD207+ and CD1a+ cells in a drop of blood sample of patients with LCH. A total of 202 blood samples from patients with LCH and 23 controls were tested using flow cytometry. A standardized cellular score was defined by quantifying CD207+ and CD1a+ expression in monocytes and dendritic cells, based on CD11b, CD14, CD11c, and CD1c subpopulations, resulting in a unique value for each sample. The scoring system was validated by a receiver operating characteristic curve showing a reliable discriminatory capacity (area under the curve of 0.849) with a threshold value of 14, defining the presence of circulating CD207+ and CD1a+ cells. Interestingly, a fraction of patients with no evident clinical manifestation at the time of sampling also showed presence of these cells (29.6%). We also found a differential expression of CD207 and CD1a depending on the organ involvement, and a positive correlation between the cellular score and plasma inflammatory markers such as soluble CD40L, soluble IL-2Ra, and CXCL12. In conclusion, the analysis of circulating CD207 and CD1a cells in a small blood sample will allow setting a cellular score with minimal invasiveness, helping with prognostic accuracy, detecting early reactivation, and follow-up.

Langerhans cell histiocytosis (LCH) is a rare inflammatory hematopoietic neoplastic disease with unknown etiology that leads to the destruction of affected tissues. LCH is characterized by inflammatory lesions containing a variety of myeloid precursors with high expression of CD1a and Langerin (CD207), infiltration of multiple immune cell types, and an inflammatory cytokine milieu (1, 2). Lesions can virtually arise in any organ system with particular affinity for bone, skin, lungs, and pituitary gland (3). The annual incidence of LCH ranges from 2.6 to 8.9 cases per million individuals, occurring mostly in children <15 years of age (1).

The clinical spectrum of LCH is broad, ranging from a single isolated lesion involving a one-organ system (single-system LCH) with good prognosis, to an acute, aggressive, and potentially lethal disseminated disease with involvement of two or more organs (multisystem LCH) presenting with fever, skin rash, anemia, thrombocytopenia, lymphadenopathy, and hepatosplenomegaly (1, 4).

The diagnosis of LCH is based on clinical findings and histological examination of lesional tissue, revealing an inflammatory infiltrate of eosinophils, T lymphocytes, neutrophils, macrophages, multinucleated giant cells, and LCH cells (5). Positive CD1a, CD207, and S100 immunohistochemical stainings of LCH cells are required for a definitive diagnosis (6, 7). A detailed molecular map of LCH lesions was recently described that supports a developmental hierarchy within LCH lesions that is seeded by progenitor-like cells and develops into more differentiated and destructive LCH cell states (8).

The disease etiology remains unknown, and no consensus has been reached whether LCH results from malignant transformation or unbalanced immune response that leads to the proliferation of pathogenic LC-like cells from putative myeloid precursors (1, 9). Relevant findings point to an association of gain-of-function driver mutations in BRAF V600E (10), or MAPK with MEK/ERK constitutively activated phosphorylation, boosting transcription factors that may promote cell division, survival, and cell maturation in LCH cells (11). This evidence gives weight to the misguided myeloid dendritic cell (DC) precursor model, which proposes that the maturation state of LCH progenitor cells with BRAF/MAPK activating events defines the extent of the disease, and these pathogenic precursor cells circulate in blood, migrate to lesion sites, and recruit inflammatory cells to establish an inflammatory milieu (1, 12, 13).

Under specific cytokine conditions, human CD14+ monocytes and DC CD1c+ can develop an LC-like phenotype in vitro that expresses CD207 and CD1a markers, suggesting a potential role for blood myeloid cells as precursor of LCH lesions (1417). In healthy conditions, blood myeloid populations do not express the surface markers CD207 or CD1a; however, in the inflammatory LCH context, we found CD207+CD1a+ expressing cells in the peripheral blood myeloid compartment of patients with active disease, indicating an important prognostic potential (16). Additionally, Quispel et al. (18) have previously shown the presence of CXCR4+CD1a+ cells in peripheral blood and/or bone marrow samples, suggesting such cells as the potential precursor of LCH cells. Furthermore, the BRAFV600E allele in PBMCs is primarily localized in myeloid populations (CD11c+ and CD14+ cells) (12, 19, 20). A mouse model with enforced expression of BRAFV600E mutation in CD11c+ cells recapitulated a phenotype with features of high-risk LCH, suggesting also that pathogenic lesional CD1a+CD207+ DCs may arise from myeloid cell precursors (12).

In this study, we propose a new method to screen circulating CD207+ and CD1a+ cells from a drop of blood sample of patients with LCH and monitoring cellular status in LCH. This new tool could allow to distinguish the cellular presence of LCH even before clinical manifestations, to help in prognostic sensitivity, to follow disease progression, and to measure response during treatment of the disease.

For this study, 68 patients with LCH and 23 controls were included from the Pediatrics Department of the Hospital de Clínicas Jose de San Martin and the Hospital de Niños Pedro de Elizalde (Buenos Aires, Argentina). All patients with LCH met the Histiocyte Society LCH III protocol criteria for diagnosis, stratification, and disease progression status (21). Median age was 8.2 y (range, 0.74–25.5 y) with 43 males and 25 females (Table I).

Table I.

Clinical characteristics of patients with LCH and controls

Patients with LCH (n = 68)Controls (n = 23)
Total no. of samples (225) 202 23 
Age (y), mean ± SD 8.2 ± 5.71 10.3 ± 7.03 
Sex, n (%)   
 Male 43 (63.24) male 17 (73.91) male 
 Female 25 (36.76) female 6 (26.09) female 
Organ compromise at 5 (7.35) MS RO (−) NA 
diagnosis, n (%) 14 (20.59) MS RO (+) 
 4 (5.88) MS RO (+) SSi 
 2 (2.94) MS RO (-) SSi 
 18 (26.47) SS (bone) 
 3 (4.41) SS MF (bone) 
 6 (8.82) SS (bone) SSi 
 4 (5.88) SS MF (bone) SSi 
 12 (17.65) SS (skin) 
Treatment at sampling, n (%)  NA 
 No treatment 40 (58.82) 
 First- and second- line treatment 28 (41.18) 
Hemogram   
 Mean ± SD (106/µl) n = 179 n = 12 
 Leukocytes 8.158 ± 2.821 8.297 ± 1.873 
 Monocytes 0.334 ± 0.281 0.315 ± 0.186 
 Lymphocytes 3.574 ± 1.481 4.144 ± 1.553 
 Neutrophils 4.034 ± 2.119 3.598 ± 0.666 
 Platelets (103/µl) 359.54 ± 117.61 339.17 ± 95.364 
 ESR (mm/h) 19.6 ± 16.53* 10.16 ± 6.78 
Patients with LCH (n = 68)Controls (n = 23)
Total no. of samples (225) 202 23 
Age (y), mean ± SD 8.2 ± 5.71 10.3 ± 7.03 
Sex, n (%)   
 Male 43 (63.24) male 17 (73.91) male 
 Female 25 (36.76) female 6 (26.09) female 
Organ compromise at 5 (7.35) MS RO (−) NA 
diagnosis, n (%) 14 (20.59) MS RO (+) 
 4 (5.88) MS RO (+) SSi 
 2 (2.94) MS RO (-) SSi 
 18 (26.47) SS (bone) 
 3 (4.41) SS MF (bone) 
 6 (8.82) SS (bone) SSi 
 4 (5.88) SS MF (bone) SSi 
 12 (17.65) SS (skin) 
Treatment at sampling, n (%)  NA 
 No treatment 40 (58.82) 
 First- and second- line treatment 28 (41.18) 
Hemogram   
 Mean ± SD (106/µl) n = 179 n = 12 
 Leukocytes 8.158 ± 2.821 8.297 ± 1.873 
 Monocytes 0.334 ± 0.281 0.315 ± 0.186 
 Lymphocytes 3.574 ± 1.481 4.144 ± 1.553 
 Neutrophils 4.034 ± 2.119 3.598 ± 0.666 
 Platelets (103/µl) 359.54 ± 117.61 339.17 ± 95.364 
 ESR (mm/h) 19.6 ± 16.53* 10.16 ± 6.78 

A total of 202 independent blood samples, considering follow-up, from 68 patients with LCH and 23 controls were included in this study.

*

p < 0.05, LCH versus healthy donors. ESR, erythrocyte sedimentation rate; MF, multifocal; MS, multisystem, NA, not applicable; RO, risk organ; SS, single system; SSi, special site.

Blood samples of patients with clinically active disease (AD) and nonactive disease (NAD) as well as controls were taken for routine blood tests in the medical centers between February 2018 and April 2021. The leftovers of peripheral blood samples from this routine procedure were used to perform the study. Control samples include healthy donors (n = 15), patients with non–Langerhans histiocytosis (n = 3), patients with disseminated juvenile xanthogranuloma (n = 1), and patients with hemophagocytic lymphohistiocytosis (n = 4). Controls are CD207- and CD1a-free samples, employed for flow cytometry analysis setup. The cohorts were matched in age and sex. Considering that in some patients with LCH a follow-up was carried out, several samplings were obtained and analyzed over the time. Altogether, from the 68 patients with LCH, 202 independent blood samples were processed and included in our approach. Supplemental Table I summarizes the amount of sampling for each patient.

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the Hospital de Clínicas Jose de San Martin; the Hospital de Niños Pedro de Elizalde, School of Medicine, University of Buenos Aires; and the National Academy of Medicine (Instituto de Medicina Experimental–Academia Nacional de Medicina). Informed consent for the use of blood samples and subsequent analysis was obtained from parents or legal guardians of the patients.

Selected Ab panel and concentrations for direct blood staining, cell density, and incubation time were evaluated to obtain the most appropriate conditions based on quality and optimal results. Briefly, 100 µl of fresh anticoagulated blood samples was collected and centrifuged at 800 × g for 5 min. After discarding the plasma, the remaining 50 µl of blood cells was incubated with six directly conjugated anti-human Abs for specific cell surface Ag, during 30 min on ice: CD11b-FITC (clone ICRF44), CD11c-PerCP/Cy5.5 (clone 3.9), CD1c-PE/Cy7 (clone L161), CD14-PE (clone HCD14), CD207-allophycocyanin (clone 10E2), and CD1a–Alexa Fluor 700 (clone HI149), all purchased from BioLegend. Finally, labeled cells were directly fixed with 100 µl of BD Cytofix buffer (BD Biosciences) for 20 min on ice and then washed with 1× PBS and centrifuged at 800 × g for 5 min. The obtained pellet of cells was treated with red blood lysing buffer (Thermo Fisher Scientific) and incubated for 10 min to eliminate erythrocytes. Cells were then washed with 1× PBS and analyzed by flow cytometry.

Stained cells were run on a FACSCanto I flow cytometer (Becton Dickinson) available from the National System of Flow Cytometry, Argentina, at the Instituto de Investigaciones Biomédicas en Retrovirus y SIDA–Consejo Nacional de Investigaciones Científicas y Técnicas de Argentina. Data were analyzed by FlowJo software (Tree Star). Bivariate dot plots with the appropriate combination of parameters were set to define the gating strategy. The threshold for positivity was set from controls. Comparative analysis between patients with NAD and AD was performed. The FlowSOM (22) R-FlowJo plugin was executed for unsupervised cytometry data analysis.

The standardized score arises from the quartiles distribution of cell percentages in each subpopulation. The value 1 represents the percentages of circulating CD1a+ and CD207+ cells up to the value of the first quartile, the score number 2 to the second quartile, the value 3 to the third, and the number 4 to the values between the third quartile and the maximum. The final score for each individual sample is the sum of each subpopulation score contribution.

Plasma levels of cytokines and chemokines were measured using preselected LEGENDplex human inflammation panel 2 (six-plex) for the simultaneous quantification of soluble (s)CD40L, sCD25 (IL-2Ra), TGF-β1, CXCL12 (SDF-1), sST2, and CX3CL1 (fractalkine). All experiments were performed following the manufacturer’s instructions (BioLegend). The LEGENDplex system is a bead-based multiplex assay panel, using fluorescence-encoded beads, which can be read by flow cytometry. The LEGENDplex system provides a standard curve to obtain concentrations of each cytokine based on the mean fluorescence intensity of the PE channel. Data were analyzed by FlowJo software.

Statistical analysis and figures were carried out using RStudio 4.1.0 (23). Violin plots and self-organizing maps were performed for visualization and interpretation of the cytometry data by FlowSOM and ggplot2 R packages. Receiver operating characteristic (ROC) curve analysis, Youden’s index, and power test and graphics were achieved by the pROC package (24). Linear model and Pearson correlation of cellular score and cytokines, graphical display, and tests were outlined using stats R package. Statistical significance was set at p < 0.05.

To assess the presence of CD207+ and CD1a+ circulating cells in LCH pediatric patients and controls, a direct peripheral blood surface staining method was set up and analyzed by flow cytometry in 225 independent samples, including patients with clinical AD, those with NAD, and controls. The clinical characteristics of each group are summarized in Table I. Control groups were used to set the negative threshold of CD207 and CD1a cells. Both markers were also validated by fluorescence minus one (Supplemental Fig. 1A, 1B). We characterized the mononuclear myeloid system based on CD11b, CD14, CD11c, CD1c, CD207, and CD1a using only 100 µl of fresh blood samples. (Fig. 1A shows the gating strategy based on side scatter/forward scatter to exclude the complex dots and death cells first and then doublets. To analyze the monocytes fraction, dendritic cells, and its precursors, a six-gating strategy was defined as follows: CD11bhighCD14high, CD11bhighCD11chigh, CD11bintCD11chigh, CD11b+CD11clow, CD11c+CD1c+, and CD11cCD1chigh. We found that CD207+ and/or CD1a+ cells are present in patients with clinically active LCH varying in a range from 1.19 to 65.7% for CD207+, 2.3 to 94.3% for CD1a+, and 1.01 to 47.3% for CD207+CD1a+ double-positives, as shown in (Fig. 1B and 1C. The proportions of CD207+ and/or CD1a+ cells were significantly higher in patients with AD compared with NAD patients or controls (Fig. 1C, Table II). CD207+ and/or CD1a+ cells are not homogeneously distributed among the six gates of each patient with AD, which also explains the overlap region of each independent gate in the total sampling compared with controls. Patients with clinical NAD mostly overlapped with control samples and did not show a significant median difference compared with controls in each subpopulation (Table II). An interesting observation was that 29.6% of patients with no clinical manifestation (NAD) at the time of sampling also showed CD207+ or CD1a+ cells, which predominantly matched with patients with bone compromise and could have transcendental implications for early detection.

FIGURE 1.

Screening of CD207+ and CD1a+ cells in the mononuclear myeloid compartment of LCH patients’ blood samples. (A) Representative dot plots, after fixation and RBCs lysis, showing gating strategies based on forward scatter (FSC) versus side scatter (SCC) and then discrimination of singlets. Representative dot plots show the six subpopulations discriminated based on the expression of CD11b versus CD14, CD11b versus CD11c, and CD11c versus CD1c. (B) Representative dot plots show comparative CD207 and CD1a expression levels in the six subpopulations analyzed (CD11bhighCD14high, CD11bhighCD11chigh, CD11bintCD11chigh, CD11b+CD11clow, CD11c+CD1c+, and CD11cCD1chigh) overlapping samples from controls and patients with active LCH. (C) Independent data for CD207+, CD1a+, and CD207+CD1a+ percentages of cells for each of the six myeloid subpopulations analyzed in patients with LCH (n = 202), NAD (n = 99), AD (n = 103), and controls (n = 23). (D) Receiver operating characteristic (ROC) curve to validate the strongest predictors of circulating CD207+, CD1a+, and CD207+CD1a+ cells based on the six analyzed subsets. The ROC curve model was performed using CD207+ and CD1a+ cell percentages from all myeloid subpopulations as the predictor, and included controls subjects (n = 23) and patients with LCH with clinical remission (n = 56), and patients with active disease (n = 57) and clinical manifestations as responses.

FIGURE 1.

Screening of CD207+ and CD1a+ cells in the mononuclear myeloid compartment of LCH patients’ blood samples. (A) Representative dot plots, after fixation and RBCs lysis, showing gating strategies based on forward scatter (FSC) versus side scatter (SCC) and then discrimination of singlets. Representative dot plots show the six subpopulations discriminated based on the expression of CD11b versus CD14, CD11b versus CD11c, and CD11c versus CD1c. (B) Representative dot plots show comparative CD207 and CD1a expression levels in the six subpopulations analyzed (CD11bhighCD14high, CD11bhighCD11chigh, CD11bintCD11chigh, CD11b+CD11clow, CD11c+CD1c+, and CD11cCD1chigh) overlapping samples from controls and patients with active LCH. (C) Independent data for CD207+, CD1a+, and CD207+CD1a+ percentages of cells for each of the six myeloid subpopulations analyzed in patients with LCH (n = 202), NAD (n = 99), AD (n = 103), and controls (n = 23). (D) Receiver operating characteristic (ROC) curve to validate the strongest predictors of circulating CD207+, CD1a+, and CD207+CD1a+ cells based on the six analyzed subsets. The ROC curve model was performed using CD207+ and CD1a+ cell percentages from all myeloid subpopulations as the predictor, and included controls subjects (n = 23) and patients with LCH with clinical remission (n = 56), and patients with active disease (n = 57) and clinical manifestations as responses.

Close modal
Table II.

Statistical Analysis of circulating CD207+, CD1a+, and CD207+CD1a+ cells in patients with active LCH (AD) versus NAD and controls

SubsetKruskal–Wallis Test/Dunn’s Multiple Comparisons Test
LCH AD versus LCH NADLCH AD versus ControlLCH NAD versus Control
SignificanceAdjusted p ValueSignificanceAdjusted p ValueSignificanceAdjusted p Value
CD207+ percentages CD11bhigh CD14high **** <0.0001 **** <0.0001 NS 0.5041 
CD11bhigh CD11chigh **** <0.0001 **** <0.0001 NS 0.3687 
CD11bint CD11chigh **** <0.0001 ** 0.0062 NS >0.9999 
CD11b+ CD11clow **** <0.0001 NS 0.2759 NS 0.0624 
CD11c+ CD1chigh **** <0.0001 **** <0.0001 NS 0.121 
CD11c CD1chigh **** <0.0001 * 0.0118 NS >0.9999 
CD1a+ percentages CD11bhigh CD14high **** <0.0001 *** 0.0009 NS >0.9999 
CD11bhigh CD11chigh **** <0.0001 *** 0.0001 NS >0.9999 
CD11bint CD11chigh **** <0.0001 * 0.0381 NS >0.9999 
CD11b+ CD11clow **** <0.0001 NS 0.2523 NS 0.1727 
CD11c+ CD1chigh **** <0.0001 *** 0.0003 NS >0.9999 
CD11c CD1chigh **** <0.0001 * 0.0329 NS >0.9999 
CD1a+CD207+ percentages CD11bhigh CD14high **** <0.0001 **** <0.0001 NS >0.9999 
CD11bhigh CD11chigh **** <0.0001 **** <0.0001 NS >0.9999 
CD11bint CD11chigh **** <0.0001 * 0.0125 NS 0.0719 
CD11b+ CD11clow **** <0.0001 ** 0.0054 NS 0.2295 
CD11c+ CD1chigh **** <0.0001 *** 0.0007 NS >0.9999 
CD11c CD1chigh **** <0.0001 * 0.0167 NS 0.7612 
SubsetKruskal–Wallis Test/Dunn’s Multiple Comparisons Test
LCH AD versus LCH NADLCH AD versus ControlLCH NAD versus Control
SignificanceAdjusted p ValueSignificanceAdjusted p ValueSignificanceAdjusted p Value
CD207+ percentages CD11bhigh CD14high **** <0.0001 **** <0.0001 NS 0.5041 
CD11bhigh CD11chigh **** <0.0001 **** <0.0001 NS 0.3687 
CD11bint CD11chigh **** <0.0001 ** 0.0062 NS >0.9999 
CD11b+ CD11clow **** <0.0001 NS 0.2759 NS 0.0624 
CD11c+ CD1chigh **** <0.0001 **** <0.0001 NS 0.121 
CD11c CD1chigh **** <0.0001 * 0.0118 NS >0.9999 
CD1a+ percentages CD11bhigh CD14high **** <0.0001 *** 0.0009 NS >0.9999 
CD11bhigh CD11chigh **** <0.0001 *** 0.0001 NS >0.9999 
CD11bint CD11chigh **** <0.0001 * 0.0381 NS >0.9999 
CD11b+ CD11clow **** <0.0001 NS 0.2523 NS 0.1727 
CD11c+ CD1chigh **** <0.0001 *** 0.0003 NS >0.9999 
CD11c CD1chigh **** <0.0001 * 0.0329 NS >0.9999 
CD1a+CD207+ percentages CD11bhigh CD14high **** <0.0001 **** <0.0001 NS >0.9999 
CD11bhigh CD11chigh **** <0.0001 **** <0.0001 NS >0.9999 
CD11bint CD11chigh **** <0.0001 * 0.0125 NS 0.0719 
CD11b+ CD11clow **** <0.0001 ** 0.0054 NS 0.2295 
CD11c+ CD1chigh **** <0.0001 *** 0.0007 NS >0.9999 
CD11c CD1chigh **** <0.0001 * 0.0167 NS 0.7612 

Kruskal–Wallis nonparametric test with Dunn’s multiple comparisons test for CD207+, CD1a+, and CD207+CD1a+ percentages into the six myeloid subsets was performed to determine statistical differences between controls (n = 23), patients with active LCH (n = 103), and patients with LCH in remission (n = 99). Significance levels and p values are detailed for each comparison.

*

p < 0.05,

**

p < 0.01,

***

p < 0.001,

****

p < 0.0001.

To determine the strongest gating predictors for monocytes and DCs, we have tested the six-gated population with a ROC curve (Fig. 1D). Considering that the six-gate strategies share markers and overlapping populations, particularly in the monocyte fraction (CD11b+ and CD11c+), we chose CD11bhighCD14high, with the strongest specificity for CD207+ and CD1a+ cells representing the monocyte subset, and renamed them as CD11bhighCD14highCD11chigh. We also selected the CD11bintCD11chigh gate for DCs, and CD11cCD1chigh for DC precursor representation (Fig. 1D). We confirm that these three myeloid subsets that arise from our gating strategy are also HLA-DR+ cells (Supplemental Fig. 1C).

Unsupervised clustering was performed to validate our results using the FlowSOM data analysis technique. This result enabled a clear overview of the behavior of markers in all cells through self-organizing maps. (Fig. 2A shows automatic unsupervised metaclustering arranged by the mean intensities of the markers in star charts. We identified metaclusters where CD14+ monocytes and CD1c+ and CD11c+ DCs are clearly represented. We observed differential CD207 and CD1a expression levels in distinct clusters in patients with LCH. CD207+ or CD1a+ monocytes and DCs clustered at a different node size, revealing a different cell percentage inside each cluster, with some of them being highly enriched compared with controls (Fig. 2B–D). The same differences were observed in patients with active skin or bone LCH disease (Supplemental Fig. 2A).

FIGURE 2.

Unsupervised metaclustering of LCH samples with active disease displays differential CD207 and CD1a expression levels. (A) Minimal spanning tree with an automatic metaclustering (FlowSOM) based on the six-Ab panel in a blood sample of a patient with active LCH (MS RO+). Each marker is indicated with a distinctive color as shown in the pie chart, and the mean intensity of each marker is shown in the star charts, where high expression reaches the border of the chart. The background color of the nodes, which result from the metaclustering, is consistent with the manual gating strategy of the different myeloid subpopulations and CD207+CD1a+ cells as indicated by CD11b+CD14+CD11c+ in pink, CD11c+ in violet, and CD1c+ in salmon. (B and C) Minimal spanning tree showing (B) CD207 and (C) CD1a expression levels in the same patient. The size of the nodes indicates the number of cells per node, and the color intensity represents the expression level of CD207 or CD1a, respectively. The arrows indicate differences in intensity and size compared with control sample shown in (D).

FIGURE 2.

Unsupervised metaclustering of LCH samples with active disease displays differential CD207 and CD1a expression levels. (A) Minimal spanning tree with an automatic metaclustering (FlowSOM) based on the six-Ab panel in a blood sample of a patient with active LCH (MS RO+). Each marker is indicated with a distinctive color as shown in the pie chart, and the mean intensity of each marker is shown in the star charts, where high expression reaches the border of the chart. The background color of the nodes, which result from the metaclustering, is consistent with the manual gating strategy of the different myeloid subpopulations and CD207+CD1a+ cells as indicated by CD11b+CD14+CD11c+ in pink, CD11c+ in violet, and CD1c+ in salmon. (B and C) Minimal spanning tree showing (B) CD207 and (C) CD1a expression levels in the same patient. The size of the nodes indicates the number of cells per node, and the color intensity represents the expression level of CD207 or CD1a, respectively. The arrows indicate differences in intensity and size compared with control sample shown in (D).

Close modal

The clinical presentation of LCH is heterogeneous, as are its treatment and recovery. At present, diagnosis is based on the histopathological analysis of tissue samples and clinical scoring system that are used as objective tools for therapeutic decision-making (25). Because CD207+ and CD1a+ cells can be detected in the circulation, and the percentages of positive cells can be quantified by flow cytometry analysis, we propose here a novel cellular score based on the addition of standardized contributions of CD207+ and CD1a+ into the three main mononuclear myeloid subpopulations, CD11bhighCD14highCD11chigh monocytes, CD11bintCD11chigh DCs, and CD11cCD1chigh DC precursors, resulting in a unique value for each blood sample analyzed (Table III). The standardized quantification has the advantage of minimizing individual variance and adjusting a range of the percentage (based on quartile distribution) to a discrete number for scoring. Scores from all blood samples were tested by ROC curve analysis, with controls and LCH clinical AD as classifiers, and with the cellular score as predictor. We obtained a ROC curve with an area under the curve of 0.849, with 68.85% of sensitivity and 86.67% of specificity (Fig. 3). To summarize the performance analysis and define a threshold for the cellular score, we conducted Youden’s statistic test and obtained a Y index: 14, with 95% confidence intervals of 12.37–15.63. Sample score values <14 were considered as patients with nonpresence of circulating LCH cells (non–cc-LCH), and score values ≥14 as patients with presence of circulating LCH cells (cc-LCH). No control sample reached the cutoff of 14.

FIGURE 3.

ROC curve to validate the cellular score of patients with LCH based on the presence of CD207+ and CD1a+ circulating cells. The ROC curve model was performed using the cellular score from the three main myeloid subpopulations (CD11bhighCD14highCD11chigh, CD11bintCD11chigh, and CD11cCD1chigh) as predictor, and included controls subjects (n = 23) and patients with LCH in clinical remission (n = 56), and patients with active disease and clinical manifestations (n = 57) as responses. The area under the curve (AUC) was 0.849, with 68.85% sensitivity and 86.67% specificity. The ROC curve power was 0.9998, and the significance level was 0.05. Confidence intervals for AUC and Youden’s J statistic are indicated in the graph (AUC, 0.849; YI, 14; 95% CI for YI, 12.37–15.63).

FIGURE 3.

ROC curve to validate the cellular score of patients with LCH based on the presence of CD207+ and CD1a+ circulating cells. The ROC curve model was performed using the cellular score from the three main myeloid subpopulations (CD11bhighCD14highCD11chigh, CD11bintCD11chigh, and CD11cCD1chigh) as predictor, and included controls subjects (n = 23) and patients with LCH in clinical remission (n = 56), and patients with active disease and clinical manifestations (n = 57) as responses. The area under the curve (AUC) was 0.849, with 68.85% sensitivity and 86.67% specificity. The ROC curve power was 0.9998, and the significance level was 0.05. Confidence intervals for AUC and Youden’s J statistic are indicated in the graph (AUC, 0.849; YI, 14; 95% CI for YI, 12.37–15.63).

Close modal
Table III.

Standardized cellular scores for the three mayor myeloid subpopulations (CD11bhighCD14highCD11chigh, CD11bintCD11chigh, and CD11cCD1chigh) based on the percentage of circulating CD207+, CD1a+, and CD207+CD1a+ cells, analyzed in a control sample, a patient with LCH in clinical remission (LCH 25.4), and a patient with active LCH disease (LCH 21.19)

CD11bhighCD14highCD11chighCD11bintCD11chighCD11cCD1chighSummary Score
CD207CD1aCD207+ CD1a+CD207CD1aCD207+ CD1a+CD207CD1aCD207+ CD1a+
Control Percentage 0.22 0.07 0.04 0.39 0.5 0.07 1.4 0.7  
Standarized Score 10 
NAD
LCH 25.4 SS (bone) 
Percentage 0.90 0.38 0.23 0.14 0.63 0.01 0.42 0.72 0.02  
Standarized Score 11 
AD
LCH 21.19 MS RO(−)SSi 
Percentage 1.77 4.08 0.61 0.69 3.49 0.54 0.52 5.09 0.00  
Standarized Score 18 
CD11bhighCD14highCD11chighCD11bintCD11chighCD11cCD1chighSummary Score
CD207CD1aCD207+ CD1a+CD207CD1aCD207+ CD1a+CD207CD1aCD207+ CD1a+
Control Percentage 0.22 0.07 0.04 0.39 0.5 0.07 1.4 0.7  
Standarized Score 10 
NAD
LCH 25.4 SS (bone) 
Percentage 0.90 0.38 0.23 0.14 0.63 0.01 0.42 0.72 0.02  
Standarized Score 11 
AD
LCH 21.19 MS RO(−)SSi 
Percentage 1.77 4.08 0.61 0.69 3.49 0.54 0.52 5.09 0.00  
Standarized Score 18 

The standardized score arises from the quartiles distribution of cell percentages in each subpopulation. The value 1 represents the percentages of circulating CD1a+ and CD207+ cells up to the value of the first quartile, the score number 2 to the second quartile, the value 3 to the third quartile, and the number 4 to the values between the third quartile and the maximum. The final score for each individual sample is the sum of each subpopulation score contribution. MS, multisystem; RO, risk organ; SS, single system; SSi, special site.

This score could be useful for recognizing cellular status of circulating cells, probably even prior to clinical manifestation, and it could become a powerful tool for use during follow-up.

Considering the heterogeneous manifestation and clinical involvement of LCH, we have performed an additional exhaustive analysis based on the cellular score and the distribution of a specific subpopulation in comparison with the organ involvement. Patients with circulating LCH cells (cc-LCH >14) and skin involvement showed an enrichment of CD1a+CD207+ double-positive cells in the CD11bhighCD14highCD11chigh monocyte fraction and the CD11cCD1chigh DC precursor (Fig. 4). However, patients with multisystem compromise showed very similar distribution of CD207 and/or CD1a among the three myeloid subpopulations analyzed. Patients with bone compromise had an equal proportion of CD207 and/or CD1a in the CD11bintCD11chigh subset, but with some trend for the occurrence of CD1a+ cells in the CD11cCD1chigh subpopulation, and CD207 on the CD11bhighCD14highCD11chigh monocyte fraction (Fig. 4).

FIGURE 4.

Differential segregation of CD207+ and CD1a+ myeloid subpopulations based on disease compromise. Relative abundance of CD207+, CD1a+, and CD207+CD1a+ based on the main three mononuclear myeloid subpopulations (CD11bhighCD14highCD11chigh monocyte fraction; CD11cintCD1chigh DC subset; and CD11cCD1chigh DC precursor) and tissue compromise (skin, indicated by black; bone, indicated by dark gray; and Multisystem [MS], indicated by light gray) to discriminate specific segregation. The abundance of a specific cell, as double-positive CD207+CD1a+, in one of the myeloid subsets is relative among the three-organ compromise. Thus, CD207+CD1a+ cells are preponderant in monocytes and DC precursors of skin involvement. CD1a is more distributed among skin and bone compromise than MS. However, considering only bone presentation, CD1a+ cells are enriched in CD11cCD1chigh and CD207 on the CD11bhighCD14highCD11chigh monocyte fraction. The MS form does not showed preponderance for a specific subset. Grouped bar plots are shown for LCH skin (n = 15), LCH MS (n = 45), and LCH bone (n = 43) samples, all of them with circulating LCH cells.

FIGURE 4.

Differential segregation of CD207+ and CD1a+ myeloid subpopulations based on disease compromise. Relative abundance of CD207+, CD1a+, and CD207+CD1a+ based on the main three mononuclear myeloid subpopulations (CD11bhighCD14highCD11chigh monocyte fraction; CD11cintCD1chigh DC subset; and CD11cCD1chigh DC precursor) and tissue compromise (skin, indicated by black; bone, indicated by dark gray; and Multisystem [MS], indicated by light gray) to discriminate specific segregation. The abundance of a specific cell, as double-positive CD207+CD1a+, in one of the myeloid subsets is relative among the three-organ compromise. Thus, CD207+CD1a+ cells are preponderant in monocytes and DC precursors of skin involvement. CD1a is more distributed among skin and bone compromise than MS. However, considering only bone presentation, CD1a+ cells are enriched in CD11cCD1chigh and CD207 on the CD11bhighCD14highCD11chigh monocyte fraction. The MS form does not showed preponderance for a specific subset. Grouped bar plots are shown for LCH skin (n = 15), LCH MS (n = 45), and LCH bone (n = 43) samples, all of them with circulating LCH cells.

Close modal

An interesting observation was the significantly higher mean fluorescence intensity levels of CD14 when patients were segregated based on the cellular score, denoting the relevance of CD14 monocytes in LCH (Supplemental Fig. 2B).

Using the LEGENDPlex system, a panel of six cytokines (sCD40L, sIL-2Ra, TGF-β1, CX3CL1, sST2, and CXCL12) was evaluated simultaneously in plasma of patients with LCH (Fig. 5A, 5B). The levels of each cytokine were quantified based on the standard curve (Supplemental Fig. 3A, 3B) and segregated based on the cellular score cc-LCH and non–cc-LCH. Our data indicate that patients with an active presence of cells have increased levels of sCD40L, sIL2Ra, and CXCL12 compared with patients with non–cc-LCH (Fig. 5C). Interestingly, ∼ 90% patients with bone involvement and cc-LCH are above the mean of CXLC12 levels of the non–cc-LCH cohort, and 70% when considering the mean plus 1 SD (Supplemental Fig. 3C). No significant differences were observed in CX3CL1 and sST2 levels comparing non–cc-LCH with cc-LCH. Free TGF-β1 was also measured in the same panel, but most samples were below the limit of detection. Most TGF-β1 exists in an inactive homodimeric form in circulation and is complexed to LAP and latent TGF-β binding protein (LTBP); very low levels of free active TGF-β1, the one that binds to the TGF-β receptor, are presented in plasma. Nonetheless, we have detected that 19% of LCH samples had circulating free TGF-β1 (Fig. 5C). We have also observed a positive correlation between the cellular score and the soluble inflammatory markers sCD40L, sIL-2Ra, and CXCL12 (Fig. 5D, Supplemental Fig. 3D), further supporting the association between inflammation and the cellular status of circulating CD207 and CD1a cells and demonstrating the power of this tool for diagnosis and follow-up.

FIGURE 5.

Cellular score correlates with sIL-2Ra, sCD40L, and CXCL12 plasma levels in patients with LCH. Plasma levels of cytokines and chemokines were measured using a LEGENDPlex assay (six-plex) in patients with LCH and segregated based on cellular status. (A) Representative dot plots showing beads A and B gating strategies based on side scatter (SSC) and forward scatter (FSC). (B) Representative dot plots showing comparative cytokine and chemokine plots from patients with the presence of circulating LCH cells (cc-LCH) and patients with noncirculating LCH cells (non–cc-LCH). Beads A include detection system for sILR2a, sCD40L, and free TGF-β1, and beads B include CX3CL1, sST2, and CXCL12. (C) Concentration levels and independent data of non–cc-LCH (n = 23) and cc-LCH (n = 46) for each cytokine and chemokine are graphed. Data are presented as mean ± SD. *p < 0.05. (D) Scatter plots showing the correlation between cellular score and sIL2Ra, sCD40L, and CXCL12 levels. Patients with non–cc-LCH are highlighted in gray, and patients with cc-LCH are highlighted in light gray. The best fit line with confidence intervals around the slope, correlation coefficient (R), and p values are shown.

FIGURE 5.

Cellular score correlates with sIL-2Ra, sCD40L, and CXCL12 plasma levels in patients with LCH. Plasma levels of cytokines and chemokines were measured using a LEGENDPlex assay (six-plex) in patients with LCH and segregated based on cellular status. (A) Representative dot plots showing beads A and B gating strategies based on side scatter (SSC) and forward scatter (FSC). (B) Representative dot plots showing comparative cytokine and chemokine plots from patients with the presence of circulating LCH cells (cc-LCH) and patients with noncirculating LCH cells (non–cc-LCH). Beads A include detection system for sILR2a, sCD40L, and free TGF-β1, and beads B include CX3CL1, sST2, and CXCL12. (C) Concentration levels and independent data of non–cc-LCH (n = 23) and cc-LCH (n = 46) for each cytokine and chemokine are graphed. Data are presented as mean ± SD. *p < 0.05. (D) Scatter plots showing the correlation between cellular score and sIL2Ra, sCD40L, and CXCL12 levels. Patients with non–cc-LCH are highlighted in gray, and patients with cc-LCH are highlighted in light gray. The best fit line with confidence intervals around the slope, correlation coefficient (R), and p values are shown.

Close modal

LCH is the most common histiocytic disorder caused by the expansion of myeloid precursors that differentiate into pathogenic CD207+CD1a+ LCH cells in the lesion. This disease has a remarkable pleiotropic clinical presentation (bone, skin, liver, lungs, bone marrow, and brain) affecting children and adults, with some patients having localized and self-limited disease and others developing fulminant leukemic-like forms (2, 26, 27). Although the current mortality rate is low for patients with single system compromise or without organ dysfunction in a multisystem, mortality rates for patients with organ dysfunction may reach 20% (1). Furthermore, reactivation frequently occurs after a long period of disease control, and the rates of permanent complications and sequelae remain high.

Currently, there is a lack of prognostic markers for identifying patients with treatment failure or disease reactivation as well as a measurable follow-up, which is priority. In this study, we report the largest LCH cohort analyzed by flow cytometry and introduce a new sensitive tracing technique to quantify the percentage of CD207+ and CD1a+ cells circulating in peripheral blood as a specific marker of LCH and readout of cellular status. This cellular score provides a unique value for each blood sample analyzed, accounting for the total contributions of CD207+ and CD1a+ into the monocytes and DCs. A sample with a score <14 was considered as sample with noncirculating CD207+ and CD1a+ cells, and scores ≥14 were patients with active cellular presence, which captured 81% of patients with LCH, with clinical manifestations with an excellent specificity (87%) and sensitivity (69%). We demonstrate in the present study that evaluating circulating CD207+CD1a+ cells in a drop of blood and the scoring system represent a consistent and a powerful tool to complement diagnosis, monitoring disease recurrence and contributing additional information to guide prognosis. Due to the nature of the disease, previously described LCH scoring systems could not take into account bone, endocrinal, or neurologic affections (25). Hence, our proposed CD207+CD1a+ cell score may fulfill this gap, identifying these cells in blood during a follow-up protocol and/or through chronicity or silent progression. Furthermore, we have identified the presence of CD207+CD1a+ circulating cells in patients with no evident clinical manifestation that could have transcendental implications in terms of early detection. First, we speculate that this method is highly sensitive and could detect circulating cells even before the clinical manifestation appears. Second, it could explain the chronicity of this disease in some patients with clinically stable conditions and later recurrence (relapsing/remitting behavior) over time. However, more studies with longitudinal cohorts are needed to support this concept.

The inflammation associated with the LCH condition is very well reported (2831), and in this study we have also evaluated the correlation between our cellular score screening and an inflammatory panel of cytokines (sCD40L, sIL-2Ra, TGF-β1, CX3CL1, sST2, and CXCL12). The presence of circulating cells and its correlation with unbalanced inflammatory cytokines in the LCH condition could explain the relapsing/remitting behavior of this disease, even after several years of nonclinical manifestation. sCD40L and sIL2Ra are classical inflammatory indicators, and the level of sIL-2Ra was proposed to have prognostic value to identify patients who are at particularly high risk of treatment failure, or for intensified or prolonged treatment protocols (28). The positive correlation reported in the present study further supports this hypothesis and validates that measuring circulating CD207 and CD1a cells provides a powerful tool for diagnosis, prognosis, and follow-up.

CXCL12 is strongly related to pathological bone loss (18, 32, 33), a recurrent event in patients with LCH who have bone compromise. In this sense, the serum levels of the decoy receptor osteoprotegerin (OPG), an important regulator of bone metabolism as well as immune responses, were reported to be elevated in LCH patients (34). Interestingly, we have found that most bone LCH with positive circulating CD207 and CD1a cells produced higher levels of the CXCL12 chemokine compared with the non–cc-LCH cohort. This strong association also demonstrates the relevance of our cellular screening system to monitor patients with LCH in a noninvasive way and particularly in those with bone compromise.

The embryonic-derived LCs arise from yolk sac progenitors and fetal liver-derived monocytes that seed the epidermis and are maintained locally, with tissue-resident precursors during the steady state (35). During inflammation, LCs precursors are recruited in waves from the circulation. The first wave derives from monocytes and develops low CD207 (Langerin) expression, and the second wave derives from CD1c DC precursors, which may lead to the long-term reconstitution of LCs (36). This is supported by the transcriptional profile of LCH CD1a+CD207+ DCs, which most closely relates to that of CD1c+ mDCs in the blood and tissue. Even though both types of precursors can arrive to tissue and potentially differentiate to LC-like cells, a recent report suggested that circulating HLA-DQB2+CD1c+ blood cells harboring the BRAFV600E mutation could be the precursor of LCH lesion CD207+CD1a+ DCs (19). The misguided model of LCH pathogenesis proposes that the extent of disease is defined by the state of differentiation at which an activating somatic MAPK pathway gene mutation arises (1, 3). This hypothesis is based on the BRAFV600E mutation, present in most of the LCH lesion, and the differentiation stage of the affected cell. Blood LC precursors of patients with LCH with an active single lesion and patients with multifocal low-risk LCH rarely present the BRAFV600E allele. Nonetheless, patients with high-risk LCH with BRAFV600E+ lesions consistently present the mutation in circulating myeloid cells (CD11c+ myeloid DC precursors and CD14+ monocytes), but with a very small percentage (<1% BRAFV600E+ cells in PBMCs), which may be due to the low sensitivity of detecting the mutation in a very small fraction of circulating cells (1, 12). In this regard, because Abs against BRAFV600E+ are widely available, testing the presence of BRAFV600E+ cells by flow cytometry by discriminating CD14+, CD1c+ and CD1a+ and CD207+ cells may be attractive to set.

The most frequent presentations of LCH are bone and skin, with half of them as a single site. Our new tracing technique to quantify the percentage of CD207+ and CD1a+ cells circulating in peripheral blood as a specific marker of cellular status provides the possibility of not only a promising tool for follow-up, but also as predictor of affected tissue based on the main subpopulation-expressing CD207 and CD1a. Thus, our preliminary results shed light on this hypothesis showing that higher double CD207+CD1a+ cells are associated with patients with the skin LCH form, and the CD1a+ cells could be more preponderant in the bone form. Multisystem compromise has a similar proportion of CD207+ or CD1a+ cells in most subpopulations, which is concordant with the early affection during ontogeny. The specific CD207+ and CD1a+ circulating subpopulations could potentially be associated with LCH organ compromise; however, these results need more extensive and further validation.

Osteoclasts, macrophages, and DCs are closely related cells of the myeloid lineage. Osteoclasts derive from cells of the monocyte/macrophage lineage present in bone marrow, but they are also present in peripheral blood as CD34+ and CD14+ precursors (3739). However, in cancer, autoimmune, and inflammatory diseases, osteoclast formation can be promoted by RANKL-expressing tumor or immune cells, which facilitate bone metastasis, pathological bone loss, and remodeling (40, 41). In this sense, we have found that patients with bone compromise have a trend on the occurrence of CD1a+ cells in the CD11cCD1chigh subpopulation, and CD207 on the CD11bhighCD14highCD11chigh monocyte fraction, which could be a key point for monitoring these patients. Screening blood could represent a minimally invasive diagnosis and prognosis strategy for patients with bone LCH.

A limitation of this work is that the analysis was carried out at a single center, and additional cores need to be included in a prospective study. Even though control samples were used exclusively for setting a negative threshold, the sample size is small compared with the LCH group. Additionally, a flow cytometer with the capacity to read at least six colors and eight parameters is mandatory, and such a complex instrument sometimes may not be readily available at health centers.

Researchers have debated for more than a century whether LCH represents a reactive immune disorder or is a result of malignant transformation of LC precursors. Nowadays, even though the debate is not closed, the most accepted definition says that LCH is an inflammatory myeloid neoplasm with unbalanced cytokines and the cellular immune profile playing a role, and where somatic mutations result in activation of the MAPK signaling pathway (42, 43). Our work suggests that we can now measure CD207+ and CD1a+ cells in the peripheral blood, set the cellular score, and open the door for monitoring LCH with minimal invasiveness, helping with prognostic accuracy, detection of early reactivation, and the follow-up therapy response. Finally, monitoring early signs of reactivation by using direct staining in peripheral blood could complement conventional radiographic methods in bone compromise.

We thank the patients and their families, the nursing staff, and field teams in each hospital. E.A.C.S. and A.E.E. thank Florencia Quiroga from INBIRS CONICET-UBA and the National System of Flow Cytometry, Argentina, for BD FACSCanto I availability, and Pamela Y. Chan for editorial support of the manuscript. We also thank Puga Andrea, Pinto Juana, and Perez Maria del Mar from the Hospital Pedro de Elizalde and Tutaglio Patricia and Rajoy Sandra from the Hospital de Clinicas Jose de San Martin, Buenos Aires, Argentina, for their kind help. We acknowledge GSK’s Trust in Science program for its continued support.

This work was supported by Agencia Nacional de Promoción Científica y Tecnológica/Fondo para la Investigación Científica y Tecnológica Grant PICT 2018-3070 (to E.A.C.S.), and by Consejo Nacional de Investigaciones Científicas y Técnicas de Argentina (CONICET) Grant PIP 2015-0567 (to A.E.E.). E.A.C.S., A.E.E., and D.A.R. are career investigators at CONICET; C.M.O. is the recipient of a Ph.D. fellowship from CONICET, and D.F. was recipient of the “Peruilh Fellowship 2019” from the School of Medicine, University of Buenos Aires. The funding sources had no involvement in the study design, in the collection, analysis, and interpretation of data, in the writing of the manuscript, and in the decision to submit the paper for publication.

C.M.O.: investigation, performing experiments, data curation, formal analysis, software, validation, writing—original draft, review and editing of the manuscript. D.A.R.: conceptualization, patient recruitment and follow-up, discussion of data and results, review and editing of the manuscript. W.N.: performing experiments, data curation, formal analysis, review of the manuscript. D.F.: patient follow-up, data curation, formal analysis, review of the manuscript. A.E.E.: conceptualization, investigation, methodology, data and result discussion, funding acquisition, project administration, resources, writing, review, and editing of the manuscript. E.A.C.S.: conceptualization, investigation, designed and supervised the research, methodology, data analysis and results, project administration, funding acquisition, resources, writing—original draft, review and editing of the manuscript.

The online version of this article contains supplemental material.

Abbreviations used in this article:

     
  • AD

    active disease

  •  
  • cc-LCH

    presence of circulating LCH cells

  •  
  • DC

    dendritic cell

  •  
  • LCH

    Langerhans cell histiocytosis

  •  
  • NAD

    nonactive disease

  •  
  • non–cc-LCH

    nonpresence of circulating LCH cells

  •  
  • ROC

    receiver operating characteristic

  •  
  • s

    soluble

1.
Rodriguez-Galindo
C.
,
C. E.
Allen
.
2020
.
Langerhans cell histiocytosis.
Blood
135
:
1319
1331
.
2.
Heritier
S.
,
J. F.
Emile
,
Z.
Helias-Rodzewicz
,
J.
Donadieu
.
2019
.
Progress towards molecular-based management of childhood Langerhans cell histiocytosis.
Arch. Pediatr.
26
:
301
307
.
3.
Allen
C. E.
,
M.
Merad
,
K. L.
McClain
.
2018
.
Langerhans-cell histiocytosis.
N. Engl. J. Med.
379
:
856
868
.
4.
Emile
J. F.
,
O.
Abla
,
S.
Fraitag
,
A.
Horne
,
J.
Haroche
,
J.
Donadieu
,
L.
Requena-Caballero
,
M. B.
Jordan
,
O.
Abdel-Wahab
,
C. E.
Allen
, et al
Histiocyte Society
.
2016
.
Revised classification of histiocytoses and neoplasms of the macrophage-dendritic cell lineages.
Blood
127
:
2672
2681
.
5.
Krooks
J.
,
M.
Minkov
,
A. G.
Weatherall
.
2018
.
Langerhans cell histiocytosis in children: history, classification, pathobiology, clinical manifestations, and prognosis.
J. Am. Acad. Dermatol.
78
:
1035
1044
.
6.
Haupt
R.
,
M.
Minkov
,
I.
Astigarraga
,
E.
Schäfer
,
V.
Nanduri
,
R.
Jubran
,
R. M.
Egeler
,
G.
Janka
,
D.
Micic
,
C.
Rodriguez-Galindo
, et al
Euro Histio Network
.
2013
.
Langerhans cell histiocytosis (LCH): guidelines for diagnosis, clinical work-up, and treatment for patients till the age of 18 years.
Pediatr. Blood Cancer
60
:
175
184
.
7.
Harmon
C. M.
,
N.
Brown
.
2015
.
Langerhans cell histiocytosis: a clinicopathologic review and molecular pathogenetic update.
Arch. Pathol. Lab. Med.
139
:
1211
1214
.
8.
Halbritter
F.
,
M.
Farlik
,
R.
Schwentner
,
G.
Jug
,
N.
Fortelny
,
T.
Schnöller
,
H.
Pisa
,
L. C.
Schuster
,
A.
Reinprecht
,
T.
Czech
, et al
2019
.
Epigenomics and single-cell sequencing define a developmental hierarchy in Langerhans cell histiocytosis.
Cancer Discov.
9
:
1406
1421
.
9.
Allen
C. E.
,
L.
Li
,
T. L.
Peters
,
H. C.
Leung
,
A.
Yu
,
T. K.
Man
,
S.
Gurusiddappa
,
M. T.
Phillips
,
M. J.
Hicks
,
A.
Gaikwad
, et al
2010
.
Cell-specific gene expression in Langerhans cell histiocytosis lesions reveals a distinct profile compared with epidermal Langerhans cells.
J. Immunol.
184
:
4557
4567
.
10.
Badalian-Very
G.
,
J. A.
Vergilio
,
B. A.
Degar
,
L. E.
MacConaill
,
B.
Brandner
,
M. L.
Calicchio
,
F. C.
Kuo
,
A. H.
Ligon
,
K. E.
Stevenson
,
S. M.
Kehoe
, et al
2010
.
Recurrent BRAF mutations in Langerhans cell histiocytosis.
Blood
116
:
1919
1923
.
11.
Chakraborty
R.
,
O. A.
Hampton
,
X.
Shen
,
S. J.
Simko
,
A.
Shih
,
H.
Abhyankar
,
K. P.
Lim
,
K. R.
Covington
,
L.
Trevino
,
N.
Dewal
, et al
2014
.
Mutually exclusive recurrent somatic mutations in MAP2K1 and BRAF support a central role for ERK activation in LCH pathogenesis.
Blood
124
:
3007
3015
.
12.
Berres
M. L.
,
K. P.
Lim
,
T.
Peters
,
J.
Price
,
H.
Takizawa
,
H.
Salmon
,
J.
Idoyaga
,
A.
Ruzo
,
P. J.
Lupo
,
M. J.
Hicks
, et al
2014
.
BRAF-V600E expression in precursor versus differentiated dendritic cells defines clinically distinct LCH risk groups. [Published erratum appears in 2015 J. Exp. Med. 212: 281.]
J. Exp. Med.
211
:
669
683
.
13.
Suh
J. K.
,
S.
Kang
,
H.
Kim
,
H. J.
Im
,
K. N.
Koh
.
2021
.
Recent advances in the understanding of the molecular pathogenesis and targeted therapy options in Langerhans cell histiocytosis.
Blood Res.
56
(
S1
):
S65
S69
.
14.
Martínez-Cingolani
C.
,
M.
Grandclaudon
,
M.
Jeanmougin
,
M.
Jouve
,
R.
Zollinger
,
V.
Soumelis
.
2014
.
Human blood BDCA-1 dendritic cells differentiate into Langerhans-like cells with thymic stromal lymphopoietin and TGF-β.
Blood
124
:
2411
2420
.
15.
Geissmann
F.
,
C.
Prost
,
J. P.
Monnet
,
M.
Dy
,
N.
Brousse
,
O.
Hermine
.
1998
.
Transforming growth factor β1, in the presence of granulocyte/macrophage colony-stimulating factor and interleukin 4, induces differentiation of human peripheral blood monocytes into dendritic Langerhans cells.
J. Exp. Med.
187
:
961
966
.
16.
Carrera Silva
E. A.
,
W.
Nowak
,
L.
Tessone
,
C. M.
Olexen
,
J. M.
Ortiz Wilczyñski
,
I. G.
Estecho
,
G.
Elena
,
A. E.
Errasti
,
D. A.
Rosso
.
2017
.
CD207+CD1a+ cells circulate in pediatric patients with active Langerhans cell histiocytosis.
Blood
130
:
1898
1902
.
17.
Picarda
G.
,
C.
Chéneau
,
J. M.
Humbert
,
G.
Bériou
,
P.
Pilet
,
J.
Martin
,
F.
Duteille
,
P.
Perrot
,
F.
Bellier-Waast
,
M.
Heslan
, et al
2016
.
Functional langerinhigh-expressing Langerhans-like cells can arise from CD14highCD16 human blood monocytes in serum-free condition.
J. Immunol.
196
:
3716
3728
.
18.
Quispel
W. T.
,
J. A.
Stegehuis-Kamp
,
L.
Blijleven
,
S. J.
Santos
,
M.
Lourda
,
C.
van den Bos
,
A. G.
van Halteren
,
R. M.
Egeler
.
2015
.
The presence of CXCR4+ CD1a+ cells at onset of Langerhans cell histiocytosis is associated with a less favorable outcome.
OncoImmunology
5
:
e1084463
.
19.
Lim
K. P. H.
,
P.
Milne
,
M.
Poidinger
,
K.
Duan
,
H.
Lin
,
N.
McGovern
,
H.
Abhyankar
,
D.
Zinn
,
T. M.
Burke
,
O. S.
Eckstein
, et al
2020
.
Circulating CD1c+ myeloid dendritic cells are potential precursors to LCH lesion CD1a+CD207+ cells.
Blood Adv.
4
:
87
99
.
20.
Milne
P.
,
V.
Bigley
,
C. M.
Bacon
,
A.
Néel
,
N.
McGovern
,
S.
Bomken
,
M.
Haniffa
,
E. L.
Diamond
,
B. H.
Durham
,
J.
Visser
, et al
2017
.
Hematopoietic origin of Langerhans cell histiocytosis and Erdheim-Chester disease in adults.
Blood
130
:
167
175
.
21.
Gadner
H.
,
M.
Minkov
,
N.
Grois
,
U.
Pötschger
,
E.
Thiem
,
M.
Aricò
,
I.
Astigarraga
,
J.
Braier
,
J.
Donadieu
,
J. I.
Henter
, et al
Histiocyte Society
.
2013
.
Therapy prolongation improves outcome in multisystem Langerhans cell histiocytosis.
Blood
121
:
5006
5014
.
22.
Van Gassen
S.
,
B.
Callebaut
,
M. J.
Van Helden
,
B. N.
Lambrecht
,
P.
Demeester
,
T.
Dhaene
,
Y.
Saeys
.
2015
.
FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data.
Cytometry A
87
:
636
645
.
23.
R Core Team
.
2021
.
R: A language and environment for statistical computing.
R Foundation for Statistical Computing
,
Vienna, Austria
.
24.
Robin
X.
,
N.
Turck
,
A.
Hainard
,
N.
Tiberti
,
F.
Lisacek
,
J. C.
Sanchez
,
M.
Müller
.
2011
.
pROC: an open-source package for R and S+ to analyze and compare ROC curves.
BMC Bioinformatics
12
:
77
.
25.
Donadieu
J.
,
C.
Piguet
,
F.
Bernard
,
M.
Barkaoui
,
M.
Ouache
,
Y.
Bertrand
,
H.
Ibrahim
,
J. F.
Emile
,
O.
Hermine
,
A.
Tazi
, et al
2004
.
A new clinical score for disease activity in Langerhans cell histiocytosis.
Pediatr. Blood Cancer
43
:
770
776
.
26.
Abla
O.
2017
.
Finding active LCH cells in the blood.
Blood
130
:
1874
1875
.
27.
Tillotson
C. V.
,
F.
Anjum
,
B. C.
Patel
.
2022
.
Langerhans cell histiocytosis.
StatPearls Publishing
,
Treasure Island, FL
.
28.
Rosso
D. A.
,
A.
Roy
,
M.
Zelazko
,
J. L.
Braier
.
2002
.
Prognostic value of soluble interleukin 2 receptor levels in Langerhans cell histiocytosis.
Br. J. Haematol.
117
:
54
58
.
29.
Murakami
I.
,
M.
Matsushita
,
T.
Iwasaki
,
S.
Kuwamoto
,
M.
Kato
,
K.
Nagata
,
Y.
Horie
,
K.
Hayashi
,
T.
Imamura
,
A.
Morimoto
, et al
2015
.
Interleukin-1 loop model for pathogenesis of Langerhans cell histiocytosis.
Cell Commun. Signal.
13
:
13
.
30.
Morimoto
A.
,
Y.
Oh
,
S.
Nakamura
,
Y.
Shioda
,
T.
Hayase
,
T.
Imamura
,
K.
Kudo
,
S.
Imashuku
;
Japan Langerhans Cell Histiocytosis Study Group
.
2017
.
Inflammatory serum cytokines and chemokines increase associated with the disease extent in pediatric Langerhans cell histiocytosis.
Cytokine
97
:
73
79
.
31.
Rosso
D. A.
,
M. F.
Ripoli
,
A.
Roy
,
R. A.
Diez
,
M. E.
Zelazko
,
J. L.
Braier
.
2003
.
Serum levels of interleukin-1 receptor antagonist and tumor necrosis factor-alpha are elevated in children with Langerhans cell histiocytosis.
J. Pediatr. Hematol. Oncol.
25
:
480
483
.
32.
Nakayama
T.
,
K.
Hieshima
,
D.
Izawa
,
Y.
Tatsumi
,
A.
Kanamaru
,
O.
Yoshie
.
2003
.
Cutting edge: profile of chemokine receptor expression on human plasma cells accounts for their efficient recruitment to target tissues.
J. Immunol.
170
:
1136
1140
.
33.
Kanbe
K.
,
J.
Chiba
,
Y.
Inoue
,
M.
Taguchi
,
A.
Yabuki
.
2016
.
SDF-1 and CXCR4 in synovium are associated with disease activity and bone and joint destruction in patients with rheumatoid arthritis treated with golimumab.
Mod. Rheumatol.
26
:
46
50
.
34.
Rosso
D. A.
,
J.
Karis
,
J. L.
Braier
,
J. I.
Henter
,
B.
Fadeel
.
2006
.
Elevated serum levels of the decoy receptor osteoprotegerin in children with Langerhans cell histiocytosis.
Pediatr. Res.
59
:
281
286
.
35.
Merad
M.
,
F.
Ginhoux
,
M.
Collin
.
2008
.
Origin, homeostasis and function of Langerhans cells and other langerin-expressing dendritic cells.
Nat. Rev. Immunol.
8
:
935
947
.
36.
Collin
M.
,
P.
Milne
.
2016
.
Langerhans cell origin and regulation.
Curr. Opin. Hematol.
23
:
28
35
.
37.
Xiao
Y.
,
S.
Zijl
,
L.
Wang
,
D. C.
de Groot
,
M. J.
van Tol
,
A. C.
Lankester
,
J.
Borst
.
2015
.
Identification of the common origins of osteoclasts, macrophages, and dendritic cells in human hematopoiesis.
Stem Cell Reports
4
:
984
994
.
38.
de Vries
T. J.
,
I.
El Bakkali
,
T.
Kamradt
,
G.
Schett
,
I. D. C.
Jansen
,
P.
D’Amelio
.
2019
.
What are the peripheral blood determinants for increased osteoclast formation in the various inflammatory diseases associated with bone loss?
Front. Immunol.
10
:
505
.
39.
Maitra
R.
,
A.
Follenzi
,
A.
Yaghoobian
,
C.
Montagna
,
S.
Merlin
,
E. S.
Cannizzo
,
J. A.
Hardin
,
N.
Cobelli
,
E. R.
Stanley
,
L.
Santambrogio
.
2010
.
Dendritic cell-mediated in vivo bone resorption.
J. Immunol.
185
:
1485
1491
.
40.
Allard-Chamard
H.
,
N.
Carrier
,
P.
Dufort
,
M.
Durand
,
A. J.
de Brum-Fernandes
,
G.
Boire
,
S. V.
Komarova
,
S. J.
Dixon
,
R. E.
Harrison
,
M. F.
Manolson
,
S.
Roux
.
2020
.
Osteoclasts and their circulating precursors in rheumatoid arthritis: relationships with disease activity and bone erosions.
Bone Rep.
12
:
100282
.
41.
Xue
J.
,
L.
Xu
,
H.
Zhu
,
M.
Bai
,
X.
Li
,
Z.
Zhao
,
H.
Zhong
,
G.
Cheng
,
X.
Li
,
F.
Hu
,
Y.
Su
.
2020
.
CD14+CD16 monocytes are the main precursors of osteoclasts in rheumatoid arthritis via expressing Tyro3TK.
Arthritis Res. Ther.
22
:
221
.
42.
Haroche
J.
,
F.
Cohen-Aubart
,
B. J.
Rollins
,
J.
Donadieu
,
F.
Charlotte
,
A.
Idbaih
,
A.
Vaglio
,
O.
Abdel-Wahab
,
J. F.
Emile
,
Z.
Amoura
.
2017
.
Histiocytoses: emerging neoplasia behind inflammation.
Lancet Oncol.
18
:
e113
e125
.
43.
Jouenne
F.
,
A.
Benattia
,
A.
Tazi
.
2021
.
Mitogen-activating protein kinase pathway alterations in Langerhans cell histiocytosis.
Curr. Opin. Oncol.
33
:
101
109
.

The authors have no financial conflicts of interest.

Supplementary data