- Top of page
The World Health Organization (WHO) lymphoma classification recognises anaplastic large cell lymphoma (ALCL), angioimmunoblastic lymphoma (AIL) and peripheral T-cell lymphoma, unspecified (PTCU) as nodal mature T-cell lymphomas. Little is known about long-term outcome and prognostic factors of these diseases. A retrospective analysis on 125 patients with ALCL, AIL or PTCU was performed to evaluate outcome parameters, taking into account histological subtype and the International Prognostic Index (IPI). Median age was 54 years (range 17–90 years). Complete remission (CR) was achieved in 51% of patients. Five-year overall survival (OS) was 43%, and 5-year relapse-free survival was 69%. Five-year OS was 61% for ALCL, 45% for PTCU and 28% for AIL. With regard to the IPI, 5-year OS was 74%, 49%, 21% and 6% for the low, low-intermediate, high-intermediate and high risk groups, respectively. In the multivariate analysis, the IPI but not the histological subtype significantly predicted survival. To a large extent, the IPI score explains the differences in survival between histological subtypes of nodal mature T-cell lymphomas. The IPI may therefore be used for risk stratification in clinical trials to identify patients who would benefit most from new treatment strategies, such as high-dose chemotherapy followed by stem cell or bone marrow transplantation.
Mature T- and natural killer-cell lymphomas represent a rare and heterogeneous group of neoplasms accounting for about 12% of all non-Hodgkin lymphomas in the Western countries and Asia. They are currently diagnosed according to the classification system of the World Health Organisation (WHO), which further separates these entities into leukaemic or disseminated T-cell neoplasms, extranodal (including cutaneous) T-cell lymphomas and nodal T-cell lymphomas (Jaffe et al, 2001). Nodal T-cell lymphomas are subdivided into peripheral T-cell lymphoma, unspecified (PTCU), angioimmunoblastic lymphoma (AIL), and anaplastic large cell lymphoma (ALCL), with PTCU being the most frequent entity among these rare diseases. The enormous diagnostic challenges, the changing classification systems (Stansfeld et al, 1988; Harris et al, 1994), and the widely differing therapeutic approaches taken to treat these diseases contribute to the uncertainty about factors that determine prognosis. The aim of this study was to investigate whether the International Prognostic Index (IPI), originally described for patients with aggressive B-cell lymphomas (The International Non-Hodgkin's Lymphoma Prognostic Factors Project, 1993), can also be used to predict the survival of patients with nodal mature T-cell lymphomas as classified according to WHO. This question is of particular importance with respect to evaluating the adequacy and success of any therapeutic intervention, including more aggressive modalities, such as autologous and allogeneic transplantation of haematopoietic stem cells (Rodriguez et al, 2001; Song et al, 2003; Morris et al, 2004; Reimer et al, 2004) or the use of monoclonal antibodies (Enblad et al, 2004).
- Top of page
This study was an analysis of 125 consecutive patients diagnosed with ALCL, PTCU and AIL and treated at the Department of Haematology, AK St.Georg, Hamburg, between 1980 and 2000. Mature T-cell lymphoma other than ALCL, PTCU and AIL were excluded as the low number of patients did not allow a meaningful analysis. Patients with precursor T-cell lymphoma were excluded because the biology and treatment strongly differs from nodal mature T-cell lymphomas.
Diagnoses were originally made according to the Kiel classification (Stansfeld et al, 1988) or – after 1994 – according to the Revised European–American classification of lymphoid neoplasms (REAL classification) (Harris et al, 1994). In this report, the corresponding entity of the WHO classification (Harris et al, 2000) is given. Diagnoses were confirmed by German lymphoma expert pathologists in 86% of cases.
Clinical stage was defined according to the Ann Arbor classification (Carbone et al, 1971). Disease spread was determined on the basis of history, physical examination, computed tomography of chest, abdomen and pelvis, and a bone marrow biopsy. In recent years, immunophenotyping of blood or marrow cells by flow cytometry was also performed.
Treatment strategies varied over time. For the purpose of this analysis, treatment was grouped into one of six categories (modified after Pellatt et al, 2002): supportive care, single agent chemotherapy, radiotherapy (RT), combination chemotherapy (CHT), combined modality treatment (CHT plus RT), and high-dose chemotherapy (HDCT) followed by autologous bone marrow (BMT) or stem cell transplantation (SCT). Treatment strategies for all patients are summarized in Table I. The vast majority of patients (88%) received CHT with or without RT, in selected cases this was followed by HDCT. First-line CHT consisted of CHOP (cyclophosphamide, hydroxy-daunorubicin, vincristine, prednisone; n = 69), or CHOP-like regimens (i.e. COP-BLAM: cyclophosphamide, vincristine, procarbazine, bleomycin, adriamycin, methylprednisolone, n = 19; or HOAP-Bleo: hydroxy-daunorubicin, vincristine, cytosine-arabinoside, prednisolone, bleomycin, n = 5). Eleven patients were treated with COPP (cyclophosphamide, vincristine, procarbazine, prednisolone). In 38 patients with poor response to first-line therapy and/or high-risk disease, initial treatment was continued with a regimen frequently used to treat T-ALL in Germany (prednisolone, vincristine, daunorubicin, asparaginase) (n = 10) (Hoelzer et al, 2002), by HOAP-Bleo (n = 14), or by DEXA-BEAM [dexamethasone, BCNU (carmustine), etoposide, cytosine-arabinoside, melphalan] chemotherapy (n = 14)(Pfreundschuh et al, 1994). Twelve patients received this latter treatment prior to BMT (n = 4) or SCT (n = 8).
Table I. Treatment strategies.
| ||Overall (n = 125)||ALCL (n = 21)||PTCU (n = 70)||AIL (n = 34)|
|Combined CHT [n (%)]||52 (42)||9 (43)||26 (37)||17 (52)|
|Combined CHT + RT [n (%)]||44 (36)||9 (43)||27 (39)||8 (24)|
|Combined CHT + SCT/BMT+/− RT [n (%)]||12 (10)||3 (14)||7 (10)||2 (6)|
|Radiotherapy [n (%)]||4 (3)||0 (0)||4 (6)||0 (0)|
|Single agent CHT, [n (%)]||8 (7)||0 (0)||3 (4)||5 (15)|
|Supportive care [n (%)]||4 (3)||0 (0)||3 (4)||1 (3)|
Extended or involved field RT was administered to 52 patients starting 4 weeks after completion of CHT using an 8-MeV linear accelerator at the Hermann Holthusen Institute for Radiotherapy at our hospital. The doses used for RT varied from 30 to 40 Gy: a boost of 10 Gy was given to areas of bulky disease.
Supportive care or single agent chemotherapy as the only treatment was given to patients who were deemed unsuitable candidates for more aggressive treatment strategies because of advanced age or severe comorbidities.
Response evaluation and statistical analysis
Response was evaluated after the third cycle of CHT and approximately 4 weeks after completion of all therapy. Complete response (CR) was defined as the disappearance of all measurable disease for at least 3 months after the end of therapy, partial response (PR) as a reduction of all tumour manifestations by at least 50%. Overall survival (OS) was defined as the time interval from diagnosis until death of any cause. The relapse-free survival (RFS) of patients who reached CR was calculated from the onset of CR until evidence of relapse. OS and RFS were calculated using Kaplan–Meier estimates. Subgroups were compared with the log rank test.
For the multivariate analysis, the time from diagnosis until death of any cause was used as the time variable in a Cox regression model. The variables, gender, histological subtype and IPI, were included in the final model. Statistical significance of the variables investigated was tested by the likelihood ratio test. The proportional hazard assumption was fulfilled for the variables tested.
The Kruskal–Wallis test was used to compare continuous variables (IPI, Ann Arbor stage, age). Categorical variables were compared with the chi-square test. All calculations were performed in STATA 8.1. P < 0·05 were considered significant.
- Top of page
Patient characteristics according to histology are given in Table II. Two patients with ALCL showed an anaplastic lymphoma kinase (ALK)-negative phenotype, the remaining 19 patients were ALK-positive. Compared with other histological subtypes, patients with ALCL were younger and less frequently presented with elevated lactate dehydrogenase (LDH) or bone marrow involvement. ALCL patients and PTCU patients presented with a better performance status (PS) and less frequently had B-symptoms. As a consequence, most patients with ALCL fell into the low or low-intermediate risk category of the IPI. In contrast, patients with AIL more often presented with unfavourable clinical and laboratory features (B-symptoms, low PS, bone marrow infiltration, elevated LDH): thus, 63% of these patients were IPI high-intermediate or high risk at the time of diagnosis. Patients with PTCU had a relatively high frequency of extranodal disease; in particular, 40% of these patients showed involvement of the skin at the time of diagnosis. Seven patients presented with autoimmune phenomena: five patients had a positive Coombs test, one patient presented with overt autoimmune haemolytic anaemia, one patient had autoimmune thrombocytopenia; another patient suffered from arthritis with positive anti-nuclear antibodies and perinuclear antineutrophil cytoplasmic antibodies.
Table II. Patient characteristics.
| ||Overall (n = 125)||ALCL (n = 21)||PTCU (n = 70)||AIL (n = 34)||P-value|
|Age (years) [median (quartiles)]||54 (40–64)||41 (33–57)||55 (39–64)||57 (44–72)||0·04|
|Female [n (%)]||51 (41)||8 (38)||30 (43)||13 (38)||0·87|
|Clinical stage [n (%)]||0·02|
| I||18 (15)||3 (14)||14 (20)||1 (3)|| |
| II||16 (13)||7 (33)||8 (12)||1 (3)|| |
| III||26 (21)||4 (19)||12 (17)||10 (29)|| |
| IV||64 (52)||7 (33)||35 (51)||22 (65)|| |
|IPI [n (%)]||0·001|
| Low Risk||43 (35)||9 (43)||28 (41)||6 (18)|| |
| Low-intermediate||31 (25)||6 (29)||19 (28)||6 (18)|| |
| High-intermediate||30 (24)||6 (29)||15 (22)||9 (27)|| |
| High risk||19 (15)||0 (0)||7 (10)||12 (36)|| |
|B-symptoms [n (%)]||58 (47)||9 (45)||26 (38)||23 (68)||0·02|
|LDH >240 [n (%)]||46 (38)||5 (25)||22 (31)||19 (59)||0·01|
|WHO PS >1 [n (%)]||28 (24)||3 (15)||12 (18)||13 (45)||0·01|
|Extranodal disease [n (%)]||81 (65)||11 (52)||47 (67)||23 (68)||0·43|
| Skin||43 (34)||5 (24)||28 (40)||10 (29)||0·30|
| Bone marrow||34 (27)||1 (5)||16 (23)||17 (50)||0·001|
| Pleura||10 (8)||2 (10)||5 (7)||3 (9)||0·92|
Response to treatment and survival
Median time of follow-up was 31 months (range 1–249 months). Overall, CR was achieved in 53% and PR in 33% of the patients. Treatment results according to histological subtype are summarized in Table III. CR rate, OS and RFS were most favourable for ALCL. The differences in CR rates were significant (P = 0·03). Relapse occurred in 27% of patients with CR, with no significant differences between the histological subgroups (P = 0·51). OS curves according to histology are shown in Fig 1. In pairwise univariable analysis, OS for AIL was significantly lower compared with ALCL (P = 0·01). The differences in OS between PTCU and ALCL as well as between PTCU and AIL were considerable but not significant (P = 0·10 and P = 0·08). Differences in RFS were not significant (ALCL versus PTCU P = 0·26; ALCL versus AIL P = 0·89; PTCU versus AIL P = 0·39).
Table III. Treatment effects according to histology.
| ||CR (%)||Relapse (%)*||5-year OS (%)||95% CI||5-year RFS (%)*||95% CI*|
|Overall (n = 125)||53||27||43||34–52||69||56–78|
|ALCL (n = 21)||71||20||61||37–78||80||49–93|
|PTCU (n = 70)||55||32||45||33–56||66||47–79|
|AIL (n = 34)||36||18||28||14–44||64||30–85|
In univariate analysis, survival for patients presenting in stage IV [5-year OS: 25%, 95% confidence interval (CI) 15%-36%] was significantly inferior (P < 0·01 all pairs) compared with all other stages. Differences between stages I (5-year OS: 71%, 95% CI 43–87%), II (67%, 95% CI 37–85%), and III (55%, 95% CI 33–72%) were not significantly different.
Figure 2 shows the OS according to the IPI. Five-year OS was 74% (95% CI 57–84%) for the low risk group, 49% (95% CI 30–65%) for the low-intermediate, 21% (95% CI 8–38%) for the high-intermediate, and 6% (95% CI 0–22%) for the high risk group. The IPI groups significantly differed from each other in pairwise univariate analysis. Only the difference in OS between the high-intermediate and the high risk groups was not statistically significant (P = 0·10).
Table IV shows the result of the Cox regression analysis. Before inclusion of the IPI, the histological subtype significantly influenced survival (P = 0·03, gender adjusted). In the final model the IPI significantly predicted survival (P < 0·0001) in contrast to the histological subtype. To avoid over-adjustment, age was not included in the final model, as this factor is represented in the IPI. Additional adjustment for age did not lead to substantial changes of the model. For the same reason, and with the same results, bone marrow infiltration and stage also were not included in the final model.
Table IV. Multivariable Cox regression analysis of overall survival.
| ||Hazard ratio||95% CI||P-value|
| ALCL (reference)||1·0||–|| |
| PTCU||1·6||[0·8–3·4]|| |
| AIL||1·7||[0·8–3·8]|| |
|International Prognostic Index||–||–||<0·0001|
| Low risk (reference)||1·0||–|| |
| Low-intermediate||2·4||[1·2–4·7]|| |
| High-intermediate||4·3||[2·2–8·3]|| |
| High risk||5·5||[2·6–11·7]|| |
Excluding patients who received supportive care or single agent chemotherapy only did not substantially change the differences between the IPI groups identified in the univariate and the multivariate survival analysis. The exclusion of patients who received an autologous BMT or SCT also had no impact on these differences.
Results of bone marrow and stem cell transplantation
Table V shows the clinical characteristics, prior treatment and outcome of 12 patients who were treated with autologous BMT or SCT. Nine of these patients received HDCT after responding poorly to initial treatment. Median age of the patients was 30·5 years (range 19–59 years). CR was achieved in 11 of 12 patients, but was followed by relapse 4–8 months later in four cases. One patient died of treatment-related toxicity (septicaemia). Figure 3 shows the survival of these patients. Three-year (5-year) OS was 74% (63%); 3-year (5-year) RFS was 67% (40%).
Table V. Patients receiving high-dose therapy plus bone marrow/stem cell transplantation.
|Pat. No.||Age (years)||Sex||Histology||Ann Arbor stage||IPI||Salvage regimen prior to TPX||TPX||Status before TPX||Status after TPX||Relapse (duration of CR, months)||Status at time of analysis (time after diagnosis, months)|
|1||21||M||AIL||3||Low||CHOP/Dexa-BEAM||Autol. BMT||Second relapse||CR||No||Alive (207)|
|2||25||M||ALCL||3||Low||T-ALL type||Autol. BMT||Poor response||CR||No||Alive (122)|
|3||25||M||PTCU||4||Low-int||T-ALL type||Autol. BMT||Poor response||CR||No||Alive (121)|
|4||25||M||PTCU||4||Low||T-ALL type||Autol. BMT||Poor response||CR||No||Dead (infection, 17)|
|5||55||M||PTCU||4||Low-int||CHOP/Dexa-BEAM||Autol. SCT||Poor response||CR||Yes (8)||Alive (76)|
|6||20||F||PTCU||4||Low-int||CHOP/Dexa-BEAM||Autol. SCT||Poor response||CR||No||Alive (71)|
|7||59||M||PTCU||4||High||CHOP/Dexa-BEAM||Autol. SCT||First relapse||CR||Yes (4)||Dead (NHL, 32)|
|8||36||M||PTCU||2||Low||CHOP/Dexa-BEAM||Autol. SCT||Poor response||CR||Yes (4)||Alive (55)|
|9||40||M||ALCL||4||Hi-int||CHOP/Dexa-BEAM||Autol. SCT||Poor response||CR||Yes (5)||Dead (NHL, 39)|
|10||19||F||ALCL||2||Low||CHOP/Dexa-BEAM||Autol. SCT||First relapse||CR||No||Alive (21)|
|11||53||F||PTCU||4||Hi-int||CHOP/Dexa-BEAM||Autol. SCT||Poor response||CR||No||Alive (35)|
|12||45||M||AIL||4||High||CHOP/Dexa-BEAM||Autol. SCT||Poor response||NR||–||Dead (NHL, 12)|
Causes of death and secondary malignancies
At the time of this analysis, 57 patients (46%) had died of lymphoma; seven patients had died of treatment-related complications (6%): two patients had experienced intracerebral haemorrhage associated with thrombocytopenia shortly after CHT, and two other patients died of pneumonia while neutropenic. The remaining three patients died of septicaemia while in CR after BMT, pulmonary embolism or mesenterial infarction. Comorbid conditions unrelated to the lymphoma were the cause of death in 13 patients (10%). Forty-two patients (34%) are still alive, 34 of them in continuous CR, the remaining eight patients after relapse. Three patients were lost to follow-up (2%).
New haematological malignancies occurred in seven cases (6%). Two patients developed secondary acute myeloid leukaemia 4 and 10 years after therapy. Diagnosis in the other five cases was marginal zone B-cell NHL (11 years after therapy), Burkitt lymphoma (3 years), multiple myeloma (8 years), acute lymphoblastic lymphoma (1 year), or myelodysplastic syndrome (2 years). Epithelial cancers occurred in seven patients (6%). These cancers were diagnosed within 1 year (n = 3), after 2 years (n = 2), and after 3 years (n = 2) of follow up.
- Top of page
The results of this single institution study confirm that patients with T-cell lymphomas differ markedly in clinical presentation, response to treatment and survival. Patients suffering from ALCL in this and other studies showed better response rates and survival when compared with other mature T- and also B-cell lymphomas (Ascani et al, 1997; The Non-Hodgkin's Lymphoma Classification Project, 1997; Tilly et al, 1997). Patients with PTCU achieved a CR in 55% of cases, and the 5-year survival was 45%, whereas patients with AIL showed a poor response to treatment. In a study of 174 patients, mostly with PTCU, ALCL and AIL, Lopez-Guillermo et al (1998) described an overall CR rate of 49% and a 4-year probability of survival of 38%. The CR rates for patients with ALCL (69%) and AIL (37%) were nearly identical to the ones seen in our analysis, while patients with PTCU had a slightly worse CR rate of 47%. The 5-year survival of patients with ALCL, PTCU and AIL were not reported separately by Lopez-Guillermo et al (1998), but the survival curves look very similar to ours. The GELA (Groupe d'Etudes des Lymphomes de l'Adulte) study on 288 patients (Gisselbrecht et al, 1998) reported CR rates of 72%, 53% and 42% for patients with ALCL, PTCU and AIL respectively. The respective OS at 5 years was 64%, 37% and 31%. Again, CR and survival from this study are similar to the ones reported here. These data are supported by further studies that presented results of chemotherapy restricted to distinct subgroups of mature T-cell lymphomas, such as ALCL or PTCU (Gascoyne et al, 1999; Pellatt et al, 2002; Reiser et al, 2002; Rudiger et al, 2002; Gallamini et al, 2004). Thus, although differences in the patient populations with respect to the histological subtypes and treatment modalities clearly exist, the larger studies show that surprisingly consistent results can be achieved by CHT.
While most authors agree that conventional chemotherapy remains the treatment of choice for patients with ALCL, there is an ongoing debate as to whether more aggressive treatment strategies, including autologous or allogeneic stem cell or BMT, should be recommended to improve treatment results for patients with PTCU or AIL (Rodriguez et al, 2001; Song et al, 2003; Morris et al, 2004; Reimer et al, 2004). Likewise, alemtuzumab, an anti-CD52 monoclonal antibody, has shown efficacy in relapsed or refractory T-cell lymphomas (Enblad et al, 2004). These new treatment modalities, however, all bear considerable risks for the patients (Lenihan et al, 2004). A treatment-related mortality higher than 10% has repeatedly been reported (Song et al, 2003; Enblad et al, 2004; Morris et al, 2004). Therefore, it is important to identify those patients with mature T-cell lymphomas who will have a poor outcome with conventional chemotherapy alone and require a more aggressive therapy. We showed that the IPI is the overriding factor determining the prognosis of patients with nodal mature T-cell lymphomas classified according to the WHO. While OS at 5 years was 74% for the IPI low risk group, it decreased to 49% for the low-intermediate, 21% for the high-intermediate and 6% for the high-risk group. Although ALCL, PTCU and AIL may differ in biology, the multivariate analysis confirmed that differences in prognosis are relatively small when adjusted for the IPI. The prognostic value of the IPI was apparent in all histological subtypes. For example, seven of nine patients with ALCL and low risk IPI were free of disease at the end of the study. In contrast, three of six ALCL patients with high-intermediate IPI never reached CR. Of the remaining three, only one did not experience a relapse. Despite the low number of patients with ALCL, the log rank test revealed significant differences in OS between the IPI groups (P = 0·02). A number of studies, all retrospective in nature and with limited numbers of patients, sought to identify risk factors indicating a poor prognosis in patients with T-cell lymphomas. Histology, older age, poor PS, B-symptoms, extranodal disease, bone marrow involvement, advanced stage, high serum LDH or high beta-2-microglobulin have all been associated with a poor prognosis. Multivariate analyses to identify the risk factors for OS of peripheral T-cell lymphoma patients have been performed by only a few authors. Lopez-Guillermo et al (1998) reported histology according to the REAL classification, the presence of B-symptoms, and the IPI (low + low–intermediate vs. high–intermediate/high) as statistically significant parameters determining survival. Gisselbrecht et al (1998) showed that the IPI significantly influenced OS in patients with non-ALCL peripheral T-cell lymphoma classified according to the updated Kiel classification: OS was 64%, 56%, 34% and 22% for patients with IPI scores of 0,1,2 or ≥ 3, respectively, at diagnosis but survival curves for patients with IPI 0 and 1 looked very similar. Ansell et al (1997) also reported that the IPI was able to predict the outcome of 78 patients with peripheral T-cell lymphoma classified using the REAL classification; however, the survival curves for patients in the low-intermediate and high-intermediate categories virtually overlapped. Recently, Gallamini et al (2004) analysed 385 patients with PTCU only and reported a 5-year OS of 58·9%, 45·6%, 39·7% and 18·3%, respectively, for patients with IPI 0,1,2, ≥3 (P < 0·0001). These authors proposed a new simplified 2-class model of IPI and a new prognostic index for peripheral T-cell lymphomas considering age, PS, LDH and bone marrow involvement. Patients in the high-risk group had a 5-year survival probability of 26% and the authors concluded that new therapeutic strategies preferentially should be explored in these patients (Gallamini et al, 2004).
Taken together, although others have shown the prognostic relevance of the IPI in patients with specific subtypes of T-cell lymphomas, or in patients diagnosed according to the REAL or Kiel classification, we for the first time show significant differences in OS between IPI categories if the WHO classification of mature T-cell lymphoma is used.
Given the retrospective nature of our study, covering a long period of time, this analysis has a number of limitations. First, changes in lymphoma classification over time may have led to misclassification in some cases. The histological subtype of cases included in this analysis was confirmed, however, by the same experienced lymphoma expert pathologist in the vast majority of cases. Secondly, changes in treatment strategies over time might have caused differences in response and survival. We do not consider this possibility to have substantially biased our results because the efficacy of conventional chemotherapy in patients with mature T-cell lymphoma was limited throughout the study period and there was no indication that more recent chemotherapy regimes significantly changed this unfortunate situation. Only 10% of the patients received high-dose therapy followed by transplantation and it is not clear from this or other studies whether BMT or SCT really improves outcome if compared with conventional chemotherapy. Thirdly, as a consequence of the relatively low numbers of patients with ALCL and AIL in the study, the multivariate analysis had limited power to detect a significant influence of the histological subtype on survival; however, the number of events, which largely determines the power of Cox regression analyses, was high in all groups. Thus, although we cannot exclude that the histological subtype may have some relevance for the outcome after chemotherapy it seems to add limited prognostic information in addition to the IPI in patients with mature T-cell lymphomas.
We conclude that patients with nodal mature T-cell lymphomas treated with CHT show considerable differences in OS depending on the presence or absence of risk factors defined by the IPI. The IPI score largely explains differences in survival between histological subtypes of nodal mature T-cell lymphomas. The IPI may be used for risk stratification in clinical trials to identify patients who would benefit most from new treatment strategies. Most probably, these will be patients with high-intermediate and high risk IPI whose particularly poor prognosis is confirmed in this analysis.