A multicenter study on patients with multisystem Langerhans cell histiocytosis developing secondary hemophagocytic lymphohistiocytosis

Abstract

BACKGROUND:

Langerhans cell histiocytosis (LCH) is a rare myeloid neoplasm characterized by the presence of abnormal CD1a⁺CD207⁺ histiocytes. Hemophagocytic lymphohistiocytosis (HLH) represents a spectrum of hyper-inflammatory syndromes typified by the dysregulated activation of the innate and adaptive immune systems. Patients with LCH, particularly those with multi-system (MS) involvement, can develop severe hyperinflammation mimicking that observed in HLH. Nevertheless, little is known about the prevalence, timing, risk factors for development and outcomes of children and young adults who develop HLH in the context of MS-LCH (hereafter LCH-associated HLH).

METHODS:

To gain further insights, we conducted this retrospective multicenter study and collected data on all MS-LCH patients diagnosed between 2000 and 2015.

RESULTS:

Of 384 patients with MS-LCH, 32 were reported by their primary providers to have met diagnostic criteria for HLH, yielding an estimated 2-year cumulative incidence of 9.3 ± 1.6%. The majority of patients developed HLH at or following MS-LCH diagnosis, and nearly a third of them (31%) had evidence of an intercurrent infection. Age < 2 years at LCH diagnosis, female gender, LCH involvement of liver, spleen and hematopoietic system, and lack of bone involvement were each independently associated with increased risk of LCH-associated HLH. Patients with MS-LCH who met criteria for HLH had significantly poorer 5-year survival than patients with MS-LCH who did not (69% vs. 97%; P < 0.0001).

CONCLUSION:

Given its inferior prognosis, further efforts are warranted to enhance the recognition and optimize the treatment of patients with LCH-associated HLH.

CONCISE ABSTRACT:

LCH-associated HLH is a rare hyperinflammatory phenotype of severe MS-LCH that is associated with inferior survival. LCH-associated HLH is most common in patients who are female, less than 2 years of age at LCH diagnosis and have LCH involvement of the liver, spleen and/or hematopoietic system, and in 31% of cases, it develops in conjunction with an intercurrent infection.

INTRODUCTION

Langerhans cell histiocytosis (LCH) is a neoplastic inflammatory disorder characterized by the accumulation of abnormal CD1a⁺CD207⁺ histiocytes in skin, bones, lungs, liver, central nervous system and other organs. LCH can affect patients of all ages and is associated with a broad spectrum of clinical manifestations and outcomes.^{1, 2} Occasionally patients with multi-system LCH (MS-LCH, i.e., LCH involving ≥ 2 organ systems) develop severe inflammation resembling that seen in hemophagocytic lymphohistiocytosis (HLH),^3–10 a heterogeneous group of disorders typified by excessive activation of the immune system.¹¹ HLH can be primary (i.e. hereditary) or secondary (i.e. non-hereditary) in origin. Primary HLH is caused by the presence of germline mutations affecting one of several genes regulating lymphocyte cytotoxic activity, such as PRF1, UNC13D, STX11, STXBP2 and RAB27A.¹¹ Historically, secondary or reactive HLH was defined as HLH in the absence of known germline mutations. The understanding and nomenclature of secondary HLH is evolving with the discovery of mutations or polymorphisms in genes that may lead to partial cytotoxic dysfunction or predilection for innate immune activation.^12–14 Secondary HLH patients present with a similar hyperinflammatory phenotype due to heterogeneous pathogenic triggers, including specific immunologic challenges, such as infection, certain malignancies, and autoimmune conditions.¹⁵ While the triggers vary, these patients all fit under the umbrella of HLH, including those with underlying autoimmune disorders with macrophage activation syndrome or MAS-HLH.¹⁶

The malignancies known to trigger secondary HLH are commonly hematopoietic in origin and include B, T and NK cell lymphomas, Hodgkin lymphoma and leukemias.^{17, 18} Although the underlying etiology of malignancy-associated HLH remains poorly understood, it is proposed that malignant as well as infiltrating immune cells promote inflammation by secreting cytokines, such as interferon γ (INFγ), which is capable of driving the activation of macrophages and T cells.^19–21 Similarly in LCH lesions, abnormal dendritic cells and infiltrating T cells produce a large array of pro-inflammatory cytokines, including IFNγ, IL-2, TNFα, and GM-CSF.^{22, 23} Collectively, these cytokines amplify a local as well as systemic cytokine cascade, supporting the notion that LCH itself may be capable of inducing an HLH-like hyperinflammation. Consistent with this notion, case reports and small case series describe patients with LCH who also exhibit features of HLH.^3–10 The largest includes 30 patients but focuses primarily on pathology with little information on confirmatory testing for HLH, treatment, and outcome.⁴ To gain further insights into the association of HLH and LCH, we performed this comprehensive retrospective multicenter study in which we estimated the prevalence and examined the clinical manifestations, timing, risk factors for development, and outcomes of LCH-associated HLH.

METHODS

Data collection

The Research Ethics Board at The Hospital for Sick Children, Toronto, Canada, approved this study, which was conducted between 1 December 2014, and 30 April 2016. Each participating institution obtained research ethics approval, and the study was conducted in accordance with the Declaration of Helsinki. A survey document was developed, approved by the Histiocyte Society and completed by investigators from 14 centers in North America, Europe and Asia (Supplement).

Patients, inclusion criteria and definitions

The study included patients aged ≤ 30 years and who had been diagnosed with multisystem (MS) involvement at LCH diagnosis between 1 January 2000 and 31 January 2015. Only patients with MS-LCH, confirmed at the time of LCH diagnosis, were included in this study. This is based on prior reports suggesting that patients with high-risk disease were those most likely to develop hyperinflammation.⁴ Local investigators were requested to complete case report forms (CRF) in which they provided information on the date of LCH diagnosis, patient demographics, presenting LCH manifestations, LCH pathology report, results of BRAF V600E mutation analysis, therapy and outcome. Local treating physicians determined whether patients had MS involvement and included this information on the CRF. The study team reviewed every CRF and confirmed the diagnosis of LCH and HLH by reviewing the submitted pathology reports for the presence of CD1a⁺CD207⁺ histiocytes and hemophagocytosis, respectively. MS-LCH was defined as involvement of ≥ 2 organ systems with or without involvement of risk organs (RO; liver, spleen, and/or hematopoietic system). RO involvement was initially defined at LCH diagnosis according to modified Lahey criteria as follows: hematopoietic system—anemia (hemoglobin <10 g/dl, infants <9 g/dl) and/or leukopenia (white blood cell count <4.0 × 10⁹/l,) and/or thrombocytopenia (platelet count < 100 × 10⁹/l); liver—enlargement of more than 3 cm below the costal margin and/or dysfunction (hypoproteinemia, hypoalbuminemia, hyperbilirubinemia and/or increased liver enzymes), spleen (enlargement to more than 2 cm below the costal margin).²⁴

Additional clinical data were collected for patients deemed to have LCH-associated HLH by their local providers, including presence of specific HLH diagnostic criteria, LCH activity at the time of HLH diagnosis and the results of HLH-specific genetic testing. The presence of five out of eight established diagnostic criteria as outlined in the HLH-2004 protocol were used to confirms the diagnosis of LCH-associated HLH (fever, splenomegaly, cytopenia (≥ 2 cell lineages), hypertriglyceridemia/hypofibrinogenemia, presence of hemophagocytosis in bone marrow or other tissue, low or absent NK activity, high ferritin (>500 ng/ml) and soluble interleukin 2 receptor >2,400 u/ml).¹

Statistical analyses

Distributions of characteristics between patients with MS-LCH who met ≥ 5 of 8 HLH diagnostic criteria and those who did not were compared using the Fisher’s exact test. The cumulative incidence of LCH-associated HLH was estimated using the Kalbfleisch and Prentice method, where death was considered a competing risk. Gray’s test was used to identify characteristics associated with developing LCH-associated HLH. Based on the statistically significant Gray’s test findings, the multi-variate Fine-Gray proportional hazards model with step-wise selection was used to further identify independent risk factors associated with LCH-associated HLH. The Kaplan–Meier estimator was used to estimate survival function; overall survival (OS) was compared using the log-rank test. Similarly, multi-variate analysis using the Cox proportion analysis model was completed with a step-wise selection method to identify independent prognostic factors for survival (all P-values are 2-sided; statistical significance was declared at P ≤ 0.05). Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

RESULTS

Patient characteristics

We collected data on 390 patients with MS-LCH. Six patients did not meet eligibility criteria and were excluded from the study (five had single-system LCH and one was diagnosed before the start of the study period). Therefore, 384 patients with MS-LCH were included in the final analysis (Table I). The median age of all patients at MS-LCH diagnosis was 1.7 years, with the majority less than 2 years (59%) and of male gender (58%). When categorized by RO and bone involvement, 180 patients (47%) had RO-negative disease with bone involvement, 129 (33%) had RO-positive disease with bone involvement, 33 (9%) had RO-negative disease without bone involvement and 41 (11%) had RO-positive disease without bone involvement.

TABLE I.

Characteristics of the Study Cohort

Clinical feature	All MS-LCH patients(n=384)	MS-LCH patients without HLH (n=352)	MS-LCH patients with HLH(n=32)	P-value
Median age at LCH diagnosis (years) (median [IQR])	1.68 (0.04–27.03)	1.74 (0.06–27.02)	1.12 (0.04–20.60)
Age at LCH diagnosis (years) Patients with data (n)	379	347	32	<0.0001
< 2	222 (58.58%)	193 (55.6%)	29 (90.63%)
≥ 2	157 (41.42%)	154 (44.4%)	3 (9.38%)
Gender (n [%]) Patients with data (n)	384	352	32	0.0079
Female	161 (41.93%)	140 (39.77%)	21(65.63%)
Male	223 (58.07%)	212 (60.23%)	11 (34.38%)
Multisystem (n [%]) Patients with data (n)	384	352	32	<0.0001
MS RO–	213 (55.47%)	209 (59.38%)	4 (12.50%)
MS RO+	171 (44.53%)	143 (40.63%)	28 (87.50%)
Bone involvement (n [%]) Patients with data (n)	383	352	31	0.0030
No	74 (19.32%)	61 (17.33%)	13 (41.94%)
Yes	309 (80.68%)	291 (82.67%)	18 (58.06%)
Multisystem with bone involvement combined (n [%]) Patients with data (n)	383	343	31	<0.0001
MS RO–, No bone	33 (8.62%)	32 (9.09%)	1 (3.23%)
MS RO–, Yes bone	180 (47.00%)	177 (51.28%)	3 (9.68%)
MS RO+, No bone	41 (10.70%)	29 (8.24%)	12 (38.71%)
MS RO+, Yes bone	129 (33.68%)	114 (32.39%)	15 (48.39%)

HLH, hemophagocytic lymphohistiocytosis; MS-LCH, multisystem Langerhans cell histiocytosis; RO, risk organ; RO-, no risk organ involvement; RO+, risk organ involvement; No bone, no bone involvement; Yes bone, bone involvement.

Thirty-two patients (8.3%) with MS-LCH met ≥ 5 of 8 HLH diagnostic criteria per their local providers with confirmation by the study team. Two patients (6.2%) met HLH diagnostic criteria > 7 days prior to the diagnosis of MS-LCH (median 217.5 days), 15 (46.9%) met HLH diagnostic criteria ± 7 days around the MS-LCH diagnosis, and 15 (46.9%) met HLH criteria > 7 days after the MS-LCH diagnosis (median 168 days after LCH diagnosis; range 9–1014 days; Supplemental Table I). Twenty-nine (90.6%) patients had active LCH and 10 (31%) tested positive for infection at the time they met HLH diagnostic criteria (Supplemental Table I). The infections included EBV (n = 8), CMV (n = 2, 1 with co-existing EBV), HHV-6 (n = 1, with co-existing CMV), adenovirus (n = 1, with co-existing EBV), rhinovirus (n = 1, with co-existing EBV) and candida species (n = 1). The majority of patients with infection (n=6) met HLH criteria >7 days after LCH diagnosis. Of those meeting HLH diagnostic criteria > 7 days after the diagnosis of MS-LCH, 6 (40%) were receiving LCH-directed chemotherapy at the time of HLH development (methotrexate, clofarabine, cytarabine, cladribine, vinblastine, 6-mercaptopurine) and there were 4 (26%) that did not have RO involvement at diagnosis. Nine of the 15 patient who met HLH criteria > 7 days after the MS-LCH diagnosis did so late in their course (≥ 6 weeks after LCH diagnosis). These patients were noted to have active infection (n=4), or recurrent (n=1) or progressive (n=4) LCH at the time of HLH diagnosis.

Most patients meeting HLH diagnostic criteria were female (n = 21; 66%; P < 0.0001), less than 2 years of age at LCH diagnosis (n = 29; 91%; P = 0.008; median age, 1.12 years; range, 15 days to 20.6 years) and had RO involvement (n = 28; 88%; P < 0.0001). Fifty-eight patients (15%) with MS-LCH—49 that did not meet HLH diagnostic criteria and nine that did—had LCH lesions tested for the presence of BRAF V600E. Of these, 40 patients (82%) without and eight patients (89%) with HLH harboured BRAF V600E. Only patients meeting HLH diagnostic criteria exhibited evidence of hemophagocytosis in the bone marrow.

Clinical and laboratory manifestations of patients with LCH-associated HLH

At least 5 of 8 HLH diagnostic criteria were fulfilled in all 32 out of 384 MS-LCH patients as shown in Table II and Supplemental Table I. Although most patients meeting HLH diagnostic criteria had ferritin levels < 2000 ng/ml (n = 25 of 29 [86%]; median, 556 ng/ml; range, 80–26,662 ng/ml), their soluble interleukin 2 receptor (sIL-2R) levels were greatly elevated (n = 17 of 18 [94%]; median, 17,295 u/ml; range, 1808–60,418 u/ml). Indeed, the sIL-2R levels appeared disproportionately elevated compared to the ferritin levels, with a median sIL-2R/ferritin ratio of 56 (range, 0.26–427) and 15 of 18 patients exhibited sIL-2R/ferritin ratios > 2. Most patients had normal natural killer cell cytotoxicity (n = 6 of 11 [55%]). Twenty-six of 32 patients (81%) with HLH were reported to have hemophagocytosis, with 25 of them demonstrating it in the bone marrow specimen and one patient in the liver specimen. Germline genetic testing was performed for at least one of the familial HLH genes in 11 patients (34%). Two patients harboured germline variants: one patient was heterozygous for PRF1 A91V and the other was heterozygous for UNC13D A2V. In the absence of additional HLH gene mutations, these variants were not thought to be causative of the HLH phenotype in these patients.

TABLE II.

Laboratory Findings in Patients with LCH-associated HLH

Laboratory Parameters	HLH patients with data (n)	Median (range) and/or number of patients that met HLH criteria
Complete Blood Count (normal values)
Hemoglobin (13.0–16.0 g/dl)	29	6.6 (5.0–9.7)
ANC (≥ 1 × 109/l)	16	0.42 × 109/l (0–0.79 × 109/l)
Platelet count (150–450 × 109/l)	28	44.5 × 109/l (6–80 × 109/l)
Cytopenia in ≥ 2 cell lineages	28	Present in 28 patients
Hepatic Panel (normal values)
AST (15–45 u/l)	27	64 (10–388)
ALT (10–40 u/l)	26	51.5(7–263)
LDH (140–333 u/l)	26	459 (131–2847)
Other (normal values)
sIL-2 receptor (<2000 u/ml)	18	17, 295 (1808–60,418) Elevated in 17 patients
Fibrinogen (170–470 mg/dl)	6	53 (36–128) Decreased in 6 patients
Ferritin (10–70 ng/ml)	29	556 (80–26,662) > 500 ng/ml in 19 patients
Triglycerides (30–140 mg/dl)	15	186 (166–258) > 265 mg/dl in 15 patients
sIL-2R/ferritin ratio	18	56 (0.26–427)
Hemophagocytosis	26	Present in 26 patients
NK cell function	11	Normal: 6 patients Depressed: 5 patients
Clinical Findings
Fever	31	Present in 31
Splenomegaly	31	Present in 31

ANC, absolute neutrophil count; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HLH, hemophagocytic lymphohistiocytosis; LDH, lactate dehydrogenase; sIL-2R, soluble IL2 receptor; NK cell, natural killer cell.

Prevalence and risk factors associated with LCH-associated HLH

The 2-year cumulative incidence of LCH-associated HLH was 9.3 ± 1.6% (Figure 1). Most patients meeting HLH diagnostic criteria did so within 6 months of LCH diagnosis (estimated incidence of 7.4% at 6 months). The 2-year cumulative incidence of LCH-associated HLH was significantly higher in MS-LCH patients with RO involvement (18.6 ± 3.4% in RO-positive patients vs. 2.5 ± 1.2% in RO-negative patients, P < 0.0001), patients aged < 2 years at LCH diagnosis (15.8 ± 2.8% in patients aged < 2 years vs. 1.3 ± 0.9% in patients aged > 2 years, P < 0.0001), female patients (14.5 ± 3.2% in females vs. 5.8 ± 1.7% in males, P = 0.0079) and patients without bone involvement (18.5 ± 4.9% in patients without bone involvement vs. 6.7 ± 1.6% in patients with bone involvement, P = 0.001; Supplemental Figures 1A–D). When considering RO and bone involvement together, patients with RO-positive LCH without bone involvement were most likely to develop HLH symptoms (31.3 ± 8.1%, P < 0.0001; Supplemental Figure 2). Based on multivariate analysis, the independent risk factor most strongly associated with developing HLH was RO involvement (hazard ratio [HR] 7.07, 95% confidence interval [CI]: 2.46–20.28, P < 0.001). Age, female gender and absence of bone involvement also increased the risk for HLH, but to a lesser extent (Table III).

Figure 1. Two-year cumulative incidence of HLH in MS-LCH.

The two-year cumulative incidence of development of HLH in patients with MS-LCH.

TABLE III.

Factors Associated with Development of LCH-associated HLH

Risk factor	HLH Development
	Hazard ratio	95% CI	P-value
Age at LCH diagnosis (< 2 years vs ≥ 2 years)	4.58	1.37–15.27	0.0132
Gender (female vs male)	2.42	1.15–5.09	0.0198
Multisystem (RO+ vs RO–)	7.07	2.46–20.28	<0.001
Bone involvement (no vs yes)	2.31	1.13–4.73	0.0224

CI, confidence interval; RO-, no risk organ involvement; RO+, risk organ involvement

Therapies used to treat HLH in patients with MS-LCH

The treatments used to control HLH varied widely (Supplemental Table 1 and Supplemental Figure 3). Thirteen of the 32 MS-LCH patients with LCH-associated HLH were treated solely with LCH therapy (front-line therapy [n=8] including vinblastine and steroids; or salvage therapy [n=5] including cytarabine and cladribine²⁵ or clofarabine). All 13 patients exhibited resolution of HLH. Three patients received LCH front-line therapy followed by other treatments (HLH front-line therapy and LCH salvage therapy), with all demonstrating resolution of HLH. Sixteen patients received front-line HLH therapy (etoposide, dexamethasone, and/or cyclosporine) with eight showing resolution of disease. An additional eight received other therapies, including HLH salvage (ATG and/or alemtuzumab) and/or LCH salvage therapy, with 5 of these eight expiring. Five patients received hematopoietic stem cell transplant. No patients received BRAF or MEK inhibitor therapy.

Survival

Survival analysis could not be performed for 43 patients due to the lack of last follow-up date. Of the 341 MS-LCH patients with available data, 13 died, 5 of those met HLH diagnostic criteria. Most deaths among patients who met HLH diagnostic criteria (n = 5) occurred within 2 years after LCH diagnosis (Figure 2; median survival 1.62 years; range, 0.7–3.95 years). Deaths were attributed to active HLH (n = 3) and/or HSCT-related complications (n = 2). Cause of death for those without an HLH diagnosis (n=8) was reported as sepsis (n=2), HSCT-related complications (n=1), surgical complications (n=1), liver failure of unclear etiology (n=1), and respiratory failure and refractory pancytopenia presumable due to LCH (n=1). There were two patients where cause of death was not reported.

Figure 2. Five-year overall survival of MS-LCH patients by HLH diagnosis.

Five-year survival of MS-LCH patients with and without HLH.

The 5-year OS for all patients with MS-LCH was 95 ± 2.2%. The 5-year OS for patients meeting HLH criteria was significantly lower than that of patients who did not meet HLH criteria (68.9 ± 19.2% vs. 97.1 ± 1.8%, P < 0.0001; Figure 2). Age < 2 years at LCH diagnosis (5-year survival for age ≥ 2 years vs. < 2 years: 91.1 ± 4.5% vs. 99.3 ± 1.1%, P = 0.004), female gender (5-year survival for males vs. females: 97.0 ± 2.2% vs. 92.1 ± 4.4%, P = 0.03) and presence of RO involvement (5-year survival for RO-positive vs. RO-negative: 92.3 ± 4.2% vs. 96.9 ± 2.3%, P = 0.033) each predicted a small but statistically significant decrease in OS. The 5-year OS was similar in patients with and without bone involvement (94.2 ± 2.7% vs. 98.1 ± 3.0%, P = 0.28). Although age < 2 years at LCH diagnosis trended towards statistical significance (HR 7.74, 95% CI 0.97–61.7, P=0.0532), multi-variate analysis revealed that meeting HLH criteria remained the only significant factor affecting OS (HR 5.18, 95% CI: 1.66–16.2, P = 0.0046).

A comparison of HLH patients with and without RO involvement at the time of LCH diagnosis was performed with non-HLH counterparts who did or did not have RO involvement at the time of LCH diagnosis. RO negative patients with LCH-associated HLH exhibited a significantly poorer survival compared to RO negative LCH patients who did not develop HLH (5-year survival for HLH RO-negative vs. non-HLH RO-negative: 33.3% ± 27.2 vs. 98.8% ± 1.4, P <0.0001). RO-positive patients with LCH-associated HLH exhibited a trend to poorer survival compared to RO+ LCH patients without HLH (5-year survival for HLH RO-positive vs. non-HLH RO-positive: 81% ± 17.7 vs. 94.1% ± 3.9, (P=0.1065). Patients meeting HLH criteria who received HLH-directed therapy showed a trend toward poorer 5-year survival than those who received LCH-directed therapy (57.4% vs. 100%, P = 0.07; data not shown).

DISCUSSION

In this study we estimate the prevalence and describe the presenting manifestations, risk factors for development, and outcomes of children and young adults with LCH-associated HLH. Among the patients in this cohort, the 2-year cumulative incidence of HLH was 9.3 ± 1.6%. Similar to malignancy-associated HLH,¹⁷ most patients developed HLH with active or progressive LCH disease, with 31% also testing positive for infection. Age < 2 years at LCH diagnosis, female gender, absence of bone involvement and presence of RO-positive disease were significant independent risk factors for development of HLH, with RO-positive disease being most strongly associated. From these data, we conclude that the association between MS-LCH and HLH is rare but should be considered in young children with active, RO-positive LCH, particularly those who lack bone involvement and develop fever, cytopenias, splenomegaly and/or hemophagocytosis in the setting of an infection. Patients may be further characterized based on timing of LCH-associated HLH (< 6 weeks vs. ≥ 6 weeks after initial LCH diagnosis) with early HLH associated with concurrent LCH and later HLH associated with infection and/or recurrent or progressive LCH.

When evaluated for outcome, there was a significantly poorer OS in patients with LCH-associated HLH (69%) versus those without (97%). Indeed, the outcome for patients with LCH-associated HLH was considerably poorer than the 5-year survival of 84% reported for RO-positive patients treated in the LCH-III clinical trial, but similar to the outcomes reported in MS-LCH patients without bone involvement by Arico et al.^{26, 27} In our cohort, age, sex, presence of RO-positive disease and lack of bone involvement were each associated with development of LCH-associated HLH, but none of these factors were independently associated with an inferior OS. Indeed, the only factor significantly associated with an inferior OS was meeting HLH diagnostic criteria. The poorer outcome is consistent with the notion that patients who develop LCH-associated HLH are those with the most severe degree of underlying inflammation, similar to what is observed for patients with systemic juvenile idiopathic arthritis who develop macrophage activation syndrome.²⁸ Additionally, 89% of LCH patients with HLH evaluated for the presence of BRAF V600E, tested positive. Previous studies suggest that this mutation reflects increased severity of LCH.^{2, 29, 30} Because our analysis was limited to a small number of patients, it is not possible to determine with statistical significance whether BRAF-positive patients are at higher risk for the development of HLH and this point warrants further investigation.

Given the poorer outcome for patients with LCH-associated HLH, and the overlap of clinical and laboratory manifestations between these two conditions, it is important to identify biomarkers that will facilitate identification of MS-LCH patients who are on the trajectory towards developing the HLH phenotype. Towards this end, patients with LCH-associated HLH had extremely high levels of sIL-2R (median 17,295 u/ml). The high sIL-2R levels may reflect the fundamental pathophysiology of LCH-associated HLH, where activated T cells play an integral role in driving an underlying cytokine storm and subsequent hyperinflammation.^22,31 Paradoxically, patients with LCH-associated HLH exhibited relatively lower levels of ferritin (median 556 ng/ml) than is typically seen in primary HLH. Based on previous data and our findings, patients with MS-LCH who exhibit elevated sIL-2R levels or increased sIL-2R/ferritin ratios (median 56 in this study) may represent those with the most severe disease and thus warrant close monitoring for the development of HLH.³² Further study of ferritin and sIL-2R levels are needed to confirm the utility of these biomarkers in predicting the prognosis of MS-LCH and/or the development of HLH.

Similar to other forms of secondary HLH, many patients with LCH-associated HLH exhibited improvement following treatment of the underlying LCH. Nevertheless, several patients required HLH-directed therapy and those that did so tended to have a poorer OS. In the absence of other evidence, the decision to treat with HLH-directed therapy versus LCH-directed therapy (including LCH salvage therapy) should be determined based on the severity of the HLH-like phenotype and the best judgement of the treating physician. Prospective studies are needed to identify the best therapies for patients with LCH-associated HLH.

It is important to note that this study has its limitations. First, it is limited by its retrospective design and the fact that not all HLH-specific tests were performed on every patient, including those who ultimately went on to meet HLH-diagnostic criteria. Consequently, physician bias remains possible, and some patients with MS-LCH may have met HLH criteria if appropriate HLH testing had been completed. Second, we did not complete a central review of all pathology samples, nor did we complete BRAF or HLH genetic testing on all patients. However, these procedures were beyond the scope and available resources for this investigation. Third, our survey was not designed to capture all of the details surrounding the determination of organ dysfunction and treatments used for patients with MS-LCH who developed HLH. Therefore, we could not draw definitive conclusions on organ dysfunction and its association with LCH-associated HLH, the efficacy of the regimens used, or determine if organ dysfunction or particular regimens influenced overall outcomes. Last, there may be some limitation to our survival analysis as RO determination was performed at LCH diagnosis (and not at the time of HLH diagnosis) and organ dysfunction was not determined as a part of this investigation.

In conclusion, we have shown that there is a phenotype of severe LCH that mirrors the typical presentation of HLH and likely has a common underlying pathophysiology. Currently, the classification of secondary HLH does not include HLH when it develops in association with LCH.¹⁶ Based on the findings of this study, we encourage the addition of this category to future HLH classification systems. Given its relative frequency and poor outcome, future studies are needed to identify more effective biomarkers and treatments for this unique subgroup of patients.