High prevalence of myeloid neoplasms in adults with non–Langerhans cell histiocytosis

Erdheim-Chester disease (ECD) is a form of histiocytosis characterized by tissue infiltration with foamy histiocytes that are CD68⁺, CD163⁺, CD1a⁻, and Langerin (CD207)⁻. Prior to 2012, there was a long-standing debate as to whether ECD represented a clonal hematopoietic disorder vs an inflammatory disease related to aberrant immune activation. However, since 2010, a series of genomic studies have uncovered BRAFV600E mutations in 55% to 70% of ECD as well as Langerhans cell histiocytosis (LCH) patients, providing evidence that these diseases represent clonal disorders driven by activated MAPK signaling.^1,2  Subsequently, activating mutations in MAP2K1,^3-6 ARAF,^6,7  and fusions in kinases including BRAF^6,8  were found in the majority of BRAFV600-wild-type ECD and LCH patients. PI3KCA⁹  and N/KRAS mutations are more frequent in ECD than in LCH. Despite the distinct clinical and histologic presentations of LCH and ECD, the previous studies identify a similar constellation of genomic alterations across both disorders. Moreover, nearly 20% of patients with ECD have a diagnosis of both ECD and LCH simultaneously (so-called mixed histiocytosis [MH]) where both lesions may contain the BRAFV600E mutation.¹⁰  For these reasons, ECD and LCH have been grouped together in the same class of conditions in a recent revision of histiocytosis classifications.¹¹ 

Although there have been great molecular advances in ECD and LCH over the last 7 years, our knowledge of the natural history of ECD and MH and their associated clinical and prognostic features remains incomplete. Nevertheless, despite great recent progress in the understanding of histiocytosis, the cellular origins of LCH, ECD, and MH are not completely understood.

In hopes of clarifying the clinical and molecular features of ECD and MH and facilitating translational research in these disorders, we assembled ECD patient data across 2 institutions in Europe and the United States with a specific research focus on ECD. Using these data, we have uncovered an unexpected high frequency of concomitant clinically diagnosed myeloid neoplasms in patients diagnosed with ECD and MH. Here we describe the clinical, pathological, and molecular features of the hematological disorders associated with ECD. Diagnosis of a concurrent myeloid neoplasm and histiocytosis had important therapeutic consequences for patients. These findings have clear clinical relevance in addition to potentially enlightening our understanding of the etiology of ECD and its disease classification.

We performed an international 2-center retrospective study of patients with biopsy-proven ECD who were referred at least once to the Internal Medicine Department of Pitié-Salpêtrière Hospital (Paris, France) or Memorial Sloan Kettering Cancer Center (MSKCC; New York, NY) between November 1981 and November 2016. A total of 189 cases of ECD were reviewed (39 from the MSKCC and 150 from the Pitié-Salpêtrière Hospital; 12 of which were also studied by whole exome sequencing previously⁶ ). Written informed consent was obtained from patients according to Helsinki convention, and this study received approval from the Ethics Committee Ile-de-France III and the Institutional Review Board at MSKCC.

ECD was diagnosed according to published criteria: (1) tissue biopsy demonstrating typical pathological signs of ECD (infiltration with foamy histiocytes and CD68⁺CD1a⁻ on immunohistochemical staining); (2) symmetric skeletal uptake on ^99mTc bone scintigraphy or 18-fluorodeoxyglucose positron emission tomographic (¹⁸F-FDG PET) or magnetic resonance imaging scan with the involvement of at least 1 other organ typically affected in ECD (xanthelasma, perinephric infiltration, “coated aorta,” pericardial infiltration, right atrial pseudotumor, or brain or dural infiltration). Patients with MH were included as well. Bone marrow (BM) aspirations and biopsies were performed according to standard of care when a complete blood count abnormality (anemia, thrombocytopenia, polycythemia, thrombocytosis, or monocytosis) was noted and not explained by nonmalignant causes such as iron, vitamin B12, or folic acid deficiencies or inflammatory syndrome. Hematological malignancies (myeloid neoplasms, lymphoma, and myeloma) were diagnosed according to the 2016 revision to the World Health Organization (WHO) classification of hematological malignancies.^12,13  For each patient, clinical parameters (age at diagnosis, sex, and main organ involvement), laboratory parameters (blood count, electrolytes, liver function tests, and C-reactive protein), and outcome (treatments and death) were obtained from medical records.

For patients from Pitié-Salpêtrière Hospital, BRAFV600E mutational status was evaluated as previously described.² JAK2V617F mutational status was evaluated in BM aspirate, peripheral blood (PB), or tissue biopsy using the TaqMan method with an Applied 7500 automaton. For patients with available material, targeted deep sequencing of the exonic regions of 24 genes (ASXL1, ATRX, BCOR, BCORL1, CBL, CEBPA, CSF3R, DNMT3A, ETV6, EZH2, FLT3, GATA2, JAK2, KDM6A, KIT, NRAS, RUNX1, SETBP1, SF3B1, STAG2, TET2, TP53, WT1, and ZRSR2) was performed on flow-sorted CD14⁺ cells (for patients with chronic myelomonocytic leukemia [CMML]) or CD14⁻CD15⁺CD16⁺ polymorphonuclear cells (for patients with classic myeloproliferative neoplasms [MPNs]) as well as T cells (CD3⁺CD19⁻CD14⁻CD34⁻) for germ-line control. This specific gene panel was used as it includes the most frequently mutated genes in myeloid neoplasms. Targeted regions were polymerase chain reaction amplified using an AmpliSeq (Life Technologies), and polymerase chain reaction products sequenced with a MiSeq (Illumina, San Diego, CA) instrument with a mean depth of 1000X (range 590-1800). We used a variant allele frequency minimal threshold of 10% to detect mutation.

For patients from MSKCC, genomic analysis of histiocytosis tissue biopsy and PB mononuclear cells was performed using the MSKCC IMPACT (Integrated Mutation Profiling of Actionable Cancer Targets) assay as previously described.¹⁴ 

Continuous variables were expressed as median and range, and categorical variables as numbers and percentages. Differences between groups of patients were tested with the Mann-Whitney U test for continuous data, and by Fischer exact test or the χ² test for categorical data. Survival analyses were performed with Kaplan-Meier curves and log-rank test. We used SAS version 9.0 (SAS Institute) and GraphPad Prism 5 for analyses.

We reviewed 189 cases of ECD, including MH. Associated hematological disorders (excluding another histiocytosis) were observed in 23 patients (12.2%). Apart from patients with lymphoproliferative or autoimmune disorders (2 patients had lymphoma, 2 had immune thrombocytopenic purpura, 1 had myeloma, and 1 developed an acute lymphoblastic leukemia [ALL] in the context of a primary myelofibrosis), hematological diseases associated with ECD were myeloid neoplasms (10.1%) (Figure 1; supplemental Table 1 and supplemental Figure 1, available on the Blood Web site). This frequency was higher in the United States (15.3%) than the French (8.6%) cohort (P = .22). Among these 19 patients, 8 had CMML; 4, ET; 2, MDSs; 2, primary myelofibrosis; 2, AML (1 secondary to MDS and 1 to PV); and 1, PV. One patient also developed an ALL in the course of his MPN. Seven patients were diagnosed with myeloid neoplasm before the ECD diagnosis (median 4 years between the 2 diagnoses, range 1-22 years), 6 were diagnosed simultaneously, and 6 were diagnosed with myeloid neoplasm after a diagnosis of histiocytosis (median 1 year, range 1-4 years).

ECD patients with a myeloid neoplasm were also more likely to have a diagnosis of an overlap histiocytosis (ECD associated with LCH [n = 6] and Rosai Dorfman disease [n = 2]) than those patients with ECD and no concomitant myeloid neoplasm (P = .02; Table 1). In addition, patients with ECD plus a concomitant myeloid neoplasm were significantly older at ECD diagnosis (68 vs 56.5 years; P = .0005) and had a lower survival (82 vs 364 months; P = .001) than ECD patients without a myeloid neoplasm (supplemental Figure 2). Deaths were mainly because of cardiac insufficiency or infections, but 2 patients died of hematological disease (patient #1 died of ALL, and patient #11 died of an AML secondary to MDS).

Given the unexpected high frequency of myeloid neoplasms associated with histiocytosis, we next sought to examine the molecular features of histiocytosis lesions and the coexisting myeloid neoplasms in these individuals. Many of these patients harbored kinase alterations characteristic of both ECD and myeloid neoplasms. Among the 19 patients with concomitant myeloid neoplasm and ECD, 12 (63.2%) harbored the BRAFV600E mutation in ECD tissue biopsy material, and 7 (36.8%) were positive for JAK2V617F in PB and/or BM. Interestingly, 4 patients (23.5%) had both BRAFV600E and JAK2V617F mutations simultaneously (Figure 2). Similarly, 1 patient with ET had a CALR mutation as well as the BRAFV600E mutation, and another had a MAP2K1 mutation in the histiocytic disease associated with a JAK2V167F mutated ET. Unlike BRAFV600E or MAP2K1 mutations, which were detected only in histiocytosis lesions, mutations in NRAS could be found in both histiocytosis and myeloid neoplasms, as exemplified by a patient who had the same NRAS mutation in ECD lesions from perirenal tissue as well as BM and PB following a diagnosis of CMML. In addition to harboring mutations in JAK2 and CALR, patients with myeloid neoplasm-associated histiocytosis also carried additional mutations in transcriptional regulatory genes common in myeloid neoplasms but rare in ECD,⁶  such as mutations in TET2, ASXL1, IDH2, U2AF1, and TP53 (Figure 1).

Although BRAF inhibition has resulted in remarkable clinical responses for patients with BRAFV600E-mutant histiocytosis,^15-18  there is a well-described risk of paradoxical activation of cytokine signaling in cells bearing kinase mutations other than BRAFV600E upon exposure to RAF inhibitors.^19-22  BRAF inhibitors (vemurafenib [n = 6] or dabrafenib [n = 1]) were given to 7/19 patients with BRAFV600E-mutant ECD and coexisting myeloid neoplasm here. In 3 cases, vemurafenib treatment resulted in an increase in blood counts, which led to treatment discontinuation. For example, patient #14 was treated with vemurafenib for BRAFV600E-mutant ECD with a partial metabolic response at 2.3 months and substantial reduction in BRAFV600E mutant allele burden as measured by analysis of urinary cell-free DNA²³  (Figure 2A-B). However, vemurafenib was discontinued at this time point because of a significant increase in monocyte count in conjunction with development of abdominal pain because of a new splenic infarct. Although PET scan at this point revealed resolution of FDG-avid femur lesions characteristic of ECD, there was a paradoxical increased in FDG avidity in all other major bones of the body (Figure 2A-C). This constellation of findings prompted a BM aspirate and biopsy that led to a diagnosis of JAK2V617F/IDH2R140Q-mutant MDS/MPN overlap disorder (Figure 2D). The patient’s monocytosis returned to normal levels following vemurafenib discontinuation further suggesting that paradoxical stimulation of JAK2-mutant cells in response to RAF inhibition was the basis for the patient’s adverse response to vemurafenib (Figure 2).

In contrast to the previous cases where distinct activating kinase mutations complicated use of a single kinase inhibitor, in other instances where the histiocytosis and myeloid neoplasm shared the same kinase mutation, use of targeted therapeutics resulted in beneficial response across both conditions. For example, patient #13 had NRAS-mutated ECD associated with a CMML-1 with the same NRAS mutation. MEK inhibition with cobimetinib in this patient led to a complete metabolic response by PET scan after 2 months in addition to improving monocyte and platelet counts (Figure 2E-G).

This study identifies a high prevalence (10.1%) of myeloid neoplasms among adults with non-LCH. This frequency of myeloid neoplasms is much higher than that encountered general population which ranges between 0.7 and 17.1 per 100 000 people in Europe, depending on the subtype.²⁴  Although prior case reports and case series^25-29  have noted additional hematological malignancies in patients with histiocytosis of the L group,¹¹  the high frequency of myeloid neoplasms in a large population of adult histiocytosis patients has not previously been recognized, likely because of the relative rarity of ECD. Nonetheless, knowledge of the presence of an associated myeloid neoplasm in ECD patients has important implications for clinical management of adult histiocytosis patients as well as the classification and biological understanding of these disorders. The difference in prevalence of myeloid neoplasms in histiocytosis patients between MSKCC (15.3%) and Pitié-Salpêtrière Hospital (8.6%) could be attributable to an institution bias, because Pitié-Salpêtrière Hospital is not a referral center for neoplastic diseases. Although this might result in a skewed estimation of the prevalence of myeloid neoplasms among histiocytosis patients, the prevalence at either center remains high.

Histiocytosis patients with an associated myeloid neoplasm seemed to be significantly older than those without a myeloid neoplasm. One hypothesis for this observation could be the well-described association between aging and clonal hematopoiesis with an increased frequency of myeloid neoplasms because of acquisition of somatic mutations in genes commonly mutated in myeloid neoplasms.^30-32  Moreover, the lower survival rate observed in histiocytosis patients with coexisting myeloid neoplasms might be attributable to the older age of those patients and not to a more severe disease.

From a clinical standpoint, the data presented here suggest that patients with ECD should be carefully evaluated at diagnosis for a coexisting myeloid neoplasm. We propose that any histiocytosis patient with a complete blood count abnormality that cannot be explained by a nonmalignant cause should undergo a BM aspiration and biopsy with histomorphologic, flow cytometric, and genetic analysis according to standard of care for myeloid malignancies. Evaluation of a concomitant myeloid neoplasm in a patient with non-LCH will be particularly important before initiation of targeted therapies for refractory ECD. As illustrated here, the myeloid neoplasm may be diagnosed only after initiation of therapy for ECD, and knowledge of both neoplasms would have influenced therapeutic decision making. Because of the paradoxical activation of RAS signaling mediated by RAF inhibitors, use of RAF inhibitors may uncover or enhance growth of a malignancy driven by a mutation other than BRAFV600E^19-22  (including mutations in NRAS, KRAS, or JAK2) as illustrated here.

In the recently revised 2016 WHO classification, ECD and LCH are classified with lymphoid neoplasms.¹²  This is largely based on several case reports describing individual patients with lymphoid neoplasms and a clonally related secondary malignant histiocytosis.^33-37  However, the high frequency of myeloid neoplasms in patients with ECD or MH suggests a nonfortuitous biological association of both conditions. Although it is well established that myeloid neoplasms such as classic MPNs, MDS, and AML originate from hematopoietic stem cells,^38-41  the cell(s) of origin of LCH and non-LCH conditions are far less clearly delineated. A common nonconstitutional trisomy 21 was recently reported in an adult patient with both LCH and mixed phenotype T/myeloid acute leukemia,⁴²  suggesting a common origin of both tumors. Recent data suggest that the BRAFV600E mutation is detectable in CD34⁺ hematopoietic progenitors in at least a portion of pediatric LCH patients.⁴³  Moreover, blood monocytes harboring the same mutations as pathological histiocytes have been reported in LCH⁴³  and non-LCH.⁴⁴  At the same time, the discovery of patients with BRAFV600E/JAK2 wild-type ECD lesions with a concomitant BRAF wild-type/JAK2V617F mutant clonal hematopoietic disorder suggests that the 2 mutations must arise at distinct points of hematopoietic development in many patients. Further efforts to functionally characterize the precise cellular origin of both conditions are therefore warranted.

B.H.D. is supported by the American Society of Hematology Senior Research Training Award for Fellows and the New York State Council on Graduate Medical Education Empire Clinical Research Investigator Program Fellowship. E.L.D. and O.A.-W. are supported by grants from the Erdheim-Chester Disease Global Alliance and the Histiocytosis Association. O.A.-W. is supported by grants from the Edward P. Evans Foundation; the Department of Defense Bone Marrow Failure Research Program (BM150092 and W81XWH-12-1-0041); the National Institutes of Health, National, Heart, Lung, and Blood Institute (R01 HL128239) and a National Cancer Institute K08 Clinical Investigator Award (1K08CA160647-01); the Josie Robertson Investigator Program; a Damon Runyon Clinical Investigator Award; an award from the Starr Foundation (I8-A8-075); the Leukemia and Lymphoma Society; and the Pershing Square Sohn Cancer Research Alliance. J.-F.E. is supported by grants from Association pour la Recherche et l'Enseignement en Pathologie.

Contribution: M.P., E.L.D., J.-F.E., D.R.-W., B.H.D., N.O., A.D., O.A.B., N.D., and Z.H.-R. performed the experiments and analyzed data; M.P., F.C.-A., E.L.D., N.G., C.S., G.A.U., R.R., J.-E.K., T.S., F.C., C.B., B.H., Z.A., and J.-F.E. collected patient material and clinical data; and M.P., E.L.D., J.-F.E., O.A.-W., and J.H. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Omar Abdel-Wahab, 1275 York Ave, New York, NY 10065; e-mail: abdelwao@mskcc.org; and Julien Haroche, Service de Médecine Interne 2, Institut E3M, Hôpital de la Pitié-Salpêtrière, 47-83 Boulevard de l'Hôpital, 75013 Paris, France; e-mail: julien.haroche@aphp.fr.