Conflict of interest: Author receives research funding from Seattle Genetics.
Disease overview: Hodgkin lymphoma (HL) is an uncommon B-cell lymphoid malignancy affecting 9,000 new patients annually and representing approximately 11% of all lymphomas in the United States.
Diagnosis: HL is composed of two distinct disease entities; the more commonly diagnosed classical HL and the rare nodular lymphocyte predominant HL. Nodular sclerosis, mixed cellularity, lymphocyte depletion, and lymphocyte-rich HL are subgroups under the designation of classical HL.
Risk stratification: An accurate assessment of the stage of disease in patients with HL is critical for the selection of the appropriate therapy. Prognostic models that identify patients at low or high risk for recurrence are used to optimize therapy for patients with limited or advanced stage disease.
Risk-adapted therapy: Initial therapy for HL patients is based on the histology of the disease, the anatomical stage and the presence of poor prognostic features. Patients with early stage disease are treated with combined modality strategies utilizing abbreviated courses of combination chemotherapy followed by involved-field radiation therapy, while those with advanced stage disease receive a longer course of chemotherapy often without radiation therapy.
Hodgkin lymphoma (HL) affects ∼9,000 new patients in the United States each year . Notably, there is an increased incidence in young adults as well as in patients 55 years and older . There are no clearly defined risk factors for the development of this disease, and the cause of HL remains unknown. Factors shown to be associated with HL include familial factors, viral exposures, and immune suppression . Same sex siblings of patients with HL have a 10-fold higher risk for developing the disease [3, 4], and a monozygotic twin of a patient with HL has a significantly increased risk for developing HL when compared with a dizygotic twin sibling of a patient with HL [5, 6]. Although familial factors may suggest a genetic cause for this disease, research also suggests that an abnormal immune response to infection may play a role in the pathogenesis of HL. Epidemiologic and serologic studies have implicated Epstein–Barr virus (EBV) in the etiology of HL, and the EBV genome was detected in tumor specimens from patients with HL . Other childhood infectious illnesses, including chickenpox, measles, mumps, rubella, and pertussis, however, are negatively associated with the risk of HL and are possibly protective . There is also an association with human immunodeficiency (HIV) infection in that HIV-infected patients have a significantly increased risk of HL when compared with the general population . Overall, HL in immunosuppressed patients, including those who are HIV positive, is associated with advanced stage of disease at presentation, unusual sites of disease, and a poorer outcome after initial therapy [10, 11].
Over the last four decades, advances in radiation therapy and the addition of combination chemotherapy have significantly increased the cure rate of patients with HL. Currently, more than 80% of all newly diagnosed patients younger than 60 years are likely to be cured of their disease.
At the time of diagnosis, the majority of patients with HL present with supradiaphragmatic lymphadenopathy. Patients are commonly presented with cervical, anterior mediastinal, supraclavicular, and axillary lymph-node involvement, while the inguinal areas are less frequently involved. Approximately one-third of patients present with systemic symptoms that include fever, night sweats, and weight loss, and many patients also present with chronic pruritis. Although the disease most commonly involves contiguous lymph-node groups, HL may also affect extranodal tissues by direct invasion or by hematogenous spread. The most commonly involved extranodal sites are the spleen, lungs, liver, and bone marrow.
The initial diagnosis of HL can only be made by a biopsy. Fine-needle aspiration or core-needle biopsies are inadequate, because the architecture of the lymph node is extremely important for an accurate diagnosis. HL is a unique malignancy in that the tumor cells constitute the minority of the cellular population, and an inadequate biopsy may fail to include malignant cells in the specimen . To confirm the diagnosis, it is necessary to identify the malignant Reed–Sternberg cell, which is of follicular center B-cell origin [13, 14], within the appropriate cellular environment of normal reactive lymphocytes, eosinophils, and histiocytes.
HL is composed of two distinct disease entities; the more commonly diagnosed classical HL and the rare nodular lymphocyte predominant HL . Nodular sclerosis, mixed cellularity, lymphocyte depletion, and lymphocyte-rich HL are subgroups under the designation of classical HL.
The presence of malignant multinucleated giant Reed–Sternberg cells within the characteristic reactive cellular background is the pathologic hallmark of classical Hodgkin lymphoma (HL; see Fig. 1A,B). Painless lymphadenopathy is the most common clinical manifestation of classical HL, although each histological subtype has its own unique clinical features . Nodular sclerosis, the most common subtype, tends to affect adolescent and young adults more commonly and usually presents with localized disease involving cervical supracavicular and mediastinal regions. Mixed cellularity HL is more prevalent in the pediatric as well as older age groups and is commonly associated with a more advanced stage of disease and a poorer prognosis. The incidence of lymphocyte depletion HL appears much lower then previously reported with many of these cases reclassified as a non-HL. This subtype occurs mainly in older age patients and in patients with acquired immune deficiency syndrome. These patients typically present with symptomatic extensive disease without peripheral lymphadenopathy. Lymphocyte rich classical HL represents a subtype similar to nodular lymphocyte predominant HL (see below) on morphologic grounds; however, the Reed–Sternberg cells have a more classical immunophenotype consistent with classical HL.
Nodular lymphocyte predominant HL
Nodular lymphocyte predominant Hodgkin lymphoma (HL) constitutes a unique clinical pathologic entity that is significantly different than classical HL. Pathologically, lymphocyte predominant HL lacks the typical Reed–Sternberg cells and instead are characterized by a neoplastic population of larger cells with folded lobulated nuclei known as lymphocytic and histiocytic cells (see Fig. 1C,D). Unlike classical HL, these cells are CD20+ and commonly negative for CD30 . Lymphocyte predominant HL is more frequently seen in men, and constitutional symptoms at presentation as well as extranodal disease are rare. Patients usually present with limited nodal disease that classically affects the neck region and spares the mediastinum. The natural history of lymphocyte predominant HL differs from classical HL in that it has an indolent course with a tendency for late recurrences .
An accurate assessment of the stage of disease in patients with HL is critical for the selection of the appropriate therapy. The staging system for patients with HL is based on whether the involved lymph nodes are on one or both sides of the diaphragm, the number of involved sites, whether the sites of involvement are bulky, whether there is contiguous extranodal involvement or disseminated extranodal disease, and whether typical systemic symptoms (B symptoms) are present. Fluorodeoxyglucose positive emission tomography (FDG-PET) scanning has emerged as an important tool in the staging of patients with HL in that it significantly adds to the staging information obtained using other standard radiographic methods .
Many patients with HL can be cured with standard treatment and are therefore at risk for potential long-term complications. Factors that identify patients at low or high risk for recurrence would therefore be most useful in optimizing therapy based on the patient's expected clinical outcome to avoid overtreatment of some and undertreatment of others. Prognostic factors for early stage HL have been developed by both the German HL Study Group (GHSG) and the European Organization for the Research and Treatment of Cancer as described in Table I [20, 21]. These prognostic models identify disease bulk, either as a single large mediastinal mass or as multiple sites of disease, as the major predictor of a poor prognosis in this subgroup of patients. In contrast, in patients with advanced HL, disease bulk and other traditional prognostic variables have been found to be less predictive of outcome. A different prognostic scoring system was therefore developed for these patients by the International Prognostic Factor Project on advanced HL (Table II) . This study identified seven variables that predicted patient outcome in a multivariate analysis. Patients with five or more factors were found to have a 5-year freedom from progression of 42% while patients with no negative prognostic factors had an 84% likelihood of being free from progression at 5 years.
Table I. Prognostic Scoring Systems for Limited Stage Hodgkin Lymphoma
EORTC, European Organization for Research and Treatment of Cancer; ESR, erythrocyte sedimentation rate; GHSG, German Hodgkin Study Group.
Unfavorable (if any criteria)
Large mediastinal mass
Large mediastinal mass
High ESR ≥ 4 sites
High ESR ≥ 3 sites
Age ≥ 50
Massive splenic disease
Table II. Prognostic Factors for Advanced Hodgkin Lymphoma
Adverse prognostic factors
Age ≥ 45 years
White blood count ≥ 15,000 cells/μl
Lymphocyte count < 600 cells/μl or < 8%
Albumin < 4.0 g/dl
Hemoglobin < 10.5 g/dl
Outcome according to prognostic score
Number of factors
5 years—freedom from progression (%)
5 years—overall survival (%)
Risk-Adapted Initial Therapy
The predominant factors that determine the initial choice of therapy for HL patients are the histology of the disease (classical HL or nodular lymphocyte predominant HL), the anatomical stage of disease (limited or advanced disease), the presence of poor prognostic features, the presence of constitutional symptoms, and the presence of bulky disease defined as a single site of disease >10 cm in diameter. During the course of treatment, FDG-PET scanning now plays a role in decisions to complete therapy as planned or to add or omit components of treatment. A positive interim FDG-PET scan after two cycles of treatment may result in intensification of therapy, and a positive PET scan at the end of treatment may result in the addition of consolidation radiotherapy to the positive sites. A positive PET scan at either point may result in a repeat biopsy to confirm or exclude persistent disease. The use of PET scans in this fashion is based on studies showing that patients with a positive FDG-PET scan at the completion of treatment have a significantly higher recurrence rate regardless of the findings on CT scan [23, 24]. Also, a FDG-PET done early in the course of treatment (after two cycles) predicted progression-free survival (PFS) and overall survival (OS) for patients with HL and was a better predictor of outcome than stage, extranodal disease, or other prognostic factors [25, 26].
The current standard of care for institutions that treat patients with HL is to have different treatment strategies for HL patients with early stage disease with favorable prognostic features, those with early stage disease but who have poor prognostic features, or those with advanced disease. As a general rule, patients with early stage disease are treated with combined modality strategies using abbreviated courses of combination chemotherapy followed by involved-field radiation therapy (IFRT), while those with advanced stage disease receive a longer course of chemotherapy without radiation therapy.
Early stage favorable HL.
Treatment strategies for early stage Hodgkin lymphoma (HL; stages I and IIA) have changed significantly in the last few decades. Previously, extended field radiation was considered the standard therapy. However, because of the recognition of high-relapse rates with significant long-term complications, extended field radiation therapy to involve adjacent lymph node areas is no longer used . A randomized comparison of patients treated with subtotal nodal radiation therapy with or without ABVD chemotherapy (doxorubicin, bleomycin, vinblastine, dacarbazine) and those who received ABVD alone found that patients who received subtotal nodal radiation therapy had a poorer OS and a higher rate of death from causes other than HL . Therefore, for favorable early stage disease, short duration chemotherapy for the control of occult lesions combined with IFRT restricted only to involved lymph-node areas is currently a standard practice. Most groups will give two to four cycles of combination chemotherapy followed by IFRT to a dose of ∼20–35 Gy . A four-arm study of 1,370 patients performed by the GHSG randomized early stage HL with favorable prognostic factors between two and four cycles of ABVD chemotherapy and 20 and 30 Gy IFRT. There was no difference between the four groups as regards response to therapy, PFS and OS. ABVD for two cycles followed by 20 Gy IFRT is therefore currently the standard treatment for early stage favorable prognosis HL .
Early stage unfavorable HL.
It is generally accepted that patients with stage I and II disease who present with adverse risk factors should be treated with chemotherapy in combination with radiation therapy. However, the optimal number of chemotherapy cycles as well as the optimal chemotherapy regimen and the dose of radiation as well as the field sizes is the subject of ongoing studies and debate. This group of patients usually consists of those with bulky mediastinal masses or those with extranodal disease. In these patients, the use of four cycles of combination chemotherapy with IFRT is generally accepted as the treatment of choice [31, 32]. In a clinical trial of 1,395 patients with stage I/IIA Hodgkin lymphoma (HL) disease with unfavorable features including large mediastinal masses, extranodal disease, high-erythrocyte sedimentation rate, or ≥3 nodal sites, patients were randomized to ABVD for four cycles or baseline doses of BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone) for four cycles plus either 20 or 30 Gy IFRT. Freedom from treatment failure was worse if 20 Gy rather than 30 Gy was used with ABVD; however, similar outcomes were seen between 20 and 30 Gy when used with BEACOPP. The conclusion was that ABVD for four cycles plus 30 Gy IFRT is the standard for these patients . A subsequent study looked at intensifying the chemotherapy used in this patient group. In this trial, 1,528 qualified patients with early stage unfavorable HL received ABVD for four cycles or escalated doses of BEACOPP for two plus ABVD for two cycles. All patients got 30 Gy IFRT. The freedom from treatment failure favored the aggressive chemotherapy arm—with a difference of 7.2% at 5 years—but there was no difference in OS, and increased toxicity was seen in the aggressive chemotherapy arm . There were no differences in treatment-related mortality or secondary malignancies. The goal of future studies will therefore be to optimize the amount of chemotherapy needed to maintain efficacy but to decrease potential toxicities in these patients.
In patients with advanced disease (Stages IIB–IV), the challenge is to increase the number of patients with durable remissions while decreasing the likelihood of long-term side effects. Initially, the MOPP regimen (nitrogen mustard, vincristine, procarbazine, and prednisone) was developed for patients who progressed after radiation therapy, and the long-term results with the MOPP regimen showed this to be an effective therapy. This combination resulted in a freedom from progression rate of 54% and an OS of 48% at 20 years . Although the MOPP regimen significantly changed the outcome of patients who previously would have died of progressive disease, approximately one-third of patients subsequently relapsed. Since then, multiple other regimens have been developed in an attempt to improve the efficacy of this regimen.
The ABVD chemotherapy regimen was then developed and also showed significant clinical activity with potentially less toxicity. This led to a randomized trial comparing alternating cycles of MOPP and ABVD chemotherapy to MOPP chemotherapy alone. The alternating regimen was found to be superior as regards the complete remission rate, freedom from progression, and OS . Several major randomized studies over the last 20 years have attempted to identify the regimen with the greatest activity and the most favorable side effect profile. Initially, MOPP, ABVD, and MOPP alternating with ABVD were compared . In this study, the complete response rate and freedom from progression were worse for patients receiving MOPP chemotherapy alone compared to those receiving ABVD or the alternating regimen. Two further studies compared the MOPP/ABVD hybrid regimen to MOPP alternating with ABVD and the regimens were found to be equivalent [37, 38]. When the MOPP/ABV hybrid regimen was compared to ABVD, the ABVD arm showed superiority with less toxicity . The results of all these trials led to ABVD chemotherapy being regarded as the treatment of choice for patients with advanced HL based on its efficacy, relative ease of administration, and acceptable side effect profile. Recent studies have added rituximab to ABVD to deplete intratumoral B-cells and clonotypic malignant cells that express CD20. Both studies demonstrated high-CR rates, and the event-free survival in both studies suggested promising activity of the combination. The efficacy of this combination will need to be confirmed in a randomized trial [40, 41].
To further minimize toxicity, the Stanford V regimen was developed, which incorporated the active agents from MOPP and ABVD into a brief dose intense regimen and combined this 12-week regimen with radiation therapy . The results initially obtained showed a 5-year freedom from progression in 142 patients of 89% and an OS of 96%. Similar results were obtained in a multi-institutional ECOG study . This regimen was tested against ABVD in a number of randomized trials. Initial studies suggested that ABVD might be superior to the Stanford V regimen. With long-term follow up, the 10-year failure-free survival was 75% and 49% for the ABVD and Stanford V regimens, respectively. The differences in outcome however may be explained by the fact that the administration of radiotherapy in the Stanford V arm differed from what was originally described . Two subsequent randomized trials comparing ABVD to Stanford V have found similar response rates, failure-free, and OS [45, 46]. The frequency of adverse events was similar between the two regimens with patients receiving ABVD experiencing more pulmonary toxicity and patients receiving Stanford V having a greater number of other toxicities.
The GHSG has also developed new regimens for patients with advanced HL including standard and dose-escalated BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone) . A large randomized trial comparing COPP (cyclophosphamide, vincristine, procarbazine, and prednisone) alternating with ABVD to dose-escalated and standard BEACOPP showed better tumor control and OS for patients receiving dose-escalated BEACOPP . The results were updated at 10 years and continued to show improved outcomes for patients treated with escalated BEACOPP . Although these results are encouraging, but it is important to note that acute myeloid leukemia or myelodysplastic syndrome was more frequently seen in patients treated with escalated BEACOPP. A further study compared six cycles of ABVD to four cycles of escalated BEACOPP followed by two cycles of standard BEACOPP. This study also had a third arm in which patients received six cycles of a multidrug intensive regimen. When the results from the ABVD arm were compared to the BEACOPP arm, there was an improved PFS with BEACOPP, but the OS in both arms was similar. More toxicity was seen in BEACOPP-treated patients, but there was a benefit for BEACOPP in poor risk patients . Further randomized studies have been done to determine the optimal number of cycles of BEACOPP needed to maintain the improved outcome and also decrease toxicity. The GHSG found that six cycles of escalated BEACOPP followed by radiotherapy administered only to PET-positive masses measuring more than 2.5 cm was more effective in terms of freedom from treatment failure and less toxic than eight cycles of the same regimen .
Recently, a randomized comparison of ABVD and BEACOPP in advanced stage HL was reported . Three hundred and thirty-one patients with poor risk HL were enrolled, and local radiotherapy was added to therapy as indicated. Patients with residual or progressive disease were subsequently treated with salvage therapy including stem-cell transplantation. The authors analyzed the outcome after initial therapy and also after salvage therapy. The freedom from first progression favored patients receiving BEACOPP and was 85% in that group compared to 73% among patients treated with ABVD (P = 0.004). However, after completion of all planned therapy including salvage therapy for those with residual or progressive disease, the 7-year rate of freedom from second progression was 88% in the BEACOPP group and 82% in the ABVD group (P = 0.12), and the 7-year OS rate was 89 and 84%, respectively (P = 0.39). Severe adverse events were more frequent in the BEACOPP patients than in the ABVD patients. These results have led some to suggest that initial therapy may not need to be highly aggressive in all patients as those who relapse may be salvaged with subsequent intensive therapy . Others have pointed out that OS was a secondary end point in this study and that the study was small compared to other similar trials . More mature follow up will be needed to confirm thisapproach.
To potentially further improve the outcome of advanced HL patients with poor prognostic features, the role of high-dose chemotherapy (HDCT) with autologous stem-cell transplantation (ASCT) has been evaluated as part of initial therapy for these patients. Patients with advanced unfavorable HL achieving a complete or partial remission after four courses of doxorubicin-containing regimens were found to have a favorable outcome with conventional chemotherapy, and no benefit from an early intensification with HDCT and ASCT was shown .
In summary, ABVD chemotherapy remains the most widely used treatment in the United States for patients with advanced stage HL. However, dose-intense regimens such as escalated BEACOPP should be considered in patients with advanced disease and multiple poor prognostic factors.
Nodular lymphocyte predominant HL.
An exception to the management approach outlined earlier is early stage patients with nodular lymphocyte predominant Hodgkin lymphoma (HL). Patients with favorable stage IA disease, with no significant risk factors, can commonly be managed with lymph-node excision followed by a “watch and wait” approach or with IFRT to a dose of ∼20–30 Gy. More advanced stage patients with nodular lymphocyte predominant HL are commonly treated with ABVD, often in combination with rituximab, because the malignant cells express CD20. Clinical trials are in progress to define the optimal therapy for this disease.
Management of relapsed/refractory disease
Despite the high-cure rate with initial therapy, ∼5–10% of HL patients are refractory to initial treatment, and 10–30% of patients will relapse after achieving an initial complete remission [56, 57]. HDCT followed by an ASCT is the standard of care for many patients who relapse following a response to initial chemotherapy.
Primary refractory disease.
Patients with primary refractory disease, defined as progression or nonresponse during induction treatment or within 90 days of completing treatment, generally have a dismal clinical course. Second-line chemotherapy for these patients produces low-response rates, with long-term disease-free survival in only 5–10% of patients [58, 59]. Therefore, in these patients, HDCT with ASCT is currently considered to be the treatment of choice. A number of retrospective analyses have suggested that patients treated with ASCT have a superior long-term outcome when compared with patients treated with chemotherapy [60, 61]. An analysis of treatment outcome in primary progressive patients showed that the 5-year freedom from failure and OS for all patients were 17and 26%, respectively, compared to 31 and 43% for those treated with HDCT and an ASCT . Other studies have further confirmed that patients receiving HDCT followed by ASCT have a better outcome than patients treated with chemotherapy [62–64]. However, a majority of these patients still relapse following HDCT and ASCT.
Between 10 and 30% of patients will relapse following an initial chemotherapy regimen. In the past, it was a standard practice for HL patients with late relapses after an initial complete response (>1 year) to be treated with the same chemotherapeutic regimen that they had received as the first-line treatment. More than 80% of patients with a late relapse achieve a second complete response, with a median survival of about 4 years . In contrast, for patients with a relapse within 12 months, different salvage chemotherapeutic regimens were tested incorporating drugs not used in the initial combination [65–67]. However, no randomized trials have been done comparing the effectiveness of different conventional salvage chemotherapeutic regimens, and no standard salvage regimen has been identified. Despite the responses to these therapeutic approaches, however, patients relapse and subsequently die of disease progression or complications of treatment. Most eligible patients with relapsed disease are now treated with an ASCT.
Initial phase II studies suggested that HDCT followed by ASCT may produce a better long-term disease-free survival than expected with conventional chemotherapy in 30–65% of patients [68, 69]. Two subsequent randomized studies confirmed an improved outcome in patients with relapsed HL treated with HDCT followed by ASCT when compared with conventional salvage chemotherapeutic regimens [70, 71]. In both studies, the event-free survival after 3 years of patients treated with HDCT was over 50%.
Not all patients are eligible for, or may benefit from, an ASCT. Elderly patients treated with an ASCT have increased treatment-related mortality and commonly have an inferior event-free survival when compared with younger patients . Some patients have relentlessly progressive disease and have been treated with tandem ASCT  or allogeneic transplantation . Preliminary results have suggested that these therapies are feasible, but toxicity and relapses have been common.
Therapeutic options following relapse after HDCT and ASCT.
Patients with progression of disease after ASCT uniformly have a poor outcome. In a study of HL patients who failed ASCT, the median time to progression after the next therapy was only 3.8 months, and the median survival after ASCT failure was 26 months . For patients who fail high-dose chemotherapy (HDCT) with ASCT, few good treatment options exist. Among cytotoxic drugs, only vinorelbine  and gemcitabine  have shown promising activity in heavily pretreated HL patients, including some cases who relapsed after HDCT. The duration of many of the responses, however, was short, and the treatment was commonly associated with significant hematological toxicity. For HL patients treated with a reduced intensity allogeneic transplant, the treatment-related mortality at 1 year was ∼20%, and the 2-year OS was 50% . The treatment-related mortality and OS were significantly worse for older patients. In patients with relapsed or refractory nodular lymphocyte predominant HL and other multiple-relapsed CD20-positive cases of HL, studies of rituximab have shown high-response rates, but, unfortunately, these have been of short duration [79, 80].
Recent studies have evaluated the efficacy of antibody-drug conjugates targeting CD30 in relapsed and refractory HL (Table III). CD30 is expressed on the Reed–Sternberg cell, and antibodies targeting this molecule have shown activity in vitro. Phase I and II clinical trials of brentuximab vedotin (SGN-35), an antibody to CD30 conjugated to an antitubulin agent auristatin, showed significant clinical activity [81, 82]. In the phase II trial of 102 patents with HL that had failed an ASCT, an ORR of 75% was seen with a CR rate of 34%. The median duration of response was 47 weeks, and the agent was relatively well-tolerated . The median duration of response for those in CR was 20.5 months.
Table III. Promising New Agents in the Treatment of Relapsed and Refractory Hodgkin Lymphoma
Other agents with promising activity in this patient population include histone deacetylase (HDAC) inhibitors, mTOR inhibitors, and immunomodulatory agents (see Table III). Panobinostat (LBH 589), an HDAC inhibitor, was tested in patients who failed a prior ASCT. In a study of 129 patients, an ORR of 27% (35 patients) was seen including 30 partial responses (PR) and 5 CRs. The median duration of response was 6.9 months, and the median PFS was 6.1 months . The mTOR inhibitor everolimus was also tested in relapsed HL. The ORR was 47% with eight patients achieving a PR and one patient achieving a CR. The median TTP was 7.2 months. Four responders remained progression free at 12 months . Lenalidomide, an immunomodulatory agent, was tested in 38 patients with relapsed or refractory HL and 19% of eligible patients had a response (1 CR and 6 PRs). The median duration of CR or PR was 6 months, and the median time to treatment failure in responders was 15 months .
Optimal management of patients with HL requires an accurate diagnosis and careful staging of the disease. Identification of poor prognostic features then allows for risk-adapted therapy to potentially increase the likelihood of cure and minimize toxicity. Future directions to further improve the outcome of patients with HL will include incorporating into frontline therapy those agents that are effective in the relapsed setting.