The immunophenotypic features of angioimmunoblastic T-cell lymphoma (AILT) have not been well described.
The immunophenotypic features of angioimmunoblastic T-cell lymphoma (AILT) have not been well described.
We retrospectively reviewed our institutional experience with the flow cytometric features of 16 cases of AILT.
Multiparameter flow cytometry was able to identify a distinct population of immunophenotypically aberrant T cells in 15 of 16 cases. In 13 lymph node specimens, the neoplastic cells ranged from 1.9 to 87% (median 23%) of cells. The ratio of reactive to neoplastic T cells ranged from 0.01 to 20 (median 1.5); reactive T cells outnumbered neoplastic in 9/13 (69%) cases. The neoplastic populations expressed CD2, CD4, CD5, and CD45RO in all cases, lacked expression of CD8 and CD56 in all cases, and showed negative or dim surface CD3 in most cases. CD10 was expressed by the neoplastic populations in 11 of 14 cases at diagnosis; in 3 of these 11 only a subpopulation of the neoplastic cells was CD10(+). CD10 tended to be absent on neoplastic cells in staging bone marrows. The neoplastic population in all but one of the 15 positive cases possessed multiple immunophenotypic abnormalities and these were generally retained during the follow-up analyses of several cases.
These results indicate the potential utility of flow cytometry in the diagnosis and follow-up of AILT. © 2006 International Society for Analytical Cytology
Angioimmunoblastic T-cell lymphoma (AILT) is a distinctive form of peripheral T-cell lymphoma (PTCL) typically characterized by generalized lymphadenopathy, hepatosplenomegaly, polyclonal hypergammaglobulinemia, and systemic symptoms (1–4). The diagnosis is often challenging because of the polymorphous nature of the cellular infiltrate, usually with prominent reactive background cell populations. Furthermore, a morphologically abnormal T-cell population, typically manifesting as clusters of clear cells, may be inconspicuous or absent in some cases. Finally, the degree of architectural effacement varies, depending on the stage of evolution of the lymphoma. Some cases are characterized by prominent reactive germinal centers and little architectural distortion, closely resembling a reactive process (5).
The diagnosis of AILT has been facilitated by the recent demonstration that neoplastic cells in most cases express CD10 (6). However, other immunophenotypic features of this lymphoma have not been well characterized, in part because of difficulty in distinguishing neoplastic from reactive T cells by immunohistochemistry. Multiparameter flow cytometry (MFC) represents a highly reproducible and objective way of assessing the antigenic profile of various T-cell populations, allowing discrimination of immunophenotypically aberrant, neoplastic T-cell populations from nonneoplastic T-cells, thus allowing their specific characterization (7).
In this study, we report our single institutional experience with the flow cytometric features of 16 cases of AILT. Our results show that the neoplastic cells in AILT may be present in very small numbers, but can be distinguished from normal T cells based on multiple immunophenotypic aberrancies. In addition, aberrancies are generally retained during follow-up, rendering MFC a potentially useful tool for monitoring residual disease.
Institutional databases were searched retrospectively from April 1994 to March 2004 for all morphologically diagnosed AILT for which MFC had been performed at diagnosis. The diagnosis of AILT was made according to WHO criteria (8) and the descriptions of Ree et al. and Attygalle et al. (5, 6), supplemented by the results of T-cell receptor (TCR) gene rearrangement analysis (see later) and, in selected cases, immunohistochemistry. This yielded 16 cases. One case evolved from a purely intrafollicular T-cell lymphoma that lacked pathologic features of AILT. The data presented for this case represent the findings at the time of development of a lymphoma with diagnostic features of AILT.
Flow cytometry tissue processing and antibody staining were performed as previously described (9, 10). MFC panels included various combinations of antibodies against: CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, CD30, CD38, CD45, CD45RO, and CD56. These antibodies were conjugated with fluorescein isothiocyanate, phycoerythrin, peridinin chlorophyll protein, and allophycocyanin. Three-color analysis and four-color analysis were performed on FACSort and FACSCalibur flow cytometers [Becton Dickinson (BD) Immunocytometry Systems, San Jose, CA], respectively. Data analysis was performed using Paint-A-Gate [software (BD)]. Nonviable cells and debris were excluded based on forward and orthogonal light scatter properties. Positivity for an antigen was defined as at least 10% lymphoma events beyond a 2% threshold based on an isotypic control tube. Percentages of cell types were determined based on total viable cellular events. Immunophenotypic aberrancy was defined as previously reported with reference to a large cohort of normal and reactive T-cell populations in various tissue sites (7), and within a given case the intensity of expression of individual antigens was assessed relative to internal normal T cells. Briefly, for most antigens, immunophenotypic abnormalities were defined as expression beyond the limits seen in nonneoplastic T-cell populations (7). For CD7, an antigen that shows a spectrum from negative to positive on normal and reactive T cells, aberrant expression was defined as either a lack of expression or a distinctly different pattern of expression than normal T cells on subpopulations defined on the basis of other immunophenotypic features (e.g., under-expression of CD3). Light scatter properties distinct from normal internal T cells were also considered to be aberrant. CD10 expression was not considered an aberrancy in this study because of data indicating the presence of small numbers of normal CD10(+) T cells in reactive lymph nodes (11).
Histologic sections of involved lymph nodes were reviewed in a blinded fashion by two observers (W.C. and S.K) and classified in two ways. Based on architectural features, the cases were divided into those with hyperplastic germinal centers (Group I) and those with either scattered residual burnt-out follicles or complete architectural effacement (Group II) (6). The lymph nodes were also graded on the extent of clear cell proliferation: Grade 0—absent or inconspicuous clear cells; Grade 1—occasional, inconspicuous, small clusters of clear cells; Grade 2—conspicuous, scattered clear cell clusters; Grade 3—extensive, sheet-like clear cell infiltration.
Polymerase chain reaction (PCR) analysis of TCR γ-chain gene rearrangements, immunohistochemistry for CD10 and CD21, and in situ hybridization analysis of Epstein-Barr virus encoded RNA (EBER) were performed as previously described (12, 13).
Comparison of continuous variables was performed using a Mann-Whitney U test.
There were 10 male and 6 female patients, aged 22–80 years (median 52) (Table 1). The specimens initially analyzed by MFC at diagnosis were 14 lymph nodes, 1 peripheral blood, and 1 bone marrow. In the latter two cases, concurrent diagnostic lymph node biopsies were not submitted for MFC. Of the 14 lymph nodes analyzed by MFC, 4 contained hyperplastic germinal centers and the remaining 10 contained inconspicuous burnt-out germinal centers or showed complete architectural effacement (Table 1). One lymph node contained no clear cell component (Grade 0), 4 inconspicuous clear cell infiltrates (Grade 1), 5 scattered conspicuous clear cell clusters (Grade 2), and 4 extensive clear cell infiltrates (Grade 3) (Table 1).
|Case||Age/sex||Source||AberrantT cellsa (%)||NormalT cells (%)||Clear cell grade||Architecture groupb||TCRrearrangement|
PCR analysis of the TCR γ-chain gene rearrangements was performed in each case (Table 1). Thirteen of the 16 cases were TCR γ-PCR positive, 2 were negative (Cases 4 and 14), and 1 showed no amplification and was thus considered indeterminate (Case 7). Immunohistochemistry for CD10 expression was performed in 6 cases (including 3 of the 4 with hyperplastic germinal centers), each of which showed a population of extrafollicular CD10(+) lymphocytes. CD21 immunohistochemistry revealed expanded follicular dendritic cell (FDC) meshworks in three cases with absent or inconspicuous germinal centers, whereas FDCs were largely confined to hyperplastic germinal centers in two cases. EBER in situ hybridization was positive in immunoblasts in each of four cases in which it was performed.
At diagnosis, 15 of 16 patients (94%) had detectable aberrant T-cell populations in the MFC samples (Table 1). One patient had negative MFC on a diagnostic lymph node (Case 14); this case morphologically lacked clear cell infiltrates. In the 13 positive lymph nodes, the median percentage of neoplastic cells was 23% (mean 27.7%, range 1.9–87%) and the median percentage of reactive T cells was 32% (mean 28.0%, range 0.63–42%). The ratio of reactive to neoplastic T cells ranged from 0.01 to 20 (median 1.5); reactive T cells outnumbered neoplastic in 9/13 (69%) cases. The median percentage of neoplastic cells by flow cytometry was 18.5% (mean 21.0%) in cases with hyperplastic germinal centers and 23% (mean 30.6%) in those without (P = 0.76). Clear cell Grade 1 cases demonstrated a median of 11.5% (mean 12.7%) neoplastic cells by flow cytometry, Grade 2 cases a median of 22% (mean 18.8%), and Grade 3 a median of 52.5% (mean 53.8%). There was no significant difference between the percentage of neoplastic cells in clear cell Grades 1 and 2 (P = 0.33), but Grade 3 cases contained significantly more neoplastic cells by flow cytometry then Grades 1 and 2 combined (P = 0.013).
Table 2 details the immunophenotypic features of the 15 cases, and examples are illustrated in Figure 1. Neoplastic cells had the same light scatter features as normal T cells in six cases and were distinctly larger in nine cases (Fig. 1A). Each aberrant population expressed CD2 (15/15; 1 dim, 4 bright), CD4 (15/15; 2 bright, 3 dim), CD5 (13/13; 5 bright), CD38 (12/12), CD45 (11/11; 4 bright), and CD45RO (10/10; 6 bright). All cases lacked CD8 (0/15), and CD56 (0/9). CD3 (Fig. 1B) was absent in 9/15 cases (60%), dim in 3 cases, and expressed at normal intensity in 3 cases. CD7 was absent in 6/15; 3/15 cases had distinct positive and negative subpopulations (Fig. 1C). CD10 (Fig. 1D) was seen in 11/14 cases (79%); in 8 cases the entire neoplastic population was positive for CD10 and in 3 cases CD10 was expressed in 31, 46, and 67% of neoplastic cells. In 1 case (Case 5) in which CD10 was negative by flow cytometry, immunohistochemistry revealed scattered clusters of extrafollicular CD10(+) lymphocytes; this was an example with hyperplastic germinal centers. CD30 (Fig. 1F) was present at low intensity in 1/12 cases (8.3%); this was corroborated by immunohistochemistry.
|Case||CD2||CD3||CD4||CD5||CD7||CD8||CD10||CD30||CD38||CD45||CD45RO||CD56||Light scatter||Total no. of abnormalities|
One to six aberrancies were found in individual cases (Table 2) with a median of four aberrancies per case. As illustrated in Table 3, the most common aberrancy was abnormal expression of CD3 (80% of cases, including 60% with a total lack of surface expression). This was followed by abnormal expression of CD7 (73%) and over-expression of CD45RO (60%). Other antigens were aberrantly expressed in a minority of cases (Table 3).
|Type of aberrations||CD2||CD3||CD4||CD5||CD7||CD8||CD30||CD45||CD45RO|
Of 13 patients with initial MFC on lymph node specimens, 6 patients had MFC on staging bone marrow specimens. All were positive for lymphoma, although the number of neoplastic cells was low (median 1.2%, range 0.08–3.0%). Of 4 cases that were CD10(+) in the primary lymph nodes, 3 were CD10(−) in the staging bone marrow. Notably, two of these three were only partially positive for CD10 in the lymph node specimens. One case (Case 7) subsequently became partial CD10(+) in a follow-up bone marrow. Two CD10(−) cases were also CD10(−) in staging bone marrows. Additional antigenic differences, were subtle with decreases in intensity of CD2, CD4, or CD7.
Three patients had positive follow-up MFC studies in bone marrow, peripheral blood, or pleural fluid. In the peripheral blood and marrow samples the number of neoplastic cells was low, comprising a median of 0.16% of total cells (range 0.04–1.2%). However, they constituted the majority of events (85%) in one fluid specimen. In addition to the case that regained CD10 expression at follow-up after showing a CD10(−) tumor population in the initial staging marrow, one case showed diminished (but not absent) CD10 in a follow-up bone marrow specimen. Additional immunophenotypic changes at follow-up were generally subtle, involving decreased or brighter intensity of CD2, CD3, CD4, CD10, or CD45RO.
In this study, we demonstrated aberrant T cell populations by MFC in 15 of 16 cases of AILT, illustrating the high sensitivity of this diagnostic approach. These included four cases with hyperplastic germinal centers, a setting where definitive histologic interpretation may be challenging. It is possible with MFC to clearly discriminate neoplastic from reactive T cells based on overt immunophenotypic aberrancy, allowing specific characterization of the neoplastic populations. We have found that AILT have a very characteristic immunophenotype. In all cases, the neoplastic cells expressed CD2, CD4, CD5, and CD45RO, and all lacked CD8 and CD56. Most cases had either dim or (more commonly) absent surface CD3 expression, and most cases expressed CD10. CD7 was variably expressed, but several of our cases demonstrated distinct CD7(+) and CD7(−) subpopulations. This pattern of CD7 expression is very unusual in PTCL in our experience. Our results are consistent with two recent smaller flow cytometry studies that assessed more limited panels of antibodies. Serke et al. reported a CD2(+)/CD3(−)/CD4(+)/CD5(+)/CD7(−)/CD8(−)/CD38(+)/CD45(+) immunophenotype in five of seven cases of AILT with circulating cells in the peripheral blood (14). Lee et al. demonstrated a CD2(+)/CD3(−) (or dim)/CD4(+)/CD7(−)/CD10(+) immunophenotype in two of three cases (15).
Each of the 15 neoplastic populations but one in the present study were aberrant in at least two respects, with a median of four aberrancies per case. Most useful from the diagnostic standpoint was under-expression or lack of surface CD3. This feature alone was sufficient to discriminate neoplastic from reactive T cells in 80% of the aberrant cases. Normal T cells maintain CD3 expression within fairly narrow limits. Although, slight differences in expression may be seen in various functional subsets (e.g., slightly brighter CD3 expression in γ–δ T cells compared to α–β T cells), these differences are insufficiently large to actually discriminate multiple populations in routine qualitative MFC analysis. Other common aberrancies in this series were abnormally bright CD45RO expression (above the range of expression seen in normal T cells), seen in 60% of cases, and abnormal expression of CD7 in 73%. Other antigens were aberrantly expressed in fewer than half of the AILTs in the present series. The frequency of specific abnormalities was similar to what we observed in a prior study of 50 PTCLs (7).
It is important to note that not all of the immunophenotypic findings considered aberrant for the present study are specific for neoplasia. For example, bright CD45RO expression by itself is insufficient to render a diagnosis of T-cell neoplasia. Nevertheless, the presence of a discrete CD45RO (bright) population should prompt careful examination of the T cells for other aberrant features. Similarly, under-expression or absent expression of CD7 as a sole aberrancy should be interpreted with extreme caution, given the fact that CD7(−) cells may rarely comprise as many as 50–60% of helper T cells in benign lymph node, peripheral blood, or bone marrow samples (7). However, if a subpopulation of T cells defined on the basis of other immunophenotypic features shows absence or under-expression of CD7, this finding supports a neoplastic interpretation. This underscores the point that overall patterns of antigen expression should be assessed, rather than focusing on expression levels of single antigens. When considered in this fashion, the neoplastic populations in this series, as well as those in most PTCLs, show overall expression patterns which deviate from those seen in normal and reactive T-cell populations (7).
Previous studies using immunohistochemical methods to investigate the immunophenotype of AILT have produced inconsistent results (1, 2, 5, 14, 16). The difficulties in interpreting immunohistochemical data arise in part from the presence of a prominent reactive T-cell component in addition to the neoplastic populations in many cases. Notably in this regard, in primary lymph node samples we found that the neoplastic component comprised as few as 1.9% of events, and in the majority of cases the neoplastic T cells were outnumbered by reactive T cells. Not surprisingly, the proportion of events comprised by the neoplastic population was related to the extent of clear cell infiltration appreciated morphologically in routine histologic sections. Notably, the single AILT lymph node that lacked an aberrant population of T cells by flow cytometry also lacked a distinct clear cell component histologically. As described previously, a clonal TCR gene rearrangement could not be demonstrated in this case, and the patient had a favorable clinical course (7). These features argue that this patient, in fact, had a reactive proliferation rather than a T-cell lymphoma. However, the consensus opinion of several local hematopathologists, corroborated by independent consultative review by an expert hematopathologist at an outside institution, was that this case was histologically diagnostic of AILT, and thus it was retained in the current series. AILT-containing reactive germinal centers has been postulated to represent an early stage of disease (6). In the present series, there was no obvious relationship between the size of the neoplastic populations and the presence or absence of hyperplastic germinal centers, although the small number of cases in the former group limits the statistical power of the analysis.
The presence of CD10 on the neoplastic cells in most of our cases is consistent with the findings in recent immunohistochemical studies (6, 17). Interestingly, while in the majority of cases the entire neoplastic population expressed CD10, in three of our cases only a subset of the neoplastic cells were positive, ranging from 31–67% of the tumor population. A single case with hyperplastic germinal centers lacked CD10 by flow cytometry, but revealed extrafollicular CD10(+) cells by immunohistochemistry; the cause of this discrepancy is not clear. The expression of CD10 by peripheral T-cell neoplasms appears to be limited to AILT (6, 17), with the exception of a rare subset of intrafollicular PTCL (18); the latter entity may be related to AILT. Of interest is the recent demonstration of small populations of primarily intragerminal center CD10(+) T cells in reactive lymphoid proliferations (11). This, in conjunction with the close association of AILT neoplastic cells with expanded follicular dendritic networks and the expression of bcl-6 by AILT cells (5, 6, 17, 19), suggests that AILT arises from germinal centers. Of note in this regard was that we observed an evolution in Case 7 from a purely intrafollicular large T-cell lymphoma to AILT (data not shown). This further supports the hypothesis of an origin of AILT from intragerminal center T cells, as well as the relationship of intrafollicular T-cell lymphoma to AILT postulated by de Leval et al. (18).
We performed MFC analysis of staging bone marrow specimens in seven cases in our series, six of which also had MFC performed on the initial diagnostic lymph node biopsy. Each contained a population of aberrant T cells, ranging from 0.08 to 3.0%. Despite their small size, these populations were appreciated because of distinct immunophenotypic aberrancies that were similar to those seen in the primary lymph node specimens. Interestingly, the neoplastic cells were CD10(−) in the marrow in three of four cases in which they had been CD10(+) in the primary tissues. This is consistent with the recent findings of Attygalle et al., who documented a lack of CD10-expressing lymphocytes by immunohistochemistry in five of six bone marrow biopsies morphologically involved by AILT, and in contrast with the retention of CD10 expression in other extranodal sites in that study (19). The distribution of CD10(+) tumor cells in other extranodal sites correlated with that of FDC meshworks; FDC meshworks were absent in the bone marrow infiltrates. This site-specific difference in immunophenotype and stromal composition bears further investigation.
MFC would seem to be a useful technique to follow minimal residual disease (MRD) because of its ability to detect very small aberrant T-cell populations. Knowledge of the stability of aberrant antigen expression is necessary to devise strategies for following MRD. We were able to compare the immunophenotypic features of the neoplastic cells in follow-up specimens in three cases in the present series. Each showed changes in antigen expression, but these were usually relatively minor. Notably, each population continued to manifest multiple immunophenotypic aberrancies.
In summary, an aberrant neoplastic T-cell population can be detected by MFC in most, if not all, cases of AILT. These may be present in very small numbers in primary lymph node specimens. However, the presence of multiple immunophenotypic aberrancies in most cases allows detection of even very small neoplastic populations, suggesting that MFC may represent a useful ancillary modality at diagnosis, for staging, and to detect MRD in AILT.
Since submission of this manuscript, Yuan et al. (Hum Pathol 2005;36:784–791) reported detection of CD10 on T cells by flow cytometry in 6 of 8 cases of AILT.