An approach to diagnosis of T-cell lymphoproliferative disorders by flow cytometry



T-cell lymphoproliferative disorders are among the most challenging diagnoses in hematopathology. Unlike the more common B-cell disorders, in which clonality is often readily discernible by surface immunoglobulin light chain restriction, there is no specific immunophenotypic signature that is diagnostic of a clonal T-cell population. Immunophenotypic criteria that are helpful in the diagnosis of T-cell neoplasms include T-cell subset antigen restriction, anomalous T-cell subset antigen expression, deletion or diminution of one of the pan T-cell antigens, a precursor T-cell phenotype, and expression of additional markers (e.g., CD30, CD20, major myeloid antigens, and TCRγδ). Analysis of the inherent forward and orthogonal light scatter properties of the cell can also provide important diagnostic clues. None of these features is 100% specific, however, for aberrant expression of pan-T antigens may be seen in viral infections, B-cell malignancies, or in reactive changes following administration of certain medications. An increased CD4:CD8 ratio is often observed in Hodgkin's lymphoma. Based on the analysis of 87 neoplastic and 80 control cases, we conclude that flow cytometric features that are most suspicious for malignancy include the loss or markedly dim expression of CD45; complete loss of one or more pan-T antigens; diminished expression of more than two pan-T antigens in conjunction with altered light scatter properties; and CD4/CD8 dual-positive or dual-negative expression (except thymic lesions). Cytometry (Clin. Cytometry) 50:177–190, 2002. © 2002 Wiley-Liss, Inc.

T-cell disorders are a heterogeneous group of lymphoid tumors that can present as adenopathy, a mass involving a variety of organs (e.g., skin, gastrointestinal [GI] tract, and liver), or as leukemia (1–18). Morphologically and clinically, they can mimic both benign conditions and nonhematopoietic malignancies. Therefore, the diagnosis may be difficult and may require the use of multiple sophisticated methodologies (1–18). Flow cytometry plays an important role in diagnosis and classification of malignant T-cell disorders (16, 19–25). In contrast to B-cell populations in which analysis of surface immunoglobulins can determine clonality, there are no specific flow cytometric markers of T-cell clonality or of malignant T cells. Therefore, the flow cytometric analysis of T-cell disorders requires a broad panel of markers and experience, as well as correlation with cytomorphology and relevant clinical and laboratory data, especially molecular studies for T-cell receptor gene rearrangement (1–5, 7–9, 13–30). We present the most useful criteria for identification of abnormal T-cell populations, including CD45 expression, altered expression of pan-T antigens (CD2, CD3, CD5, CD7, with special attention to dual-parameter analysis of CD3 and CD7), T-cell subset restriction (CD4 with respect to CD8), the presence of additional markers (e.g., T-cell receptors [TCRs], blastic markers CD34 and terminal deoxynuleotidyl transferase [TdT], CD30, CD56, CD57), and cellular changes that are reflected by altered light scatter properties. Also portrayed are characteristic immunoprofiles associated with T-cell acute lymphoblastic leukemia/lymphoma (T-ALL), T-chronic lymphocytic leukemia (CLL)/prolymphocytic leukemia (PLL), T-cell large granular lymphocyte leukemia (LGL), adult T-cell leukemia/lymphoma, anaplastic large cell lymphoma (ALCL), hepatosplenic γδ T-cell lymphoma, and natural killer cell (NK) neoplasms.


From September 2000 through September 2001, flow cytometric samples containing abnormal T-cell populations were submitted for this study. The samples were analyzed at IMPATH (New York) and at the Westchester Medical Center (Valhalla, NY). Flow cytometry data were reanalyzed and correlated with cytomorphology and/or molecular studies. All cases without firm morphologic and/or molecular confirmation of a malignant T-cell disorder were excluded. Precursor T-cell lymphoblastic lymphoma/leukemia was defined by blastic cytomorphology and expression of TdT and/or CD34 with cytoplasmic CD3, CD1a, or CD10. Lesions fulfilling the criteria for bilineage (or biphenotypic neoplasms) were not included. T-cell LGL was defined by the presence of LGL lymphocytosis for more than 6 months, which showed mature CD56+ and/or CD57+ T-cell phenotype, CD4 and/or CD8 subset restriction, or the presence of TCRαβ or TCRγδ. NK neoplasms were defined by negative expression of surface CD3, lack of CD4, CD8, TCRαβ, and TCRγδ and the presence of NK-associated markers, CD56 or CD57.

The control population consisted of peripheral blood from patients with acute viral syndromes (5 cases), relative or absolute lymphocytosis (10 cases) or anemia (10 cases), lymph nodes with reactive changes (10 cases), Hodgkin lymphomas (10 cases), diffuse large B-cell lymphomas (DLBCL; 10 cases) and bone marrow from patients during or after chemotherapy for acute myeloid leukemia (10 cases), for staging of B-cell lymphomas (5 cases), and from patients without specific hematologic diagnosis (10 cases, including patients with adenopathy, a monoclonal gammopathy, or reactive granulocytic leukocytosis.

Flow Cytometry Analysis

Heparinized bone marrow aspirate, peripheral blood, and fresh tissue specimens were used for flow cytometry analysis. The specimens were processed within 24 h of collection. A leukocyte cell suspension was obtained from peripheral blood and bone marrow specimens after red blood cell (RBC) lysis with 0.008% ammonium chloride lysing solution, followed by 10 min of centrifugation. The cell pellet was suspended with an appropriate amount of RPMI 1640 media (Gibco, Grand Island, NY). Fresh tissue samples were disaggregated with a sterile blade, followed by passage through a mesh filter (<100 μm). The cells were washed in RPMI media and centrifuged at 1,200 rpm for 10 min. To minimize nonspecific binding of antibodies, the cells were incubated in RPMI media supplemented with 10% heat-inactivated fetal calf serum in a 37°C water bath for 30 min. The samples were then washed with 0.1% sodium azide phosphate-buffer saline buffer (PFA). Viability was assessed using both trypan blue and 7-aminoactinomycin D (Sigma, St. Louis, MO) exclusion assays.

Immunophenotypic analysis was performed on FACSCalibur system instruments equipped with a 15-mW 488-nm air-cooled argon-ion laser supplemented with a 635-nm red diode laser (Becton Dickinson Immunocytometry System [BDIS], San Jose, CA). Because the panel of antibodies evolved during the study and additional antibodies were used later on, specimens were analyzed using three or four-color antibody panels (BDIS; Pharmingen, San Diego, CA; Dako, Carpinteria, CA) that varied between 24 and 30 antibodies. Three and four-color directly labeled antibody combinations consisting of fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyl (PerCP), and allophycocyanin (APC) were used for surface staining of fresh tissue, bone marrow, and peripheral blood cell suspensions. Internal negative controls within each tube and isotype controls for IgG1, IgG2a, and IgG2b were used as negative controls.

The settings and calibration of the instrument fluorescence detectors were monitored according to manufacturer's recommendations, using CaliBRITE beads (BDIS) and the system linearity was evaluated using Sphero Rainbow Beads (Spherotech, Libertyville, IL). Flow cytometry data were collected in list mode and analyzed using CellQuest and CellQuest Pro software (BDIS). A six-gate strategy was employed, using CD45 PerCP versus side scatter (SSC) to characterize the lymphocyte, monocyte, granulocyte, blast, hematogone, and nucleated red cell precursor (erythroid) populations. Five to six-parameter analysis (forward scatter [FSC], SSC, FL1, FL2, FL3, and FL4) or multiparameter data analysis of antibody staining patterns was used to assess specific antigen expression. Various combinations of T-cell–associated antibodies were used. When an abnormal T-cell population was identified on routine analysis, various custom-made combinations of the following antibodies were used: CD1a, CD2, CD3, CD4, CD5, CD7, CD8, CD25, CD30, CD56, CD57, TCRαβ/γδ, and TdT.

The daily instrument setup for intensity and color compensation using CaliBRITE beads and FACSComp software (version 4.0; BDIS) placed the lymphocyte population between 200 and 400 on a linear channel on FSC. Fluorescence beads were used to monitor the consistency of the instrument's optical alignment as part of daily instrument quality control, which assured the reproducibility of the data generated. CD45 expression of benign lymphocytes was designated arbitrarily as bright. Dimmer expression was designated as moderate (up to one decalog below normal) or dim (one to two decalogs below normal lymphocytes). The expression of pan-T antigens and CD4 and CD8 antigens of normal (benign) lymphocytes was designated as moderate and the expression below that level was designated as dim (one or more decalogs below) or negative (identical or with minimal overlap with isotypic control). This grading system is somewhat akin to the guidelines of the U.S.-Canadian Consensus Conference (25). However, we differ by assigning values of bright CD45 and moderate pan-T antigens to normal lymphocytes. The orthogonal (right angle) SSC of normal (benign) lymphocytes was designated as regular; slightly above that level (overlapping) was called increased and overtly above the level of normal lymphocytes (without significant overlap) was designated as markedly increased. The FSC of normal (benign) lymphocytes was designated as regular (cells located between 200 and 400 on a linear scale). Signals located above that level (approximately 400 on an FSC axis) were designated as increased and signals located at approximately 600 were designated as markedly increased. Regular, increased, and markedly increased FSC correspond cytologically to small, medium-sized, and large lymphocytes, respectively.


Of 87 cases analyzed, 30 were bone marrows, 24 were peripheral blood samples, 25 were lymph nodes, and one each of liver, small intestine, spleen, tonsil, soft tissue, omentum, lung, and skin. The age of the patients ranged from 6 to 92 years (average 59.2 ± 23). The percentage of abnormal cells in the sample ranged from 2% to 92% (average 50 ± 23). Sixteen cases were classified as precursor T-cell neoplasms (T-cell lymphoblastic leukemia/lymphoma) and 71 as peripheral (mature/postthymic) T-cell disorders. The age of patients with precursor neoplasms ranged from 6 to 78 years (average 31.3) and patients with peripheral (mature/postthymic) T-cell lesions ranged in age from 19 to 92 (average 66.4) years.

CD45 Expression

Of 87 cases, CD45 expression was bright in 68.9% (60 cases) similar to normal T lymphocytes, moderate (one decalog lower than benign T cells, blastic region on CD45 versus SSC) in 21.8% (19 cases), and dim (overlapping with negative control) in 5.7% (5 cases). Two cases with dim CD45 expression showed a subset of cells with moderate CD45. Of 87 cases, 3 (3.4%) were negative for CD45: all were peripheral T-cell disorders: two were diagnosed as CLL/PLL and the other as peripheral T-cell lymphoma, unspecified. Precursor populations (lymphoblastic lymphoma/leukemias) more often showed dimmer expression of CD45 (12.5% bright, 87.5% moderate/dim) when compared with peripheral (mature/postthymic) disorders (81.6% bright, 14.1% moderate/dim, and 4.2% negative).

Light Scatter Properties

A majority of T-cell lesions (76.9%) displayed regular orthogonal SSC (similar to normal T lymphocytes). 23.1% cases had increased or markedly increased SSC. FSC, which corresponds to cell size, was regular (similar to normal lymphocytes) in 45.9% of cases, increased in 36.8%, and markedly increased in the remaining 17.3%. All precursor populations had increased FSC, whereas peripheral (mature/postthymic) tumors showed both regular (56.3%) or increased FSC (22.5% increased and 21.1% markedly increased).

Pan-T Antigens

Seventy-seven cases (88.5%) showed either diminution or deletion of one or more of the pan-T antigens. Among precursor lesions (16 cases), those with down-regulation of one, two, or three pan-T markers accounted for 50%, 12.5%, and 37.5% of cases, respectively (the complete lack of expression of one, two, or three pan-T antigens was seen in eight, one, and two cases, respectively). Among peripheral (mature/postthymic) lesions (71 cases), those with down-regulation of one, two, three, or four pan-T markers accounted for 40.8%, 35.2%, 15.4%, and 2.8% of cases, respectively (the complete lack of expression of one, two, three, or four pan-T antigens was seen in 32.4%, 15.5%, 11.3%, and 0%, respectively).

Among precursor neoplasms, our data show that CD2 was lacking in six cases (37.5%) and dim in two cases (12.5%); CD3 was lacking in nine cases (56.2%) and dim in three cases (18.7%); CD5 was lacking in four cases (25%) and dim in five cases (31.2%); and CD7 was lacking in one case (6.2%) and dim in zero cases.

Among peripheral (mature/postthymic) neoplasms, our data show that CD2 was lacking in 7 cases (9.8%) and dim in 4 cases (5.6%); CD3 was lacking in 21 cases (29.6%) and dim in only 2 cases (2.8%); CD5 was lacking in 19 cases (26.8%) and dim in 15 cases (21.1%); and CD7 was lacking in 22 cases (31%) and dim in 16 cases (22.5%).

In summary, among all cases, CD2 was least commonly down-regulated or absent (22%); the other pan-T antigens were aberrantly lacking or diminished in roughly equal proportions: CD3 in 40%, CD5 in 49%, and CD7 in 45%. Diminished pan-T antigen expression was observed less frequently than complete loss in peripheral (mature/postthymic) lesions.

Only three cases (3.5%) showed aberrantly bright expression of at least two pan-T markers. One case (CLL/PLL) showed bright expression of CD2, CD3, and CD5 and coexpression of CD4 and CD8. CD7 was moderate. The remaining two cases showed bright expression of CD2 and CD5. One also had dual CD4/CD8 expression and coexpression of CD56 and CD57 (T-cell LGL). The other was CD4+ and showed strong expression of CD30 by flow cytometry analysis (morphologic evaluation revealed ALCL).

In the remaining seven cases (8%), the expression of all pan-T antigens did not differ significantly from that of benign T lymphocytes (i.e., all pan-T antigens were moderately expressed). The indication that one may be dealing with an abnormal T-cell process was the presence of one or more of the following: anomalous T-cell subset restriction (see CD4/CD8 below), aberrant or down-regulated expression of CD45, CD30 expression (on a significant number of T cells), and/or abnormal light scatter properties. In one of seven cases, the presence of an abnormal T-cell population was identified only by markedly increased FSC (reflecting large cell size), whereas all other parameters (CD45 and pan-T antigen expression) were similar to T cells within a benign lymph node. Due to the presence of both neoplastic and residual (benign) T cells, the CD4:CD8 ratio was not helpful (histologic examination of the lymph node in this case revealed angioimmunoblastic lymphodenopathy (AILD)-like T-cell lymphoma). The other cases without aberrant pan-T antigen expression were diagnosed as CLL/PLL (there was no CD45 expression); LGL leukemia (there was coexpression of CD4 and CD8 as well as CD56 and CD57), peripheral T-cell lymphoma, unspecified (one case had dim CD45 and CD4 restriction, a second case was CD8+, and a third case had markedly increased FSC and increased SSC); and peripheral T-cell lymphoma of the skin (CD8+, increased FSC and SSC).

CD4 Versus CD8: T-Cell Subset Restriction and Anomalous T-Cell Subset Expression

Among precursor lesions, eight cases (50%) were CD4-/CD8-, five cases (31.2%) were CD4+/CD8+, two cases (12.5%) were CD4+, and one case (6.2%) was CD8+. Among all peripheral (mature/postthymic) cases, 11 cases (15.5%) were CD4-/CD8- and 4 cases (5.62%) were CD4+/CD8+ (1 case of T-CLL/PLL showed unusual variable expression of CD4 and CD8, similar to thymocytes; no true coexpression of both antigens was present). Twenty-eight cases (39.4%) were CD4+ and 27 cases (38%) were CD8+ (18 of the CD8+ cases were diagnosed as T-LGL leukemia).

Presence of Additional Markers

CD56 was expressed in 23 cases (26.4% of total; 5 precursor and 18 peripheral neoplasms) and CD57 in 23 peripheral (mature/postthymic) neoplasms (32.4% of peripheral/mature tumors), most of them classified as LGL; 8 of the 23 LGL lesions (34.8%) coexpressed both CD56 and CD57.

Among 74 tumors analyzed for TCRαβ/TCRγδ in our series, 39 (52.7%) were TCRαβ+ (38 were peripheral T-cell disorders and 1 was a precursor neoplasm) and 11 (14.9%) were TCRγδ+ (4 precursor and 7 peripheral/mature neoplasms). The remaining 24 cases (32.4%) were negative for both markers (7 precursor and 17 peripheral/mature). Only 5 of 12 precursor T-cell lymphoproliferative disorders analyzed for TCR expressed the marker. All but one case expressing either TCRαβ or TCRγδ were positive for surface CD3. The CD3-/TCRαβ+ case was diagnosed as T-ALL in soft tissue in a 78 year-old patient. It was CD4+/CD8+ and showed moderate expression of all other pan-T antigens.

CD33/CD13 (major myeloid antigens), CD11b, CD11c, and CD16 were expressed in 5 (precursor T-cell neoplasms), 13 (peripheral), 11 (peripheral), and 8 (peripheral) cases, respectively. CD117 was expressed in four cases (three precursor and one peripheral), CD103 in two peripheral neoplasms, CD25 in six peripheral neoplasms, CD10 in eight cases (four precursor and four peripheral), and CD20 in one peripheral T-cell lymphoma, unspecified. All of the ALCLs analyzed by flow cytometry for CD30 showed bright expression.

Analysis of Control Cases

Peripheral blood from patients with confirmed infectious mononucleosis (five cases) showed activated HLA-DR+ T cells (42%–67%, average 51%) with an inverted CD4:CD8 ratio (0.12–0.63, average 0.33). No pan-T cell antigenic deletion was present, but all cases displayed down-regulation of CD7 (the average intensity of CD7 expression was one decalog below that of normal T cells). Peripheral blood from patients with reactive lymphocytosis (10 cases) or anemia (10 cases; morphologic evaluation and cytogenetics/molecular studies did not reveal a clonal B-cell population, T-cell disorder, myelodysplastic syndrome, or acute leukemia) showed that the percentage of T cells ranged from 32% to 81% (average 48%) with an average CD4:CD8 ratio of 3.6 (range 0.8–9.3, with one case showing an inverted CD4:CD8 ratio). None of these cases showed aberrant expression of pan-T antigens or diminished CD45 expression; CD56+ NK cells were not increased (i.e., <10%).

Among 10 reactive lymph nodes, T cells (range 27%–79%; average 54%) did not display aberrant expression of pan-T antigens. The CD4:CD8 ratio ranged from 2.1 to 8 (average 3.7). Lymph nodes in Hodgkin lymphoma cases (classical type; 10 cases) showed increased numbers of T cells (average 74%; range 51%–87%) with an increased CD4:CD8 ratio (average 11.2; range 3.9–28) and no aberrant pan-T antigen expression. Among 10 lymph nodes in DLBCL cases, the CD4:CD8 ratio ranged from 0.2 to 3.1; 6 cases showed a reversed ratio (ranging from 0.2 to 0.5). T cells accounted for 10%–40% of cells. Only one case displayed aberrant expression of a pan-T marker (dim CD5 expression).

Among 25 benign bone marrows, the percentage of T cells ranged from 9% to 83% (average 28%). The bone marrows immediately during or after chemotherapy for acute myeloid leukemias often showed relative T-cell lymphocytosis. The CD4:CD8 ratio ranged from 0.1 to 12.7 (average 3.4). All but one case showed moderate expression of pan-T antigens. This case showed an atypical CD8+ T-cell population displaying dim expression of CD5 and CD7 and increased FSC. This was from a 22-year-old female who presented with adenopathy and a history of long-term phenytoin administration. Molecular studies and follow-up flow cytometric analysis of this patient's bone marrow and peripheral blood 3 months after cessation of medication were negative for a T-cell disorder, and CD5 and CD7 expression normalized.

In summary, among 80 control cases (negative for T-cell neoplasms), none showed loss of pan-T antigens or diminished/absent CD45 expression. Six cases (7.5%) showed aberrant (dim) expression of one of the pan-T antigens (CD7 in five cases and CD5 in one): five patients were diagnosed with acute viral infection (mononucleosis) and one patient with DLBCL. Only one case (1.25%) showed aberrant expression of two pan-T markers (dim expression of CD5 and CD7) with increased FSC and predominance of CD8+ lymphocytes. A reversed CD4:CD8 ratio was commonly observed in the control population, especially in peripheral blood from patients with acute viral syndromes and in lymph nodes in DLBCL patients. Increased numbers of T cells (up to 87%) and an increased CD4:CD8 ratio were usually associated with Hodgkin lymphoma.

Comparison of Flow Cytometric Data in Neoplastic T-Cell Disorders With Control Population

None of the 80 control cases showed aberrant CD45 expression, whereas 27 out of 87 neoplastic T-cell disorders (31%) displayed overtly diminished (24 cases) or absent (3 cases) CD45 expression. Increased FSC, although observed more frequently in neoplastic processes (54% of cases), is not entirely specific and was observed in the case of phenytoin-induced adenopathy. Neoplastic T-cell disorders, in contrast to reactive/benign conditions, lacked one or more of the pan-T antigens. The cases in the control group showed only a very small subset of T cells with an absence of the CD7 antigen (average 1.5%; SD 2.8). The true absence of one or more pan-T antigens in a significant or major lymphoid population is compatible with a neoplastic process. This is not the case with diminished expression of any of the pan-T markers, because such aberrant expression was observed in benign conditions (viral infections, medication-associated cases, or in lymph nodes involved in DLBCL). The predominance of CD4+ or CD8+ T cells is not specific for neoplastic processes: acute viral infections or DLBCL were often associated with CD8+ predominance and classical Hodgkin lymphomas often showed CD4+ predominance. The presence of increased numbers of T cells with the TCRγδ phenotype is associated with malignancy: none of the reactive T-cell populations revealed this phenotype.

In conclusion, based on the analysis of 87 neoplasms and 80 control cases, flow cytometric features that are most suspicious for (essentially diagnostic of) malignancy include loss or markedly dim expression of CD45, complete loss of one or more pan-T antigens, diminished expression of more than two pan-T antigens in conjunction with altered light scatter properties, or CD4/CD8 dual-positive or dual-negative expression in a major population (except thymic lesions). The differences between neoplastic T-cell processes and control cases regarding the above-mentioned four parameters were statistically significant (P < 0.001). Table 1 presents a comparison between controls and those with neoplastic disorders.

Table 1. Comparison of Phenotypic Characteristic in Control Population and T-Cell Disorders
 Control cases (n = 80)Precursor T-cell neoplasms (n = 16)Peripheral (mature) T-cell neoplasms (n = 71)
  • a

    Control group showed both CD4 and CD8+ T-cells, with occasional CD4 predominance. The number of cells coexpressing CD4 and CD8 ranged from 0% to 1.5% (average 0.43%, SD 0.49). The number of cells with CD4/CD8 phenotype ranged from 0% to 8% (average 0.9%, SD 0.49). The number of cells with CD4/CD8 phenotype ranged from 0% to 8% (average 0.9%, SD 1.29).

  • b

    Bright expression is also included in this group.

  • c

    One case showed unusual variable expression of CD4 and CD8 without true coexpression (similar to that observed in thymocytes).


Immunophenotype of Specific T-Cell Disorders

A summary of the most characteristic immunophenotype of different T-cell disorders is presented in Table 2. Precursor T-lymphoblastic leukemia/lymphomas (16 cases) usually showed dimmer expression of CD45 when compared with normal (benign) T lymphocytes: only two cases (12.5%) had bright CD45; the remaining cases (87.5%) had either moderate or dim CD45 expression. All cases showed lack or diminished expression of at least one of the pan-T antigens (eight cases showed lack of one antigen, two cases showed aberrant expression [lack and/or down-regulation] of two antigens, and the remaining six cases showed aberrant expression of three antigens). Surface CD3 was most often and CD7 least commonly absent (56.25% and 6.25%, respectively). CD2 and CD5 were lost in 41.2% and 29.4%, respectively. Diminished expression of CD2, CD3, CD5, and CD7 was observed in 12.5%, 18.75%, 31.25%, and 0% of cases, respectively. All cases were cytoplasmic CD3+, compatible with T-cell ontogeny up to the common thymocyte stage. Only one case was CD8+ and two cases were CD4+; the remaining precursor neoplasms showed either a CD4+/CD8+ (31.25%) or a CD4-/CD8- (50%) phenotype. Among additional antigens analyzed, CD10, CD13/CD33, CD34, CD56, CD117, and TdT were positive in 18.75%, 37.5%, 37.5%, 31.25%, 18.75%, and 81.25% of cases, respectively. Figure 1 depicts an example of T-lymphoblastic lymphoma/leukemia.

Table 2. A Summary of the Characteristic Immunophenotype of Selected T-Cell Disorders*
 T-ALL (n = 16)T-CLL/PLL (n = 6)T-LGL (n = 21)NK-LGL (n = 4)ATLL (n = 3)ALCL (n = 4)Peripheral T-cell lymphoma (unspecified) (n = 29)
  • *

    Numbers represent actual cases with given phenotype.

  • a

    One case showed unusual variable expression of CD4 and CD8 without true coexpression (similar to that observed in thymocytes).

CD4 versus CD8       
 Not done41 1 15
OtherOccasional cases positive for CD13/33 (≈30%) and/or CD117 (≈18%) Often CD11b (≈43%), CD11c (≈33%) and/or CD16 (≈29%) positiveOften CD11b (≈75%), CD11c (≈50%), and/or CD16 (≈50%) positiveCD25+ (all)CD30+ (all)Occasional cases positive for CD10 ((≈10%), CD11c (≈14%), CD25 (≈14%) and CD103 (≈3%)
Figure 1.

Upper panels: T-lymphoblastic lymphoma with characteristic blastic cytomorphology in lymph node (H&E, 1,000×). Cytograms show the following phenotypes: CD2+, CD3-, CD7+, TdT+, and TCRαβ-/TCRγδ- (asterisks indicate residual benign T lymphocytes). Lower panels: Extensive bone marrow infiltration by lymphoblasts (H&E, 400×). The cytograms reveal CD34+ T-ALL with absent CD2 and surface CD3, positive expression of CD7 and cytoplasmic CD3, and aberrant expression of CD33 and CD56.

T-CLL/PLL was diagnosed in six cases (the ages ranged from 50 to 92 years): five cases occurred in men (four in blood and one in bone marrow) and one case occurred in a woman (lymph node). CD45 expression was normal (bright) in three cases, moderate in one, and two lesions were CD45-. Phenotypically, one case showed moderate expression of all pan-T antigens (CD45-), one case showed abnormally bright expression of three pan-T antigens and moderate expression of CD7 (CD4+/CD8+), and the remaining cases showed dim or absent expression of at least one of the pan-T antigens. Three neoplasms were CD4+, one case was CD4+/CD8+, one case was CD4-/CD8-, and one case showed unusual variable expression of CD4 and CD8 (without true CD4/CD8 coexpression; Fig. 2), which is similar to that observed in thymocytes. Five lesions were analyzed for TCR: four were TCRαβ+ and one was TCRαβ-/γδ-. CD56 was positive in one and CD20 (dim expression) was observed in one. CD25 was negative in all cases. In summary, this clinical entity has no characteristic immunophenotypic profile in our series.

Figure 2.

T-CLL/PLL with characteristic convoluted nuclear cytomorphology (Wright-Giemsa, 1,000×). Cytograms demonstrate aberrant loss of CD3 and TCRαβ/TCRγδ and unusual CD4/CD8 expression (most T-CLL/PLL are CD4+).

T-LGL (21 cases; 11 peripheral blood and 10 bone marrow) revealed a range of 14%–61% abnormal T cells with bright (normal) CD45 (all cases), regular to slightly increased SSC, and regular FSC (all cases). Only one case did not show any loss of pan-T antigens (the cells were CD4+/CD8+) and one case showed abnormally bright expression of CD2 and CD5 (CD3 and CD7 were normally expressed). The remaining 19 cases showed aberrant expression of the pan-T markers: one, two, or three of the pan-T antigens were lacking or diminished in 10 (47.6%), 7 (33.3%), and 2 (9.5%) cases, respectively. Of the T-cell antigens, CD2 was least often affected: only two cases (9.5%) showed dim expression and one case showed bright expression. CD5 was most often lost or diminished (17 cases, 80.9%: absent in 6 cases, dim in 11 cases). CD3 was absent in three cases (14.3%) and CD7 was absent in two (9.5%) and dimly expressed in six (28.6%) cases. A majority of cases showed a CD4-/CD8+ phenotype (18 cases, 85.7%): one case was CD4-/CD8- (surface CD3+ coexpressing CD56 and CD57) and two cases were CD4+/CD8+ (CD8 expression was dim and CD4 moderate). Fourteen cases (66.7%) were TCRαβ+, five cases (23.8%) were TCRγδ+, and the remaining two cases (9.5%) were negative for TCRαβ and TCRγδ. All cases expressed CD57 and/or CD56 (eight cases showed coexpression of both markers). CD11b, CD11c, and CD16 were positive in nine (42.8%), seven (33.3%), and six (28.6%) cases, respectively (three cases were both CD11b+ and CD11c+). Figure 3 shows an example of a T-LGL leukemia.

Figure 3.

LGL with characteristic cytoplasmic granules in the cytoplasm (Wright-Giemsa, 1,000×). Cytograms show dim CD5 expression (arrows), loss of CD7, CD8 restriction, and expression of NK markers CD56 and CD57 (asterisks indicate residual benign T lymphocytes).

NK cell LGL leukemias (two in the peripheral blood and two in the bone marrow) were negative for CD3 and CD5. CD56 was expressed in three and CD57 in the remaining one case (no coexpression was present). CD11b, CD11c, and CD16 were usually expressed.

Adult T-cell leukemia/lymphoma (HTLV1+; three cases) with confirmed seropositivity for HTLV-1 were seen in the blood or bone marrow. All three cases were CD4+/CD8- and showed bright (normal) expression of CD45, moderate expression of CD25 and CD5, and slightly increased SSC and FSC. CD2 was negative in one case, CD3 was negative in two cases, and all were negative for CD7, CD56, and CD57.

ALCL, T/null cell (four cases) occurred in the lymph nodes of two females and two males, whose ages ranged from 61 to 71 years. Abnormal cells ranged from 18% to 51%. CD45 was bright in three cases and moderate in one case. One case, which was CD4+/CD8-, did not show loss or diminution of pan-T antigens. Instead, CD2 and CD5 expression were slightly brighter than that observed in normal (benign) T cells. The malignancy was suspected in this case by markedly increased FSC. Three cases showed loss of CD3, CD5, and CD7 (two cases) and loss of CD3 and CD7 (one case). A CD4+/CD8- phenotype was present in two cases; one case was CD4-/CD8- (Fig. 4) and the other was CD4-/CD8+. TCR was analyzed in three cases: two were negative for both TCRαβ and TCRγδ and one case was TCRαβ+. CD56 and CD57 were not expressed. CD30 was moderately expressed in all cases.

Figure 4.

T-cell ALCL showing characteristic pleomorphic cytomorphologic features of hallmark (Reed-Sternberg-like) cells (upper left panel; H&E, 200×) and strong CD30 and Alk-1 expression (upper middle and left panels; peroxidase staining, 400×). Cytograms demonstrate large cell size by FSC with aberrant loss of CD3, CD5, and CD7 (arrows) and absent CD4/CD8 (asterisks indicate residual benign T lymphocytes).

Hepatosplenic γδ T-cell lymphoma (two cases) occurred in the liver of a 35-year-old man and in the bone marrow of a 37-year-old woman. Both cases showed a similar immunoprofile: CD45+(bright), CD2+, CD3+, CD5-, CD7+(dim in one case), CD4-/CD8-, TCRγδ+, CD56+, CD57-, and CD16+. Figure 5 shows an example of hepatosplenic T-cell lymphoma involving the liver with characteristic sinusoidal distribution of atypical cells.

Figure 5.

Hepatosplenic γδT-cell lymphoma showing sinusoidal distribution of atypical lymphoid cells within liver parenchyma (H&E, 400×; inset, 1,000×). The cytograms exhibit expression of CD3 (arrow), CD7 (dim), TCRγδ, CD16, and CD56 and negativity for CD5 and CD4/CD8. Note increased FSC reflecting larger cell size (asterisks indicate residual benign T lymphocytes).

Peripheral T-cell lymphoma, unspecified (29 cases), were from the peripheral blood (4 cases), bone marrow (8), lymph nodes (13), and one each of the omentum, skin, lung, and spleen. The ages ranged from 19 to 90 years (average age 65.4). The expression of CD45 was bright (normal) in 23 cases (79.3%), moderate in 4 cases (13.8%), dim in 1 (3.4%), and absent in 1 case. FSC was regular in 11 cases (37.9%), increased in 8 cases (27.6%), and markedly increased (large cell size) in the remaining 10 cases (34.5%). Four cases (13.8%) did not reveal any loss or diminution of pan-T antigens (all were TCRαβ+). An atypical phenotype was suspected in cases based on CD8 subset restriction (2 cases), increased FSC indicating large cell size (2 cases), and overtly dim expression of CD45 (1 case). The remaining 25 cases showed either diminished expression (9 cases), lack (19 cases), or both dim and/or absent expression of one or more of the pan-T antigens. Eleven cases showed loss of one, four cases loss of two, and four cases loss of three of the pan-T antigens. CD7 was most frequently absent (13 cases; 44.8%). CD2, CD3, and CD5 were lost in 5 (17.2%), 8 (27.6%), and 4 (13.8%) cases, respectively. A majority of peripheral T-cell lymphomas, unspecified, were CD4+ (18 cases, 62%). Eight cases were CD8+ (27.6%) and three cases (10.3%) had a CD4-/CD8- phenotype. None of the neoplasms analyzed showed dual CD4+/CD8+ coexpression. TCR was analyzed in 24 cases: 18 of them (75%) were TCRαβ+ and the remaining 6 cases (25%) were negative for both TCRαβ and TCRγδ. Of additional markers analyzed, CD10, CD11c, CD25, CD56, and CD103 were occasionally expressed. CD30 was not expressed. Morphologic evaluation revealed Lennert's lymphoma in 2 of 25 cases. The one skin lesion analyzed was a CD8+ T-cell neoplasm, and was consistent with an unspecified cutaneous T-cell lymphoma, rather than the specific category of mycosis fungoides/Sezary's syndrome. Due to the paucicellularity and relatively poor viability of skin samples, no clear-cut examples of mycosis fungoides were seen in our cohort. Only one case represented Sezary's syndrome: peripheral blood analysis from a patient with known mycosis fungoides revealed an atypical T-cell population with aberrant loss of CD7 and CD2. Figure 6 presents an example of peripheral T-cell lymphoma involving lymph node (upper panels) and bone marrow (lower panels).

Figure 6.

Upper panels: peripheral T-cell lymphoma in lymph node (H&E, 400×); insets show peroxidase staining (200×) for CD3 antigen (positive expression; upper) and CD7 (negative expression, lower). Cytograms depict CD4+ cells of medium size, expressing CD5 and CD3 with aberrant loss of CD7 and unusual expression of CD10 (asterisks indicate residual benign T lymphocytes). Lower panels: peripheral T-cell lymphoma involving the bone marrow (H&E, 200×). Cytograms reveal medium to large cells by FSC (arrow) with the following phenotype: CD2+, surface CD3-, cytoplasmic CD3+, and CD4+.

Enteropathy-type T-cell lymphoma (one case) was seen in the jejunum of a 40-year-old male patient. Classical clinical presentation showed bright CD45, moderate CD2, CD3, CD4, CD5, and CD103, and dim CD7. FSC was markedly increased. Angioimmunoblastic T-cell lymphoma (one case, Fig. 7) did not reveal aberrant expression of CD45 or pan-T antigens. Atypicality was suspected only by light scatter abnormalities of the T-cell population: FSC was markedly increased and SSC was slightly increased when compared with normal (residual) T-cells, which were also present and gave the false impression of a normal CD4:CD8 ratio (atypical cells were CD4+).

Figure 7.

Angioimmunoblastic-like T-cell lymphoma (hypervascular pleomorphic lymphoid infiltrate; H&E, 400×). Cytograms show no aberrant expression of pan-T antigens or TCR. The CD4:CD8 ratio was within the normal range. The only flow cytometric indication of an abnormal T-cell population was increased FSC (backgating demonstrated CD4 restriction by larger T cells).


T-cell lymphomas are a heterogeneous group of precursor (immature/thymic) and peripheral (mature/post-thymic) neoplasms. They have a broad spectrum of clinical, immunophenotypic, and morphologic features, making them difficult to differentiate from reactive conditions and non–T-cell tumors (1–18). Therefore, application of a spectrum of methodologies, including histologic evaluation of adequate tissue sections, flow cytometry, cytogenetics, and molecular studies, is sometimes necessary to diagnose T-cell lymphomas (1–30); cytogenetics may offer prognostic information as well. In contrast to B-cell lymphoid proliferations, in which flow cytometric analysis of immunoglobulins for light chain restriction can determine clonality, neoplasia cannot be determined easily by flow cytometry analysis of T-cell or NK cell proliferations. This may change, however, with the advent of the development of antibodies to the family of T-cell antigen receptor beta chains (31, 32). Large panels of monoclonal antibodies to the TCR-Vβ repertoire can detect expansion of one or more TCR-Vβ families and can predict clonality in some instances (32). However, the large panel is unwieldy.

As presented in this paper, a combination of several parameters, including altered light scatter properties, diminished CD45 (leukocyte common antigen) expression, pan-T antigenic diminution or deletion, T-cell subset antigen (CD4/CD8) restriction, and additional markers (e.g., associated with cell immaturity or activation), is the most practical and sensitive way to identify and characterize abnormal T-cell populations that may be neoplastic in origin.

A lymphoid proliferation composed of predominantly T cells raises the possibility of a T-cell lymphoproliferative process, keeping in mind that some reactive conditions and tissues involved in Hodgkin's lymphoma often contain a predominance of T cells. The percentage of T lymphocytes in our group of control cases (see above) ranged from 9% to 83%, which is in agreement with published data (33, 34). Therefore, the predominance of a T-cell population is not diagnostic per se of malignancy. Among 87 neoplasms analyzed in this study, only 11 cases contained an overt predominance of T cells (80% or more T cells in analyzed samples); 10 cases had 20% or less T cells.

Abnormal CD4/CD8 subset restriction (abnormal CD4:CD8 ratio) is one of the more obvious criteria to detect an atypical T-cell proliferation, but again this cannot be equated with clonality. An abnormal CD4:CD8 ratio was observed in a subset of control cases in this study. This included both a markedly increased ratio (up to 28:1 in Hodgkin's lymphomas) and a reversed ratio (viral infections, DLBCL, and atypical T-cell proliferation associated with phenytoin treatment). These observations are in agreement with published reports: an increased number of CD4+ T lymphocytes is observed in Hodgkin's lymphoma or in dermatopathic lymphadenitis. Conversely, an increased number of CD8+ T cells in relation to CD4+ cells can be seen in immunocompromised individuals (human immunodeficiency virus [HIV]+), as well as in individuals with tuberculosis, pediatric immunodeficiency disorders, or with certain viral infections (4, 21, 35–38). A reversed helper-to-suppressor T-cell ratio may also be observed in patients with untreated hairy cell leukemia (HCL), most likely as a result of spontaneous T-cell activation (39–41). The abnormal CD4:CD8 ratio is more prominent in patients with HCL who underwent splenectomy (40, 41). None of the control cases showed a CD4+/CD8+ or a CD4-/CD8- phenotype (CD4+/CD8+ cells averaged 0.43% ± 0.49 and CD4-/CD8- cells averaged 0.9% ± 1.29), whereas occasional peripheral T-cell neoplasms and the majority of precursor lesions displayed a double-positive or double-negative phenotype. Therefore, the CD4+/CD8+ or CD4-/CD8- phenotype strongly suggests malignancy when present in high numbers, keeping in mind that deficiency of both CD4 and CD8 expression may be observed occasionally in patients with HIV or with the autoimmune lymphoproliferative syndrome (ALPS; 4,21,25,26,33–35,37). ALPS is a disorder of lymphocyte homeostasis associated with Fas (CD95) gene mutations that regulate apoptosis; the lymphocytes are characteristically CD3 and TCRαβ+ and are CD4/CD8- (42). Normal adults may have small subsets (<3%) of CD4+/CD8+ cells (usually with one antigen being dimmer) in peripheral blood (33–35). Only in lesions located in the mediastinum may such a phenotype represent benign thymocytes, either within a reactive thymus or within a thymoma (21). A variable pattern of CD4 and CD8 in lesions located in the mediastinum may cause a problem in distinguishing thymocytes from a neoplastic T-cell population. Correlation with the expression of other antigens (e.g., variable expression of CD3 would favor nonneoplastic thymocytes), as well as cytomorphologic evaluation, is necessary for accurate diagnosis. Molecular studies for TCR gene rearrangement may be indicated to establish a definite diagnosis. Aberrant (dim) expression of either CD4 or CD8 is also helpful in detecting an atypical T-cell process, especially when residual (benign) T lymphocytes are still present and can be identified easily as a separate (control) population.

One of the most useful criteria to diagnose a T-cell neoplasm is aberrant lack or dim expression of one or more of the pan-T antigens (4, 7, 16, 19–21, 25, 26, 43). In our series, among peripheral (mature/postthymic) disorders, CD2 was least frequently absent (9.8%) and the other pan-T antigens were lost in approximately the same number of cases (CD3, CD5, and CD7 were lost in 29.6%, 26.8%, and 31% cases, respectively). Knowles (4) reported CD7 to be the antigen that is most often absent. Jones and Dorfman (16) observed the absence of CD2, CD3, CD5, and CD7 in 18%, 26%, 30%, and 68%, respectively. Chu et al. (43) reported a lack of CD2, CD3, CD5, and CD7 in 14%, 59%, 40%, and 64%, respectively. Jamal et al. (33) detected aberrance in the expression of CD7, CD5, and CD2 in 58%, 42%, and 28% of cases. The source of the differences between studies may be related to different methodologies and different types of T-cell neoplasms analyzed (33). Although aberrant expression of pan-T antigens is suspicious for T-cell neoplasms, aberrantly dim expression of one or more of the pan-T antigens can be observed, however, in benign/reactive T lymphocytes (see above). Our data indicate that lack of at least one of the pan-T markers or diminished expression of more than two antigens favors a neoplastic process, especially when accompanied by other flow cytometric abnormalities (e.g., increased FSC). Again, increased FSC of T cells, even with dim expression of pan-T antigens, cannot be equated with malignancy in all cases, as illustrated by phenytoin-induced adenopathy in one of our control cases.

Finally, analysis of additional markers, including TCRαβ/TCRγδ, blast-associated antigens (CD34 or TdT), NK-associated antigens (CD56, CD57), markers associated with activation (CD30) or aberrant expression of antigens not typical for T lymphocytes, e.g., myeloid antigens (CD13, CD33) or B-cell antigens, may be very helpful in detection of atypical T-cell populations by flow cytometry (7, 11, 19–21, 25, 26, 29, 30, 43–45). Expression of CD30 antigen by a significant number of T cells, especially when they display increased FSC, is characteristic for ALCL, keeping in mind that occasional activated lymphocytes might express CD30 (immunoblasts). Lack of CD30 expression by flow cytometry, on the other hand, does not exclude the diagnosis of ALCL (unpublished observations).

Once an aberrant immunophenotype has been detected by flow cytometry, cytomorphologic correlation is essential for diagnosis. If the lymph node or other tissue is sufficiently effaced and T cells display atypical cytomorphologic features, this may be all that is necessary for adequate correlation and diagnosis. Often, however, a second methodology is required for confirmation that an atypical T-cell phenotype is indeed clonal and malignant, given the many reactive conditions that may contribute to an atypical T-cell immunophenotypic profile and mimic T-cell lymphoma. If there is fresh tissue, both karyotypic analysis by standard cytogenetics and molecular analysis by Southern blot technique for TCR gene rearrangement are available methodologies. At present, Southern blot is considered the gold standard for assessment of clonality. Reverse transcription-polymerase chain reaction (RT-PCR) for T-cell gene rearrangement can also be done on paraffin-embedded tissue, providing it is not decalcified.

As noted, distinguishing malignant from benign T-cell proliferations is difficult due to the variability in T-cell antigenic expression and the lack (at present) of a readily identifiable clonal phenotype. The flow cytometric features that are most suspicious for (essentially diagnostic of) malignancy are (1) the loss or markedly dim expression of CD45, (2) complete loss of one or more pan-T antigens, (3) diminished expression of more than two pan-T antigens in conjunction with altered light scatter properties, or (4) CD4/CD8 dual-positive or dual-negative expression (except thymic lesions). A combination of methodologies (flow cytometry, immunohistochemistry, cytogenetics, and molecular techniques), coupled with cytomorphologic correlation, may be required to achieve the most definitive and accurate diagnosis for patients with T-cell lymphoproliferative disorders. Once the atypical T-cell population is correctly recognized, an attempt should be made to categorize and define the type of the T-cell lymphoid proliferation. This is important because of the different prognoses and treatment options for patients with different T-cell neoplasms, ranging from clinical observation to bone marrow transplantation (46).