Diffuse large B-cell lymphoma (DLBCL) is the most commonly occurring lymphoid malignancy. While a series of trials support R-CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone, plus rituximab)-21 as the standard of care for all patients, DLBCL has substantial biological and clinical heterogeneity, leading to marked differences in outcomes for disease subgroups. We examine clinical, biological, and functional imaging techniques for risk-stratifying patients, and we review approaches for dose intensification in the rituximab era that are aimed at improving outcomes for poor-risk patients. Together, the results achieved with these measures indicate no particular benefit for administering R-CHOP-14 vs R-CHOP-21 in older or younger patients with DLBCL, highlight opportunities for future studies of young patients with high-risk DLBCL, and suggest the promise of biologic risk stratification. Such approaches will provide key opportunities for further advances in the treatment of DLBCL, given that chemotherapy intensification appears to provide limited additional benefits over the current standard of care.
Diffuse large B-cell lymphoma (DLBCL) is the most commonly occurring lymphoid malignancy and accounts for one-third of adult cases of non-Hodgkin lymphoma (NHL). It is regarded as an aggressive lymphoma, characterized by rapid growth and limited survival in the absence of treatment or with inadequate treatment. DLBCL is both biologically and clinically heterogeneous, with at least three important subtypes (activated B-cell [ABC], germinal center B-cell [GCB], and primary mediastinal large B-cell lymphoma [PMBL]), features of which carry predictive and prognostic relevance. A substantial number of patients are cured with standard anthracycline-based chemotherapy, and with the addition of rituximab (R; Rituxan) to cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP), there have been significant improvements in clinical outcomes, including overall survival (OS).[1-4] However, a significant portion of high-risk patients still fail to achieve the desired response to induction therapy and have a poor prognosis with current standard treatments. We examine clinical and biological features of DLBCL patients with poor outcomes, and we review recent studies addressing alternatives to standard front-line management strategies together with unresolved questions.
Defining Poor-Risk Subgroups of DLBCL
Clinical prognostic factors
Developed in the 1990s, the international prognostic index (IPI) is the most functional clinical tool used to predict outcome for patients with DLBCL. This model was derived from a pooled analysis of more than 2000 patients with aggressive lymphoma (mainly DLBCL) treated with an anthracycline-containing chemotherapy regimen between 1982 and 1987. Clinical features that were predictive of OS and relapse-free survival were identified and include: age, stage, serum lactate dehydrogenase (LDH), performance status, and number of extranodal disease sites. Patients are scored based on these clinical features and can then be stratified into one of four discrete groups with 5-year predicted survival rates of 73%, 51%, 43%, and 26%. In addition, this index has enabled comparisons of published study results, and it supports trials designed to classify patients with similar characteristics and expected outcomes. A question that has been raised is whether the IPI remains prognostic in the rituximab era. Sehn et al performed a retrospective analysis of patients with DLBCL treated with R-CHOP and redefined three outcome categories, termed the revised IPI, which was a more clinically useful predictor of outcome than the IPI in this population. All groups had a predicted 4-year OS greater than 50%. A more recent analysis of more than 1000 patients treated with chemo-immunotherapy suggests that the IPI is useful in the rituximab era for predicting event-free survival (EFS), progression-free survival (PFS), and OS. The IPI continues to be an important tool for clinicians to use to risk-stratify patients and to provide the prognostic information that is often requested by patients in consultation. At present, this index remains central to the design and interpretation of DLBCL clinical trials, by facilitating identification and classification of discrete patient populations.
Other clinical factors also may have prognostic significance. In addition to its use in the IPI, age has been one of the most robust adverse prognostic features in NHL. Numerous studies have associated older age with inferior outcomes, including OS.[8-11] One may speculate that reasons for worse survival among older individuals with DLBCL include the higher incidence of comorbid conditions, poorer performance status, and possibly inferior treatment prescribed, given concerns about the ability of older individuals to tolerate aggressive regimens. A retrospective analysis of very elderly patients (> 80 years; median, 83 years) reported that this patient population presented with similar clinical and prognostic features as younger patients, but they had significant differences in disease management compared with their younger counterparts. The authors concluded that the majority of deaths were attributed to progressive disease and that age was the limiting factor in prescribing effective treatment. With the expanding life expectancy and aging population, the significance of age and its implications for outcomes will only increase in importance, and clinical trials are needed to address this question.
Race also may influence the prognosis of DLBCL. Notably, African-American patients with DLBCL in the US present at a younger age and more advanced stage, and they have inferior survival.[13,14] Among 37,009 DLBCL cases diagnosed from 1992 to 2005 in the Surveillance, Epidemiology, and End Results (SEER) registry, the majority of African-American patients presented with stage III/IV disease (54%), and 5-year survival rates were 38% for black vs 46% for white patients (P = .02). Disparities in outcomes have been observed even when the same treatment is administered to black vs white patients. What has not been clearly elucidated to date is whether racial and age variances are a result of differences in tumor biology that can explain inferior outcomes or are a reflection of disparities in socioeconomic status or other nonclinical factors that influence outcome. Future studies are needed to examine interactions between these and other clinical factors and biological predictors of differences in presentation and outcome for patients with DLBCL.
Biological factors and functional imaging
Despite their practical value in distinguishing DLBCL patients, prognostic clinical features are likely surrogate markers for biological heterogeneity or composite measures that mix the influences of tumor biology, clinical factors, and nonclinical factors. Over the past decade there have been significant efforts to define biologically relevant subgroups of DLBCL that can lead to rational therapeutic strategies. Distinct gene-expression patterns have been defined using hierarchical clustering from DNA microarrays that reflect differences in cell of origin, proliferation rate, and host immune response to the tumor, and which can be used as molecular predictors of survival. The encompassing diagnosis of DLBCL can now be subdivided by gene-expression profiling (GEP) into at least three molecular subtypes: ABC, GCB, and PMBL DLBCL. When treated with CHOP-like regimens, patients in the GCB subgroup had better 5-year survival rates than patients with ABC DLBCL (OS, 60% vs 35%, P < .001), independent of IPI risk. Additional studies using GEP as predictors of survival have confirmed these findings and expanded upon them.[17-19] An analysis from the Lymphoma/Leukemia Molecular Profiling Project performed GEP on 233 tumor biopsy samples from patients with DLBCL treated with R-CHOP. A multivariate gene-expression-based survival model predicted a favorable prognosis for a GCB signature and paralleled the distinction between ABC and GCB in studies prior to the ubiquitous use of rituximab. In addition, differences in immune cells, fibrosis, and angiogenesis in the tumor microenvironment also may influence survival in DLBCL.
With improved technology that allows for GEP of paraffin-embedded tissue, more comprehensive analysis may be possible in the future, with further exploration of clinically significant biomarkers. However, routine clinical application of GEP is not currently widespread; instead, extrapolation of gene-expression results to immunohistochemistry (IHC) algorithms is more practical and appears to determine the cell of origin in the majority of cases.[21-23] Individual biomarkers, including BCL2, BCL6, p21, and c-myc, also have prognostic significance in DLBCL.[24-27] Present in 10% to 20% of cases, c-myc expression or amplification portends a poor prognosis, with OS less than 30% at 2 years. When it is associated with additional unfavorable biomarkers, such as BCL2, it results in an extremely poor prognosis. Discerning cell-of-origin phenotypes is particularly important for identifying poor-risk patients, so that rational therapeutic strategies can be developed and employed.
Functional imaging may be another approach for risk-stratifying patients at diagnosis or early in the course of therapy. While positron emission tomography (PET) with [18F] fluorodeoxyglucose (FDG) is often part of staging and assessment of DLBCL, there has been interest in using interim PET assessment to identify patients who are at high risk for refractory disease or relapse after standard therapy. National Comprehensive Cancer Network (NCCN) guidelines currently include the use of interim PET/CT after 2 to 4 cycles of therapy. Evidence for the use of PET assessment during chemotherapy has been limited by study heterogeneity and conflicting results. Further exploration of this question has raised additional questions regarding the value of interim PET scans for defining prognosis and supporting decision making. Ghesquires et al investigated whether the use of interim or post-treatment completion PET had an association with prognosis. Consistent with previous findings, a positive PET following completion of initial therapy was associated with a poorer prognosis compared with patients with a negative PET (5-year OS, 50% vs 84%, P = .001). These findings are in contrast to a prospective trial from Memorial Sloan-Kettering in which patients with residual FDG-PET–positive disease after 4 cycles of accelerated R-CHOP underwent a repeat biopsy. The results of the interim FDG-PET assessment did not correlate with PFS, and only 5 of 38 patients with positive scans had positive biopsies; the remaining tissue specimens showed inflammation. A phase II trial conducted by Stewart et al (n = 67) tested the use of PET/CT to guide the use of high-dose sequential induction therapy with autologous stem cell transplant (ASCT). Patients with an unfavorable interim PET/CT after two cycles of R-CHOP received R-DICEP (rituximab, dose-intensive cyclophosphamide, etoposide, cisplatin) followed by R-BEAM (rituximab, carmustine [BCNU], etoposide, Ara-C [cytarabine], melphalan)/ASCT while those with a favorable interim PET/CT received an additional 4 cycles of R-CHOP. With a median follow-up of 24 months, there was no significant difference in 2-year PFS between patients with unfavorable or favorable interim PET/CT, suggesting that PET-directed aggressive therapy for poor-risk patients may overcome the expected poor outcomes for this group. However, these findings remain premature and conflict with the results of the Memorial Sloan-Kettering study. Another limitation of interim PET assessment is reproducibility. The results of a blinded, independent review of the Eastern Cooperative Oncology Group (ECOG) study E3404 revealed that one-third of the time, the expert panel of nuclear medicine physicians disagreed on the interpretation of the interim scans despite using consensus criteria. In the absence of clear evidence that supports the routine practice of interim PET outside of a clinical trial, caution is advised regarding the use of this information to define prognosis or as a basis for treatment decisions.