How to cite this article: Song JY, Filie AC, Venzon D, Stetler-Stevenson M, Yuan CM. Flow cytometry increases the sensitivity of detection of leukemia and lymphoma cells in bronchoalveolar lavage specimens Cytometry Part B 2012; 82B: 305–312.
Article first published online: 26 JUL 2012
Published 2012 Wiley Periodicals, Inc.
Cytometry Part B: Clinical Cytometry
Volume 82B, Issue 5, pages 305–312, September 2012
How to Cite
Song, J. Y., Filie, A. C., Venzon, D., Stetler-Stevenson, M. and Yuan, C. M. (2012), Flow cytometry increases the sensitivity of detection of leukemia and lymphoma cells in bronchoalveolar lavage specimens . Cytometry, 82B: 305–312. doi: 10.1002/cyto.b.21033
This article is a US government work and, as such, is in the public domain in the United States of America.
- Issue published online: 6 SEP 2012
- Article first published online: 26 JUL 2012
- Manuscript Accepted: 31 MAY 2012
- Manuscript Revised: 8 MAY 2012
- Manuscript Received: 7 DEC 2011
- Intramural Research Program of the NIH, NCI
- bronchoalveolar lavage;
- flow cytometry;
- hematolymphoid neoplasm
Recent studies have definitively determined that flow cytometry (FC) is significantly more sensitive than cytomorphology (CM) in detection of hematolymphoid neoplasms (HLNs). However, its utility in paucicellular bronchoalveolar lavage (BAL) specimens has not been established.
FC was performed on BAL specimens submitted from 44 patients with a prior diagnosis of HLN. Panels chosen were based upon cellularity of specimen and patient history. FC results were compared with concurrent CM evaluations.
All 44 BALs were deemed satisfactory for FC and yielded informative results that assisted in diagnosis. Diagnoses included 22/44 B-cell neoplasms, 16/44 T-cell neoplasms, four/44 myeloid neoplasms, and two/44 plasma cell neoplasms. Overall concordance was demonstrated between FC and CM in 77% (34/44) of cases. In nine/44 cases (20%), one technique (FC or CM) clearly detected malignant cells when the other did not. FC was more sensitive than CM in detecting a HLN in eight/nine discordant cases. In only one case (one/44, 2%) were malignant HLN cells suspected by CM, but not identified by FC (one/44, 2%).
We demonstrate, in the largest series published to date, that FC can be performed on BAL specimens. FC is indicated in evaluation of BAL for HLN and improves sensitivity of detection of HLN over CM alone. An integrated FC and CM approach is superior to either technique alone in diagnostic evaluation of BAL. Published 2012 Wiley Periodicals, Inc.
Flow cytometry (FC) analysis is clearly indicated in the diagnosis of hematolymphoid neoplasms (HLNs) (1). Nevertheless, diagnosis of HLN is frequently based upon evaluation of paucicellular specimens such as fine needle aspirates and body fluids [i.e., cerebrospinal fluid (CSF), vitreous humor, and effusions]. When sufficient numbers of cells are present in these paucicellular samples, FC can be more sensitive than morphology. FC has been demonstrated to increase sensitivity of detection of HLN in fine needle aspirates and to assist in both detection and diagnostic subclassification of lymphoma in these specimens (2–4). Involvement of the CSF by hematopoietic malignancies may also be difficult to document by morphology alone. In studies assessing FC in evaluating CSF for involvement by non-Hodgkin lymphoma, FC was significantly more sensitive than cytology alone in disease detection and provided prognostication in specific circumstances (5–8). FC is also useful in identifying CSF involvement with leukemia and increases the detection rate over cytology alone (6, 8, 9). Thus, FC is crucial in the evaluation of CSF for hematolymphoid malignancies (5–7, 10). Bronchoalveolar lavage (BAL) allows evaluation of diffuse pulmonary infiltrates of infectious or neoplastic etiology with minimal complications; as such, it is a relatively useful and safe diagnostic tool for detection of disseminated HLN affecting the lung (11–14). Although cytomorphology (CM) is the traditional means by which a BAL is evaluated, CM may not be the optimal modality to differentiate reactive lymphoid cells from neoplastic cells, especially in the setting of low grade lymphoma/leukemia. Previous studies have demonstrated the utility of FC evaluation of normal lymphocyte subsets in BAL specimens from patients with non-neoplastic processes (e.g., sarcoidosis) (15–17). Although isolated reported case studies exist regarding FC detection of HLN in BAL (18), little is generally known about the utility of FC in analysis of BALs. In this study, we performed diagnostic FC evaluation of 44 BAL specimens to determine if specimens were satisfactory for FC (e.g., adequate cellularity) and if sensitivity of detection of HLN cells could be enhanced.
MATERIALS AND METHODS
We reviewed 44 cases of BALs submitted from patients with a confirmed diagnosis of a HLN (according to the 2008 WHO classification) (19) for diagnostic FC between 1995 and 2010. Only one specimen per patient is included in this study. Cases were distributed as follows: four/44 (9%) patients with myeloid neoplasm (three acute myeloid leukemia (AML), one chronic myeloid leukemia in myeloid blast crisis); 22/44 (50%) patients with B-cell neoplasm (six chronic lymphocytic leukemia, two hairy cell leukemia, one Burkitt lymphoma, five diffuse large B-cell lymphoma, three follicular lymphoma, one extranodal marginal zone lymphoma of mucosa associated lymphoid tissues type, two B lymphoblastic leukemia, one lymphomatoid granulomatosis, and one B-cell lymphoma-NOS); 16/44 (36%) patients with T-cell lymphoma (10 adult T-cell lymphoma/leukemia, two peripheral T-cell lymphoma-NOS, three cutaneous T-cell lymphoma, and one gamma-delta T-cell lymphoma), and two/44 (5%) patients with a plasma cell dyscrasia (Table 1). The diagnosis of a HLN was confirmed by review of original diagnostic biopsy as well as analysis of blood and bone marrow. Internal diagnostic review was performed by tissue histology alone (23/44, 52%), FC alone (three/44, 7%), or by both FC and tissue histology (18/44, 41%). All patients had a lung lesion detected by radiographic methodology that was suspicious for a neoplastic process. In all 44 cases, concurrent CM evaluation was also available. All specimens were submitted as part of routine protocol evaluation at the National Cancer Institute, NIH (Bethesda, MD). All patients signed IRB-approved informed consents to be evaluated.
|Acute myeloid leukemia||3 (7)|
|Chronic myeloid leukemia, blast crisis||1 (2)|
|Total myeloid cases||4 (9)|
|B-cell lymphoma, not otherwise specified||1 (2)|
|Chronic lymphocytic leukemia/Small lymphocytic lymphoma||6 (14)|
|Hairy cell leukemia/lymphoma||2 (5)|
|Burkitt lymphoma||1 (2)|
|Diffuse large B-cell lymphoma||5 (11)|
|Follicular lymphoma||3 (7)|
|Marginal zone lymphoma, MALT type||1 (2)|
|B-lymphoblastic leukemia||2 (5)|
|Lymphomatoid granulomatosis||1 (2)|
|Total B-cell cases||22 (50)|
|Adult T-cell leukemia/lymphoma||10 (23)|
|Peripheral T-cell lymphoma, not otherwise specified||2 (5)|
|Cutaneous T-cell lymphoma||3 (7)|
|Gamma-delta T-cell lymphoma||1 (2)|
|Total T-cell Cases||16 (36)|
|Plasma cell myeloma||2 (5)|
|Total Plasma Cell Cases||2 (5)|
|Total Number of Cases||44 (100)|
The BAL specimens were divided for CM and FC. For FC the specimens were stained within 24 h (h) of collection with a panel of antibodies. When present, erythrocytes were lysed by incubating with lysing solution (150 mM NH4Cl, 10 mM KHCO3, and 0.1 mM EDTA) for 10 min at room temperature (maintained at 21–23°C) at a ratio of 1:9 (volume of sample: volume of lysing solution). Specimens were then washed with phosphate buffered saline before determining cell number. Specimens were stained for 30 min at room temperature (maintained at 21–23°C) with a cocktail of four antibodies (antibody concentrations used per manufacturer's recommendations) according to Clinical Laboratory Standards Institute document H43-A recommendations (20) and as previously described (21). Antibody combinations were chosen based on the number of cells, diagnosis, and previous immunophenotypic data. All cells were fixed in 1.0% paraformaldehyde post staining and stored at 4°C for up to 12 h before acquisition. Cellularity was determined manually, using a hemocytometer, and Trypan Blue exclusion was used to determine viability of the cells. Specimens were acquired with six-parameter four-color FC on the FACSCalibur (BD Biosciences, San Jose, CA) using CellQuest Pro software (sensitivity of fluorescent detectors monitored using standard beads according to the manufacturer's recommendations). No BAL cases were excluded based on limited cellularity. FC was performed on every BAL specimen received; for the majority of the BAL specimens, two–four tubes were prepared for FC.
Data (collected in list mode) were analyzed with CellQuest Pro (BD Biosciences, San Jose, CA) or FCS Express version 3 (De Novo Software, Los Angeles, CA). For analysis, cell populations were gated according to characteristic forward and side scatter properties, in conjunction with antigen back-gating. Normal lymphoid cells within specimens served as internal positive and negative controls (e.g. B cells served as negative controls for T-cell directed antibodies) as well as for antibody intensity. Bright expression was defined as higher than normal while dim expression was defined as lower than that observed in normal lymphocytes of the same lineage, in concordance with the 1997 U.S.-Canadian consensus guidelines (22). Analysis of FC data for involvement by hematolymphoid neoplasia was performed by a hematopathologist.
An aliquot from the fresh BAL specimens were received and refrigerated. All specimens were processed within 12 h of collection, with the vast majority of specimens processed within 2–4 h of receipt. Cytospins from each sample were prepared by centrifugation of undiluted specimen using a Cytospin 4 centrifuge (Thermo Fisher Scientific, Waltham, MA) at 500 revolutions per minute for 5 min (55g). The slides were then stained with Diff-Quik (Dade, Aguada, Puerto Rico) and Papanicolaou. Samples were concentrated or diluted in RPMI-1640 (Gibco BRL, Grand Island, NY) when required.
Each CM case was independently reviewed by a single cytopathologist and classified as positive (disease present), negative (disease absent), or atypical (suspicious cells present, but diagnostic criteria for HLN are not met). Likewise, each FC case was independently reviewed separately by a single hematopathologist and classified similarly.
All 44 BAL specimens received in the laboratory were processed for FC. No cases were excluded due to low viability or cellularity. FC detected HLN cells in BAL specimens, providing definitive immunophenotypic evidence of leukemia/lymphoma in 25/44 (57%) specimens and was able to rule out involvement by a HLN in the remaining cases (19/44, 43%). CM evaluation was available for all BAL specimens studied and a HLN was detected in 18/44 cases (41%) by CM. In six cases CM was atypical but non-diagnostic. Only reactive normal appearing cells were observed in the CM negative cases with no definitive evidence of infectious or other etiology.
FC was useful in evaluating BAL for involvement by myeloid neoplasms, detecting neoplasia in four/four (100%) of myeloid cases (Table 2). Between 0.05 and 80% of the mononuclear cells were neoplastic. In case 1, over 70% of the cellularity was composed of immature myeloid cells, expressing bright CD13, moderate CD33, and CD34 negative, along with 4% blasts expressing CD13 and CD34, consistent with involvement by AML. In case 2, CD34 positive, CD56 positive blasts with dim to negative CD45 expression were detected, consistent with the patient's diagnostic AML blast immunophenotype. In case 3, approximately 80% of the cellularity was composed of CD14 positive CD13 positive monocytic cells, along with 3% CD34-positive blasts. These findings, supported involvement by the patient's acute myelomonocytic leukemia. In case 4, a small population of CD34 and CD56 positive myeloblasts was detected, consistent with the patient's CML in myeloblast crisis.
|Case #||Known HLN||Initial diagnosis||FC diagnosis (% HLN Cells)||CM diagnosis|
|1||AML||Tissue, FC||Positive (70%)||Positive|
|3||AML (myelomonocytic)||Tissue||Positive (80%)||Positive|
|4||CML (blast crisis)||Tissue, FC||Positive (0.5%)||Negative|
|7||B-cell, NOS||Tissue, FC||Negative||Negative|
|10||CLL||Tissue, FC||Positive (70%)||Positive|
|11||CLL||Tissue, FC||Positive (19%)||Positive|
|12||CLL||Tissue, FC||Positive (31%)||Positive|
|13||CLL||Tissue, FC||Positive (10%)||Positive|
|19||DLBCL||Tissue, FC||Positive (0.1%)||Positive|
|32||ATLL||Tissue, FC||Positive (86%)||Positive|
|33||ATLL||Tissue, FC||Positive (10%)||Atypical|
|35||ATLL||Tissue, FC||Positive (26%)||Atypical|
|37||CTCL||Tissue, FC||Positive (80%)||Positive|
|41||PTCL||Tissue, FC||Positive (19%)||Positive|
|Plasma cell cases|
FC detected B-cell neoplasia in eight/22 specimens from patients with a history of a B-cell lymphoproliferative disorder (Table 2) with neoplastic cells comprising between 0.03 and 70% of the cells. These included six CLL (cases 9–14), one DLBCL (case 19), and one HCL (case 25). Abnormalities detected by FC included monotypic surface light chain expression (seven/eight cases), absence of surface light chain expression (one/eight cases), and abnormal antigen expression (seven/eihgt cases). Abnormal antigen expression included aberrant expression of CD5 on B-cells (in six/six CLL cases), diminished expression of CD20 on B-cells (five/six CLL, CD20 not performed in one/six CLL), diminished expression of CD22 on B-cells (three/six CLL, CD22 not performed in three/six CLL), bright expression of CD20, bright expression of CD11c, and aberrant expression of 103 on B-cells (one/one HCL case).
Involvement of BAL with T-cell neoplasia was demonstrated by FC in 12/16 (75%) cases, with neoplastic cells comprising between 4% and 86% of cells (Table 2). Malignant T cells were detected based upon abnormal intensity of antigen expression (six/12 cases), and failure to express normal T-cell antigens (12/12 cases). Representative FC cases of BAL are shown (Figs. 1 and 2). Figure 1 demonstrates involvement by a B-cell lymphoma immunophenotypically consistent with the patient's diagnosis of CLL/SLL. Figure 2 demonstrates involvement by a T-cell lymphoma, immunophenotypically consistent with the patient's diagnosis of ATLL.
FC detection of abnormal surface antigen expression and intracellular monoclonal light chain expression was helpful in diagnosis of BAL involvement with plasma cell neoplasms. Monoclonal plasma cells were detected in one/two plasma cell cases by FC (3% of cells) (Table 2). By FC, the plasma cells expressed bright CD38, dim CD45, and restricted intracytoplasmic kappa light chain. This was consistent with the patient's previous history of plasma cell myeloma.
Excellent correlation was demonstrated between FC and CM (Table 3). We found 34/44 cases (77%) were concordant. In nine/44 (20%) cases, one method was superior to the other in detecting malignant cells (Table 3). FC was more sensitive than CM, detecting HLN in eight/nine discordant cases (three CM negative cases and five CM atypical cases, including one CML in blast crisis, one CLL, one HCL, and five ATL). CM detected HLN in one FC negative CTCL case. In only one/44 cases (2%) were malignant HLN cells suspected, but not definitively identified by either CM or FC; this was a case of DLBCL, where atypical cells were suspected by CM but not detected by FC. Both FC and CM were equal in their detection of the two cases of BAL where involvement with plasma cell neoplasm was suspected. FC appeared to be more sensitive than CM in detection of myeloid malignancies (100% positive results by FC compared to 75% by CM alone) and B-cell leukemia/lymphoma (36% positive results by FC compared to 27% by CM alone). FC also appeared more sensitive than CM in the detection of T-cell lymphoma/leukemia, detecting diagnostically aberrant T-cells in 12/16 (75%) cases, while CM positively identified only eight/16 (50%) cases. Among our entire set of cases, the eight/nine cases that were positive by FC and not by CM, in comparison to the one/nine cases positive by CM and not by FC shows that FC is significantly more sensitive than CM (P = 0.039 by the exact two-tailed McNemar's test); however, this is largely due to the contribution of T-cell cases, and a study population with different proportions of HLNs might have a different outcome.
|Cytology (+)||Cytology Atypical|
|Flow cytometry (+)||17 5 CLL, 1 DLBCL, 1 PTCL, 1 CTCL, 5 ATLL, 1 PCM, 3 AML||5 (4 ATLL, 1 HCL)|
|Flow cytometry (−)||1 (CTCL)||1 (DLBCL)|
Clinicians rely on BAL specimens as an adjuvant to radiologic studies when the differential diagnosis includes neoplastic versus reactive lesions. These specimens are typically sent for cytologic examination; however, differentiating reactive from neoplastic lymphoid cells on morphology may be difficult, especially when the neoplastic cells are sparse. This difficulty in differentiating reactive from neoplastic lymphoid cells is also encountered when evaluating similarly paucicellular specimens such as CSF or fine needle aspirates. FC has been demonstrated to be extremely useful in diagnostic assessment of CSF and fine needle aspirates; therefore, a potential role for FC in BAL testing is clearly indicated. The utility of FC in diagnosis of HLN in BAL specimens, though reported, is not well established, as the literature is limited to isolated case reports and a single recently published series that included 31 BALs from patient's with suspected HLN. Cases positive for HLN were primarily of B-cell lineage with only one AML and four T-cell lymphomas detected (23,24). In this study, FC was able to detect neoplastic leukemia/lymphoma cells in 25/44 (56%) of BAL specimens. Furthermore, an unusually high proportion of T-cell lymphoproliferative disorders (16/44 cases, Table 1) was studied.
FC definitively diagnosed B-cell lymphoma/leukemia in eight/22 (36%) of BAL studied by demonstrating the presence of B-cells with monoclonal light chain expression, lack of surface light chain expression or abnormal antigen expression. FC detected neoplastic T-cell populations in BAL specimens from 12/16 patients with T-cell lymphoma or leukemia. The primary immunophenotypic findings were abnormal levels of antigen expression (abnormally dim or bright) and failure to express normal T cell antigens. 10/12 positive T-cell cases were ATL, a neoplasm rarely encountered in the typical FC laboratory. The immunophenotypic features of this disease; however, mimic those of CTCL and the criteria for flow cytometric diagnosis are generally applicable to T-cell neoplasia (25). FC was also useful in evaluation of BAL specimens for abnormal plasma cells (based upon abnormal surface antigen expression and monoclonal intracellular light chain expression) and myeloid neoplasia (presence of blasts and cells with immature myeloid features and abnormal antigen expression). In all of our cases, the patient's original HLN diagnosis, as well as the immunophenotype (either from prior tissue immunohistochemistry or from FC), were known. The availability of this information was advantageous, as immunophenotypic panels could be selected for each BAL specimen specifically with the patients known immunophenotype in mind. For the majority of BAL specimens, two–four tubes were processed for FC. This is in contrast to peripheral blood, where greater than 10 tubes are routinely processed for FC. Thus, this approach optimizes the immunophenotypic information obtained from the BAL, possibly leading to increased sensitivity.
We observed a high concordance between FC and CM results (77%), despite the fact that a greater proportion of our cases were composed of diagnostically challenging T-cell lymphoproliferative disorders (36% of cases in this study in comparison to 13%, as reported by Cesana et al), as well as myeloid and plasma cell neoplasms (Table 1). Our concordance of 77% of total cases appears to be slightly higher than that previously described (26), although a direct comparison is difficult to make, given the different study design, and the observation that only a minor proportion of the BAL cases reported by Cesana et al. (six/31 cases) had detectable disease by FC. FC was more sensitive than CM for the detection of malignant cells in BAL fluids, detecting HLN in eight CM negative or atypical cases. CM detected HLN in one FC negative case, indicating that maximal sensitivity is achieved when utilizing both FC and CM testing concurrently (Table 3), resulting in an approximate 20% increase in sensitivity.
The presence of neoplastic cells in leukemic patients raises the issue of possible peripheral blood contamination. This scenario, however, equally affects both CM and FC (as well as molecular) evaluation of BAL specimens and thus does not impact solely upon the utility of FC in diagnosis. Rather, interpretation of results achieved by any method of testing is potentially affected by the possibility of peripheral blood contamination in the specimen and should be addressed in the diagnostic report. Examination of a concurrent peripheral blood specimen in situations suspected to have extensive peripheral blood involvement is an additional strategy that may be helpful. BAL samples can also contain debris, which may produce artifactual staining patterns; however, different strategies can be used to help deal with these issues. Receipt of a fresh specimen, and processing within a few hours of collection to maximize viability is optimal. The use of FSC-Area and FSC-Height as parameters can be used to include only single events and exclude clumps and doublets (doublet discrimination). The inclusion of CD45 can be very helpful in “cleaning up” the sample in terms of gating and removing nonrelevant events. Mature lymphoid cells are nearly always present in these samples, and examining CD45 with forward scatter, as well as CD45 with side scatter, can help to identify nonrelevant smaller CD45 negative events that can be excluded from the analysis. A time parameter can also be included as a time gate can be drawn to exclude any “noise” events collected if a tube should run dry.
With such a high concordance observed between FC and CM and with cost considerations becoming an increasing concern for most pathology practices, one might be tempted to believe that performing CM evaluation alone in BALs, as is the traditionally accepted and standard practice, may be sufficient. However, the clinical question that is commonly being asked is whether the sample from the lung lesion represents a reactive/infectious or malignant process. The treatment decision stemming from the answer to that question, administration of antimicrobials versus chemotherapy, represents an critical fork in the clinical decision-making road. Thus, this is an important question to answer correctly and accurately; the 20% decrease in sensitivity that would result from testing with a single modality could be a serious issue. It may be possible to first screen with CM for obvious HLN and then only perform FC on CM negative or atypical cases. This would reduce the number of FC evaluations from 44 to 26, thus cutting costs. Studies on the preservation of viability of neoplastic cells in BALs over time would be necessary before such a delay could be considered. As the cost of misdiagnosis far outweighs the expense of a diagnostic test, a combination of FC and CM is recommended until the issue of specimen stability is addressed.
Several factors may contribute to the greater sensitivity of FC. Immunophenotyping is especially useful when non-neoplastic cells undergo reactive changes making them appear morphologically atypical (27,28) and when low grade lymphoproliferative disorders exhibit minimal cytologic atypia. In addition FC has excellent sensitivity in detecting the neoplastic population when the known HLN has a distinct immunophenotype, such as detection of HCL, in Patient 25 (Table 2). Similarly in Patients 27–36, all of the patients had a known history of ATLL, which also has a distinct immunophenotype (i.e., dim CD3 and positive for CD25). These are five cases in which FC was “positive” while CM was negative or atypical. A recent article by Shao et al. found that FC was highly sensitive in ATLL, detecting as low as 0.29% neoplastic cells in peripheral blood using multiparameter flow cytometric analysis (25). This further underscores the sensitivity of FC in HLN with a distinct immunophenotype. T-cell lymphomas present a more difficult diagnostic challenge in that multiple antibodies need to be examined for loss or diminished expression of T-cell antigens. However, in instances where the HLN has a characteristic immunophenotype that can be exploited by specific T-cell antibody combinations in a few tubes, FC is successful in detecting HLN. FC can identify very small populations of atypical T-cells in other compartments, such as the peripheral blood and bone marrow (25, 29, 30). Until recently, there was no definitive way of determining T-cell clonality by FC. With the availability of antibodies directed against the V region of the T-cell receptor (TCR) β chain (Vβ), FC evaluation has proven its utility in determining whether the T-cells are clonal (31, 32). Furthermore, T-cell clonality by V-beta can be utilized in low cellularity specimens (such as CSF and fine needle aspirates) using a combination of antibodies in a single, patient-specific TCR V-beta cocktail to delineate the atypical T-cells, as long as an initial full TCR V-beta analysis was performed in a diagnostic/screening specimen. The immunophenotypic information from the initial screening specimen informs the choice of antibodies used for the patient-specific TCR V-beta cocktail, and also the gating strategy used to identify the neoplastic T-cells (33).
In conclusion, FC demonstrates a clear utility in BAL evaluation for HLN. The addition of FC testing results in a more sensitive method than CM testing alone for detection of HLN in BAL. Thus, optimal evaluation of BAL fluids should incorporate both CM and FC. Each modality produces valuable, complementary information, that when integrated, is superior in rendering a diagnosis in BAL specimens than either modality alone.
We offer sincere thanks to David Liewehr (Biostatistics and Data Management Section, National Cancer Institute) for his expertise and assistance with statistical analysis. The authors have no conflicts of interest to declare. This work was supported by the Intramural Research Program of the NIH, NCI.
- 12006 Bethesda International Consensus recommendations on the flow cytometric immunophenotypic analysis of hematolymphoid neoplasia: Medical indications. Cytometry B Clin Cytom. 2007; 72B(Suppl 1): S5–S13., , , , , , , , , , , ,
- 8Flow Cytometric Immunophenotyping of Cerebrospinal fluid. Hoboken, NJ: Wiley, 2008., , , , , , , , , .
- 11Guidelines of the Papanicolaou Society of Cytopathology for the examination of cytologic specimens obtained from the respiratory tract. Papanicolaou Society of Cytopathology Task Force on Standards of Practice. Diagn Cytopathol 1999; 21: 61–69., , , , , , , , , , , , , , , ,
- 18DNA analysis of bronchoalveolar lavage in the diagnosis of pulmonary lymphoma. Respiration 1995; 62: 359–360., , ,
- 19WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Lyon, France: International Agency for Research on Cancer, 2008., , , , , , , .
- 20Clinical flow cytometric analysis of neoplastic hematolymphoid cells. In: Stetler-Stevenson MAE, Barnett D, editors. Clinical and Laboratory Standards Institute. Approved Guideline-Second Edition. Wayne, Pennsylvania: Clinical and Laboratory Standards Institute, 2005., , , , , , , , , , ,