review
Allogeneic transplantation for Hodgkin’s lymphoma Paolo Corradini,1,2 Barbara Sarina3 and Lucia Farina1 1
Division of Hematology, Fondazione IRCCS Istituto Nazionale dei Tumori, 2Chair of Hematology, School of Medicine – Universita` degli Studi di Milano, and 3Division of Hematology, Istituto Clinico Humanitas, Milano, Italy
Summary Hodgkin’s lymphoma (HL) can be cured in most of the patients, but in case of refractory disease or relapse after autologous stem cell transplantation (SCT) the prognosis becomes very poor. In these patients a consensus about the standard approach has not been achieved so far and only allogeneic SCT has shown a long-term disease control. The postulated graft-versus-Hodgkin’s lymphoma is a matter of controversy, but the clinical responses observed after donor lymphocyte infusions may explain the superiority of alloSCT over standard chemo-radiotherapy. The results of conventional myeloablative alloSCT had a relevant non-relapse mortality (NRM), discouraging its widespread application as salvage treatment. In the last 10 years, reduced intensity conditioning (RIC) significantly decreased the NRM, widening the application of alloSCT also to heavily pretreated patients. Taking into account all phase II studies, 20–30% of patients receiving RIC alloSCT are disease-free and probably some of them are cured. Keywords: Hodgkin’s lymphoma, allogeneic, reduced-intensity. Hodgkin’s lymphoma (HL) has been always considered a disease with a favourable prognosis, since more than 90% of the patients in early stages and about 75% of patients in advanced stages can be cured after ABVD (adryamicin, bleomicin, vinorelbine and dacarbazin) chemotherapy cycles ± radiotherapy (Bonfante et al, 1992). About 50% of patients relapsing after first line therapy may be rescued by high-dose chemotherapy followed by autologous stem cell transplantation (autoSCT) (Josting et al, 2002, 2005; Schmitz et al, 2002; Tarella et al, 2003; Lavoie et al, 2005; Viviani et al, 2010). In particular, patients who experienced a late relapse after the first-line chemotherapy can do significantly better than those with primary refractory disease (survival of 70–80% versus 20–30%) (Lazarus et al, 1999; Sweetenham
Correspondence: Paolo Corradini, MD, Division of Hematology, Fondazione IRCCS Istituto Nazionale dei Tumori, Via Venezian 1, 20133 Milano, Italy. E-mail:
[email protected]
et al, 1999; Josting et al, 2005). Overall, despite the peculiar chemo- and radio-sensitivity demonstrated by tumour cells, 10–15% of all the patients affected by HL may experience disease relapse after an autoSCT. For these patients the prognosis is very poor and salvage chemo-radiotherapy produces only short-lasting responses with a median overall survival ranging between 8 and 36 months (Vose et al, 1992; Varterasian et al, 1995; Bolwell et al, 1997; Shamash et al, 2000; Kewalramani et al, 2003). From the late eighties increasing data showed that allogeneic stem cell transplantation (alloSCT) was a therapeutic approach able to produce prolonged remissions in case of chemorefactory or relapsed HL patients (Appelbaum et al, 1985; Phillips et al, 1989). During the last few years phase I-II trials have been studying new drugs, such as monoclonal antibodies, histone deacetylase inhibitors, and mTOR inhibitors, but although promising, preliminary results have suggested only a limited and/or transient efficacy (Younes et al, 2003; Dickinson et al, 2009; Johnston et al, 2010). Thus, at the moment, alloSCT can be considered the only potentially curative treatment in this patient setting. The advantage provided by allografting is the postulated graft-versus-tumour effect, which may be beneficial in patients in which the curative mechanism of chemo-radiotherapy showed to be ineffective. Nevertheless, comparative studies between myeloablative alloSCT and autoSCT in refractory and relapsed HL patients provided only a weak evidence of the supposed graft-versus-HL (GVHL) effect (Jones et al, 1991; Anderson et al, 1993; Milpied et al, 1996; Akpek et al, 2001). For these reasons the widespread use of alloSCT in HL patients is still a matter of controversy and was quite limited until the advent of reduce intensity conditioning (RIC). In this review, we will discuss the following critical points about alloSCT in HL: (i) which are the evidences supporting the existence of the GVHL effect?; (ii) how is the clinical outcome of the patients undergoing an alloSCT in terms of disease response, toxicity and survival?; (iii) how can we select the patients who may benefit the most from an alloSCT? May we identify poor prognosis patients and anticipate the timing of allografting to further improve the curability of HL?
ª 2010 Blackwell Publishing Ltd, British Journal of Haematology, 152, 261–272
First published online 14 December 2010 doi:10.1111/j.1365-2141.2010.08492.x
Review
Does the graft-versus-Hodgkin’s lymphoma effect really exist? Nowadays, it is well-known that the results obtained after an alloSCT are related not only to the cytotoxic activity of the conditioning regimen, but also to the antitumor effects of the immune cells contained in the graft. The so called ‘graftversus-tumour’ effect is mediated by donor T and natural killer cells and the rationale of alloSCT is to provide an adoptive immunotherapy. However, as compared to other diseases, data supporting a specific GVHL effect are few and mainly based on clinical observations.
Biological base of the GVHL effect The target antigens of the GVHL effect are not known, but in half of the HL cases the Epstein Barr virus (EBV) infection can be demonstrated thus allowing the hypothesis that donor T cells may recognize EBV specific antigens (Weiss et al, 1987; Mueller et al, 1989). The role of EBV in the pathogenesis of HL has been extensively studied and is supported by the fact that the virus is able to transform B lymphocytes in vitro through the expression of EBV proteins among which latent membrane protein-1 (LMP-1) is thought to be the most important since it is responsible of the increased NF-kB activity (Huen et al, 1995). Several observations suggest that EBV antigens may be appropriate targets for immunotherapy. Many of the available evidences come from the literature regarding the post-transplant lymphoproliferative disorder (PTLD) which is an aggressive lymphoma that may occur after an alloSCT. PTLD is caused by the reactivation of EBV virus in latently infected B-cells and is favoured by the absence of the immune surveillance by cytotoxic T cells. The overall frequency of PTLD after an alloSCT is approximately 1%, but the incidence increases in case of T-cell depleted transplants, with the vast majority of events occurring during the first year (Ocheni et al, 2008). The findings that support a possible role of EBV as a target for post-transplant immunotherapy are the following: (i) most of the patients who develop PTLD can achieve a complete remission (CR) after donor lymphocyte infusions (DLI) (Papadopoulos et al, 1994; Small et al, 1999); (ii) pre-emptive DLI may prevent PTLD after alloSCT in high risk patients (DiNardo & Tsai, 2010); (iii) autologous EBVspecific cytotoxic T cell infusions have been studied in HL patients, as a potential curative approach (Rooney et al, 1998). Despite these evidences, Peggs et al (2005) reported that the majority of the DLI-responsive patients having their diagnostic specimens tested were negative for EBER (Epstein–Barr-virusencoded-RNA) transcripts. These results suggest that EBV may provide antigens able to induce an immune response, but probably other still unknown mechanisms contribute to mediate the GVHL effect. Recent studies have shown that microenvironment plays a role in tumour cell survival and proliferation (Chantrain et al, 262
2008; Burger et al, 2009). The microenvironment of HL cells is characterized by an extensive infiltration of non-malignant cells that may favour the escape of tumour cells from immune surveillance (Koenecke et al, 2008; Ku¨ppers, 2009; Aldinucci et al, 2010). In this context, Diepstra et al (2008) showed that in about half of the HL cases tumour cells expressed the histocompatibility-leukocyte-antigen (HLA)-G, that may cause an inhibition of cytotoxicity especially when associated to the loss of class I HLA molecules, which in turn protects tumour cells from death. Moreover, Marshall et al (2004) found that the infiltrating lymphocytes in HL were highly enriched of regulatory T-cells which contribute to an immune suppressive microenvironment and provide the explanation for the ineffective immune clearance of Reed–Stenberg cells. On the contrary, other authors demonstrated that an increased number of tumour infiltrating FOXP3-positive cells positively influenced overall survival (OS) and failure free survival (FFS) of HL patients (Tzankov et al, 2008). From these data, regulatory T-cells emerge as an important lymphoma-host microenvironment modulator whose function is still unclear, but deserve further investigations. To support the role of microenvironment in HL Steidl et al (2010) recently showed that an increased number of tumourassociated macrophages was strongly correlated with a poor survival in patients affected by classical HL. Although preliminary and somehow controversial, all these findings may lead to the hypothesis that the immune reconstitution following an alloSCT could offer the opportunity to modify tumour microenvironment and the resistance to immune surveillance, thus exerting an immune-mediated attack towards HL cells.
Direct evidences of the GVHL effect The most significant data supporting the existence of the GVHL effect come from the responses observed either after DLI or after the development of graft-versus-host disease (GVHD). Many studies reported the use of DLI for persistent or relapsed HL, although the number of patients was limited and in many cases DLIs were preceded by chemotherapy, thus confounding the real benefit strictly correlated to immunotherapy (Peggs et al, 2005, 2007; Alvarez et al, 2006; Anderlini et al, 2008; Thomson et al, 2008; Robinson et al, 2009; Sureda et al, 2009). The best results were observed by Peggs et al (2005) who administered DLI after a Campath-based RIC alloSCT in 16 patients with residual or progressive disease. Nine patients (56%) showed a disease response and in five patients, who did not receive chemotherapy prior to DLI, CR was achieved at median time of 675 d from the last infusion. Successively, other authors reported response rates ranging from 30% to 55% (Table I). In a large retrospective study by Robinson et al (2009), 64 patients received DLI for persistent or progressive disease and 13 out of 41 assessable patients (32%) achieved CR. Interestingly, 8 responses (44%) were also
ª 2010 Blackwell Publishing Ltd, British Journal of Haematology, 152, 261–272
Review Table I. Responses to donor lymphocyte infusions in Hodgkin lymphoma patients.
Reference
Patients
Preceding CT
Overall response (CR/PR)
Response rate (%)
Response rate (%) (DLI only)
Peggs et al (2007)
14 (MFA) 11 (MF) 14 41 15 25
5 3 11 23 1 9
8 6 6 (3/3) 13 8 11
57 54 43 32 53 47
43 NA 33 44 NA 40
Anderlini et al (2008) Robinson et al (2009) Thomson et al (2008) Sureda et al (2009)
CT, chemotherapy; CR, complete remission; PR, partial remission; DLI, donor lymphocyte infusion; MFA, cyclosporine/alemtuzumab; MF, cyclosporine/methotretaxe; NA, not available.
observed in 18 patients receiving DLI alone. These data were confirmed by the M.D. Anderson group who reported an overall response rate (ORR) of 43%, with 33% of the patients responding to DLI alone (Anderlini et al, 2008). The data about an association between the clinical signs of GVHD and the occurrence of a significant GVHL effect are more controversial. A number of studies showed a correlation between the development of GVHD and the clinical response (Milpied et al, 1996; Akpek et al, 2001; Peggs et al, 2005; Alvarez et al, 2006; Sureda et al, 2008, 2009; Sarina et al, 2010). However, Milpied et al observed that this did not confer a survival benefit due to the high mortality rate of patients affected by ‡grade 2 acute GVHD. A more recent EBMT analysis on 168 HL patients treated with an alloSCT found that patients with chronic GVHD experienced a lower risk of relapse (P < 0Æ05) (Sureda et al, 2008). This effect translated into a trend towards a better progression free survival (PFS) with no significant impact on non-relapse mortality (NRM) rate. Similarly, the GITMO (Gruppo Italiano Trapianto Midollo Osseo) study found that the occurrence of chronic GVHD was associated with a better OS and PFS (Sarina et al, 2010). On the other side, HL has been always considered a disease in which the GVHL effect was not able to induce a significant tumour response because some studies did not observe a correlation between GVHD and disease control (Anderson et al, 1993; Gajewski et al, 1996; Anderlini et al, 2008). The discrepancies among the published studies may be explained by several factors, such as the limited number of events, the heterogeneity of the patients and the fact that it is sometime difficult to assess the correlation between the onset of GVHD and disease response.
Indirect evidences of the GVHL effect Indirect evidences of the GVHL effect are based on the lower relapse rate after alloSCT as compared to autoSCT. In fact, since bone marrow is often not involved by HL the advantage provided by an allograft has been thought to reside on the GVHL effect rather than on the stem cell graft free of residual tumour cells.
In the first comparative study published by the Johns Hopkins group, the relapse rate of chemosensitive patients was 18% vs. 46% after allo- and autoSCT, respectively (P = 0Æ02) (Jones et al, 1991). The long-term data of the same authors confirmed a trend towards a lower relapse incidence in the allografted group, that did not reach a statistical significance probably due to the small number of patients enrolled in the study (Akpek et al, 2001). Similarly, the Seattle group reported a lower relapse rate in case of alloSCT from a HLA matched donor as compared to autoSCT, but no significant difference was observed in the event-free survival (EFS) (Anderson et al, 1993). On the contrary, an EBMT analysis did not support these findings, describing a relapse rate of both 61% in 90 autografted and allografted patients. In this study the two groups were well matched for age, sex, disease status, conditioning regimens and time from diagnosis to transplant in order to overcome the limitations of a retrospective analysis (Milpied et al, 1996). It is noteworthy that the transplantrelated mortality (TRM) of all these studies ranged between 32% and 54%, thus offsetting any beneficial effect on survival in allografted patients. More recently, the use of RIC transplants, by decreasing the TRM, further supported the idea that alloSCT can provide a real advantage over salvage chemo-radiotherapy (Thomson et al, 2008; Castagna et al, 2009a; Sarina et al, 2010). Thomson et al (2008) were the first to compare retrospectively the outcome of HL patients who failed an autoSCT and received either a RIC alloSCT or chemo-radiotherapy; in this study, OS at 10 years was 48% vs. 15% (P < 0Æ001), respectively. Nevertheless, the results might be questionable because
