Clinicopathological characteristics of histiocytic sarcoma affecting the central nervous system in dogs

Abstract Background Histiocytic sarcoma affecting the central nervous system (CNS HS) in dogs may present as primary or disseminated disease, often characterized by inflammation. Prognosis is poor, and imaging differentiation from other CNS tumors can be problematic. Objective To characterize the clinicopathological inflammatory features, breed predisposition, and survival in dogs with CNS HS. Animals One hundred two dogs with HS, 62 dogs with meningioma. Methods Retrospective case series. Records were reviewed for results of cerebrospinal fluid (CSF) analysis, CBC, treatment, and outcome data. Results Predisposition for CNS HS was seen in Bernese Mountain Dogs, Golden Retrievers, Rottweilers, Corgis, and Shetland Sheepdogs (P ≤ .001). Corgis and Shetland Sheepdogs had predominantly primary tumors; Rottweilers had exclusively disseminated tumors. Marked CSF inflammation was characteristic of primary rather than disseminated HS, and neoplastic cells were detected in CSF of 52% of affected dogs. Increased neutrophil to lymphocyte ratios were seen in all groups relative to controls (P <.008) but not among tumor subtypes. Definitive versus palliative treatment resulted in improved survival times (P < .001), but overall prognosis was poor. Conclusions and Clinical Importance Clinicopathological differences between primary and disseminated HS suggest that tumor biological behavior and origin may be different. Corgis and Shetland Sheepdogs are predisposed to primary CNS HS, characterized by inflammatory CSF. High total nucleated cell count and the presence of neoplastic cells support the use of CSF analysis as a valuable diagnostic test. Prognosis for CNS HS is poor, but further evaluation of inflammatory mechanisms may provide novel therapeutic opportunities.


| INTRODUCTION
Histiocytic sarcomas (HS) are tumors that arise typically from interstitial dendritic cells 1,2 and may occur as localized disease or as disseminated disease affecting multiple organs. Several dog breeds are consistently overrepresented in HS case series, and a familial association has been reported in Bernese Mountain Dogs. [3][4][5] Defined breed predisposition data for HS are limited, but Bernese Mountain Dog, Golden Retriever, Flat Coat Retriever, Shar Pei, Miniature Schnauzer, Labrador Retriever, and Pembroke Welsh Corgi breeds have been shown to be overrepresented based on defined control populations, 6,7 and Rottweilers are conspicuously overrepresented in several case series. 1,[8][9][10][11] Histiocytic sarcoma involving the central nervous system (CNS) is relatively uncommon representing 2.2% (n = 2) of primary and 3.4% (n = 7) of secondary intracranial neoplasms diagnosed at necropsy, 12 and prognosis regardless of treatment is anecdotally poor. [13][14][15] Intracranial CNS HS commonly are reported as extra-axial, uniformly contrastenhancing masses on magnetic resonance (MR) imaging, and differentiation from more commonly occurring, and generally more benign extra-axial tumors such as meningioma can be challenging. 13,16,17 The majority of published data about CNS HS relates to primary, localized CNS HS in small case series, [13][14][15][16][17][18][19] with limited availability of defined breed predisposition data. 13 Pembroke Welsh Corgis, Shetland Sheepdogs, Labrador Retrievers, and Golden Retrievers were overrepresented relative to total hospital population in 1 study, 13 and Corgis also were noted at higher frequencies in other uncontrolled studies. 18,19 Central nervous system HS appear to differ from HS in most other organs in the severity of inflammatory infiltrate, 2,13,15,18,19 but it is unclear whether this inflammatory phenotype is common to all CNS HS or if it is restricted to primary localized disease. Markedly increased total nucleated cell counts (TNCC) have been reported in cerebrospinal fluid (CSF) from CNS HS cases. 13,20,21 This observation likely reflects the inflammatory component of the disease and is in contrast to most other primary CNS tumors in which normal to mildly increased CSF cell counts are typical. [22][23][24] To further evaluate the inflammatory nature of CNS HS in dogs, a clinicopathological review of a series of primary and disseminated CNS HS cases and meningiomas was done. Markers of inflammation in CSF (TNCC) and peripheral blood (leukocyte neutrophil to lymphocyte [N:L] ratio) were defined in the context of tumor subtype (primary or disseminated CNS HS, meningioma) to determine potential relevance to diagnostic and prognostic criteria.

| CSF analysis
Cerebrospinal fluid analyses were included from cases in which samples were acquired at first presentation, before specific treatment.
The CSF collection site was noted when available. Any corticosteroid administration within 1 week of CSF analysis was noted. Total nucleated cell counts, total protein concentration (TP), RBC counts, and differential counts were performed within 30 minutes of collection of CSF as previously described. 23 Three cases of HS CSF analysis (primary brain cases 14, 30; primary spinal cord case 5) and 56 reference CSF analyses from intracranial meningioma cases were reported previously. 21

| Neutrophil to lymphocyte (N:L) ratio
Inclusion criteria for N:L ratios were the same as for CSF analysis.
Complete blood counts from cases that had received corticosteroids within 1 week of presentation were excluded from analysis. Only VMTH cases were included, and reference values for N:L ratios were calculated from raw CBC reference interval data from the UC Davis VMTH clinical laboratory.

| Treatment and outcome
Treatment groups were defined as definitive if specific treatment modalities including surgical resection, radiation treatment, or chemotherapy (specifically lomustine) were used in any combination, and palliative if anti-inflammatory (non-steroidal), corticosteroid, antiepileptic, or other analgesic medications were the only therapeutic interventions.
Survival was defined for all cases from time of presentation with neurological signs, and only for cases with survival times >1 day to exclude animals euthanized at the time of imaging or diagnosis. Vertebral bone involvement was documented in 14/25 of the disseminated tumors with extradural lesions. No apparent regional predisposition was seen with primary HS (2 cervical, 3 thoracic, 2 lumbar, 1 multifocal), but disseminated HS was most common in the thoracic region (3 cervical, 11 thoracic, 6 lumbar, 5 multifocal of which 3 involved thoracic segments). One primary and 2 disseminated tumors also involved brain.
Neoplastic cells were identified in 52% (9/18 primary CNS HS, 4/7 disseminated CNS HS) of cytological preparations for which slides were available for review ( Figure 3). Neoplastic cells were identified in 3 samples in which they had not been identified in the initial report.
Mitoses were present in 10 samples and only were seen in cases where neoplastic cells also were identified (Figure 3).

| Neutrophil to lymphocyte ratio
Among dogs that did not receive corticosteroids, N:L did not differ sig-  Corgis have been reported to have a high frequency of HS affecting tissues outside the CNS 7 and also HS affecting the CNS. 13,18,19 Although limited in numbers, these data and ours suggest that CNS MHCII) may fail to distinguish dendritic cells from monocytes, macrophages, and even activated microglia, particularly in non-lymphoid organs. 30 Interstitial dendritic cells generally are not thought to be present within the brain, but dendritic cells have been reported in the meninges, choroid plexus, and circumventricular organs. 31 Adding to this complexity, infiltrating inflammatory monocytes or macrophages and subpopulations of resident microglia also may differentiate into dendritic cells within the CNS. 32,33 Defining the cellular origin of specific CNS histiocytic tumor subgroups will be important for appropriate tumor phenotyping for future genetic studies and for development of appropriately targeted novel therapeutic approaches with respect to what may be 2 distinct tumors.

The previously reported predisposition in Labrador Retrievers
was not confirmed in our study, but the suspected predisposition in Shetland Sheepdogs as proposed previously 13 36,37 making anatomical location unlikely to be the only factor. In the absence of a defined intracranial subdural space, both primary and disseminated HS involve both the meninges and the brain parenchyma, and most other intracranial neoplasia is not associated with marked CSF pleocytosis, [22][23][24] also supporting an argument for inherent differences in the pro-inflammatory nature of the primary versus disseminated HS subtypes. Articular or periarticular HS also has been associated with a marked inflammatory response, 8 and associations have been made between prior injury and inflammation involving the joint and the development of HS. 38,39 No direct evidence supports underlying inflammation as a predisposing factor for the development of HS, but the link between inflammation and a variety of cancers is well documented. 40 Further investigation is warranted to determine whether preexisting inflammation may be a precursor for primary CNS HS and whether specific breeds may have genetic predispositions consistent with this pathogenic mechanism.
Systemic inflammatory responses associated with cancer are well documented, and increases in peripheral blood N:L ratios have been used widely as an indicator of tumor-associated inflammation. 41 Increased N:L ratios have been associated with poorer overall survival in a variety of cancers in humans, 42 including intracranial gliomas. 43 The N:L ratios were significantly increased in all tumor subgroups in our study compared to normal reference ranges, but the marked pro-  44,45 Limited data are available to define the incidence of primary versus secondary CNS HS in humans, but most reports involve primary tumors. Interestingly, a marked inflammatory infiltrate also has been described for these primary tumors in humans similar to their counterparts in dogs, [46][47][48][49] suggesting that they may have some common underlying characteristics.
Our study conclusions are limited by relatively small sample sizes in tumor and treatment subgroups, variability in defined survival end points and potential for inappropriate categorization of primary tumors, even with comprehensive necropsy examination. However, the significant differences defined between primary and disseminated CNS HS particularly in terms of breed predisposition and inflamma-