Introduction

Histiocytic neoplasms constitute a group of rare hematologic neoplasms characterized by the accumulation of neoplastic histiocytic/dendritic cells with associated inflammatory infiltrate [1]. Central nervous system (CNS) involvement of these neoplasms is associated with increased morbidity and mortality [2, 3]. Contrast-enhanced brain and spine MRI is the current standard for evaluating for the presence of CNS disease and monitoring response to treatment [4]. Neurofilament light chain (NfL) is a recently identified biomarker of axonal degeneration that has been utilized in monitoring disease activity in a wide variety of neurologic diseases [5]. It has extensive potential clinical utility as it can be tested in blood. The aim of this study was to evaluate if elevated NfL in patients with histiocytic neoplasms is associated with CNS involvement and active CNS disease.

Methods

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Mayo Clinic, Rochester, Minnesota (Institutional Review Board 08-006647). All patients gave written consent to the use of medical records and blood samples for research purposes. All methods were performed in accordance with relevant guidelines and regulations.

Patient identification

All patients seen at Mayo Clinic, MN with a diagnosis of histiocytic disorder by biopsy and appropriate clinical phenotype who were enrolled in the Center for Multiple Sclerosis and Autoimmune Neurology (CMSAN) or Lymphoma Specialized Program of Research Excellence (SPORE) and had stored blood available were included. All cases were centrally reviewed by the Mayo Clinic-University of Alabama at Birmingham Histiocytosis Working Group to ensure appropriate classification [6]. Clinical data were abstracted by reviewing the electronic medical records and associated images.

Collection of clinical data

Medical records were reviewed by two neurologists (SAB and WOT). BRAF mutation status was tested as part of clinical care in 74 patients (multiple methods were used for testing in 27 patients) via immunohistochemistry in 44 patients, next-generation sequencing in 31 patients, cell-free DNA in 19, and digital droplet polymerase chain reaction in 5 patients. Parenchymal involvement of the brain and spinal cord was defined as T2 hyperintensity or gadolinium enhancement on MRI seen in the parenchyma of the brain and/or spinal cord, not better explained by another disease (eg. multiple sclerosis, non-specific leukoaraiosis). CNS involvement was defined as parenchymal brain or spinal involvement, enhancement of meninges or intracranial cranial nerve on post-gadolinium T1 MRI, and/or pituitary involvement by imaging or biochemical testing. Systemic involvement, including orbital, dermatologic, bone, renal, cardiac, and pulmonary, was noted. CSF analysis for oligoclonal bands was available in 18 of 27 patients; bands were considered positive if there were 2 or more CSF unique bands. Patients were evaluated for comorbidities that may impact NfL levels, including neurologic, cardiac [7, 8], and renal disease [9].

Neurofilament light-chain testing

Two different assays were used for neurofilament light-chain assessment. NfL was tested as part of clinical care in serum in 7 patients using the Simoa Quanterix assay with standardized levels determined by patient age [10]. In 76 patients with stored plasma available, NfL was tested using the Protein Simple Ella Simple Plex assay according to manufacturer protocol. Specifications provided by the manufacturer for the Ella NfL lab assay on serum provided a mean value of 10.6 pg/mL, standard deviation of 4.73 pg/mL, and a range of 6.23–22.2 pg/mL (n = 10). NfL values assessed by the SIMOA assay were transformed to the Ella assay as described previously (y = 2.97 + 1.61x) [11].

Radiographic review

Brain MRI was available in 26 of 27 (96%) (one patient had CT head) with CNS involvement, and spine was available in 18 of 27 (67%). Among those without CNS involvement, MRI brain was available in 38 of 56 (68%), and spine in 6 of 56 (10%). MRI of the brain with and without gadolinium was available for 23/27 (85%) patients within one month of the initial NfL blood sample draw (18 on 3 Tesla MRI, 5 on 1.5 Tesla MRI). Enhancement at the time of NfL was defined as parenchymal enhancement on post-gadolinium T1 images within 1 month of NfL test (Fig. 1).

Fig. 1: Examples of MRI features evaluated and summary boxplots.
figure 1

A, B T2 hyperintensity in bilateral middle cerebellar peduncles extending into cerebellum and pons (A, arrows) with associated gadolinium enhancement on T1 weighted images (B, arrows), defined as parenchymal brain involvement, with enhancement. C Gadolinium enhancement on T1 weighted images in subarachnoid spaces suggestive of leptomeningeal disease (C, arrow), defined in the study cohort as non-parenchymal CNS involvement, without parenchymal enhancement. D Large dural-based lesion with homogenous enhancement on T1 post-gadolinium sequences (D, arrow) with compression of brainstem and encasement of the vertebral artery, defined as non-parenchymal CNS involvement, without parenchymal enhancement. E Boxplot comparing cases to controls; neurofilament light chain (NfL) by Simoa assay transformed to Ella level plus Ella assay in patients with central nervous system (CNS) histiocytosis were significantly elevated compared to those with histiocytosis without CNS involvement (P < 0.0001). F NfL by Simoa assay transformed to Ella level plus Ella assay in patients with histiocytic disease without CNS involvement compared to those with CNS involvement with and without parenchymal involvement. NfL was significantly higher in those with parenchymal involvement than non-CNS histiocytosis (P < 0.0001). G NfL by Simoa assay transformed to Ella level plus Ella assay in patients with histiocytic disease without CNS involvement compared to those with CNS involvement with and without parenchymal enhancement. NfL was significantly higher in those with parenchymal enhancement than non-CNS histiocytosis (P < 0.0001).

Full size image

Statistical analysis

Patient characteristics were compared across groups using the chi-square test for categorical variables and the Kruskal–Wallis test for continuous variables. Log transformations were performed as appropriate. In all analyses, P values < 0.05 were considered statistically significant.

Results

Demographic, clinical, and laboratory features

This cohort of 83 patients included 42 patients (51%) with ECD, 23 patients (28%) with LCH, 13 with RDD (16%), 2 with mixed ECD/RDD (2%), 2 with Histiocytic neoplasm (2%), and 1 with mixed LCH/ECD (1%). Patients with and without CNS histiocytosis did not differ significantly in demographic features or comorbidities (Supplemental Table 1). BRAF V600E mutation was present in 43% (32/74) in whom it was tested, including 54% of those with CNS involvement (14/26) and 38% without (18/48). Other mutations found in those with CNS involvement included: BRAF non-V600E mutations (2), CSF1R mutations (2), MAP2K1 mutation (1), NF1 loss of function (1), and GAB2-BRAF fusion (1). In those without CNS involvement, other mutations included: BRAF-LMTK2 fusion (1), UBR2-BRAF fusion (1), RNF1-BRAF fusion (1), MAP2K1 mutation (1), and PTPN11 gain of function (1).

CNS histiocytosis features

There was a median follow-up of 56 months (6–216) from neurologic symptom onset in those with CNS disease. MRI features are summarized in Supplemental Table 2. Oligoclonal bands were positive in CSF in 4/18 (22%). Systemic, non-CNS involvement was observed in 23/27 (85%) CNS histiocytosis patients including bone (14, 52%), renal (9, 33%), mediastinal or cardiac (4, 15%), vascular (4, 15%), pulmonary (4, 15%), lymph node (3, 11%), orbital (3, 11%), dermatologic (2, 7%), and gastrointestinal (2, 7%). Diabetes insipidus was found in 6 (22%), and anterior hypopituitarism in 4 (15%).

Neurofilament light chain in histiocytosis

The median NfL of 56 control patients with histiocytosis without CNS involvement was 17.8 pg/mL, (range 5.5–265.5 pg/mL). Patients with CNS histiocytic neoplasms had significantly higher NfL levels than those without CNS involvement regardless of assay used (Ella median 40.2 pg/mL, range 17.8–917.0; Ella assay plus Simoa corrected to Ella assay 47.3, range 15.2–917, P < 0.0001) (Table 1 and Fig. 1). Patients with parenchymal brain or spinal cord involvement had numerically higher NfL when compared with patients with CNS histiocytic neoplasms without parenchymal involvement (median parenchymal involvement 56.0 pg/mL, range 15.2–917.0; median without parenchymal involvement 37.0 pg/mL, range 17.8–167.2; P = 0.31) (Supplemental Table 3 and Fig. 1). Patients who had parenchymal enhancement on MRI within 1 month of NfL level testing had numerically higher NfL than those without enhancement (median parenchymal enhancement 56.0 pg/mL, range 17.8–917.0; median without parenchymal enhancement 38.2 pg/mL, range 15.2–86.0; P = 0.14) (Supplemental Table 4 and Fig. 1). One patient had multiple NfL measurements summarized in the Supplemental Figure.

Table 1 Blood neurofilament light chain in CNS vs non-CNS histiocytosis.
Full size table

Discussion

This study demonstrates that NfL is a useful biomarker of active CNS histiocytic neoplasms. In this cohort of 83 patients with histiocytic neoplasms, NfL was significantly higher in patients with CNS involvement than those without. Further, in those with parenchymal disease and enhancement within 1 month of NfL measurement, NfL was numerically higher than in those without. Although this did not reach statistical significance, the magnitude of the difference was high.

This cohort of adult patients with histiocytosis predominantly included patients with ECD. BRAF V600E mutations were present in nearly half of those in whom it was tested in, and presence did not differ significantly between groups. Comorbidities known to impact NfL did not differ between the groups.

Current staging for CNS involvement of histiocytosis consists of neuraxial MRI scan with and without contrast. However, small lesions may go undetected, are difficult to monitor over time, and some patients cannot tolerate the MRI imaging. NfL is a readily available blood test that can be routinely measured on patients being monitored for and treated with histiocytosis. NfL has been studied in neurodegenerative and neuroinflammatory disorders and is a promising biomarker in histiocytic neoplasms. Previous work evaluating the utility of CSF NfL in 12 patients with LCH found levels to be elevated in half of the patients tested 4–12 years from disease onset, particularly in those with worsening radiographic features [12]. More recently, a study found CSF and plasma levels correlated in 10 patients with LCH [13]. The authors concluded that NfL could be a useful tool to screen for neurodegeneration. However, our study shows these values are most elevated in patients with actively enhancing radiographic disease, suggesting that NFL is most useful to identify and monitor active neoplasm in the CNS. A recent study evaluating Mitogen-activated protein kinase inhibitors for treatment of neurodegeneration using CSF NfL as a biomarker of treatment response in five children with LCH [14], found CSF NfL levels tended to decrease with treatment.

This study carries the limitations of a retrospective design. Two different assays were used as one was available on a research basis prior to commercial availability, though these assays have both been validated and directly compared [11], and the findings were supported in both assays. CNS involvement was defined by the presence of MRI changes and small lesions could have been missed. MRIs were not available in all the non-CNS histiocytosis cohort, however none had symptoms to suggest CNS involvement.

In conclusion, elevated blood NfL is a marker of neurologic involvement in histiocytic neoplasms and may represent a sensitive marker of active parenchymal disease. Further studies are needed to evaluate whether NfL can be used to monitor treatment response in patients with histiocytic neoplasms.