Clinical, laboratory and genetic features of Erdheim-Chester disease patients: analysis of a retrospective cohort of two reference centers in Latin America

ABSTRACT

Objectives and Methods: Erdheim-Chester disease (ECD) is a rare histiocytic neoplasm with a heterogeneous clinical course, ranging from localized and asymptomatic bone lesions to a multisystem disease, causing significant morbidity and mortality. There are few cohorts published, mainly from North America and Europe. We retrospectively collected clinical data on sixteen biopsy-proven ECD patients diagnosed and treated at two Brazilian reference centres for histiocytic disorders from January 2006 to February 2020.

Results: Median time from onset of symptoms to diagnosis was 13 months (0.1–142). The main organ involved in ECD was bone (75%) and also 75% of the patients presented involvement of more than one organ, characterizing a multi-organic form. BRAF status was available in 81.2% of patients and BRAF V600E mutation was detected by Sanger sequencing in only 18.8%, which can be explained by the low sensitivity of this technique. All patients were treated due to symptomatic disease and a median of two lines (range: 1–7) of therapy were needed. The most common first-line therapy used was α-interferon (75%). The median progression-free survival was 7.5 months, and the median OS was not reached.

Discussion and Conclusion: In the largest Latin American cohort of patients with ECD reported to date, we observed findings which resemble demographic characteristics, sites of involvement and treatment choices reported by other groups. The outcomes may be better with target therapies such as BRAF and MEK inhibitors in patients with mutation and with the adoption of recently published consensus recommendations for the management of ECD patients.

KEYWORDS:

• Erdheim-Chester disease
• BRAF mutation
• Latin America patients

Erdheim-Chester disease (ECD) is a rare malignancy, characterized by tissue proliferation of anomalous histiocytes CD68+/CD1a- and systemic inflammation [1,2]. Until recently, ECD was included in the non-Langerhans cell histiocytosis (nLCH) group. However, according to the World Health Organization’s (WHO) most recent classification of Histiocytic Disorders, it has been assigned within the Langerhans group disease (group ‘L’) [3].

ECD is associated with BRAF V600E mutation and other genetic abnormalities involving the RAS-RAF-MEK-ERK pathway [4]. Moreover, it presents a heterogeneous clinical course, from localized and asymptomatic tissue infiltration to a systemic disorder with multiple organ involvement, leading to significant morbidity and mortality [2,3]. The treatment of ECD includes target agents such as BRAF and MEK inhibitors, biological agents such as α-interferon (α-IFN) and cytotoxic chemotherapy, especially 2-chloro-2’-deoxyadenosina (2-CdA) and cytosine-arabinoside (ara-C) and other approaches to control ‘cytokine storms’ which are a pathophysiological hallmark of this disease [2].

As it is a very rare disease, there are few published series of cases involving ECD, mainly from Asia [5,6], Europe [7] and North America [8]. Owing to the lack of robust ECD data in Latin America [9], we established a registry aiming to collect clinical and laboratory data, as well as biological material from patients with ECD in Sao Paulo, Brazil. We aimed to provide data containing clinical features, outcomes of therapy and survival of Brazilian patients with ECD. Herein, we report the first data of Brazilian ECD patients captured retrospectively. These patients were diagnosed and treated in two referral centers for cancer treatment: a public healthcare service – Hospital das Clínicas at Sao Paulo University and a private service – Hospital Sírio-Libanês, from January 2006 to February 2020.

Clinical and laboratory data were captured at diagnosis, before starting the next therapy line and at the response assessment time. Categorical variables were displayed as absolute frequencies and percentages, and continuous variables as medians (range: maximum and minimum). Progression-free survival (PFS) was calculated from the diagnostic biopsy date until disease progression finding. Overall survival (OS) was defined as the interval between the date of diagnosis and death for any cause or last follow-up. OS and PFS were estimated using the Kaplan–Meier method. The SPSS software for Windows, v 25.0 was used.

Sixteen patients with biopsy-proven ECD diagnosis were included, with male predominance (75% – 12/16), and a median age of 47.7 years (19.9–84.3). The median follow-up time was 49.9 months (6.6–162.8 months), and the median time between onset of the first symptoms and the ECD diagnosis was 13.2 months (95% CI: 7.8–61.2 months), reflecting the difficulty to establish the diagnosis due to delay in the clinical awareness of this rare disease and lack of pathognomonic histologic features.

In our cohort, the main ECD involved organ was bone (75% – 12/16), followed by skin (43.8% – 7/16), central nervous system (CNS) (43.8% – 7/16), lymph node (25% – 4/16), lung (12.5% – 2/16), liver (6.3% – 1/16), spleen (6.3% – 1/16), muscle (6.3% – 1/16) and gastrointestinal tract (6.3% – 1/16). The majority of CNS lesions occurred in the pituitary gland (86% – 6/7). Twelve patients (75%) presented involvement of more than one organ, characterizing a multi-organic form. Xanthelasma and xanthomas were the most common skin lesions. The most frequent clinical manifestations were bone pain (43.8% – 7/16) and neurogenic diabetes insipidus (37.5% – 6/16). Osteosclerotic lesions occurred in 75% (12/16) of cases, retroperitoneal fibrosis and thickening of the renal fascia (‘hairy kidney’ or ‘perinephric straining’) in 37.5% (6/16), 25% (4/16) presented coated aortic sign, and orbital infiltration was found in 25% (4/16), constituting the highly specific features for the diagnosis of ECD. Table 1 summarizes the main clinical-molecular characteristics, therapeutic modalities and responses of the 16 Brazilian patients with ECD included in our analysis.

Table 1. Clinical–molecular characteristics and therapeutic responses in 16 Brazilian patients with Erdheim-Chester disease.

N = 16	Age	Gender	Sites of involvement	BRAF status	Primary therapy	Response
Patient 1	84	Male	Skin, bone, lung, serous membranes	Wild-type	Vemurafenib	Partial response
Patient 2	24	Male	Skin, muscles	Wild-type	Surgery	Partial response
Patient 3	28	Male	Bone, liver, spleen, lymph nodes, CNS	Wild-type	Steroids	Partial response
Patient 4	47	Female	Kidney ‘hairy kidney’	Wild-type	Interferon	Stable disease
Patient 5	64	Female	Skin, bone	Not acessed	Interferon	Stable disease
Patient 6	54	Male	Bone	Wild-type	Interferon	Partial response
Patient 7	52	Female	Skin, bone, CNS and ‘coated aorta’	Wild-type	Steroids and thalidomide	Progressive disease
Patient 8	33	Male	Skin, bone, CNS	Wild-type	Interferon	Partial response
Patient 9	29	Male	Bone, lymph nodes, lung and GIT	Not acessed	Interferon	Partial response
Patient 10	42	Male	CNS, ‘coated aorta’, ‘hairy kidney’	Not accessed	Interferon	Partial response
Patient 11	55	Male	Bone, CNS, ‘hairy kidney’	Wild-type	Interferon	Partial response
Patient 12	55	Female	Bone, lymph nodes, ‘coated aorta’ ‘hairy kidney’	Mutated	Interferon	Partial response
Patient 13	53	Male	Bone, ‘coated aorta’, ‘hairy kidney’	Wild-type	Interferon	Progressive disease
Patient 14	63	Male	Skin, bone, CNS, ‘hairy kidney’	Mutated	Interferon	No response
Patient 15	19	Male	CNS	Wild-type	Interferon	Partial response
Patient 16	55	Male	Skin, bone, lymph nodes	Mutated	Interferon	No response

CNS, central nervous system; GIT, gastrointestinal tract.

Absence of bone involvement is rare in ECD, as well as presence of lymph node and/or spleen/liver involvement [2]. In our series, 25% (4/16) of patients had no bone involvement, as well as evidence of involvement of the reticuloendothelial system was observed in 25% (4/16) of cases. Therefore, Table 2 summarizes the main clinical, molecular and histopathological characteristics of these ‘atypical’ cases of ECD.

Table 2. Diagnosis and characteristics of Brazilian Erdheim-Chester disease patients without bone involvement and/or with involvement of the reticuloendothelial system.

N = 16	Sites of involvement	BRAF status	Biopsy site	Histology and IHC	RE involvement	Diagnostic criteria
Patient 2	Skin, muscles	Wild-type	Skin	Foamy histiocytes, Touton giant cells CD1a-, S100-, CD68+	No	Classic histopathologic and IHC findings
Patient 3	Bone, liver, spleen, lymph nodes, CNS	Wild-type	Bone, lymph node	Foamy histiocytes and fibrosis CD1a-, S100-, CD68+	Yes Lymph node biopsy with classic findings; Spleen/liver presumed by PET-CT	Classic histopathologic and IHC findings plus typical symmetric bone involvement
Patient 4	Kidney ‘hairy kidney’	Wild-type	Kidney	Foamy histiocytes CD1a-, S100+, CD68+, XIIIa+	No	Classic histopathologic and IHC findings plus perinephric stranding
Patient 9	Bone, lymph nodes, lung and GIT	Not acessed	Bone	Histiocytic proliferation, CD1a-, S100- CD68+	Yes, Lymph node involvement presumed by PET-CT	Histopathologic and IHC plus symmetric bone involvement of lower extremities
Patient 10	CNS, ‘coated aorta’, ‘hairy kidney’	Not accessed	CNS	Foamy histiocytes, CD1a-, S100-, CD68+	No	Classic histopathology plus periaortic infiltrates and perinephric stranding
Patient 12	Bone, lymph nodes, ‘coated aorta’, ‘hairy kidney’	Mutated	Bone	Histiocytic proliferation, fibrosis, CD1a-, S100-, CD68+	Yes, Lymph node involvement presumed by PET-CT	Classic histopathology, molecular criteria plus classic clinical findings
Patient 15	CNS	Wild-type	CNS	Histiocytic proliferation, Touton cells, CD1a-, S100-, CD68+	No	Classic histopathologic and IHC findings plus atrophic cerebellum involvement
Patient 16	Skin, bone, lymph nodes	Mutated	Skin, bone	Foamy histiocytes, CD1a-, S100-, CD68+, XIIIa+	Yes, lymph node involvement presumed by PET-CT	Classic histopathology, molecular criteria plus symmetric bone involvement

IHC: immunohistochemistry, RE: reticuloendothelial involvement (lymph nodes, spleen, liver or bone marrow), CNS: central nervous system.

In our cohort, biopsies from the involved organ showed that no case was positive for CD1a, 94% (15/16) were positive for CD68 and 18.7% (3/16) for S-100 protein. At the baseline, the median hemoglobin was 118 g/L (79–147 g/L), the median leukocyte count was 8.36 × 10⁹/L (4.63–16.06 × 10⁹/L) and the median platelet count was 354 × 10⁹/L (186–558 × 10⁹/L). Additional laboratory features such as albumin, total bilirubin and creatinine were mostly unremarkable. Positron emission tomographies with 18-fluorodeoxiglucose (18-FDG-PET-CT) were performed for all patients and 81.3% (13/16) presented avid lesions. BRAF status was available in 13 of 16 (81.3%) patients and BRAF V600E mutation was detected by Sanger sequencing in 18.8% (3/13), lower than reported in developed country cohorts (57–70%) [4].

Unexpectedly, we showed lower frequency of BRAF V600E mutation in our cases of ECD than previously described by others authors (18.8% versus 50–60%). However, we were not able to perform molecular tests searching for this mutation in 3 of 16 (18.8%) patients. Thereafter, this could have collaborated to underestimate the real frequency of this mutation in our cohort. Moreover, according to Melloul et al. [10], the BRAF V600E mutation in histiocytic malignancies shows low allelic fraction (< 5%) which impairs its identification in less sensitive techniques such as Sanger sequencing. In addition, in our cohort we did not use more sensitive techniques such as PCR (dd PCR) or next-generation sequencing.

All patients were treated due to symptomatic disease and a median of two lines (range: 1–7) of therapy were needed. The median time between diagnosis and first line of therapy was 1.1 months (95% CI: 0.4–5.2 months). The most common first-line therapy used was αα-interferon (75% – 12/16), followed by corticosteroids (31% – 5/16), thalidomide (12.5% – 2/16), anti-BRAF/vemurafenib (6% – 1/16) and tumor excision (6% – 1/16). First-line treatment was discontinued in 18.8% (3/16) of patients due to toxicity, particularly fever, myalgia and flu-like symptoms associated with α-INF.

Cytotoxic chemotherapy was the most common second-line treatment used; 2-CdA monotherapy was indicated for 4/16 (25%) patients and LCH-like polychemotherapy regimens containing etoposide, vimblastine, methotrexate and 6-mercaptopurine in 2 of 16 (12.5%) patients. Other therapeutic modalities administered during the follow-up included radiotherapy (4/16 – 25% of patients), and cobimetinib, imatinib and infliximab, in 1/16 (6%) patient each.

The therapeutic management of Brazilian patients was heterogeneous, although most of them received α-IFN as first-line therapy. Since 2020, international recommendations for management of patients with ECD have been published [2], and their applications will allow more homogeneous strategies for the ECD therapeutic approach and a reliable comparison between results obtained in different treatment centers worlwide.

The median PFS was 7.5 months (95% CI: 5.1–10.0 months) and the median OS was not reached. PFS at 4 years was 27.3% (95% CI: 3.8–50.8%) and OS at 4 years was 92.9% (95% CI: 79.2–100%). One patient died of infectious complication at 50 months, after a single cycle of rescue chemotherapy with 2-CdA. The overall response rate (ORR) at first-line therapy was 62.5% (10/16), with no one reaching complete response, and 12.5% (2/16) were refractory. In the subgroup treated with α-IFN in the first line, ORR was observed in 58.3% (7/12), with median PFS of 8.6 months (95% CI: 5.3–12.0 months) and 7.2 months (95% CI: 0–15.4 months) for those who did not receive α-IFN.

To our knowledge, this is the largest Latin American cohort of patients with ECD reported to date. Our findings resemble demographic characteristics, sites of involvement and treatment choices reported by other groups [5–9], although it is clear that the proportion of ECD patients showing BRAF mutation (18.8%) seemed to be lower than previously reported (50%) [4]. Owing to the small sample size of our cohort, it was not possible to look at the impact of the use of α-IFN on survival as it has been shown in other larger series [11]. Our α-IFN ORR was lower than that described by another study (58.3% vs. 80%) [12]. A small proportion of patients had access to target therapies, a treatment modality known to be associated with better PFS [13]. The high cost of these medications justifies the low availability in resource middle-income countries, such as Brazil.

Although the Brazilian cohort is small and shares many characteristics similar to other ECD case series, the main difference observed is highlighted compared to Estrada-Veras et al. [8] and Cohen-Aubart et al. [14] since we found higher frequency of cases without bone involvement (25% vs. 5-10%). Similarly, we showed higher rate of reticuloendothelial system involvement (25% vs. 8–10%), no cases with ECD/LCH overlapping (0% vs. ∼10%) and lower frequency of BRAF V600E mutation (18.8% vs. ∼50-60%). In addition, we found lower response to α-IFN (58.3% vs. ∼70–80%) and higher OS (92.9% at 4 years vs. ∼80% at 5 years).

Our study has several limitations, including a small sample and those intrinsic to a retrospective analysis, but we believe that it can contribute to the clinical and laboratorial knowledge of Latin American patients with this rare histiocytic neoplasm. We described difficulties inherent to ECD diagnosis, particularly in the scenario of low and middle-income countries, where molecular diagnostic techniques are not universally accessible. We thus demonstrate that although ECD causes significant morbidity, the mortality was low and satisfactory clinical control was obtained in more than half of patients with α-IFN. These outcomes may be better with target therapies use as BRAF and MEK inhibitors in patients with mutation.

In conclusion, ECD is a rare disease with non-specific signs and symptoms and consequently not promptly recognized by many physicians, which contribute to keeping this disease sub-diagnosed. Our data highlights the need of continuous medical education on ECD and the establishment of reference centers with availability of diagnostic tools. Importantly, the establishment of local and national registries of rare diseases is essential to enlarge the cohort and to confirm this preliminary data. Only cooperative groups will provide reliable information and guide recommendations for therapies for this entity which has just over 2000 cases cataloged in different registries worldwide.

Acknowledgements

We thank Dr Sheila Aparecida Coelho de Siqueira (Hospital das Clínicas da Faculdade de Medicina da USP, São Paulo, Brazil) for performing tissue biopsies analysis. We thank Prof. Dr Israel Bendit (Hospital das Clínicas da Faculdade de Medicina da USP, São Paulo, Brazil) for analyzing the BRAF V600E mutation in tissues samples.

Disclosure statement

No potential conflict of interest was reported by the author(s).