Improved remission rates and survival of patients with cancer may increase the incidence of central nervous system (CNS) disease recurrence.1-4 Before the development of effective chemotherapy regimens, CNS recurrence was relatively uncommon in patients with leukemia and lymphoma. As therapies improved, CNS recurrence rates increased, and CNS involvement occurred in up to 50% of children with acute lymphoblastic leukemia (ALL).3, 5, 6 The subsequent incorporation of CNS prophylaxis into treatment regimens dramatically reduced the rate of CNS recurrence in childhood ALL.7 Until recently, however, CNS prophylaxis in childhood ALL routinely included craniospinal radiotherapy, thereby increasing the risk of long-term neurological side effects.8 Although these risks may vary for adults and those with other hematological malignancies, there has been much debate regarding whether universal CNS prophylaxis is required for patients with leukemia and lymphoma who have a lower risk of CNS recurrence.9, 10 For example, only approximately 5% of patients with histologically aggressive non-Hodgkin lymphoma (NHL) will develop CNS recurrence.10 However, because CNS recurrence is nearly always fatal in these individuals, careful monitoring for symptoms and signs of CNS disease is crucial in patients who have aggressive NHL and risk factors for CNS recurrence.9, 11
Neoplastic meningitis (NM), a particular manifestation of CNS recurrence, results from the infiltration of metastatic cells into the cerebrospinal fluid (CSF) and meninges. NM is also referred to as lymphomatous meningitis or leukemic meningitis in patients with lymphoma and leukemia, respectively. Patients with NM often experience disabling multifocal neurological deficits with a median survival that rarely exceeds 6 months.12 Although NM is nearly always fatal, early diagnosis may allow more treatment options and may increase the time to neurological disease progression and improve prognosis.12 Because intrathecal chemotherapy can penetrate a depth of only 1 to 2 mm, later diagnosis of NM when the depth of neoplastic coating exceeds 4 mm makes intrathecal chemotherapy much less likely to be effective.13 If diagnosed late, as occurs in a large number of cases, patients are more likely to have multiple fixed neurological deficits, and palliative care is often the only option.14, 15
Until recently, the detection of NM has relied on standard diagnostic tools, including periodic follow-up for clinical manifestations, magnetic resonance imaging (MRI), and cytological evaluation of the CSF.16 However, each method is limited by suboptimal sensitivity and/or specificity. For example, clinical signs and symptoms, which are almost always the first indication of NM, may be difficult to distinguish from symptoms of the primary disease or neurological side effects of treatment.17 Contrast-enhanced MRI of the brain and spine detects NM in only approximately one-half of cases.18 Cytological detection of malignant cells in the CSF is highly specific for NM. However, conventional cytology has a low sensitivity because of a paucity of cells in the CSF19 and morphological similarities that can make it difficult to distinguish benign from malignant cells.20 Depending on the extent and spread of leptomeningeal disease, cytology has detected between 38% (focal disease) and 66% (disseminated disease) of cases found to be positive for NM at autopsy.21 Because of the low detection rate, lumbar punctures are often repeated up to 3 times in patients with suspected NM. However, even after repeated CSF sampling, false-negative cytology reportedly occurs in 10% to 20% of patients with NM.22-24 Improved diagnostic tools leading to earlier treatment could enhance quality of life, broaden treatment options, and improve outcomes, and represent an unmet need. Herein, we review recent studies that evaluated the role of flow cytometry in the diagnosis of NM. Because T-cell lymphoma constitutes only 5% of primary CNS lymphoma, the diagnosis of NM in this subgroup is rare and to our knowledge there are only minimal data that systematically address this entity.25
Sensitivity of Flow Cytometry
Flow cytometry is a highly sensitive and specific technique that when combined with multicolor fluorescent antibody labeling can detect malignant cells in samples with very low cell counts. A review of studies reported over the past 6 years evaluating flow cytometry for the detection of the infiltration of leukemia and lymphoma cells into the CSF demonstrated the superior sensitivity of flow cytometry compared with standard cytology for the detection of CSF infiltration in patients with leukemia or lymphoma.19, 20, 26 The majority of these studies involved patients with newly diagnosed, aggressive NHL with clinical features indicative of an increased risk of CNS involvement (Table 1).19, 20, 26-28 Commonly accepted risk factors for CNS disease include the involvement of ≥ 2 extranodal sites, the testes, or the bone marrow, and elevated serum levels of lactate dehydrogenase (Table 1). In a study of 51 patients with aggressive lymphoma who were at an increased risk of CNS involvement, Hegde et al19 reported that flow cytometry detected an abnormal B-cell population in the CSF of 11 of 51 (22%) patients, whereas only 1 (2%) of these patients was found to have positive CSF involvement by conventional cytology (P = .002) (Table 1). Patients found to be positive by flow cytometry were mostly asymptomatic for CNS involvement, indicating that CNS disease is detectable before the manifestation of clinical symptoms. The 1 patient who was found to be positive by both conventional cytology and flow cytometry had the highest CSF concentration of tumor cells, which constituted 99% of the cells in the CSF.
|Positive FC||Positive CC|
|Study||No.||No.||No.||Patient Population (Inclusion Criteria)|
|Hegde 200519||51||11 (22%)||1 (2%)||High risk of CNS disease:|
|• DLBCL with either ≥2 extranodal sites and ↑ LDH or bone marrow involvement|
|• AIDS-related lymphoma|
|Di Noto 200826||42||11 (26%)||4 (9.5%)||High risk of CNS disease:|
|• DLBCL, BV-MCL, B-LBL, or T-LBL with either ≥2 extranodal sites and ↑ LDH or bone marrow involvement|
|Quijano 200920||123||27 (22%)||7 (6%); suspicious in 3 (2%)||High risk of CNS disease:|
|• Aggressive B-NHL with either infiltration of extranodal sites (testis, breast, paranasal sinus, and/or bone marrow), neurological symptoms, or ↑ LDH|
|Bromberg 200727||219||44 (73%)a||19 (32%)||Patients who underwent CSF evaluation for hematological malignancy|
|Schinstine 200628||32||19 (59%)||Repeat cytology: 9 (47% of 19)||Patients with an initial CSF diagnosis of “atypical” or “suspicious”|
Two subsequent studies have documented a 2-fold to 3-fold higher detection rate with flow cytometry compared with conventional cytology in patients with newly diagnosed, aggressive NHL and a high risk of CNS involvement (Table 1).20, 26 In an Italian study of 42 patients, Di Noto et al26 observed CSF involvement by flow cytometry in 11 (26%) patients, only 4 (9.5%) of whom were found to be positive with conventional cytology. In what to our knowledge was the largest series to date of high-risk patients, which included 123 patients from 29 hospitals in Spain (Table 1), flow cytometry was positive in 27 (22%) patients, whereas conventional cytology was positive in 7 (6%) and suspicious in another 3 (2%) patients.20 Six of the 7 patients found to be positive by conventional cytology were also positive by flow cytometry; the 1 discrepant case was found to be a false-positive result on repeat cytology. It is interesting to note that the cell count and the percentage of neoplastic B cells identified by flow cytometry were higher in patients who had positive cytology and flow cytometry results compared with those found to be positive by flow cytometry alone (P < .0001).20 A CSF sample was considered positive by both conventional cytology and flow cytometry only when neoplastic B cells accounted for > 20% of CSF cells, whereas flow cytometry was considered positive at much lower levels (P < .0001) of infiltration.20 A more recent study supported the ability of flow cytometry to detect malignant cells in CSF specimens with low cellularity. In 2 cases, the diagnosis of NM was made with < 100 events in the cell population of interest.29
Similar detection rates by flow cytometry were observed in a retrospective review of CSF samples from 219 patients with leukemia/lymphoma at a single center (Table 1).27 CNS disease was diagnosed in 60 (27%) of these patients by either flow cytometry, cytomorphology, or both. On the first CSF sampling, 44 (73%) of the 60 diagnosed patients were found to be positive by flow cytometry, but only 19 (32%) were found to be positive by conventional cytology. Compared with patients who were found to be positive by flow cytometry alone, patients who had positive cytology results had a higher incidence of clinical symptoms of CNS disease (58% vs 95%, respectively) and CSF pleocytosis (25% vs 84%, respectively). The ability of flow cytometry to detect NM in its early stages provides the impetus to use flow cytometry to diagnose CSF involvement before the onset of clinical symptoms. In turn, this may improve treatment outcomes.
Flow cytometry can also be useful in diagnosing patients whose initial cytology results are ambiguous. In a study of 32 patients with “atypical” or “suspicious” CSF cytology who were followed for 1 year, subsequent cytology and/or flow cytometry analysis confirmed malignancy in 19 (59%) patients.28 Cytology findings were positive in only 9 (47%) of the 19 patients who were found to be positive by flow cytometry. In a more recent series, 29 patients with hematological malignancies had a cytology result that was scored as “doubtful.”30 Flow cytometry raised the diagnostic sensitivity from 73% (cytology) to 96% (flow cytometry), the specificity from 94% to 97%, the positive predictive value from 88% to 96%, and the negative predictive value from 76% to 97%.30 Flow cytometry can thus help establish the diagnosis in suspicious or uncertain situations and eliminate the need for repeated CSF sampling, thereby decreasing cost and patient discomfort.
In general, the detection of malignant cells using conventional cytology requires that neoplastic cells constitute a minimum of 5% of the cells in a sample.20, 25 In contrast, flow cytometry is capable of detecting malignant cells when they comprise as few as 0.2% of the total cell count (median, 7 % [range, 0.2%-99%]).19 Flow cytometry can detect monoclonal B-cell populations in samples with very low numbers of B cells and is accurate even in the presence of normal polyclonal B cells.19 Antigen expression intensity, monotypic light chain expression, and differential antigen expression patterns can be analyzed simultaneously with cell size and granularity using forward and side scatter, enabling the detection of lymphoma cells in samples that contain mixtures of monoclonal and polyclonal B cells.19 For example, it is particularly well suited for the detection of aberrant antigen expression patterns in B cells (such as the partial loss or absence of CD19, CD20, and CD22 expression; the presence of light chain clonal excess; and aberrant overexpression of antigens such as CD5 and CD10), which would not be expected in reactive CSF B-cell populations.
False-negative results are rare with flow cytometry. No incidences of CNS disease that was found to be positive with cytology and negative with flow cytometry were reported in 3 of the 5 studies we reviewed that evaluated flow cytometry.19, 26, 28 Quijano at al20 observed 1 case that was initially found to be positive with cytology and negative with flow cytometry. Repeat cytology indicated that this case was false positive by conventional cytology. However, Bromberg et al27 reported that 7% (4 of 60) of patients with confirmed CSF infiltration were found to be positive by cytology but negative by flow cytometry. Thus, false-negative results that can occur with flow cytometry may be detectable by cytology. For this reason, the National Comprehensive Cancer Network guidelines recommend that conventional cytology be performed in conjunction with flow cytometry when diagnosing CSF leukemia or lymphoma.31
Application of Flow Cytometry and Technical Limitations
Flow cytometry screening for CSF infiltration of lymphoma has several technical requirements. First, a CSF sample of sufficient volume must be obtained via lumbar puncture. A total CSF volume of approximately 2.0 mL appears to be sufficient for flow cytometry,20 whereas a minimum of 10 mL has been recommended for accurate cytology results.32 Because CSF samples must be processed within 1 hour of CSF sampling to avoid cell deterioration, flow cytometry facilities and experienced technicians must be readily available.33 The samples are collected without anticoagulant and transported to the laboratory as quickly as possible. The samples are typically transferred to the laboratory without fixative; however, the use of a fixative (eg, TransFix/ethylenediamine tetraacetic acid [EDTA]; Immunostep SL, Salamanca, Spain) may help prevent cellular degradation in CSF samples for hours and even days.20 In the study by Quijano et al, a fixative permitted the overnight shipment of CSF samples from 29 different hospitals to a central flow cytometry laboratory.20
To the best of our knowledge, no consensus exists regarding the optimal antibody selection and combination. Most investigators will, based on the number of cells provided, label from 1 to 3 tubes, each containing a cocktail of antibodies.19 Previous histologic diagnosis and the patient's clinical history help guide the selection of antibodies. In addition, to our knowledge no consensus has been reached regarding the best combination of antibodies with which to detect NHL in CFS, but listed in Table 2 are various 3-color to 9-color combinations that have been used for patients with either B-cell or T-cell malignancies.19, 20, 26-28 Quijano et al20 reported that the use of a standard 6-color combination independent of neoplastic B-cell phenotype was equal, or superior, to approaches that use combinations of reagents adapted to the specific phenotype of malignant cells (Fig. 2) (Table 2). 1
|Study||Antibody Panel||Criteria for Positive FC|
|Hegde 200519||CD19, CD20, CD22, CD38, CD45, CD5, or CD10 paired with κ and λ light chain antibodies (sIg), based on systemic diagnosis||Not given|
|Schinstine 200628||3-color and 4-color combinations:||Not given|
|• T cells: CD3/CD4/CD7/CD25; κ and λ sIgs paired with CD19, CD20, CD22, CD38, CD45, CD5, and CD10 based on systemic diagnosis|
|Bromberg 200727||Same as Hegde 200519||Positive: >25 events meeting malignant criteria|
|Suspicious: <10 events and <25 events meeting malignant criteria|
|Negative: <10 events meeting malignant criteria|
|Di Noto 200826||4 colors:||>30 events meeting|
|• B-cell NHL: CD20/CD10/CD45/CDl9 or sIgκ/sIgλ/CD45/CDl9 if insufficient cell number||malignant criteria|
|• T-cell NHL: CD5/CD3/CD7/CD2 6 colors:|
|• B-cell NHL: κ/λ/CD45/CD5/CD19/CD20|
|• T-cell NHL: CD4/CD7/CD45/CD5/CD8/CD3|
|Quijano 200920||6 colors with 9 antibodies (CD8-sIgλ/CD56-sIgκ/CD4-CD19/CD3/CD20/CD45) and scatter data||>10 events meeting malignant criteria|
|• Monoclonal vs polyclonal B cells: sIgκ/sIgλ or abnormal forward scatter/side scatter/CD19/CD20 pattern|
In addition to differences in the antibody panels used, flow cytometry results are defined and interpreted differently between studies (Table 2). A sample of typical data output from a flow cytometer is presented in Figure 2.19 Interpreting this type of data requires the definition of the minimum number of aberrant events deemed necessary for a positive result. Although some reports have not defined this minimum,19, 28 others vary widely with regard to their criteria. For example, DiNoto et al26 used a threshold of at least 30 events or cells to define a positive result, whereas a minimum cluster of 10 events was considered positive by Quijano et al.20 The variation in the protocols and criteria used to indicate positive results limits the interpretation of flow cytometry data across studies. Standard flow cytometry protocols with accepted antibody panels and uniform definitions of positivity are needed to advance the state of the art.
Flow Cytometry for the Detection of Intraocular Lymphoma
Flow cytometry of vitreous fluid samples has proven useful in the diagnosis of intraocular lymphoma, a subset of primary CNS lymphoma.34 In a retrospective study of 20 patients with suspected intraocular lymphoma on slit lamp examination, 7 of 10 patients ultimately diagnosed with intraocular lymphoma by vitrectomy had positive flow cytometry findings, whereas only 3 had positive cytology.35 Flow cytometry immunophenotyping also distinguished patients with uveitis or intraocular infections (n = 10) from those with intraocular lymphoma (n = 10) by indicating the lack of a monotypic cell population and, in some cases, by abnormal CD4:CD8 ratios and a high percentage of activated cells.35 A subsequent series of 28 patients with suspected intraocular lymphoma, 15 of whom were tested by flow cytometry, confirmed the ability of flow cytometry to distinguish intraocular lymphoma from immunologically mediated uveitis.36 Vitreous cytology was positive in only 4 of the 13 (31%) patients who had a final diagnosis of intraocular lymphoma. No false-positive cases were noted.36 The sensitivity of flow cytometry varied with the marker evaluated, with CD22 being the most accurate marker of intraocular lymphoma. A CD22 level > 20% indicated an 88% probability of intraocular lymphoma, whereas levels < 20% indicated a 71% likelihood that the disease was not present.36 Overall, flow cytometry had higher positive and negative predictive values compared with cytology. Despite these findings, cytology is still considered the gold standard for diagnosing intraocular lymphoma.34 If readily available, however, flow cytometry of vitreous fluid performed with cytology has the potential to enhance diagnostic accuracy significantly.
In patients with aggressive NHL and risk factors for CNS recurrence, careful monitoring for CNS disease is imperative. The specificity of both cytology and flow cytometry approaches 100%; false-positive findings are rare. The sensitivity of flow cytometry is several-fold higher than that of conventional cytology for the detection of infiltration of leukemia or lymphoma into the CSF. Because flow cytometry can detect CNS disease before the manifestation of clinical symptoms and CSF pleocytosis, the routine use of flow cytometry and cytology may permit the earlier detection of NM and thus afford more treatment options and possibly improve prognosis. However, there remains some uncertainty regarding the optimal protocol for detecting CSF lymphoma, and efforts to standardize the procedure and data interpretation are needed to permit broader clinical application. Future directions in the diagnosis of leptomeningeal lymphoma may include techniques such as microRNA assays, which in 1 report demonstrated a sensitivity and specificity of 96% and 97%, respectively.37