You are here
Evaluating Cell-of-Origin Subtype Methods for Predicting Diffuse Large B-Cell Lymphoma Survival: A Meta-Analysis of Gene Expression Profiling and Immunohistochemistry Algorithms
Clinical Lymphoma Myeloma and Leukemia
Patients with DLBCL exhibit widely divergent outcomes despite harboring histologically identical tumors. Currently, GEP and IHC algorithms assign patients to 1 of 2 main subtypes: germinal center B cell-like (GCB), or activated B cell-like (ABC), the latter of which historically carries a less favorable prognosis. However, it remains controversial as to whether these prognostic groupings remain valid in the era of rituximab therapy.
Materials and Methods
A systematic literature review identified 24 articles from which meta-analyses were conducted, comparing survival outcomes for patients assigned to either GCB or ABC/non-GCB subtype using GEP and/or Hans, Choi, or Muris IHC algorithms.
Patients designated as GCB DLBCL using GEP fared significantly better in terms of overall survival than those with ABC DLBCL (hazard ratio, 1.85; P < .0001). In contrast, the Hans and Choi algorithms failed to identify significant differences in overall survival (P = .07 and P = .76, respectively) between GCB and non-GCB groups.
Our study illustrates a lack of evidence supporting the use of the Hans and Choi algorithms for stratifying patients into distinct prognostic groups. Rather, GEP remains the preferred method for predicting the course of a patient's disease and informing decisions regarding treatment options.
Keywords: Chemoimmunotherapy, Non-Hodgkin lymphoma, Prognosis, Rituximab, Systematic review.
Although diffuse large B-cell lymphoma (DLBCL) is the most common form of non-Hodgkin lymphoma, clinical presentation, response to therapy, and long-term prognosis varies considerably across patients, suggesting underlying differences in the biology of this disease despite a uniform histologic appearance. 1 Alizadeh et al classified DLBCL tumors into germinal center B cell-like (GCB) and activated B cell-like (ABC) subtypes based on distinct patterns of gene expression profiling (GEP) that share features with corresponding normal B-cell counterparts. 2 These subtypes use distinct mechanisms of oncogenesis and exhibit divergent expected outcomes with standard treatment.2 and 3
Because GEP has not yet been adopted as a standard approach in most clinical settings, researchers have attempted to recapitulate the prognostic grouping of GEP with immunohistochemistry (IHC). Several IHC algorithms have been developed4, 5, 6, 7, and 8 to assign patients into GCB and non-GCB/ABC subtypes.8, 9, 10, and 11 The first such algorithm was described by Hans et al, in which staining for Cluster of differentiation 10, B-cell lymphoma 6, and Multiple myeloma 1 is performed in a stepwise manner. The Choi and Muris methods, among others, were devised using additional cell surface markers in an attempt to improve the prognostic value of IHC. One report measured the concordance of the Hans, Choi, and Muris algorithms with GEP to be 86%, 87%, and 77% respectively. 8
Although classifying DLBCL as GCB or ABC lends insight into lymphoma biology, clinically useful classification schemas must identify subgroups with meaningful differences in survival. Historically, patients with DLBCL have been treated with chemotherapy regimens containing CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone), resulting in a 5-year overall survival (OS) and failure-free survival of 57% and 39%, respectively. The inclusion of rituximab with CHOP (R-CHOP) improved 5-year OS and failure-free survival to 67% and 52%, respectively. 12 Before the era of rituximab therapy, patients with an ABC subtype had a 5-year OS of 35% compared with 60% for those with GCB DLBCL (P < .001). 3 Since the inclusion of rituximab with standard therapy, outcomes for GCB and ABC subtypes have improved. In one study, the 3-year OS rates for patients treated with R-CHOP were 92% and 44% for GEP-defined GCB and ABC cases, respectively (P < .001), and 87% and 44% for Choi IHC-defined GCB and ABC cases, respectively (P < .001). 6 However, the prognostic significance of DLBCL subtype designation has been called into question, particularly when IHC is used.13 and 14 To further examine whether GEP and IHC are appropriate prognostic tools in the era of rituximab therapy, we performed a systematic review of the literature and meta-analyses evaluating OS and progression-free survival (PFS) in patients assigned to either GCB or ABC/non-GCB subtype using GEP and/or IHC.
Materials and Methods
Systematic Literature Review
Studies that directly compared OS and PFS for GCB and ABC/non-GCB subtypes were identified in the MedLine database using a series of searches using combinations of the medical subject heading terms, “Lymphoma, large B-cell, diffuse,” “gene expression profiling,” “immunohistochemistry,” “survival rate,” “survival analysis,” and “prognosis,” generating a pool of 361 articles that were assessed for suitability as shown in Figure 1 . Four reviewers (JAR, JNW, JLK, and JBC) independently performed study selection and quality assessment using a predefined format. Any disagreement was resolved by another reviewer (LJN or CRF), each of whom independently confirmed that inclusion criteria were met for all selected studies and that exclusion criteria were met in a random sample (15%) of the remaining studies. Studies were carefully screened for possible duplication of study population based on the author list, participating institutions, and period of patients' diagnosis and accrual.
Meta-Analysis Inclusion Criteria, Study Selection, and Data Extraction
Criteria for inclusion in the meta-analysis were: (1) articles published in English between January 1, 2007 and November 30, 2013, with a full set of experimental details; (2) a cohort of patients with de novo DLBCL, analyzed separately, who had been treated exclusively with a regimen including rituximab and anthracycline-based chemotherapy; (3) a direct comparison between GCB and ABC/non-GCB subtypes reported as Kaplan–Meier survival data with the outcome expressed in terms of hazard ratio (HR), or data from which the HR could be calculated; and (4) patients included in an analysis not duplicated in any other article included in our meta-analysis. When more than 1 article featured patients from the same study or reported the same authors and institutions during the same recruitment period, then only the study with the largest number of participants was included in the meta-analysis. We extracted data for patient characteristics (age, sex, Ann Arbor stage, and International Prognostic Index score), follow-up time, and HR. HRs compared PFS and OS for patients in the GCB group with those in the ABC (GEP data) or non-GCB (IHC data) group. When multiple IHC algorithms were used to analyze survival data, each set of data was extracted and analyzed separately.
In cases for which HR was not reported, it was calculated indirectly from the data available, according to the methods described by Parmar et al. 15 In cases in which the only available data were presented in the form of graphical survival curves, the freely available Engauge Digitizer software version 4.1 (SourceForge, http://digitizer.sourceforge.net/ ) was used to extract survival rates at specified time points, assuming that the rate of patient censoring was constant throughout the follow-up period. HR was then calculated using data points for each group.
Data Analysis and Statistical Methods
We used the freely available Review Manager (RevMan) software version 5.2 (The Cochrane Collaboration, http://ims.cochrane.org/revman/download ) to perform all statistical analyses. Forest plots were generated using the rmeta package in R (The R Foundation for Statistical Computing, http://cran.r-project.org/package=rmeta ). Individual estimates of HR for OS and PFS were combined using a fixed effects or random effects model based on the degree of heterogeneity detected. To test for the presence of study heterogeneity, a χ2 test was performed and the corresponding I2 value calculated. 16 Significant heterogeneity was detected when the χ2 value was greater than the degrees of freedom, and in these cases a random effects model was used rather than a fixed effects model. Forest plots were generated for each analysis to display the result of combining individual study estimates. To test for the presence of bias in the studies collected, funnel plots were generated for each analysis and visually evaluated for asymmetry. To assess the sensitivity of the results to the effect of individual studies, the meta-analyses were iteratively repeated with each study omitted.
Characteristics of the Study Population
The search strategy generated an initial pool of 361 articles, which was narrowed to 24 using the inclusion criteria depicted in Figure 1 . Of these, 3 articles included only GEP data,17, 18, and 19 18 included only IHC data,8, 13, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, and 36 and 3 included GEP and IHC data.9, 37, and 38 The included reports were published between 2008 and 2013. In total, 1120 patients with GEP data and 2618 patients with IHC data were included in the meta-analyses. Patient characteristics, treatment protocols and experimental details from each study are presented in Table 1 . For each meta-analysis, the results were unchanged when repeated with each study removed iteratively (data not shown), indicating that the results were insensitive to the effects of individual studies. Funnel plots of OS and PFS end points were symmetric for each meta-analysis and did not demonstrate evidence of publication bias (Supplemental Figure 1 and Supplemental Figure 2 in the online version). In each meta-analysis, the χ2 measure was greater than the degrees of freedom, and thus each analysis was performed using a random effects model.
|Reference||Patients, n||Median Age, Years||Male, n (%)||Median Follow-Up, Years||Stage III/IV, n (%)||IPI >2, n (%)||Method||Treatment Protocol|
|Barrans et al, 2012 37||140||67||88 (63)||–||70 (56)||–||GEP, Hans||R-CHOP|
|Green et al, 2012 20||193||64||113 (59)||4.7||92 (48)||65 (34)||Hans||R-CHOP|
|Gutierrez-Garcia et al, 2011 9||157||65||77 (49)||4.3||94 (60)||–||GEP, Hans, Choi, Muris||R-CHOP|
|Hong et al, 2012 21||70||59||33 (47)||2.5||34 (49)||20 (29)||Hans, Muris||R-CHOP|
|Hu et al, 2013 17||466||–||272 (58)||4.8||228 (51)||75 (47)||GEP||R-CHOP|
|Huang et al, 2012 22||97||59||60 (62)||4.7||40 (41)||10 (10)||Hans||R-CHOP|
|Ilic et al, 2009 31||92||51||41 (45)||3.1||63 (68)||40 (43)||Hans||R-CHOP or R-CHOP like|
|Jais et al, 2008 18||67||69||33 (49)||3.3||–||148 (36)||GEP||R-CHOP|
|Koh et al, 2013 33||120||60||75 (63)||3.5||59 (49)||39 (33)||Choi||R-CHOP|
|Kojima et al, 2013 32||100||67||56 (56)||4.2||40 (40)||33 (33)||Hans, Choi, Muris||R-CHOP or R-CHOP like|
|Lanic et al, 2012 38||57||65||30 (53)||2.3||–||28 (64)||GEP, Hans||R-CHOP|
|Lenz et al, 2008 19||233||–||–||2.1||121 (54)||–||GEP||R-CHOP or R-CHOP like|
|Meyer et al, 2011 8||262||62||–||–||–||–||Hans, Choi, Muris||R-CHOP or R-CHOP like|
|Niitsu et al, 2011 23||101||51||54 (53)||3.5||72 (71)||–||Hans||R-CycloBEAP|
|Nyman et al, 2009 25||117||63||62 (53)||2.4||74 (63)||45 (39)||Hans||R-CHOP or R-CHOEP|
|Nyman et al, 2009 24||88||67||42 (48)||3.1||47 (53)||33 (38)||Muris||R-CHOP|
|Ott et al, 2010 13||313||68||171 (55)||–||161 (51)||132 (42)||Hans||R-CHOP|
|Porrata et al, 2012 26||99||61||54 (55)||2.7||67 (68)||54 (55)||Hans||R-CHOP|
|Seki et al, 2009 28||279||68||142 (51)||2.1||139 (50)||101 (36)||Hans||R-CHOP|
|Song et al, 2010 29||136||61||72 (53)||3||58 (43)||54 (40)||Hans||R-CHOP|
|Varoczy et al, 2012 34||51||53||21 (41)||–||–||–||Hans||R-CHOP|
|Wilson et al, 2012 35||43||60||24 (56)||5.2||29 (67)||18 (43)||Hans||DA-EPOCH-R|
|Wilson et al, 2008 36||63||50||40 (63)||4.5||45 (71)||25 (40)||Hans||DA-EPOCH-R|
Abbreviations: DA-EPOCH-R = dose accelerated etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab; GEP = gene expression profiling; IHC = immunohistochemistry; IPI = International Prognostic Index; R-CHOEP = rituximab, cyclophosphamide, doxorubicin, vincristine, etoposide, and prednisone; R-CHOP = rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CycloBEAP = rituximab, cyclophosphamide, vincristine, bleomycin, etoposide, doxorubicin, and prednisolone.
Meta-Analysis of GEP Data
Six articles were included in the meta-analysis for GEP data. Forest plots summarizing results of the meta-analyses comparing survival rates in individuals with GCB and ABC DLBCL are presented in Figure 2 . Patients diagnosed with GCB DLBCL using GEP had significantly better survival than patients diagnosed with ABC subtype. The pooled HR was 1.85 (95% confidence interval [CI], 1.46-2.35; P < .0001) for OS and 1.80 (95% CI, 1.36-2.38; P < .0001) for PFS. The I2 values for the OS and PFS data were calculated as 3% and 25%, respectively, indicating no heterogeneity using the criteria recommended by Higgins et al. 16
Meta-Analysis of IHC Data
Twenty-one articles were included in the meta-analyses of IHC data. Forest plots summarizing results of the meta-analyses comparing survival rates in individuals with GCB and non-GCB DLBCL defined using IHC methods (Hans, Choi, or Muris algorithms) are shown in Figure 3 .34 and 35 There was no significant difference in OS between patients designated as GCB or non-GCB subtype using the Hans method (n = 1970; Figure 3B ), although a significant difference in PFS was observed between the 2 groups ( Figure 3A ). The pooled HR was 1.27 (95% CI, 0.98-1.64; P = .07) for OS and 1.49 (1.09-2.03; P = .01) for PFS. The I2 values for the OS and PFS data were calculated as 37% and 39% respectively, implying low heterogeneity. The results of these meta-analyses were not dominated by any single study; the study by Meyer et al exerted the greatest influence on the results with a weight of 12.3%. 8
For patients assigned to GCB or ABC subtype using the Choi method (n = 594), there was no significant difference in OS or PFS ( Figure 3C and D ). The pooled HR was 1.04 (95% CI, 0.69-1.58; P = .85) for OS and 1.12 (0.54-2.32; P = .76) for PFS. The I2 values for the OS and PFS data were calculated as 26% and 71% respectively, indicating low to moderate heterogeneity.
When patients assigned as GCB subtype using the Muris method were compared with ABC patients (n = 640), significant differences in OS and PFS were observed ( Figure 3E and F ). The pooled HR was 2.12 (95% CI, 1.50-3.00; P < .0001) for OS and 2.07 (1.32-3.24; P = .002) for PFS. The I2 values for the OS and PFS data were calculated as 0% and 51%, respectively, indicating zero to moderate heterogeneity.
The meta-analyses conducted based on this systematic review assess the ability of GEP and the most commonly used IHC algorithms to classify DLBCL patients into subtypes with divergent PFS and OS. Our results demonstrate that the prognostic differences first described between ABC and GCB subtypes in the era before rituximab therapy remain consistent across more recent studies using GEP to classify patients treated with rituximab and anthracycline-based chemotherapy regimens. Patients assigned to the GCB group using GEP exhibited significantly better OS and PFS compared with those assigned to the ABC group. In contrast, the IHC methods produced widely variable results. When patients were stratified into GCB and non-GCB subtypes using the Hans algorithm, no significant difference was observed for OS, and only a small difference in PFS was seen. These results echo those of a recent meta-analysis performed with individual patient data on a more limited set of studies. 14 Our analysis identified low heterogeneity between studies using the Hans algorithm, which directly opposes the idea that IHC methods predominantly suffer from inconsistency when applied by different pathologists or in different patient populations. Similarly, pooled estimates of DLBCL subtype using the Choi algorithm failed to demonstrate a significant difference in OS or PFS.
The Muris method divided DLBCL patients into GCB and ABC subtypes with significant differences in HRs for PFS and OS, as seen with corresponding subtypes assigned by GEP. Although the Muris algorithm appears comparable to GEP in terms of prognostic ability, it is clear that the two are not equivalent. GEP and the Hans algorithm both assign patients to GCB or ABC/non-GCB subtypes in an approximately 1:1 ratio. In contrast, the Muris algorithm assigns more patients to the GCB than the ABC subtype, in a ratio of 1.9:1, and the Choi algorithm assigns more patients to the ABC subtype, in a ratio of 1:1.5. However, these ratios are aggregates drawn from the meta-analyses, and an inspection of individual studies reveals inconsistency in the distribution between subtypes. This likely reflects the variability in the population being studied and the variability inherent in the IHC methods. Overall, these findings support the notion that GEP remains the most robust method for assigning patients to distinct prognostic groups based on underlying DLBCL biology. Furthermore, the Hans and Choi algorithms lack prognostic value, at least in part because of misclassification of subtype.
The potential limitations of the current study are those inherent to carrying out any meta-analysis and include: publication bias, study selection bias, and synthesis of data derived from studies with heterogeneous methodology. To avoid the latter issue, our search strategy was focused to include only studies that compared survival data from patients treated with R-CHOP or similar immunochemotherapy regimens. It remains possible that our narrow inclusion criteria biased our results; however, from the qualitative examination of funnel plots produced from the meta-analyses (Supplemental Figure 1 and Supplemental Figure 2 in the online version), any publication bias appears to be minimal.
Much research has been conducted to better understand the genetic aberrations that drive the difference in survival between ABC and GCB subtypes. It is clear that GCB and ABC subtypes of DLBCL use distinct mechanisms of oncogenesis. For instance, ABC subtypes are dependent on nuclear factor (NF)-κB signaling as indicated by more frequent mutations in genes for TNFAIP3 (Tumor necrosis factor, alpha induced protein 3), CARD11 (Caspase recruitment domain-containing protein 11), CD79B (Cluster of differentiation 79B) and MYD88 (Myeloid differentiation primary response gene 88), and recent studies have demonstrated that mutations to BCL2 (B-cell lymphoma 2), MYC and EZH2 (Enhancer of Zeste homologue 2) occur more frequently in patients with the GCB subtype.39, 40, 41, and 42 Consequently, therapies targeting the NF-κB pathway appear more likely to be effective in patients with ABC DLBCL. 43 Examples of such therapies include bortezomib (a proteasome inhibitor that might weakly inhibit NF-κB activation), ibrutinib (an upstream B-cell receptor signaling pathway inhibitor), and lenalidomide, which acts by downregulating the transcription factors IRF4 (Interferon regulatory factor 4) and SPIB (Spi-B transcription factor), thus enhancing B-cell interferon-β production.1 and 44 The findings of this study indicate that GEP approaches should be used to define the optimal subgroup of DLBCL patients who require interventions that might improve on outcomes expected with R-CHOP.
Recent studies have argued that the prognostic difference between GCB and ABC subtypes might be attributed to concomitant MYC and BCL2 overexpression and/or translocation.17, 20, and 43 Although it has been firmly established that patients with translocations affecting MYC and BCL2 genes (so-called “double-hit” lymphoma) have an extremely poor prognosis, these patients are thought to represent only 5% of all cases of DLBCL. 20 In contrast, simultaneous overexpression of MYC and BCL2 as determined by IHC is more common, representing approximately 46% of patients with ABC DLBCL and 22% of those with the GCB subtype. 17 In a study by Hu et al, patients were divided into 2 groups: 1 group whose lymphoma was found to overexpress MYC and BCL2 and a second that did not overexpress both gene products. 17 When the authors examined patients within the MYC and BCL2 positive expression group, they found no difference in survival between patients who were further divided into GCB and ABC subtypes (P = .3163 for OS; P = .4291 for PFS). Similarly, there was no difference in survival between patients with GCB and ABC subtype lymphomas in the group that did not express MYC and BCL2 (P = .4114 for OS; P = .7020 for PFS). Further study will be necessary to determine how best to integrate MYC and BCL2 expression with ‘cell of origin’ subtyping when risk-stratifying patients newly diagnosed with DLBCL. Unfortunately, the effect of MYC and BCL2 expression could not be assessed in our meta-analysis because of limitations in the number of available IHC studies.
The meta-analyses performed in this study demonstrated that the Hans and Choi methods for classifying DLBCL subtypes do not hold significant prognostic value for patients treated with R-CHOP. Although these algorithms might be subject to inconsistency in their application by individual pathologists, the low to moderate heterogeneity observed for these studies suggests that the lack of prognostic value might relate to the IHC methods themselves rather than variability in the study populations or implementation of the IHC algorithms. Although the Muris method allows classification of DLBCL patients into subtypes with distinct prognoses, its accuracy is called into question because it assigns 2/3 of all patients to the GCB group. Thus, further study is necessary to determine whether the Muris method remains relevant in terms of predicting response to new therapies developed on the basis of GEP analysis.
At present, GEP remains the gold standard for classifying DLBCL into biologically and prognostically meaningful groups. A recent study evaluated the Lymphoma/Leukemia Molecular Profiling Project's Lymph2Cx assay, a digital gene expression test that uses 20 genes to assign GCB and ABC DLBCL subtypes using formalin-fixed paraffin-embedded tissue. 45 This test had high concordance with GEP performed on frozen tissue from the same cases and > 95% concordance between 2 independent laboratories. This test and similar approaches might promote the integration of GEP into broader clinical practice as technological advances make GEP faster, simpler, and more affordable.
Clinical Practice Points
- Gene expression profiling allows division of DLBCL patients who are treated with anthracycline-based chemoimmunotherapy into biologically distinct subtypes that have significant differences in survival.
- The Hans and Choi IHC algorithms divide DLBCL into subgroups that lack prognostic significance when studies of chemoimmunotherapy are combined.
The authors have stated that they have no conflicts of interest.
This work was supported by National Cancer Institute grant R21 CA158686 to Dr. Flowers and an American Society of Hematology Scholars Award to Dr. Nastoupil.
- 1 C.R. Flowers, R. Sinha, J.M. Vose. Improving outcomes for patients with diffuse large B-cell lymphoma. CA Cancer J Clin. 2010;60:393-408
- 2 A.A. Alizadeh, M.B. Eisen, R.E. Davis, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000;403:503-511 Crossref
- 3 A. Rosenwald, G. Wright, W.C. Chan, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med. 2002;346:1937-1947 Crossref
- 4 C.P. Hans, D.D. Weisenburger, T.C. Greiner, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103:275-282 Crossref
- 5 J.J. Muris, C.J. Meijer, W. Vos, et al. Immunohistochemical profiling based on Bcl-2, CD10 and MUM1 expression improves risk stratification in patients with primary nodal diffuse large B cell lymphoma. J Pathol. 2006;208:714-723 Crossref
- 6 W.W. Choi, D.D. Weisenburger, T.C. Greiner, et al. A new immunostain algorithm classifies diffuse large B-cell lymphoma into molecular subtypes with high accuracy. Clin Cancer Res. 2009;15:5494-5502 Crossref
- 7 H. Nyman, M. Adde, M.L. Karjalainen-Lindsberg, et al. Prognostic impact of immunohistochemically defined germinal center phenotype in diffuse large B-cell lymphoma patients treated with immunochemotherapy. Blood. 2007;109:4930-4935 Crossref
- 8 P.N. Meyer, K. Fu, T.C. Greiner, et al. Immunohistochemical methods for predicting cell of origin and survival in patients with diffuse large B-cell lymphoma treated with rituximab. J Clin Oncol. 2011;29:200-207 Crossref
- 9 G. Gutierrez-Garcia, T. Cardesa-Salzmann, F. Climent, et al. Gene-expression profiling and not immunophenotypic algorithms predicts prognosis in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. Blood. 2011;117:4836-4843 Crossref
- 10 C.H. Lawrie, E. Ballabio, E. Soilleux, et al. Inter- and intra-observational variability in immunohistochemistry: a multicentre analysis of diffuse large B-cell lymphoma staining. Histopathology. 2012;61:18-25 Crossref
- 11 R. Coutinho, A.J. Clear, A. Owen, et al. Poor concordance among nine immunohistochemistry classifiers of cell-of-origin for diffuse large B-cell lymphoma: implications for therapeutic strategies. Clin Cancer Res. 2013;19:6686-6695 Crossref
- 12 T.M. Habermann, E.A. Weller, V.A. Morrison, et al. Rituximab-CHOP versus CHOP alone or with maintenance rituximab in older patients with diffuse large B-cell lymphoma. J Clin Oncol. 2006;24:3121-3127 Crossref
- 13 G. Ott, M. Ziepert, W. Klapper, et al. Immunoblastic morphology but not the immunohistochemical GCB/nonGCB classifier predicts outcome in diffuse large B-cell lymphoma in the RICOVER-60 trial of the DSHNHL. Blood. 2010;116:4916-4925 Crossref
- 14 J.J. Castillo, B.E. Beltran, M.K. Song, et al. The Hans algorithm is not prognostic in patients with diffuse large B-cell lymphoma treated with R-CHOP. Leukemia Res. 2012;36:413-417 Crossref
- 15 M.K. Parmar, V. Torri, L. Stewart. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med. 1998;17:2815-2834 Crossref
- 16 J.P. Higgins, S.G. Thompson, J.J. Deeks, D.G. Altman. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557-560 Crossref
- 17 S. Hu, Z.Y. Xu-Monette, A. Tzankov, et al. MYC/BCL2 protein coexpression contributes to the inferior survival of activated B-cell subtype of diffuse large B-cell lymphoma and demonstrates high-risk gene expression signatures: a report from The International DLBCL Rituximab-CHOP Consortium Program. Blood. 2013;121:4021-4031 Crossref
- 18 J.P. Jais, C. Haioun, T.J. Molina, et al. The expression of 16 genes related to the cell of origin and immune response predicts survival in elderly patients with diffuse large B-cell lymphoma treated with CHOP and rituximab. Leukemia. 2008;22:1917-1924 Crossref
- 19 G. Lenz, G. Wright, S.S. Dave, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008;359:2313-2323 Crossref
- 20 T.M. Green, K.H. Young, C. Visco, et al. Immunohistochemical double-hit score is a strong predictor of outcome in patients with diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30:3460-3467 Crossref
- 21 J. Hong, S. Park, J. Park, et al. CD99 expression and newly diagnosed diffuse large B-cell lymphoma treated with rituximab-CHOP immunochemotherapy. Ann Hematol. 2012;91:1897-1906 Crossref
- 22 Y. Huang, S. Ye, Y. Cao, et al. Outcome of R-CHOP or CHOP regimen for germinal center and nongerminal center subtypes of diffuse large B-cell lymphoma of Chinese patients. ScientificWorldJournal. 2012;2012:897178
- 23 N. Niitsu, J. Tamaru, T. Yoshino, et al. A study on nm23-H1 expression in diffuse large B-cell lymphoma that was treated with CyclOBEAP plus rituximab therapy. Ann Hematol. 2011;90:185-192 Crossref
- 24 H. Nyman, M. Jerkeman, M.L. Karjalainen-Lindsberg, A.H. Banham, S. Leppa. Prognostic impact of activated B-cell focused classification in diffuse large B-cell lymphoma patients treated with R-CHOP. Mod Pathol. 2009;22:1094-1101 Crossref
- 25 H. Nyman, M. Jerkeman, M.L. Karjalainen-Lindsberg, A.H. Banham, G. Enblad, S. Leppa. Bcl-2 but not FOXP1, is an adverse risk factor in immunochemotherapy-treated non-germinal center diffuse large B-cell lymphomas. Eur J Haematol. 2009;82:364-372 Crossref
- 26 L.F. Porrata, K. Ristow, T.M. Habermann, et al. Absolute monocyte/lymphocyte count prognostic score is independent of immunohistochemically determined cell of origin in predicting survival in diffuse large B-cell lymphoma. Leuk Lymphoma. 2012;53:2159-2165 Crossref
- 27 J. Ruan, P. Martin, R.R. Furman, et al. Bortezomib plus CHOP-rituximab for previously untreated diffuse large B-cell lymphoma and mantle cell lymphoma. J Clin Oncol. 2011;29:690-697 Crossref
- 28 R. Seki, K. Ohshima, T. Fujisaki, et al. Prognostic impact of immunohistochemical biomarkers in diffuse large B-cell lymphoma in the rituximab era. Cancer Sci. 2009;100:1842-1847 Crossref
- 29 M.K. Song, J.S. Chung, Y.M. Seol, et al. Influence of low absolute lymphocyte count of patients with nongerminal center type diffuse large B-cell lymphoma with R-CHOP therapy. Ann Oncol. 2010;21:140-144 Crossref
- 30 Z.G. Xia, Z.Z. Xu, W.L. Zhao, et al. The prognostic value of immunohistochemical subtyping in Chinese patients with de novo diffuse large B-cell lymphoma undergoing CHOP or R-CHOP treatment. Ann Hematol. 2010;89:171-177 Crossref
- 31 I. Ilic, Z. Mitrovic, I. Aurer, et al. Lack of prognostic significance of the germinal-center phenotype in diffuse large B-cell lymphoma patients treated with CHOP-like chemotherapy with and without rituximab. Int J Hematol. 2009;90:74-80 Crossref
- 32 M. Kojima, H. Nishikii, J. Takizawa, et al. MYC rearrangements are useful for predicting outcomes following rituximab and chemotherapy: multicenter analysis of Japanese patients with diffuse large B-cell lymphoma. Leuk Lymphoma. 2013;54:2149-2154 Crossref
- 33 Y.W. Koh, H.S. Hwang, S.J. Jung, et al. Receptor tyrosine kinases MET and RON as prognostic factors in diffuse large B-cell lymphoma patients receiving R-CHOP. Cancer Sci. 2013;104:1245-1251 Crossref
- 34 L. Varoczy, E. Zilahi, A. Gyetvai, et al. Fc-gamma-receptor IIIa polymorphism and gene expression profile do not predict the prognosis in diffuse large B-cell lymphoma treated with R-CHOP protocol. Pathol Oncol Res. 2012;18:43-48
- 35 W.H. Wilson, S.H. Jung, P. Porcu, et al. A Cancer and Leukemia Group B multi-center study of DA-EPOCH-rituximab in untreated diffuse large B-cell lymphoma with analysis of outcome by molecular subtype. Haematologica. 2012;97:758-765
- 36 W.H. Wilson, K. Dunleavy, S. Pittaluga, et al. Phase II study of dose-adjusted EPOCH and rituximab in untreated diffuse large B-cell lymphoma with analysis of germinal center and post-germinal center biomarkers. J Clin Oncol. 2008;26:2717-2724 Crossref
- 37 S.L. Barrans, S. Crouch, M.A. Care, et al. Whole genome expression profiling based on paraffin embedded tissue can be used to classify diffuse large B-cell lymphoma and predict clinical outcome. Br J Haematol. 2012;159:441-453 Crossref
- 38 H. Lanic, S. Mareschal, F. Mechken, et al. Interim positron emission tomography scan associated with international prognostic index and germinal center B cell-like signature as prognostic index in diffuse large B-cell lymphoma. Leuk Lymphoma. 2012;53:34-42 Crossref
- 39 J. Zhang, V. Grubor, C.L. Love, et al. Genetic heterogeneity of diffuse large B-cell lymphoma. Proc Natl Acad Sci U S A. 2013;110:1398-1403 Crossref
- 40 R.D. Morin, M. Mendez-Lago, A.J. Mungall, et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature. 2011;476:298-303 Crossref
- 41 L. Pasqualucci, V. Trifonov, G. Fabbri, et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet. 2011;43:830-837 Crossref
- 42 J.G. Lohr, P. Stojanov, M.S. Lawrence, et al. Discovery and prioritization of somatic mutations in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci U S A. 2012;109:3879-3884 Crossref
- 43 R. Sinha, N. DeJoubner, C. Flowers. Novel agents for diffuse large B-cell lymphoma. Expert Opin Investig Drugs. 2011;20:669-680 Crossref
- 44 W.H. Wilson, F.J. Hernandez-Ilizaliturri, K. Dunleavy, R.F. Little, O.A. O'Connor. Novel disease targets and management approaches for diffuse large B-cell lymphoma. Leuk Lymphoma. 2010;51(suppl 1):1-10 Crossref
- 45 M.Y. Choi, T.J. Kipps. Inhibitors of B-cell receptor signaling for patients with B-cell malignancies. Cancer J. 2012;18:404-410 Crossref
1 Emory University School of Medicine, Atlanta, GA
2 The University of Texas M.D. Anderson Cancer Center, Houston, TX
∗ Address for correspondence: Christopher R. Flowers, MD, MS, Lymphoma Program, Oncology Data Center, Bone Marrow and Stem Cell Transplantation, Winship Cancer Institute, 1365 Clifton Rd NE, Building B, Suite 4302, Emory University, Atlanta, GA 30322
© 2014 Elsevier Inc., All rights reserved.
Comments by the editor:
A systematic literature review on 24 articles that compared treatment outcomes in patients with Diffuse Large B-cell Lymphoma (DLBCL) assigned to either GCB or ABC/non-GCB subtype by GEO and /or immunohistochemistry (Hans, Choi or Murrs alghorithms) shows that cell of origin distinction by GEP but not IHC can predict a better outcome (in GCB patients in comparison to ABC DLBCL patients)