ALK-positive histiocytosis: a new clinicopathologic spectrum highlighting neurologic involvement and responses to ALK inhibition

Histiocytic disorders are a group of rare diseases characterized by the accumulation of macrophage-, dendritic cell–, or monocyte-differentiated cells in various tissues and organs.^1,2 With the advent of molecular technologies, recurrent genetic alterations have been identified in several histiocytoses,^3-17 reframing the conceptualization of these disorders from one of primary inflammatory conditions to that of clonal neoplastic diseases.^18,19 These alterations primarily affect genes encoding protein kinases of the intracellular mitogen-activated protein kinase (MAPK) signaling pathway, leading to constitutive activation of this pathway (Figure 1).¹¹ However, downstream extracellular signal-regulated kinase (ERK) activation has also been observed in histiocytoses without detected MAPK pathway mutations,^3,11,13 leading to the notion that histiocytic neoplasms are uniformly characterized by dependence on MAPK signaling.²⁰ This has led to the clinical implementation of pharmacological inhibitors of the MAPK signaling pathway (BRAF and/or MEK inhibitors) for the treatment of patients with severe and/or relapsed/refractory histiocytic disease.^20-27 In addition to MAPK pathway mutations, recurrent activating genetic alterations in genes encoding protein kinases of the PI3K/AKT/mTOR signaling pathway (ie, PIK3CA) or receptor tyrosine kinases (ie, CSF1R, NTRK1, RET, or ALK) have been identified in a subset of histiocytic neoplasms.^6,13 Furthermore, significant associations between specific kinase alterations and clinicopathologic phenotypes have recently been observed.⁶ These findings pave the way for molecular (sub)classification of histiocytic neoplasms,^1,6,8,28 and highlight the potential for targeted therapy beyond BRAF or MEK inhibition in histiocytosis patients.

In 2008, Chan et al first described 3 infants with a novel type of systemic histiocytosis characterized by ALK immunoreactivity in large CD163⁺ histiocytes with variable expression of other histiocyte/dendritic cell markers (S100, fascin, factor XIIIa).²⁹ Clinically, the disease was characterized by liver and hematopoietic involvement and could be confused with a storage disorder or leukemia. The disease tended to resolve slowly, sometimes even without any specific treatment but only supportive care. One case demonstrated a TPM3-ALK translocation. Over the following years, several individual case reports and small case series about ALK⁺ histiocytosis have been published,^30-49 confirming the rare but indefinite occurrence of this emerging histiocytic entity. Most notably, Chang et al described a series of 10 cases,⁴³ including the original 3 cases reported by Chan et al, with 8 having molecular confirmation of ALK rearrangements. Together, these reports have documented ALK-positive histiocytosis in both older children and adults with single- or multisystemic disease, expanding beyond the initial observation of the disease as a systemic disorder of infancy. Moreover, the frequent presence of KIF5B-ALK fusions has been reported,⁴³ and successful treatment of some patients with ALK inhibition has been described.^{6,27,33,34,39,40,43,50,51} Due to the modest number of reported patients, and particularly few cases with clinical information and/or successful molecular analysis, the full clinicopathologic and molecular spectra of ALK-positive histiocytosis remain, however, incompletely characterized. Accordingly, here we describe the results of the largest study of ALK-positive histiocytosis to date, outlining the clinicopathologic features of 39 patients, including 37 cases with proven ALK rearrangements.

Fifty-two cases from different hospitals throughout North America, Europe, and Australia were compiled for this international retrospective study. Six cases were previously published^{6,30,31,39,44,47} and are reported with updated follow-up (when available). All cases underwent central review of pathology slides (J.P. and J.-F.E.) for confirmation of diagnosis. Additional immunohistochemical and/or molecular analyses were performed when deemed appropriate. Criteria for a diagnosis of ALK-positive histiocytosis and inclusion in the study were the following: (1) histologic confirmation of a histiocytosis,¹ with expression of at least 2 macrophage/histiocyte markers (including CD163, CD68, CD14, and/or CD4) by the lesional cells and (2A) molecular confirmation of ALK rearrangement and/or (2B) classic infantile systemic disease with diffuse strong ALK immunoreactivity as previously described.²⁹ Cases with ALK rearrangements in which immunohistochemistry revealed histiocytes to be disparate from lesional ALK⁺ cells without immunoreactivity for macrophage/histiocyte markers were termed “atypical ALK-rearranged histiocyte-rich tumors” and not included in the primary study cohort. Also, histiocytosis cases with ALK immunoreactivity but without ALK rearrangements by comprehensive molecular analysis (ie, RNA sequencing [RNA-seq]) were not included in the primary study cohort. Cases excluded for these 2 reasons are characterized separately. Written informed consent for this study was obtained from the patients and/or their legal representative when required, or a waiver of consent was obtained from the relevant institutional review board for retrospective research.

Clinical information was extracted from the medical records by the treating physicians and provided in a pseudonymized fashion according to a standardized deidentified case report form. Retrieved data included demographic characteristics, presenting symptoms, sites of disease involvement, treatments, and treatment outcomes. First- and second-or-further line treatments were categorized as observation with supportive care, surgical resection, radiotherapy, conventional systemic therapy (ie, cytotoxic chemotherapy, corticosteroids, immunosuppression, and/or interferon-α), ALK inhibition, or a combination of these. As previously published for Erdheim-Chester disease (ECD) and Langerhans cell histiocytosis (LCH),^27,52,53 responses to treatment as documented in the medical record were classified as complete response (complete resolution of disease), partial response (partial resolution of disease), stable disease (no significant change in lesions), or progressive disease (growth of known lesions or appearance of new lesions). Responses were extracted from the clinical medical records and based on formal clinical interpretations of ultrasound, computed tomography (CT), magnetic resonance imaging (MRI) and/or positron emission tomography (PET)-CT when available. The less favorable radiologic response was prioritized if multiple imaging methods were used that showed discrepant responses. Objective response was defined as partial or complete response. Whether or not progression or relapse of disease was subsequently observed was dichotomously captured. Progression was defined as progressive disease or death from any cause.

Histopathologic information was collected in a standardized deidentified case report form. Morphology and immunophenotype were reviewed for all cases (J.P. and J.-F.E.). Immunohistochemistry (IHC) was performed as part of initial diagnosis at participating institutions and consequently was executed using varying staining techniques and monoclonal antibodies. If ALK, CD163, CD68, S100, and/or CD1a IHC was not originally performed, these stains were done in the process of central review when sufficient tumor tissue was available. The ALK clones that were used for each case are shown in supplemental Table 1, available on the Blood Web site. The phosphorylated-ERK (p-ERK) immunostains were all performed using clone D13.14.4E (Cell Signaling).

Molecular pathologic information was collected in a standardized deidentified case report form. Molecular analysis was performed as part of initial diagnosis at referring institutions or done in the process of central review, when sufficient tumor tissue was available. For analyses performed during central review, fluorescence in situ hybridization (FISH) analysis was performed using CytoCell dual color ALK break-apart probes (Oxford Gene Technology) as previously described.⁵⁴ Targeted RNA-seq was performed using customized Archer FusionPlex next-generation sequencing (NGS) panels, which allow the detection of both fusions with known or unknown fusion partners. For NGS analysis at the European referral center Ambroise Paré Hospital (Paris, France), RNA was isolated using the Maxwell Rapid Sample Concentrator RNA Formalin-Fixed Paraffin-Embedded Kit (Promega), and RNA concentrations were measured using the Qubit RNA High Sensitivity Assay Kit (Thermo Fisher Scientific). Successful NGS required 20 to 250 ng RNA per reaction. Generated libraries were sequenced using a MiSeq system (Illumina), and produced reads were analyzed with Archer Analysis software. Archer RNA NGS at the North American referral center Memorial Sloan-Kettering Cancer Center (New York, NY) was performed as previously described.⁵⁵ Clonality assessment of T-cell receptor or immunoglobulin (Ig) gene rearrangements was performed using standardized multiplex polymerase chain reaction assays as previously described.^56,57

Data were analyzed using descriptive statistics. Continuous variables were summarized with medians and ranges, and categorical variables were summarized with frequencies and proportions. Frequency of objective responses were summarized for treatment modality by patient group and for the entire cohort, and the frequency of disease progression or relapse following therapy was summarized.

Thirty-nine patients meeting criteria for ALK-positive histiocytosis were identified and included in the study (Tables 1-3; supplemental Tables 1-3). In addition, 3 patients with atypical ALK-rearranged histiocyte-rich tumors were recognized (Table 3), and 10 patients were identified with histiocytoses demonstrating ALK immunoreactivity but no ALK rearrangement by RNA-seq (supplemental Table 4). These 13 patients without complete criteria for ALK-positive histiocytosis diagnosis were excluded from the primary study cohort and are described separately.

The 39 patients with ALK-positive histiocytosis comprised 24 females and 15 males and included 31 children and 8 adults (Table 1). Their age at presentation ranged from 0 days to 41 years: 1 infant had hepatosplenomegaly, anemia, and thrombocytopenia only 7 hours after birth.⁴⁷ Moreover, the 31 pediatric cases consisted of 21 children of 3 years or younger, including 13 infants (age < 1 year). The 39 patients had diverse organ system manifestations (Figures 2-6) and were classified into 2 distinct clinical phenotypic groups: patients with multisystem disease (Group 1, n = 16) and patients with single-system disease (Group 2, n = 23). Within Group 1 were infants with systemic disease with liver and hematopoietic involvement as originally described²⁹ (Group 1A, n = 6) and other patients with multisystemic disease, not fitting the original infantile disease presentation (Group 1B, n = 10).

Patients in Group 1A ranged from 0 days to 5 months of age and presented with hepatomegaly, anemia, and thrombocytopenia. In addition, 5/6 had splenomegaly, 5/6 displayed leukocytosis, and some had measurable liver dysfunction, such as elevated liver enzymes, high bilirubin, and/or decreased serum protein and albumin levels. In 3 cases, focal lesions were seen in the liver (Figure 2). Coagulation profiles were available for 4/6 patients (supplemental Table 2), revealing prolonged prothrombin time, activated partial thromboplastin time, and decreased serum fibrinogen in Case 3. Small numbers of ALK⁺ histiocytes were observed in the bone marrow of all 5 patients that had a bone marrow biopsy performed. Regarding additional sites of disease, 2/6 had renal involvement,³¹ 1/6 had interstitial lung involvement requiring prolonged oxygen supplementation,⁴⁷ and 1/6 had skin involvement (Table 3).

Group 1B contained 5 children and 5 adults who presented with disseminated disease involving various organs (Table 1), including 3 infants without hematopoietic involvement. The most commonly involved organs were the nervous system (7/10), bone (7/10), lungs (7/10), liver (5/10), skin (4/10), and lymph nodes (4/10). All 3 pediatric cases with neurologic involvement had multiple masses in the brain (Figure 4A-E); adult cases had lesions of cranial and/or spinal nerves (3/4; Figure 4G-I), intramedullary spinal cord and/or leptomeningeal enhancement (2/4; Figure 4F), and/or parenchymal brain masses (1/4). Reminiscent of the localization of bone lesions in ECD, bone involvement included bilateral lesions in femora and/or tibiae in 7/7 patients (Figure 3A-D). Lung involvement was uniformly characterized by pulmonary nodules, ranging in size from micronodules to significant tumors (Figure 3H-K). All 5 patients with liver involvement had focal lesions on imaging.

Patients in Group 2 were 20 children and 3 adults, including 12 cases with neurologic involvement (52%). Patients with neurologic involvement presented with diverse neurological symptoms, including seizures (3/12), ataxia (3/12), headaches (3/12), vomiting (2/12), hypotonicity (2/12), unclear paroxysmal neurologic symptoms (“spells”) (1/12), torticollis (1/12), trigeminal neuralgia (1/12), paresis (1/12), and diplopia (1/12). Tumoral lesions were widely distributed in the central and peripheral nervous system (Figure 4J-S; Table 3). In addition, 2 patients had cerebrospinal fluid monocytosis, including 1 case with a very low level of ALK rearrangements in the cerebrospinal fluid by FISH analysis. The 11 patients with nonneurologic disease comprised 1 case with a single lung tumor, 2 cases with solitary bone lesions, and 8 children with localized skin lesions or soft tissue tumors in varying locations (Figure 5; Table 3).

Two of 6 patients did not receive histiocytosis-directed therapy but were managed with supportive care including transfusions (Table 2). In both cases, spontaneous resolution of disease was observed. Three other patients were treated with chemotherapy. Two experienced progressive disease, including 1 patient that had worsening of ascites and died of coagulopathy and sepsis. The other received second-line systemic therapy with dexamethasone, anakinra, and azathioprine and obtained a complete response. The third patient achieved a partial response to first-line chemotherapy, relapsed with biopsy-proven renal involvement at 5 months after completion of treatment, and subsequently received a second round of chemotherapy.³¹ The remaining patient was lost to follow-up.

First-line therapy consisted of conventional systemic therapy alone in 6/10 patients and ALK inhibition with or without conventional therapy in 4/10 patients (Table 2). Conventional systemic therapy led to an objective response in 3/6 patients, 2 of whom had sustained complete responses to chemotherapy and 1 of whom died of (likely immunosuppression-related) sepsis following a partial response to high-dose corticosteroids (Table 3). The 3/6 patients with stable or progressive disease following conventional systemic therapy were treated with ALK inhibition as second-line therapy, with objective responses in all 3. ALK inhibition, with or without chemotherapy, led to an objective response in 4/4 patients in the first-line setting. Responses to ALK inhibition were durable in 7/7 patients, with no events of progression or relapse on treatment (Figure 7).

First-line treatment consisted of surgical resection in 13/23, radiation in 1/23, conventional systemic therapy with or without local therapy in 4/23, and ALK inhibition in 1/23 patients (Table 2). Furthermore, 1/23 patients was observed without treatment, 1/23 was not treated yet, and 2/23 were lost to follow-up. Of those with local therapy, 2/14 had progressive disease. Of patients treated with conventional systemic therapy, 3/4 had an objective response, and 1/4 had progressive disease. The patient treated with ALK inhibition had an objective response. Second-line treatment was given to 4 patients. ALK inhibition was administered in 2/4 patients with objective response, whereas the 2 others received chemotherapy with progressive disease. Third-line treatment with ALK inhibition was initiated in 1 of these patients with objective response (Figure 5A-D). Responses to ALK inhibition were durable in 4/4 patients (Figure 7).

The morphology of ALK-positive histiocytoses varied and included classic xanthogranuloma features with plump foamy histiocytes and variable Touton giant cells in 12/39 (31%) patients (Figure 8A-B) to a more densely cellular, monomorphic appearance without lipidized histiocytes in the majority of cases (Figure 8C,E), some with more spindled/streaming or epithelioid histiocyte morphology (supplemental Table 1). Nuclear features included ovoid nuclei that frequently displayed either a slight fold or indentation (ie, “cup shape”) of its nuclear contour. Mild atypia was limited to 6/39 cases, often with an epithelioid phenotype, which can be confused with malignant histiocytosis; however, mitotic counts and Ki-67 proliferation indexes were low to moderate. Bone marrow involvement was typically a small amount and not a diffuse infiltrative process as in leukemia. The morphology was that of plump histiocytes with indented nuclei and a juvenile xanthogranuloma (JXG)-like phenotype in normo- to hypercellular bone marrow with relatively preserved hematopoiesis and occasional eosinophilia, mild myelofibrosis, and/or megakaryocytic hyperplasia.

The ALK staining pattern did not appear to correlate with molecular alteration as the frequent KIF5B-ALK fusion was noted with all stain patterns; however, no (convincing) nuclear ALK staining was observed. Four cases displayed focal weak ALK staining and 3/39 had exclusive cytoplasmic Golgi dot-like staining in the lesional histiocytes (Figure 8I; Table 4) despite confirmed ALK fusions. Almost half of cases had variable Rosai-Dorfman disease–like histologic features including S100 expression (18/39) and emperipolesis (13/39). When tested for OCT-2 expression,⁵⁸ 14/23 (61%) were positive. When tested for p-ERK (18/39) and Cyclin D1 (19/39) expression (supplemental Table 1), the lesional histiocytic cells often showed corresponding, strong expression (Figure 8K-L).

ALK rearrangements were confirmed in 37/39 patients (Table 1). In 2 infants from Group 1A with classic systemic disease, similar to the cases described by Chan et al,²⁹ no ALK rearrangements were detected by FISH; however, the material was insufficient for comprehensive analysis by RNA-seq. KIF5B was the most common ALK fusion partner and identified in 27 cases, with exon 24 of KIF5B fused to exon 20 of ALK in 25/27 cases (supplemental Table 3). In addition, CLTC-ALK, TPM3-ALK, TFG-ALK, EML4-ALK, and DCTN1-ALK fusions were detected in single cases (supplemental Figure 1).

Three cases were identified without macrophage/histiocyte marker expression by the ALK⁺ tumor cells but with prominent intermixed CD163-expressing (reactive) macrophages (supplemental Figure 2), which can be mistaken for ALK-positive histiocytosis. Moreover, by hematoxylin and eosin stain, the tumor cells often had similar morphology as those seen in ALK-positive histiocytosis, consisting of plump mononucleated cells with eosinophilic cytoplasm. In addition, the lesional cells demonstrated p-ERK and/or Cyclin D1 expression (supplemental Figure 2). Despite extensive immunohistochemical and molecular pathologic analysis, no definitive diagnosis could be made for the 3 cases. The ALK⁺ cells lacked immunoreactivity for many lineage markers, including CD30 and smooth muscle actin (supplemental Table 1). Moreover, clonality assessment of T-cell receptor and Ig gene rearrangements did not reveal high monoclonal peaks (supplemental Methods), arguing against a type of lymphoma. CD34 immunostains did reveal a remarkable nested, endocrinoid to hemangiopericytoma-like pattern with abundant blood capillaries in all 3 cases (supplemental Figure 2). None had the KIF5B-ALK fusion but rather harbored ALK fusions involving EML4 or SQSTM1.

Clinically, 1 of the 3 patients was an adult with Group 1B–like multisystemic disease involving the CNS, left lung, and bone (Case A1; supplemental Figure 3A-D). This patient initially had an objective response to ALK inhibition but then demonstrated disease progression and switched to chemotherapy (Figure 7; Table 3). The remaining 2 cases had Group 2–like single-system disease with localized soft tissue tumors. Both were treated with surgery; however, 1 relapsed with subcutaneous and liver metastases (Case A2; supplemental Figure 3E-F) and received systemic treatment with ALK inhibition with stable disease after 3 months of crizotinib.

Ten patients were identified with histiocytoses demonstrating variable ALK immunoreactivity (supplemental Figure 4) but no ALK rearrangement by RNA-seq. Both clinically and histologically, these cases represented the full spectrum of histiocytic neoplasms, from L-, C-, and R- to M-group (supplemental Table 4).¹ When compared with ALK-positive histiocytoses, they more often showed nuclear ALK staining. Other MAPK activating somatic mutations were identified in 5/10 patients.

We present here a collaborative international study of ALK-positive histiocytosis with detailed clinicopathologic data of 39 cases, including 37 cases with confirmed ALK rearrangements. This study defines a new clinicopathologic spectrum and highlights frequent neurologic involvement. Histology included classic xanthogranuloma features in almost one-third of patients, whereas the majority displayed a more densely cellular, monomorphic appearance without lipidized histiocytes. We affirm the frequent occurrence of KIF5B-ALK fusions⁴³ and expand the molecular spectrum by describing single cases with CLTC-ALK, TPM3-ALK, TFG-ALK, EML4-ALK, or DCTN1-ALK fusions. Finally, we show dramatic and durable responses in 11/11 patients treated with ALK inhibition (Figure 7), 10 with neurologic involvement.

We noted an apparent predilection for females in our cohort of ALK-positive histiocytosis cases, consistent with our overview of the existing literature (supplemental Table 5) and the female predilection observed in ALK-rearranged tumors other than non–small cell lung cancer,⁵⁰ such as ALK-rearranged inflammatory myofibroblastic tumors.⁵⁹ This female predilection is in clear contrast to the male predominance in LCH,^60,61 ECD,^62,63 and JXG,^64,65 particularly in BRAF p.V600E–associated CNS lesions with JXG histology.²⁸ 

Our Group 1A patients were similar to the cases originally described by Chan et al.²⁹ In addition to liver, spleen, and hematopoietic involvement, our patients had lesions in the lungs, kidneys, and/or skin. Unlike in LCH,⁶⁶ in which liver, spleen, and/or hematopoietic involvement constitutes “risk organ” involvement associated with increased mortality, this clinical phenotype in ALK-positive histiocytosis does not necessarily comprise high-risk disease as 2 of 6 patients from Group 1A had spontaneous regression of disease with only supportive care. However, of the 3 other patients with clinical follow-up, 3/3 required second-line systemic therapy and 1 died, underscoring the life-threatening condition that this disease can still represent.²⁹ 

Other patients with multisystemic ALK-positive histiocytosis (Group 1B) have been variably reported previously (supplemental Table 5), sometimes as ECD or systemic JXG. The nervous system was the most frequently involved organ in our patients from this group, followed by the lungs, bone, liver, skin, and lymph nodes. Although there are similarities between patients from this group and patients with ECD, particularly bilateral bone lesions in the legs, many other stigmata of ECD were not observed. For example, perinephric soft tissue thickening (“hairy kidneys”), periaortic encasement, diabetes insipidus, and right atrial infiltration are frequent in ECD^62,67 but were absent in our cohort. Moreover, ECD patients are generally older, with a median age of 55 years,^2,63 as opposed to a median age of 14.5 years of our Group 1B patients. Apart from the specific phenotypic entity that is ECD, there exists the constellation of “extracutaneous/systemic JXG,”^68,69 referring to the rare disease in childhood with JXG histology involving extracutaneous tissues. In a recent series of “systemic JXG,” associations were observed between specific molecular alterations and clinical phenotypes, with ALK rearrangements overrepresented in female cases and cases with lung involvement.⁶⁸ These findings are compatible with ours, raising the question whether ALK-positive histiocytosis is distinct from “systemic JXG” without ALK rearrangements. Altogether, although there is overlap in histomorphology and immunophenotype, we believe that sufficient evidence is available for the designation of ALK-positive histiocytosis as a separate histiocytic entity from both ECD and “extracutaneous/systemic JXG” as previously proposed by others.⁴⁵ 

The findings in our patients with single-system disease (Group 2) were similar to that which has been observed previously (supplemental Table 5), with the nervous system and skin/soft tissue as common sites of disease and surgical resection often being definitive. We expand the clinical spectrum by describing patients with localized bone involvement. When compared with previously reported cases with single-system disease, our cohort includes more cases with isolated neurologic involvement (52% vs 19%) and more children (87% vs 38%). Given the retrospective nature of our study, we cannot rule out some selection bias influencing the clinical spectrum of the disease in our cohort. For example, our cohort did not include adult female patients with isolated breast masses as recently described by Kashima et al³⁴ and Osako et al.⁴¹ Yet, our study did include 2 patients with multisystemic disease including breast masses (Table 3; Figure 3R), supporting the recurrent involvement of this anatomic site.

Eleven patients in our cohort were treated with diverse ALK inhibitors, either as first- or second- or further-line therapy, and sustained objective responses were observed in 11/11 patients (Figure 7). These data corroborate the favorable outcomes of ALK inhibition observed in 7 ALK-positive histiocytosis patients outside of our cohort.^{33,34,40,43,50,51} Our patients were treated with various ALK inhibitors at variable doses; however, the impression of their efficacy is unequivocal. The responses to ALK inhibition stand in contrast to outcomes with conventional systemic therapy, which conferred an objective response in only 7/13 (54%) patients as first-line therapy and 2/5 (40%) patients as second-line therapy (Table 2). However, given that this is a retrospective study, it remains undefined whether ALK inhibition should be implemented as first-line treatment or when disease is refractory to conventional therapies. Also, our study does not address the optimal treatment regimen or duration for ALK inhibition, which is an enduring question for targeted therapy of histiocytosis more broadly, as cessation of BRAF inhibition in histiocytosis patients has been shown to lead to early relapses in most cases.^22,26,70

As illustrated by the 3 “atypical ALK-rearranged histiocyte-rich tumors,” stringent histopathologic assessment is essential before making a diagnosis of ALK-positive histiocytosis. Diagnosis requires diffuse expression of macrophage/histiocyte markers, preferably CD163, by the lesional ALK⁺ tumor cells. ALK-rearranged tumors with prominent intermixed histiocytic infiltrates disparate from ALK⁺ cells without immunoreactivity for macrophage/histiocyte markers are a potential diagnostic pitfall and should be assessed for the possibility of representing another ALK-rearranged entity, such as inflammatory myofibroblastic tumors, which typically do not harbor KIF5B-ALK fusions.^71,72 Future studies may elucidate whether “atypical ALK-rearranged histiocyte-rich tumors” without characteristics of established (ALK-rearranged) entities comprise a new entity. In all 3 of our cases, the CD34 immunostain revealed a remarkable nested pattern with abundant blood capillaries (supplemental Figure 2), which was also observed in a case previously reported as ALK⁺histiocytosis with unusual morphology and a TRIM33-ALK fusion.³² Clinically, this patient had a large mesenteric tumor, highly similar to Case A2 from our study (supplemental Figure 3E-F). Finally, the monomorphic dense infiltration of histiocytes seen in the majority of our ALK-positive histiocytosis cases should be distinguished from epithelioid fibrous histiocytoma.⁷³ 

Supported by our 10 excluded histiocytosis cases demonstrating variable ALK immunoreactivity but no ALK rearrangement by RNA-seq, molecular confirmation of ALK rearrangement should be performed whenever possible to confirm the diagnosis of ALK-positive histiocytosis. This is important because cases with ALK staining but without ALK rearrangements may not benefit from ALK inhibition.^74,75 ALK IHC should particularly be applied in the diagnostic workup of infants with classic multisystemic disease and patients with tumorous lesions in the nervous system, lungs, bone, and/or soft tissue with non-LCH histology. Yet, pathologists should be aware of the variable staining intensity and pattern, with variations further compounded by the ALK clone and protocol used. Comprehensive molecular analysis (ie, RNA-seq) should be performed in ALK IHC–negative cases when the detection of an ALK rearrangement could have treatment consequences as we noted in 7 confirmed ALK-rearranged cases weak or exclusive Golgi dot-like cytoplasmic ALK staining that could be misinterpreted as negative.

Overall, this study defines a new clinicopathologic spectrum of ALK-positive histiocytosis with frequent neurologic involvement and durable responses to ALK inhibition. The findings support the recognition of ALK-positive histiocytosis as a distinct entity from both ECD and (extracutaneous/systemic) JXG. Comprehensive genomic analysis of patients with histiocytic neoplasms is now a necessary adjunct to detailed morphologic and immunophenotypic assessment as this may directly affect disease classification and therapeutic decision making.

The authors acknowledge the International Rare Histiocytic Disorders Registry (IRHDR, ClinicalTrial.gov number NCT02285582), which included some patients described in this study. The authors thank Chris Woods (Lead Analyst, Imaging Informatics) and the Research Pathology Laboratory at Cincinnati Children’s Hospital Medical Center, John DeCoteau (molecular pathologist) at University of Saskatchewan, Sven van Kempen (senior laboratory technician) and Lennart Kester (clinical molecular biologist in pathology in training) at Princess Máxima Center for Pediatric Oncology, and Demi van Egmond (laboratory technician), Tom van Wezel (clinical molecular biologist in pathology), and Arjen Cleven (pathologist) at Leiden University Medical Center for technical assistance. The authors thank Els Wauters (pulmonologist and respiratory oncologist) at University Hospitals Leuven, Andreas Rosenwald (pathologist) at the Institute of Pathology of the University of Würzburg, and Bipin Mathew (pathologist) at Leeds Teaching Hospitals for clinical care and/or diagnostics of patients described in this study. The authors thank Les Laboratoires Servier for allowing the use of their medical illustrations (Servier Medical Art, https://smart.servier.com) in the visual abstract. Figure 6 was created with BioRender.com.

P.G.K. received a grant from Leiden University Medical Center. A.G.S.v.H. is supported by Stichting 1000 kaarsjes voor Juultje. E.L.D. is supported by the National Institutes of Health/National Cancer Institute Core Grant (P30 CA008748), National Cancer Institute (R37 CA259260-01), the Frame Fund, Applebaum Foundation, and Joy Family West Foundation. J.-F.E. is supported by the Programme de Recherche Translationnelle sur le Cancer from the French National Cancer Institute (PRT-K19-143).

Contribution: P.G.K., J.P., E.L.D., and J.-F.E. conceived the study and drafted the manuscript; J.P. and J.-F.E. performed the central pathology review and provided histologic images; P.G.K., J.P., B.H.D., Z.H.-R., L.H.-J., and J.-F.E. performed additional immunohistochemical and/or molecular analyses; P.G.K. made the figures; A.G.S.v.H. participated in data discussions; A.G.S.v.H. and O.A. provided supervision; and all authors contributed cases, and read, revised, and approved the manuscript (meeting the International Committee of Medical Journal Editors criteria for authorship).

Conflict-of-interest disclosure: E.L.D. discloses unpaid editorial support from Pfizer Inc and paid advisory board membership from Day One Biopharmaceuticals outside the submitted work. M.E. discloses honoraries and travel grants by Jazz Pharmaceuticals, Merck Sharp & Dohme, and Amgen. V.P. discloses giving lectures supported by Sanofi, R-farma, and Roche and participation in an expert board supported by Novartis. J.L.H. is a consultant to Aadi Biosciences and TRACON Pharmaceuticals outside the scope of the manuscript. The remaining authors declare no competing financial interests.

Correspondence: Jennifer Picarsic, Division of Pathology, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Ave, ML 1035, Cincinnati, OH 45229; e-mail: jennifer.picarsic@cchmc.org; Eli L. Diamond, Department of Neurology, Memorial Sloan-Kettering Cancer Center, 160 East 53rd St, 2nd Floor Neurology, New York, NY 10022; e-mail: diamone1@mskcc.org; and Jean-François Emile, EA4340 Laboratory and Department of Pathology, Ambroise Paré Hospital, 9 Avenue Charles de Gaulle, 92104 Boulogne, France; e-mail: jean-francois.emile@uvsq.fr.