Brain stereotactic biopsy flow cytometry for central nervous system lymphoma characterization: advantages and pitfalls

Primary central nervous system lymphoma (PCNSL), a rare hematological disease, represents a big challenge for clinicians and researchers [1, 2]. Despite improved therapeutic approaches, the diagnostic - prognostic characterization has documented little improvement and long-term response to treatment is rare [3, 4]. More than 90 % are diffuse large B-cell lymphomas (DLBCL) derived from a late germinal center B cell [5]. PCNSL is usually a widespread - diffusely infiltrating lymphoma. Thereafter, efforts at resection are discouraged, although resection may be associated with prolonged survival in surgically safe, single, PCNSL lesions [6]. Hence, stereotactic biopsy (SB) followed by conventional histopathology and immunohistochemistry (IHC) is the gold standard diagnostic approach useful to perform a differential diagnosis among other brain lesions, such us gliomas [7‐11].

Flow cytometry (FCM) is a proven valuable diagnostic tool for the diagnosis and monitoring of hematological malignancies in routine clinical practice [12, 13]. More recently, FCM immunophenotype of fine-needle aspiration and/or core needle biopsy from lymphoid tissues has shown to improve diagnostic accuracy in B-cell non-Hodgkin lymphomas classification [14‐18] and FCM significantly increases the sensitivity and specificity of cerebrospinal fluid (CSF) infiltration detection in onco-hematology [19‐22]. However, no biological data have been published on FCM immunophenotype of fresh single-cell suspension obtained from SB of brain lymphomas.

We performed a feasibility study to evaluate the diagnostic and potential added value of FCM characterization of a single SB taken in vivo from intra-cerebral suspected lymphomas. FCM was compared with conventional IHC results in terms of diagnosis reliability and sensitivity in cell markers identification. The membrane CD20 mean fluorescence intensity (MFI) was evaluated to document, in PCNSL, the intensity of expression of a surface molecule potentially related to clinical response to Rituximab [23]. Finally, the tumor lymphoid side-population was characterized and compared to the peripheral blood (PB) lymphocytes phenotype, to investigate the blood-brain barrier permeability to non-malignant lymphocytes in CNS lymphomas.

This is a feasibility, prospective study that describes for the first time the use of FCM applied on a cohort of brain SB samples, providing evidence that a single SB is a valid source for identification and characterization of brain lymphomas and reactive infiltrating leucocytes, with a significant agreement between the FCM and IHC diagnosis (P = 0.0034). Despite the magnetic resonance strongly suggestive for PCNSL in 29 selected cases, 18 PCNSL, 10 gliomas and 1 PNET were diagnosed by IHC. FCM identified 16 out of 18 (89 %) PCNSL cases. Thereafter, 2 brain lymphomas (11 %) were missed by FCM. The overall experience indicates that FCM has difficulty detecting high-grade lymphomas, particularly DLBCL. Failure must be assigned to necrosis, cell fragility or inadequate sample size [15]. Moreover, necrosis can represent a major problem in PCNSL diagnosis after corticosteroid administration, as shown by a recently published study [26], considering that a large portion of newly diagnosed PCNSL patients often underwent a steroid treatment just before diagnosis, for clinical need. According to the study design, a single biopsy was analyzed by FCM while 6–8 samples were utilized for the IHC characterization. Nevertheless, the manual processing technique for tissue disaggregation allowed a successful FCM identification of brain lymphomas in 89 % of cases, confirming the limits of a correct diagnosis in patients pre-treated with steroids. Our study shows that a single SB is an adequate sample size for a comprehensive PCNSL characterization in absence of tissue necrosis. In fact, in the 2 PCNSL cases not identified by FCM the necrosis was the main factor. A diffuse necrosis with a small residual foci of lymphoma was documented by IHC in both cases, confirming the relevance of necrosis in hampering the diagnosis in large cell lymphomas. It is likely that performing FCM analysis on more than a single bioptic sample the rate of success in FCM brain lymphomas identification would increase.

FCM is a proven valuable diagnostic tool in hematological CSF infiltration detection [27, 28], increasing the sensitivity and specificity of leptomeningeal disease identification in CNS lymphomas [22]. We have recently documented that FCM can discriminate between reactive and neoplastic plasma cells in CSF samples with very low cell counts, confirming to be significantly more sensitive than standard approaches [21]. CSF FCM was performed on 5 out of 18 PCNSL cases and, despite the low cell count, a clonal B cell population was identified in one case. However, CSF cytology was negative and, although FCM was indicative of a B-cell clonal disorder, the data was not strong enough to reach the diagnosis of PCNSL. These findings are in agreement with previously published data [22].

A very wide panel of markers could be tested by FCM in a single brain SB, FCM demonstrate to be more sensitive than IHC for the identification of relevant markers. CD10 has been reported to be 10 times lower in PCNSL comparing to systemic diffuse large B-cell lymphomas [29]. In our study, the CD10 expression was documented in 3/16 PCNSL (19 %) by FCM and 1/18 (5 %) by IHC. The reason for this discordance is uncertain. Probably the different clones utilized for FCM (clone HI10a) and IHC (clone 56C6) could be involved in this matter. However, a different sensitivity between FCM and IHC in antigen identification is widely documented. Our report documents a possible underestimation of CD10 frequency of expression in PCNSL by IHC staining, highlighting the added value provided by performing FCM in terms of a more reliable technique for the identification of selected markers.

The panel of antigens used was in accordance with the WHO guidelines [25]. However, more recently, a significant differential expression of CD44 between Burkitt lymphoma and CD10 DLBCL has been documented, making CD44 an excellent candidate for rapid immunophenotypic discrimination between these two types of lymphoma [30]. Flanking IHC studies, FCM characterization, enriched by new relevant diagnostic markers, can significantly improve the diagnostic approach to PCNSL.

Surface Ig light chain restriction was documented in 11 cases. Immunoglobulin light chains restriction is an important adjunct to the use of surface markers in onco-hematology [13]. However, light chain restriction is not always documented on high grade B-cells NHL. In our series 5 cases were both surface and intra-cytoplasmic negative. Despite the lack of light chain restriction, the high FSC/SSC scatter of the B cell population (large cells) compared to the small T lymphocytes as well as the prevalence of B cells (85 %, range 60-96) compared to the T subset (10 %, range 1-38), strongly supported the diagnosis of a B cell neoplasm, confirmed by histopathology in all the cases. Moreover, since Ig light chain expression in non-Hodgkin lymphomas (NHL) is difficult to identify by IHC, cytometry can easily represent the method of choice for clonality detection in PCNSL.

The presence of a reactive immune response and perivascular T-cell infiltrate within the lymphoma has been shown to represent a prognostic biomarker and influence disease outcome in PCNSL [31, 32]. Antigen co-expression, mandatory for the classification of T and NK lymphocyte sub-populations, is easy to identity by FCM while difficult to assess by IHC. In this study, we documented a significant difference in lymphocyte sub-set distribution between brain lymphoma and peripheral blood. A prevalence of CD8 reactive T lymphocytes was documented in the majority (81 %) of PCNSL cases, while a prevalence of CD4 positive lymphocytes was identified in all but one peripheral blood samples (90 %) (P < 0.001). CNS is an immunogical sanctuary with restricted access and an unique microenvironment. The present study consistently shows that the blood-brain barrier actively selects a sub-population of CD8 non-malignant cells in PCNSL, providing a promising rationale for the investigation of cellular immunotherapy in brain tumors. To date, there is an increasing interest in engineered chimeric antigen receptor (CAR) T-cells in hematologic malignancies, also in NHL [33]. In this point of view, exploiting their migration to pathological lesions, reactive T-cells could be used for adoptive immunotherapy strategies based, for example, on targeting brain lymphoma CD19 positive B-cells by bi-specific T-cell engager monoclonal antibodies or engineered CAR T-cells in preclinical models.

Tumor associated macrophages, commonly present in NHL, have been related to poor prognosis and are potentially involved in the pathogenesis of brain NHL [34]. We identified a minority of CD4 dim CD14 positive monocytes among the B lymphoma cells and a great amount in the PNET case. Biological studies, aimed at a better understanding of the role of the tumor associated reactive population, would easily benefit from the FCM approach.

The identification of new therapeutic strategies is of prominent importance in PCNSL and FCM characterization can be a valuable tool for the detection of potential therapeutic targets and orientate tailored antibody-based therapies. The CD20 B cell-specific antigen is a membrane-bound protein whose expression is very heterogeneous among different B lymphoma subtypes [35] and significantly associated with disease prognosis [36, 37]. This difference may correlate with clinical response to Rituximab, a monoclonal antibody directed against the CD20 antigen that, in combination with conventional chemotherapy, has dramatically improved the survival of patients with diffuse large B-cell lymphoma [3]. By contrast, reduced CD20 expression has been associated with an inferior survival, suggesting that the evaluation of cell surface antigen intensity of expression can improve treatment tailoring of antibody-based therapy [38]. Anti-CD20 antibody therapy has shown to improve disease control and outcome in patients with primary CNS lymphoma [23, 39]. No biological data, however, is available regarding the intensity of surface antigen expression in intra-cranial B lymphomas. An objective quantification of antigen intensity of expression can be easily performed by FCM by the difference between antigen MFI and negative control. This technique is practical for use in routine setting. Utilizing this approach, our study documents for the first time a marked variability between surface CD20 expression level in brain lymphomas, ranging from very “bright” to “dim” in MFI with one CD20 negative case by FCM. CD20 (L26) staining is not usually graded by pathologists, and hematopathologists report all lymphomas that react with L26 antibody as “CD20 positive” regardless of expression level [40]. Therefore, the determination of CD20 MFI could represent a novel biomarker for a better stratification of PCNSL patients more likely to benefit from Rituximab treatment, for a more effective tailoring of the first line therapy. Actually, this hypothesis should be confirmed in further prospective studies. A prevalence of CD45 negative CD56 positive cells, consistent with a neuro-ectodermal population [41], was documented in 13 cases, with a strong CD34 expression identified in the PNET case, suggesting a potential role of FCM in non hematological tumors.

This paper provides evidence that a single brain stereotactic core biopsy is a valid source for rapid identification and accurate FCM characterization of CNS lymphomas. FCM has shown to be more sensitive than IHC for selected markers, easily applicable for antigen intensity quantification and, documenting a prevalence of CD8 reactive T lymphocytes in brain lymphomas, suggests an active mechanism of lymphoid migration through the blood-brain barrier, potentially providing relevant information for further therapeutic decision making.

In the contest of the gold standard approach, the advantages of FCM over IHC include an excellent sensitivity, a short execution time, an objective quantification of antigen intensity of expression, a reliable approach for the evaluation of antigens co-expression on both tumor and tumor-side populations. The presence of necrosis remains a major disadvantage, as well as the requirement of fresh tissue for analysis. Finally, establishing a provisional diagnosis of brain lymphoma, FCM could positively affect the cost-benefit ratio, contributing in a shortening of time-to-treat interval with reduction of hospitalization indirect costs. Prospective studied will establish the synergic and complementary role of these two techniques for a better diagnostic approach and clinical management of brain lesions.