Cancer Cell Biology
Angioimmunoblastic T cell lymphoma is derived from mature T-helper cells with varying expression and loss of detectable CD4
Article first published online: 24 OCT 2002
Copyright © 2003 Wiley-Liss, Inc.
International Journal of Cancer
Volume 103, Issue 1, pages 12–20, 1 January 2003
How to Cite
Lee, S.-S., Rüdiger, T., Odenwald, T., Roth, S., Starostik, P. and Müller-Hermelink, H. K. (2003), Angioimmunoblastic T cell lymphoma is derived from mature T-helper cells with varying expression and loss of detectable CD4. Int. J. Cancer, 103: 12–20. doi: 10.1002/ijc.10758
- Issue published online: 26 NOV 2002
- Article first published online: 24 OCT 2002
- Manuscript Accepted: 23 AUG 2002
- Manuscript Revised: 5 AUG 2002
- Manuscript Received: 6 FEB 2002
- Korea Science and Engineering Foundation
- angioimmunoblastic T-cell lymphoma;
- double labeling;
- clinicopathologic correlation
Angioimmunoblastic T cell lymphoma (AILT) is a rare lymphoma that is regarded as a clinicopathologic entity but shows considerable histomorphologic diversity, variable immunophenotypes and inconsistent T cell receptor (TCR) gene rearrangement. One hundred four paraffin blocks of AILT were investigated defining tumor cell lineage by triple immunostains with a confocal laser scanning microscope and correlating morphology, immunophenotype and TCRγ gene rearrangement to clinical outcome. Ninety-nine cases were CD4+, some of them showing a mixture of CD4+ and CD4− tumor cells. The remaining 5 specimens were CD3+/CD4−/CD8−. A considerable number of T cells of different subtypes could always be found, but even in 13 cases predominated by CD8+ cells, proliferation could be attributed to atypical CD4+ cells. TCRγ gene rearrangement was monoclonal in 48 cases (69%) among 70 tested. In 29 of these semi-quantitative gene scan analysis resulted in a median proportion of monoclonal peak of 35% of PCR-products. Clinical outcome was identical grouping patients by clonality of TCRγ, absence or presence of clear cell clusters and international prognostic index. We conclude that AILT is mainly derived from CD2+CD3+CD4+CD5+CD7− mature T-helper cells with varying expression and partial loss of detectable CD4. A significant number of non-neoplastic T cells (resting CD4+ T cells and activated small or medium-sized CD8+ lymphocytes) may coexist with a minor neoplastic T cell population. Clinicopathologic correlation suggests AILT to be a well defined homogeneous entity with poor prognosis. Currently no prognostic factors can be derived. © 2002 Wiley-Liss, Inc.
Angioimmunoblastic lymphadenopathy with dysproteinemia was originally described by various groups1, 2, 3 as a systemic lymphoproliferative disorder characterized by generalized lymphadenopathy, hepatosplenomegaly, fever, skin rash, anemia and immunologic abnormalities. Histologically, the lymph node structure is diffusely effaced by a hypocellular infiltrate including a mixture of small, medium-sized and large atypical lymphoid cells, plasma cells, eosinophils and histiocytes and a proliferation of high endothelial venules.4
For some years it could not be clarified whether the disease should be considered a lymphoma1, 5, 6 or an abnormal hyperimmune reaction,2, 4 because the fatal outcome was mostly attributed to severe infection rather than lymphoproliferation.2, 4 Subsequent studies, however, showed evidence of a clonal process demonstrating chromosomal abnormalities characteristic for T cell lymphoma7, 8, 9 or a clonal rearrangement of T cell receptor genes.10, 11, 12, 13, 14, 15, 16 Based on these and immunophenotypic studies, the disease was incorporated into the updated Kiel classification of non-Hodgkin's lymphoma as “T cell lymphoma of angioimmunoblastic type”.17
Angioimmunoblastic lymphoma (AILT) has also been recognized by the REAL18 and the World Health Organization (WHO) classification19 as a separate disease entity of peripheral T cell lymphoma (PTCL). A morphological variant, AILT with hyperplastic germinal centers has been described recently.20
It remains controversial whether the disease represents a homogeneous entity or should be divided into a preneoplastic and a neoplastic phase because some cases lack definite cytologic atypia and approximately 20–30% of AILT showed no obvious evidence of clonality.12–15, 21–25 Criteria to predict an indolent clinical course have been proposed and were based on the absence of clear cells and lack of clonal T cell receptor gene rearrangement.6, 12, 13, 26–29 Based on clinical data, others suggested the presence of clustered or sheet-like proliferations of immunoblasts or pale cells of T cell nature as a distinguishing criterion for AILD-like T cell lymphoma from reactive angioimmunoblastic lymphadenopathy.6, 28 The predictive value of these prognostic factors has not been settled so far.
Ongoing research on AILT has also focused on the functional immunophenotype of putative tumor cells. These investigations proved to be rather difficult because of the morphologic heterogeneity of the cellular composition in AILT. Additionally, the nuclear morphology is not well preserved in frozen tissue sections. Although in most series the majority of cases seemed to be derived from CD4+ T cells, a minority of tumors was regarded to be CD8+.12, 13, 16, 20, 23, 24, 27, 30–39. In some studies, a considerable number of T cells was thought to be part of the reactive inflammatory background infiltrate of T cell lymphoma.24, 33 Therefore, the quantitatively predominant T cell subset might not necessarily represent the tumor cell lineage.
We addressed the question of tumor cell phenotype evaluating a large series of AILT with triple immunostaining on paraffin sections using a confocal laser scanning microscope. These observations were correlated to a semi-quantitative analysis of T cell receptor gene γ-chain rearrangement in the involved tissue to define the amount of the neoplastic infiltration and clinical follow-up data.
MATERIAL AND METHODS
Case selection, histologic examination and immunohistochemistry
The files of the Institute of Pathology, Würzburg University, Germany, were searched for cases diagnosed as angioimmunoblastic lymphadenopathy with dysproteinemia or angioimmunoblastic T cell lymphoma and 104 cases presenting primarily in the lymph nodes were included in our study. All cases were stained with H&E, Giemsa, periodic acid-Schiff and silver impregnation (Gömöri) to assess morphology. Routine immunohistochemical stains were carried out using mouse monoclonal antibodies (MoAb) against CD3, CD4, CD5, CD56, Ki67 (Novocastra, Newcastle upon Tyne, United Kingdom), CD2, CD8, CD20, CD21, kappa, lambda, CD30 and LMP (all Dako, Hamburg, Germany), Granzyme B (Monosan, Uden, The Netherlands), CD7, CD57 and TIA-1 (Coulter/Immunotech, Marseille, France).
Diagnoses were established morphologically and immunophenotypically relying on established criteria.3, 19, 40 Lymph node architecture was effaced by dispersed and variegated cellular infiltrates of atypical lymphoid cells that expressed T cell markers and often included characteristic clear cells. These were accompanied by a marked increase of arborizing venules, small lymphocytes, plasma cells, plasma cell precursors (B-immunoblasts), admixed with eosinophils and histiocytes. Characteristic proliferations of CD21+ follicular dendritic cell networks could be demonstrated outside follicles.40, 41, 42
Detailed morphological assessment included those features listed in Table I. Particular cell types (e.g. basophilic blasts, plasma cells, CD20+ blasts) were scored as “many” if they were easily discernible and occurred in groups. Summarizing these, 2 morphological groups (M1 and M2) were defined attempting to recapitulate the distinction between prognostic groups as published previously.6, 29 Groups were distinguished according to the distribution of morphologically recognizable clear cell clusters and immunoblast-like cells of T cell nature. Whereas group M2 included cases showing easily identifiable clear cell clusters or large blastic T cell with significant nuclear atypia, the infiltrates in group M1 was more reminiscent of reactive changes showing only scattered large atypical T-blasts or absent or few scattered clear cells.
|Radiating follicular dendritic cells (CD21+)||102||98|
|Marked vessel proliferation||99||95|
|Preserved sinus pattern||99||95|
|Basophilic blasts, many||93||89|
|Plasma cells, many||90||87|
|Perinodal spread to fat||77||74|
|Clear cell clusters||69||66|
|Total effacement of structures||59||57|
|Marked lymphocytic atypia||45||43|
|Epithelioid cell clusters||24||23|
Triple or double immunostaining with confocal laser microscopic examination
Immunohistochemical triple or double staining was attempted for 81 cases with combination of CD4/CD8/Ki67, CD4/CD8, CD4/Ki67 and CD8/Ki67. Triple staining for CD4/CD20/Ki67 and CD4/CD3/Ki67 was additionally carried out if necessary. Interpretable results could be obtained in 56 well-fixed cases and surface expression of CD4 or CD8 of the proliferating cells was defined.
To obtain strong signals and overcome autofluorescence of formalin-fixed paraffin-embedded tissue sections, we applied an indirect immunohistochemistry method using streptavidin conjugated with fluorescence dyes (Cy3, Cy5) for the first 2 steps and direct FITC-labeling of Ki67 for the last step. All primary monoclonal antibodies except Ki67, were in routine dilutions. Four micrometer sections were cut and dewaxed overnight in xylene. The following steps were carried out at room temperature, unless stated otherwise:
- 110% goat blocking serum; 20 min
- 2primary MoAb against CD4 (DAKO, Hamburg, Germany) (1:5); 3 hr (or optimal dilutions of other primary MoAb for 1 hr)
- 3biotinylated goat anti-mouse IgG (Biogenex supersensitive); 1–2 hr
- 4Cy3-conjugated streptavidin (Jackson ImmunoResearch laboratories, Inc. West Grove, PA); 2 hr
- 5unconjugated streptavidin; 30 min
- 6AffiniPure Fab-unconjugated fragments goat antimouse IgG (H+L) (Jackson ImmunoResearch Laboratories); 2 hr
- 710% goat blocking serum; 15 min
- 8Anti CD8 (1:10); 1 hr
- 9biotinylated secondary antibody (Zymed); 40 min
- 10Cy5-conjugated streptavidin (Jackson ImmunoResearch Laboratories); 30 min.
- 11Anti-Ki67, directly conjugated with FITC (Dianova, Hamburg) (1:3); overnight; 4°C.
Steps 5 and 6 were included to avoid cross-reaction of the detection systems. For each fluorochrome label, negative control sections were included. For positive control for CD4/CD8/Ki67 triple staining, paraffin sections of T-lymphoblastic lymphoma and reactive tonsil were used.
Confocal laser scanning microscope analysis.
Confocal fluorescence images were obtained on a Leica TCS NT (Leica Microsystems, Heidelberg, Germany) confocal system, equipped with Ar/Kr laser. Images were taken using a 40× 1.25 NA objective. Possible cross-talk between FITC, Cy3 and Cy5, which could give rise to false-positive signals, was avoided by careful selection of imaging conditions. The standard FITC/Cy3/Cy5 filter combination and Kalman averaging was used. Color photomicrographs were taken from electronic overlays.
TCRγ gene rearrangement by PCR-based gene scan analysis or TGGE-PCR
Seventy cases were successfully examined for TCRγ gene rearrangement by PCR analyzing the product with either gene scan (PCR-GSA; 44 cases) or temperature gradient gels (26 cases). For the other cases, no material was available after immunostains or DNA quality was poor. In 44 cases analyzed by PCR-GSA, the correlation between the magnitude of the monoclonal TCRγ DNA peak and the amount of tumor cells in the tissue sections estimated morphologically was analyzed.
For molecular studies, paraffin blocks were dewaxed and DNA was extracted as described previously.43 PCR-GSA for detection of clonal T cell populations was carried out with specific V and J segment primers as described by Trainor et al.,44 however, all J segment primers were labeled with the fluorescent dye FAM (Perkin-Elmer, Weiterstadt, Germany). On every case, 3 separate PCRs with 3 different DNA template concentrations were carried out. Aliquots of the PCR reactions were then mixed with size standard and formamide, denatured and subjected to electrophoresis on a ABI377 DNA Sequencer (Perkin-Elmer, Weiterstadt, Germany). The automatically collected data were analyzed using Genescan software (ABI 671, Applied Biosystems, Perkin Elmer) as described in manufacturer's manual.
All PCRs were run with a negative and a positive (Jurkat-cell line) control. Monoclonality was defined as a presence of an identical dominant peak twice exceeding the second highest peak, in at least 2 separate PCRs. Biallelic rearrangement resulted in duplicate peaks differing in their size by no more than 30%. For the calculation of the malignant-cell percentages in biallelic samples, we assumed that 2 rearrangements occurred in the same cell population. Biallelic pattern was therefore regarded monoclonal. Oligoclonal samples showed 3–8 isolated bands. More than 8 peaks indicated polyclonality.
In monoclonal cases the area under the monoclonal peak was calculated as percentage of the total area of all peaks from automatically calculated tabular data for each peak in each of the 3 dilutions. Data were taken to be representative if the deviation from the mean was ≤5% in at least 2 dilutions. This calculated percentage was taken as a rough estimate for the percentage of clonally rearranged T cells among all T cells in the lymph node.
To correlate this quantitative analysis to morphology, the proportion of cells with nuclear atypia (clear cells or large T-blasts) was estimated independently by 2 observers (SSL and TR) in 10% steps on sections of blocks stained for H&E and CD3, because CD3 would stain those cells giving an amplificate in the TCRγ PCR reaction. Patterns in corresponding slides stained for CD4, CD8 and Ki67 were additionally considered but not entered in this quantitative estimation.
Clinical evaluation and statistical analysis
Clinical data could be obtained from patient records for 47 patients. Regarding the distribution of the morphological parameters evaluated here, this cohort did not differ from the patients for whom detailed clinical data could not be obtained. Estimates of overall survival distribution and failure-free survival were calculated using the method of Kaplan and Meier;45 time-to-event distributions were compared using the log-rank test. The individual items for histologic features, immunologic and clinical parameters including international prognostic index (IPI)46 were estimated in correlation with the other items and clinical outcome using χ2 tests or t-tests as appropriate.
Histological findings of 104 patients with AILT are summarized in Table I. Characteristic irregular radiating proliferations of follicular dendritic cells were detected by CD21 immunostains in all but 2 cases. These latter showed all other characteristic features of AILT, such as marked vessel proliferation with arborizing pattern, heterogeneous cellular composition with a proliferation of B-blasts and plasma cells, easily identifiable clear cell clusters, PAS+ fibers and spread into perinodal fat. In the other cases the extension of CD21+ FDC networks varied widely from only focal irregular to highly exuberant. The FDC proliferation could also be recognized in 78 cases (75%) in H&E stains.
In most cases follicles were absent or seen as regressing remnants. Seven cases, however, showed hyperplastic germinal center B-cells with ill-defined borders without mantle zone as described by Ree et al.,20 which were different from normal or reactive follicles. Eleven percent of 104 AILT showed high amount of CD20+ B-blasts, which did not show any correlation with other morphologic variables.
Clusters or sheets of clear cells were observed to a variable extent in 66% of the cases. Morphological grouping (Table II) was mainly based on the amount of clear cell clusters and T-blasts. Seventy-four cases with prominent clear cell clusters or T-blasts were grouped as M2 possibly reflecting a higher morphological grade. Thirty cases lacking these features were regarded to represent a more low-grade morphology (Group M1). Morphological group M2 (high-grade) were more often monoclonal and showed a significantly higher proliferation index of the tumor cells when compared to M1 (low-grade) group (Table II). The amount of CD20+ B-blasts was variable within both groups. Hyperplastic germinal center cell proliferation tended to occur more often in M2 cases.
|M1 (low-grade) (n = 30)||M2 (high-grade) (n = 74)||p-value|
|Reactive CD8+ cells|
|Low||13 (43%)||48 (65%)||NS|
|High||17 (57%)||26 (35%)|
|Proliferation-index (mean)||33||46||p < 0.05|
|Monoclonal||12 (55%)||36 (75%)||p < 0.01|
|Polyclonal||10 (45%)||12 (25%)|
|Median overall survival (months)||14.5||12.5||NS|
|Median failure-free survival (months)||8.7||8.0||NS|
Tumor cell phenotype
T-lymphocytes highlighted by stains for CD3 or CD5 displayed a wide morphological spectrum reaching from small (resting) over activated, slightly atypical medium-sized to large atypical lymphoid cells. T-immunoblasts and cells with abundant cytoplasm and medium-to-large atypical nuclei, corresponding to clear cells in H&E or Giemsa stains also occurred. Most clear cell clusters and large atypical T-blasts were positive for CD4, whereas CD8 was expressed mainly in scattered small or medium-sized lymphocytes. CD8+ T cells occasionally showed mildly atypic, irregular and enlarged nuclei.
The proportion of each T cell subset was variable. In 61 cases (59%), CD4+ cells by far exceeded CD8+ cells (Group A). In 13 cases (13%), CD8+ cells outnumbered CD4+ cells, 3 cases of which were dominated by an CD8+ infiltrate comprising more than 70% of all T cells (Group B). Thirty cases (29%) had about equal numbers of CD4+ and CD8+ T cells (Group C).
In the cases containing a predominant CD4+ population (Group A), most clear cells and T-immunoblasts could be easily detected with CD4 in routine immunohistochemistry. With CLSM examination (Fig. 1), applying double or triple immunostains, most proliferating cells proved to be of CD4+ lineage. These tended to aggregate forming sheet-like groups. CD8+ T cells were distributed in a scattered reactive pattern and showed a very low proliferation rate. Small CD4+ T cells showed no nuclear atypia in conventional stains and their proliferation rate was comparable to that of CD8+ lymphocytes.
In the cases with a more prominent CD8+ population (Groups B,C), it was difficult to define the functional immunophenotype by single immunohistochemistry. With aid of the CLSM, double or triple immunostains showed that most proliferating cells were large in size and expressed CD4+. These cells presented in aggregates or isolated. The CD8+ population seemed to be smaller in size and less proliferating when compared to the CD4+ population. Even in those 3 CD8 predominant cases, CD4+/Ki67+ clusters of large cell were found multifocally (proliferation rate 80%) in the sea of CD8+ T cells with low proliferation (proliferation rate <10%) (Fig. 2).
In line with these observations, only 50% of our AILT cases were clonal in Group B with a predominance of CD8+ cells, whereas 71% of the evaluated cases were clonal in both Groups A and C. This difference was not statistically significant, however and probably not all CD4+ cells in AILT are tumor cells.
CD4 expression was considerably weaker on atypical cells than on reactive small lymphocytes in 61 of 99 CD4+ AILT. In some cases, the atypical cells were weakly CD4+ in some areas of the lymph node but CD4−/CD8− in others. Therefore, these latter cases were supposed to contain mixture of many CD3+/CD4+/CD8− and some CD3+/CD4−/CD8− tumor cells. Additionally, in 5 cases large atypical cells and clear cells were all positive for CD3 and CD5, but all double negative for both CD4 and CD8.
In summary, applying both routine immunohistochemistry and immunofluorescent triple stains, atypical cells could be demonstrated to be CD4+ in 99 (96%) of 104 AILT cases and CD4−/CD8− (double negative) in the remaining 5 cases (4%). There were neither double positive (CD4+/CD8+) nor CD8+ cases on atypical cell population in our series.
Identifying the possible atypical cells by their nuclear irregularity in matching areas of differently stained serial sections, their immunophenotype for additional markers was determined (Table III). Pan-T cell markers of CD3, CD5 and CD2 were positive in the atypical cells, in 97%, 98% and 100% of examined cases, respectively. CD7 was almost always negative in atypical cells except 2 cases, however, there were variable numbers of reactive CD7+ cells in each case.
|Monoclonal antibody||Positive/examined (n)||Positive (%)|
In most cases, TIA1 expression paralleled that of CD8. The number of CD8+/TIA1+ lymphocytes was variable in each case. Granzyme B, staining activated cytotoxic T cells, was variably expressed, usually staining much fewer cells than CD8 or TIA1. Medium-sized lymphocytes with slightly irregular nuclei seemed to represent activated CD8+/TIA1+/Granzyme B+ cytotoxic T cells, although small lymphocytes corresponded to both CD8+ cells and CD4+ reactive cells. Cytotoxic markers were not expressed in the atypical cells in any of our cases.
CD10 expression could be defined in 51% of the cases and did not correlate to any of the other morphological or phenotypical markers.
TCRγ gene rearrangement, analyzed by either PCR-GSA or TGGE, was monoclonal in 48 cases (69%) among 70 tested, but 22 cases (31%) were polyclonal or oligoclonal. Most TCRγ-polyclonal cases showed the more reactive pattern M1 with only scattered or low population of CD4+ atypical cells on histology (Table II). Cases having easily identifiable atypical cells by morphologic examination of H&E and immunostaining, displayed monoclonality with quantitative correlation of atypical cell amount and the percentage of monoclonal peak by PCR-GSA. Clonality of TCRγ was therefore found in 36 of 48 cases (75%) of the morphologic subgroup M2 (high-grade) and in 12 of 22 cases (55%) of the M1 (low-grade) group.
To estimate the proportion of monoclonal cells in the total T cell population, the amount of the PCR-products in the monoclonal peak was calculated as a percentage of all amplified TCRγ specific PCR-products by PCR-GSA of the rearranged gene. In 29 cases given as monoclonal by PCR-GSA, the amount of PCR-products representing monoclonal rearranged TCRγ ranged from 11–78% (median 35%) of the whole PCR products. Four cases (14%) showed a biallelic rearrangement (Fig. 3). In these cases, taking the 2 peaks as a biallelic monoclonal rearrangement, the amount of identical PCR-products correlated to the morphologically estimated amount of atypical cells among all T cells.
Clinical presentation and follow-up
The 104 patients (57 men and 47 women) were mostly elderly, ranging in age from 33–90 years (median 68 years). Detailed clinical data and follow-up could be obtained in 47 patients (Table IV).
|Age (years)||Median (range)||68 (33–90)|
|Overall survival (median)||14 months|
|Failure-free survival (median)||8 months|
Most patients presented with generalized lymphadenopathy (84%). B-symptoms and elevated serum LDH were detected in 76% each. Clinical presentation was not significantly different by morphological groups (M1 vs. M2) or by clonality of TCRγ rearrangement.
Primary therapy regimens comprised prednisolone (11 patients), single agent chemotherapy (2 patients), combination chemotherapy (CHOP, 26 patients) and high-dose chemotherapy (COP-BLAM, 2 patients). Two patients refused to be treated and 4 were treated with surgery, radiation or immunotherapy only.
The survival curves (Fig. 4) show rapid downhill course within first 2 years (2-year overall survival, 40%; 2-year survival, 19%).
Statistical analysis showed no prognostic significance for any of clinical, histologic, immunophenotypic and immunogenotypic characteristics. Morphologic subgrouping by the presence of clear cell clusters and overall T cell atypia (low-grade M1 vs. high-grade M2) did not show any difference in terms of clinical presentation or outcome (Fig. 5). Also there was no significant difference in outcome whether or not a clonal TCRγ rearrangement could be detected (Fig. 6). Even the international prognostic index (IPI) failed to predict patient outcome in our cohort (Fig. 7). Comparison between the therapy regimen also did not show any prognostic benefit in this retrospective series.
All AILT are derived from CD4+ T-lymphocytes
Because AILT has been recognized as a T cell lymphoma, most cases have been described to be CD4+. The proportion of reported CD8+ AILT varied from 0–30%, averaging 12% of all reported cases.12, 13, 16, 20, 23, 24, 27, 30–34, 36, 47, 48 The functional phenotype of the tumor cells, however, is especially difficult to interpret in AILT because the infiltrate is characteristically heterogeneous, containing abundant reactive T cells in addition to the tumor cells.
Most previous investigations, including all that were based on double staining,16, 24, 47 were carried out on frozen sections in which morphology is usually not well-preserved and surface staining probably could not be readily correlated to nuclear morphology. Additionally, CD4 expression is not restricted to T cells but histiocytes are also stained.49 Because these latter are usually non-proliferating the proliferation of putative tumor cells may be greatly underestimated.
In studies carried out on paraffin sections20, 34 no double staining was carried out for lineage and proliferation markers. Additionally, CD4 may stain inconsistently and be difficult to interpret in material that is not optimally fixed.
A numerical dominance of CD4+ or CD8+ subsets can probably not be accepted as an evidence of a neoplastic clone in AILT, because even reactive T cell proliferations can show subset predominance.50 Moreover, we have observed 30 cases (29%) lacking any predominance of CD8 or CD4 cells. In addition to these, a possible double expression of CD4 and CD8 antigen on the same cells requires the simultaneous visualization of CD4, CD8 and proliferation antigen in the same tissue section. This latter has never been studied in previous reports.
Our own series included 43 cases (41%) with a sizable CD8+ T cell population, a feature also reported previously.12, 13, 23, 30, 32–35, 47, 48. As these CD8+ T cells in part showed mild to moderate nuclear irregularity, some cases in our series were therefore initially misinterpreted to be CD8+ neoplasms. Triple or double immunostaining by CLSM examination, however, showed an unexpected multifocal presence of highly proliferating large CD4+ cells in a background of low-proliferating, small to medium-sized CD8+ cells (Fig. 2). In corresponding conventional immunostains, these latter showed minimal nuclear atypia but still there was no significant proliferative activity and both proliferation and nuclear pleomorphism were marked in CD4+ cells. Therefore, we regarded the CD8+ cells to be part of a reactive inflammatory infiltrate. These represent cytotoxic T-lymphocytes with some nuclear atypial similar to those found in the inflammatory background infiltrate of T cell rich B-cell lymphoma.51 These results are comparable to those of Nakamura et al.23 Interestingly, only 50% of the cases proved to be clonal by PCR in these CD8-dominant cases, although 71% were positive among the CD4-dominant ones.
In previous studies, CD4+ cells might not have been detected sufficiently, not only due to their small number but also to their weak expression of the antigen. The observed co-existence of a CD4+ and a double negative population of proliferating large atypical cells could be supported by a FACS analysis of native lymph node tissue in 2 cases showing an abnormally high proportion of double negative (CD4−CD8−) T cells (12.1% and 22.5% of all CD3+ T cells, respectively) (data not shown), but to clarify this finding, additional studies need to be done.
Based on these considerations, we concluded that in the majority of our cases (95%) the tumor cells were CD4+ whereas a minority (5%) was negative for both CD4 and CD8. In addition, CD4 was weakly expressed and partially lost from tumor cells in 61 of 99 CD4+ cases suggesting a possible downregulation of this molecule also in our negative cases. There was no CD8+ AILT in our series.
Tumor cells may be in the minority among T-lymphocytes
We quantitatively analyzed TCRγ PCR products by genescan to estimate the percentage of neoplastic and therefore monoclonal T cells in all T cells. Although we did not perform strictly quantitative PCR studies,52 the PCR products at least roughly correspond to the amount of the different rearranged TCRγ genes in the T cell infiltrate. Therefore monoclonal populations can be detected in a polyclonal background. Clonal populations varied widely, but represented a minority of PCR products in most (25/29) cases. In consequence, a polyclonal result may represent a very low number of tumor cells that is below the detection threshold of this method. This supports our impression that a significant number of non-neoplastic T cells does coexist with a minor neoplastic T cell population. The proportion of tumor cells in the T cell compartment, as estimated in H&E and CD3 stains, was correlated to this percentage of monoclonal PCR products defined by genescan analysis. This correlation was not very strong, however, probably because our PCR-based detection of monoclonal populations is not precisely quantitative or because our morphological criteria for tumor cells in AILT are not precise enough. It was also difficult to appreciate regional variations in tumor cell density. The more proliferating large atypical T cells and clear cells designated as tumor cells could be found, the higher the proportion of the monoclonal peak within the polyclonal background was. Concordant with our results a clonal TCR rearrangement could be detected by the analysis of single cells in a case that failed to be monoclonal by whole tissue DNA analysis.53
As also suggested in other reports,13, 23, 24 poly- or oligo-clonality in 31% of our cases may be attributed to relatively few tumor cells. Our cases can thus be regarded to represent a spectrum reaching from polyclonal over oligoclonal to monoclonal ones. Comparing clonality to the T cell subpopulations, there was no significant difference between tumors with dominant CD4+ or CD8+ populations (Groups A–C), suggesting that also among the CD4+ population the tumor cells may be in the minority. The same mechanism may have resulted in polyclonality of 25% of cases with detectable tumor cell clusters (Table II).
Our results strongly suggest that most, if not all, AILT are neoplasms derived from CD2+CD3+CD4+CD5+CD7−CD8− mature T-helper cells. The spectrum of CD4 expression in the tumor cells reaches from strong to a loss of detectable expression in some or all tumor cells. The CD3+ T cell compartment in AILT comprises highly proliferating CD4+ clear cells or CD4+ T-immunoblasts (the tumor cells), resting CD4+ small lymphocytes without atypia and activated CD8+ medium-sized or small lymphocytes with or without mild nuclear irregularity in keeping with the physiological spectrum of reactive T cells. The tumor cells are located in groups or sheets in a highly reactive infiltrate containing small and activated T-lymphocytes, B-cells, histiocytes, eosinophils and activated high endothelial venules. A study based on single cell analysis also supports our view. Separately analyzing CD4+ and CD8+ populations picked from AILT cases Willenbrock et al.53 found clonal rearrangements only in the CD4+ compartment. They could only investigate 7 cases with this technique, however, and did not correlate their results to morphology.
It has been well documented that peripheral T cell lymphomas (PTCL) often lack pan T cell antigens and, in addition, occasionally express a CD4+CD8+ or CD4−CD8− phenotype that is regarded as abnormal.54, 55 In our study, almost all AILT cases tested showed pan-T cell antigen loss. All but 2 cases (86/88) manifested CD7 antigen loss, an additional 2 cases were CD5− and 5 cases showed CD4−CD8− double negative phenotype.
As in other studies,2, 12, 22, 23, 56, our AILT patients had a poor prognosis. It was impossible to define any prognostic factors in our study, which is interesting even if patients were heterogeneously treated. Clinical parameters at presentation combined in the International prognostic index (IPI) score failed to predict outcome. IPI score was shown to be relevant in other peripheral T cell lymphomas.57, 58 These studies analyzed patients with PTCL-NOS and AILT together, however, and the latter were in minority. Also, the IPI score was not a significant predictor of survival in the international lymphoma classification project for 17 cases of AILT, but the prognostic power increased for all cases of nodal peripheral T cell lymphoma as a group after AILT was excluded.59 To the best of our knowledge, no prospective study has tested the predictive capacity of the IPI in AILT independently from other PTCL.
Case numbers were very low in the various groups of our study, but currently available therapy did not seem to influence survival significantly. Additionally, neither clonality of tumor cells nor grouping lymphoma according to proposed criteria for the distinction of indolent from aggressive AILT6, 12, 13, 26–29 were of prognostic relevance.
The nature of AILT has been controversial since the time of its first description and remains so today. Although AILT was incorporated in the REAL classification as a distinct type of peripheral T cell lymphoma named “angioimmunoblastic T cell lymphoma” with fatal clinical outcome,18 heterogeneity in some aspects of morphology or molecular data has lead to a discussion whether AILT is a single disease entity or a heterogeneous disease group and whether prognostic factors can be defined within this entity. Some investigators,12, 25, 27 and even textbooks,60 suggested that AILT should be regarded as an arbitrarily defined morphologic portion of a spectrum of atypical immunoproliferative disorders ranging from probably reactive and reversible to clearly neoplastic and highly aggressive.
This view might be attributed to a consistent existence of a minority of AILT showing a reactive histological pattern without easily identifiable tumor cells, lacking monoclonal rearrangement of TCR genes. Attempts to separate “prelymphoma” from lymphoma in AILT has been based on the absence of clear cells and molecular clonality.12, 13, 26–28 It has become increasingly apparent, however, that a distinction between these lesions, if possible at all, is far from being made.11–13, 22, 23, 26, 32 Because of its fatal outcome it is crucial to distinguish AILT from reactive and possibly curable diseases, especially lymphadenopathies in the state of immunosuppression or autoimmune lymphadenitis. This differential diagnosis requires the synopsis of all available data.
Tobinai et al.13 suggested the presence of focal or sheet-like proliferations of immunoblasts or pale cells of T cell nature as a distinguishing diagnostic criterion for T cell lymphoma from reactive angioimmunoblastic lymphadenopathy. They also stated that some cases with less prominent such diagnostic lesions may be histologically indistinguishable from what others called angioimmunoblastic lymphadenopathy. In contrast to their data our accordingly defined morphological groups M1 and M2 were not different with regard to anything else but proliferation index (Fig. 5). Especially, these criteria cannot predict the outcome of the patient and should therefore not be used in the differential diagnosis.
Our study also demonstrated identical outcome grouping patients by clonality of TCRγ gene rearrangement or absence and presence of clear cell clusters, respectively. Therefore, the absence of detectable clonal gene rearrangement does not exclude neoplastic growth and fatal outcome in patients with suspected AILT. Concordant to other studies,12, 13, 22, 23, 26 evidence of clonality or the detection of clear cell clusters are surely helpful for diagnosis of AILT but, however, cannot be regarded as essential diagnostic criteria. The first description of AILT in the early seventies was quite characteristic and remains sufficient to make a diagnosis of AILT.3, 4 According to our clinicopathological evaluation, we suggest to regard AILT as a well-categorized, homogeneous, distinct entity as originally described.
O'Connor et al.11 proposed 2 hypotheses for the etiology of AILT based on Southern blotting: 1) that a monoclonal proliferation of T cells may always be present (but not always detectable) and that T and B cells proliferate secondarily to this underlying monoclonal population; or 2) that AILT may initially be a polyclonal hyperreactive disorder, affecting both T and B cells and a monoclonal T cell population may arise as a secondary event. Many investigators supported his second hypothesis. The reactive histological appearance and minor nuclear atypia of AILT suggested a hyperplastic/dysplastic state of T cells,61 from which lymphoma could develop due to a decreased immunosurveillance and this process is illustrated by an evolution from hyperplasia to lymphoma.27
For several reasons, we strongly prefer the first hypothesis, because 1) we could show that the neoplastic clone may comprise a minority of T cells, 2) the reactive features may result from defined but so far unknown paracrine effects exerted by the tumor cells. Additionally, 3) our clinicopathologic analysis showed an identical outcome not only between the so-called angioimmunoblastic lymphadenopathy of Tobinai et al.13 (Group M1) and AILT (Group M2), but also with or without detectable clonal gene rearrangement and 4) occasionally a patient may show different histology in the different lymph nodes at the same time or with a very short interval.20, 27 In a cytogenetic study the equivalent cytogenetic aberrations were demonstrated in atypical hyperplasia and lymphoma with AIL-feature.31
- 19World Health Organization classification of tumours. Pathology and genetics of tumours of hematopoietic and lymphoid tissues. Lyon: IARC Press, 2001., , , .
- 22Angioimmunoblastic lymphadenopathy type of T cell lymphoma and angioimmunoblastic lymphadenopathy: a clinicopathological and molecular biological study of 13 Chinese patients using polymerase chain reaction and paraffin-embedded tissues. Virchows Arch 1994;424: 593–600., , , , , .
- 29Surgical pathology of the lymph nodes and related organs. In: LivolsiVA, ed. Major problems in pathology, 2nd ed., vol. 16. Bethesda: W.B. Saunders Company, 1995. 391–409..
- 40Histopathology of non-Hodgkin's lymphomas (based on the updated Kiel classification), 2nd ed. New York: Springer-Verlag, 1992., .
- 60Ackerman's surgical pathology, 8th ed. New York: Elsevier, 1996..