Demographics and disease characteristics
A total of 254 adult patients with LCH were included, including 168 males (66.1%) and 86 females (33.9%), with a male to female ratio of 1.95:1 (Table 1). The median age at diagnosis was 34 years (range, 18-69 years). Eight patients had mixed LCH/ECD. Ten patients had other tumors diagnosed before or concurrently with their LCH, including thyroid cancer in six, pituitary microadenoma in two, and lung adenocarcinoma and chronic myelomonocytic leukemia (CMML) in one patient each.
Among the entire cohort, 239 patients (94.1%) had symptomatic disease at diagnosis (Figure 1A). The most common manifestations were bone pain (33.1%, n=84), followed by diabetes insipidus (27.6%, n=70), rash (10.6%, n=27), nonspecific soft tissue masses (9.1%, n=23) and pneumothorax (8.7%, n=22). Patients were classified as having MS disease (64.2%, n=163), unifocal LCH (20.1%, n=51), SS-M disease (13.4%, n=34), or SS-P (2.4%, n=6) (Figure 1B). All the mixed LCH/ECD patients were MS LCH patients. Unifocal sites were bone (76.5%, n=39), lymph nodes (n=3), skin (n=3), pituitary gland (n=3), liver (n=1), eyelid (n=1) and digestive tract (n=1) (Figure 1C). All SS-M patients had bone lesions. The three most commonly affected organs in the 163 MS LCH patients were bone (73.6%, n=120), lung (52.8%, n=86) and pituitary gland (52.1%, n=85) (Figure 1D). MS patients were further classified as MS with (MS-RO+, 30.1%, n=49) or without (MS-RO-, 69.9%, n=114) risk organ involvement. Diagnostic imaging included full-body CT (79.9%, n=203), FDG-PET (64.6%, n=164), and brain MRI (32.7%, n=83), with all patients undergoing at least one full-body radiological exam.
Molecular status among adults with LCH
Overall, NGS of lesional tissue was performed in 237 patients. The remaining 17 patients were found to have BRAFV600E mutation by qPCR ((Figure 2, Table S2). Alterations in the MAPK/PI3K pathway were observed in 77.6%(n=197)of patients, most commonly BRAFV600E (30.7%, n=78), followed by BRAFindel (18.1%, n=46) and MAP2K1 mutation (12.6%, n=32). The types of BRAFindel included BRAFN486_P490del (n=38), BRAFL485_P490delinsF(n=2), BRAFN486_T491delinsK (n=3), BRAFV487_T491del (n=2) and BRAF487_492delinsA (n=1). For MAP2K1 mutation, the most common type was MAP2K1E102_I103del (n=18). Other BRAF mutations detected included BRAFV600D, BRAFG466E, BRAFE501K, BRAFT599I and BRAFV487Tfs*10 in one patient each (0.4%), and fusion in two patients (0.8%): BICD2--BRAF and PICALM--BRAF. One patient had both BRAFindel and BRAFA366P, and another had both BRAFV600L and BRAFY472C. Other mutations detected were KRAS (6.7%, n=17), PIK3CA (3.9%, n=10), MAPK1 (2.8%, n=7), NRAS (2.4%, n=6), ERBB3 (2.4%, n=6), ALK (2.0%, n=5), ERBB4 (2.0%, n=5), PIK3CD (1.6%, n=4), CSF1R (1.6%, n=4), CSF3R (1.6%, n=4). Additionally, NGS detected myeloid-associated mutations including IDH1 (3.9%, n=10), TET2 (3.5%, n=9) and RUNX1 (2.8%, n=7) (Figure S1). No mutations were detected by NGS in 37 patients; biopsy sites in these patients were bone (n=21), skin (n=5), lung (n=4), liver (n=3), pituitary (n=2), lymph node (n=1), and digestive tract (n=1).
Among the six patients with LCH and thyroid cancer, three carried BRAFindel, one carried a BRAFV600E mutation, and two had no MAPK/PI3K pathway mutations. One of the two patients with LCH and pituitary microadenoma carried a BRAFV600E mutation, while the other had no detected MAPK/PI3K pathway mutations. The one patient with LCH and CMML carried a KRAS mutation, and the one patient with LCH and lung adenocarcinoma carried both KRAS and NRAS mutations. Among the eight patients with mixed LCH/ECD,seven carried BRAFV600E and one carried a PIK3CA mutation.
Clinical presentation in relation to molecular status
The distribution of MAPK/PI3K pathway alterations based on age at diagnosis and disease classification are shown in Figure 3. Age did not impact the proportion with MAPK/PI3K mutations. In single system disease, BRAFV600E, BRAFindel and MAP2K1 mutation occurred in 34.1%, 6.6% and 11.0% respectively. While in MS LCH, frequencies were 28.8%, 24.5% and 13.5%. BRAF deletion was significantly more common in MS versus single system disease (24.5% vs 6.6%, p<0.001). No significant differences occurred in BRAFV600E, MAP2K1, IDH1, TET2, or RUNX1 mutations rates between MS and single system groups.
In patients with single-system disease, compared to those without bone involvement, BRAFV600E mutation was most frequent among patients with bone lesions (p=0.005). Half of patients with SS-P LCH had BRAFindel (Figure 4A). There were two patients with unifocal LCH who carried BRAFindel mutation. One of them was unifocal bone involvement LCH, the other was unifocal pituitary involvement.
In MS LCH patients, BRAFV600E mutation was most common among those with bone involvement (p<0.001) compared to patients without bone involvement. BRAFindel predominated in liver, thyroid gland and pituitary (Figure 4A). The difference was statistically significant in the pituitary (p=0.037), liver (p=0.004) and thyroid (p=0.001) groups compared to patients without involvement of these organs (Figure 4B).
Of the patients with IDH1, TET2, and RUNX1 mutations, 70.0%, 66.7% and 71.4% had MS LCH, respectively. However, IDH1, TET2 or RUNX1 mutations did not correlate with specific disease sites, likely due to the small sample size. No correlation was found between blood counts and these three myeloid-associated mutations. Additionally, no correlation was observed between and BRAFV600E, BRAFindel, or MAP2K1 mutations and IDH1, TET2 or RUNX1 mutations.
First-line therapy
Initial treatment is outlined in Figure S2. Patients with unifocal LCH were mainly treated with local therapy, including surgery in 30 patients (58.8%) and radiotherapy in 8 (15.7%). Two patients with unifocal LCH (3.9%) received cytarabine monotherapy therapy. Eleven patients with unifocal LCH (21.6%) were untreated. All the six patients with SS-P LCH were asked to strictly quit smoking. Patients with SS-M and MS LCH were mainly treated with systemic therapy, including cytarabine monotherapy in 91 (46.2%), methotrexate-cytarabine regimen in 27 (13.7%), vinblastine-prednisone (VP) based regimen in 22 (11.2%), other chemotherapy regimens in 15 (7.6%), BRAF inhibitors in 9 (4.6%), and MEK inhibitors in 4 (2.0%). Thirteen (6.6%) patients with MS LCH only received surgery, and seven (3.6%) patients with MS LCH only received radiotherapy. One patient with LCH/ECD was treated with IFN-α.Of the eight MS patients who were untreated, seven had not started treatment at last follow-up, and one terminal stage patient lost treatment opportunity.
Prognosis in relation to molecular status
Follow-up data were available for 252 patients, excluding 2 lost to follow-up. The median follow-up duration was 32.9 months (range, 0.4-259.8 months). No new second malignancies were found in these patients during the follow-up period. Eight patients died, including five from disease progression, two from post-chemotherapy infection, and one from accident. All deaths occurred in those with MS LCH. The estimated 3-year OS for the entire cohort was 96.3% (Figure 5A). By subgroup, the estimated 3-year OS was 94.6% for MS LCH and 100% for single system disease (Figure 5B). BRAFindel was associated with inferior OS for all patients (the estimated 3-year OS 90.0% vs 98.0%, p=0.038) (Figure 5C). The mutation status of BRAFV600E and MAP2K1, LCH classification, and treatment did not impact OS.
Seventy patients had disease progression, 56 (80%) with MS LCH, eight with SS-M LCH and six with unifocal LCH. The estimated 3-year PFS was 71.9% for the entire cohort (Figure 5A). By subgroup, the estimated 3-year PFS was 66.5% for MS LCH, 66.1% for SS-M, 83.7% for unifocal disease and 100% for SS-P (Figure 5D). MS LCH had worse PFS than single system disease (3-year PFS 66.5% vs 83.4%, p=0.046). BRAFindel was associated with inferior PFS (3-year PFS 46.7% vs 77.6%, p<0.001) (Figure 5E). Conversely, BRAFV600E mutation correlated with improved PFS (3-year PFS 83.2% vs 67.3%, p=0.032) (Figure S3). Multivariate analysis confirmed BRAFindel as an independent adverse prognostic factor for PFS (p<0.001, HR=2.706, 95%CI 1.555-4.709). MAP2K1 mutation status or treatment did not impact PFS.
Within disease subgroups, BRAF deletion was associated with inferior PFS in MS LCH (3-year PFS 44.6% vs 74.2%, p<0.001) (Table 2) (Figure 5F). BRAFV600E mutation correlated with improved PFS (3-year PFS 76.7% vs 62.5%, p=0.048) (Figure S3). Thyroid involvement also predicted worse PFS (3-year PFS 37.3% vs 73.9%, p=0.002). Other factors with p<0.1 were bone (p=0.082) and lymph nodes involvement (p=0.079). Multivariate analysis confirmed BRAFindel as an independent adverse prognostic factor for PFS (p=0.012,HR=2.205, 95%CI 1.188-4.095). No factors impacted PFS in unifocal or SS-M LCH.