Molecular Risk Stratification in Aggressive B-Cell Lymphomas

In recent years, it has become increasingly apparent that diffuse large B-cell lymphoma (DLBCL) is not a single pathologic entity. Widely available diagnostic techniques such as immunohistochemical staining and fluorescence in situ hybridization (FISH) have helped to subcategorize DLBCLs by their pathologic features and identify those associated with a lower probability of cure after standard first-line rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP). This recognition has prompted increased use of diagnostic testing to identify patients with high-risk aggressive lymphoma, the results of which have, at times, led clinicians to offer such patients alternative first-line treatment regimens in hopes of increasing their likelihood of achieving cure. Here, we further explore this concept of molecular risk stratification among patients with DLBCL or high-grade B-cell lymphoma (HGBL), consider potential limitations to performing such stratification in current practice, and suggest how risk stratification may be more optimally implemented now and in the future.

Perhaps the best example of how the widespread use of molecular diagnostics has had an impact on clinical management of DLBCL is the entity of double-hit lymphoma (DHL). DHL is defined in the 2016 WHO classification of lymphoid neoplasms as HGBL with rearrangements of MYC and BCL2 and/or BCL6.¹ Although this classification does not depend on morphology other than requiring a diffuse aggressive large B-cell lymphoma, there is some evidence to suggest that it may further discriminate outcomes among patients with DHL.^2,3 A decade after the first reports of a poor prognosis for patients with DHL who were treated with R-CHOP were published,^4,5 the lymphoma community has implemented two practice changes in an attempt to improve DHL outcomes: (1) attempting to capture more DHL cases by routinely performing FISH for MYC rearrangements along with either routine or reflexive FISH for BCL2 and BCL6 rearrangements at diagnosis and (2) treating fit patients who have DHL with intensive first-line immunochemotherapy based upon the results of retrospective analyses that demonstrate an apparent survival benefit to these regimens compared with R-CHOP therapy.

Although it seems as if these two practices have increased our ability to cure patients with newly diagnosed DHL, it is also possible that we are ultimately identifying a greater number of patients with low-risk DHL who may also be cured with R-CHOP and consequently are exposing them to unnecessary toxicity from intensive first-line immunochemotherapy regimens. The retrospective analyses that have supported the use of such regimens in DHL are prone to biases, most notably selection bias in terms of which patients were tested for DHL as well as which were treated with intensive first-line regimens as opposed to R-CHOP. Furthermore, while the prospective phase II clinical trial of first-line treatment with dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (DA-EPOCH-R) demonstrated favorable survival outcomes for newly diagnosed patients with MYC-rearranged DLBCL and HGBL (including DHL⁶), it lacked a comparator arm treated with R-CHOP. In addition to potentially overtreating some patients with DHL in the first-line setting, relying only on DHL for risk stratification might also be suboptimal in that it could miss additional patients with molecularly defined high-risk DLBCL or HGBL who are likely to have a poor prognosis after treatment with R-CHOP and could be considered for alternative first-line therapies.

Because predicting which patients have DHL from their presenting clinicopathologic features may not be possible,⁷ our institution and others^8-10 have adopted the practice of performing FISH for MYC rearrangement on all newly diagnosed patients with DLBCL and HGBL to identify all patients with DHL. We implemented routine testing (RT) in January 2015 using the MYC break-apart probe, along with reflex testing for BCL2 and BCL6 rearrangements if an MYC rearrangement was identified. Before this, selective testing (ST) was performed at the discretion of the treating physician or the interpreting hematopathologist. Despite there being no relevant differences in measured baseline clinicopathologic features or in rates of treatment with intensive first-line immunochemotherapy, patients with DHL who were identified by ST as opposed to RT had a 2-year relapse-free survival (RFS) of 43% versus 70% and a higher risk of 2-year relapse and death, even when accounting for clinical, pathologic, and treatment factors in a multivariable model.¹¹ These results did not change when considering only the 80% of patients treated with intensive immunochemotherapy in each group (2-year RFS, 38% v 70% and 2-year overall survival, 38% v 74%). In addition, for patients in the RT cohort, those who received R-CHOP compared with those who received intensive first-line immunochemotherapy experienced similar 2-year RFS (80% v 70%) as well as 2-year overall survival (67% v 74%), suggesting a possible lack of survival benefit from using intensive first-line immunochemotherapy in the RT cohort.

Although similarly limited by potential biases in treatment selection, both the differences in outcomes experienced by patients with DHL after RT was implemented despite no change in the use of intensive first-line treatment and the similar survival outcomes seen in the RT group regardless of first-line treatment intensity suggest that even though RT may identify a greater number of patients with DHL, this practice may also capture a collective group of patients with DHL who have a relatively better prognosis. Consequently, RT may not provide optimal prognostic information to inform treatment decisions. Additional evidence that this may be the case comes from an article by Sha et al,¹² who identified a molecular high-grade (MHG) gene expression profile (GEP) in patients with DLBCL. Patients with DHL with and without MHG features had a 3-year progression-free survival of approximately 30% and approximately 90%, respectively, with the latter exhibiting a prognosis that was comparable to that in patients lacking both DHL and MHG features.

Recent evidence has also suggested that focusing on DHL for risk stratification may not capture additional patients with molecularly defined high-risk DLBCL or HGBL who are likely to have a poor prognosis if treated with R-CHOP. Recent data have suggested that three major techniques may improve the molecular risk stratification of patients with DHL: identification of TP53 alterations, characterization of the MYC translocation partner, and combinations of genomic sequencing and GEP.

Studies in patients with MYC-rearranged aggressive B-cell lymphomas have suggested that TP53 alterations can act as a second hit and predict a poor prognosis similar to that for patients with DHL. In one study, patients with high P53 expression by immunohistochemistry along with MYC rearrangements showed a worse prognosis than with either alone.¹³ Similar results were seen in another cohort in which TP53 mutations were detected by molecular sequencing.¹⁴ In addition, Chapuy et al¹⁵ identified frequent TP53 mutations and 17p loss in a distinct cluster of patients with DLBCL who exhibited a poor prognosis and a distinct trajectory of continued progression and mortality events over time. Thus, in patients with MYC rearrangements, testing for TP53 mutations and deletions and/or P53 overexpression may capture additional high-risk patients, even in the absence of BCL2 or BCL6 rearrangements.

It also seems that not all MYC rearrangements carry the same poor prognosis. Studies of patients with aggressive B-cell lymphomas have suggested that when MYC is fused with an immunoglobulin (IG) heavy or light chain locus, the prognosis is worse.^4,16 Building on this work, Rosenwald et al¹⁷ recently demonstrated that in a large cohort of patients with DLBCL treated with R-CHOP, having DHL only portended a poorer prognosis when the MYC translocation partner was an IG locus.

Taking this information into account, we are proposing an algorithm that risk stratifies patients on the basis of the results of molecular testing currently available for clinical use (Fig 1). For patients identified as being molecular standard risk, available data do not clearly suggest inferior survival outcomes after receiving first-line R-CHOP therapy, and we believe that R-CHOP should be considered for these patients. For those identified as being molecular high risk and those with DHL, available data suggest that outcomes may be poor after receiving first-line R-CHOP. However, further studies are needed to clinically validate such a testing-driven treatment strategy. Therefore, alternative regimens, specifically in the context of a clinical trial, should be considered for these patients. For example, patients with MYC/BCL2 DHL should be considered for enrollment onto the US Cooperative Group Protocol A051701 (ClinicalTrials.gov Identifier: NCT03984448), which is a randomized comparison of first-line therapy with DA-EPOCH-R with or without venetoclax in this patient population.

FIG 1.

Suggested algorithm for risk stratification using currently available molecular testing methods. DLBCL, diffuse large B-cell lymphoma; FISH, fluorescence in situ hybridization; HGBL, high-grade B-cell lymphoma; IGH, immunoglobulin heavy locus; IGK, immunoglobulin kappa locus; IGL, immunoglobulin lambda locus; IHC, immunohistochemistry.

Moving beyond relying solely on DHL for risk stratification in aggressive B-cell lymphomas, studies that used genomic sequencing and GEP have identified additional molecular risk groups that could inform future practice.^12,15,18,19 Two recent studies examined the GEP of tumors from patients with and without DHL. Ennishi¹⁹ and colleagues investigated the prognostic value of a double-hit signature in patients with DLBCL or HGBL, which identified nearly twice as many patients who experienced inferior survival outcomes as FISH for DHL rearrangements did. Similarly, the MHG GEP described by Sha et al¹² predicted a worse prognosis when present, regardless of DHL status.

This focus on molecular risk stratification exists in the context of a more generalized investigation into subclassifying aggressive B-cell lymphomas by molecular features. Chapuy et al¹⁵ and Schmitz et al¹⁸ have used genomic sequencing and GEP to identify distinct subtypes of DBLCL classified by genetic events. This has revealed molecular pathways responsible for lymphomagenesis and informed prognosis after treatment with R-CHOP. Their work provides diagnostic and prognostic information that could inform the design of clinical trials of first-line therapy that test the rational targeting of these pathways with novel agents.

Even assuming that these assays, or assays with similar complexity, were proven to be prognostic or theragnostic in the context of clinical trials, there would be challenges to their widespread implementation. Turn-around time for testing must be rapid enough that results would routinely be available to clinicians when treatment decisions are being made. Because most patients with non-Hodgkin lymphoma are treated in the community setting,²⁰ restricting these panels to select academic medical centers or reference laboratories would remain a barrier to their widespread use. Therefore, the challenge for future investigations into the management of patients with newly diagnosed DLBCL or HGBL may be as much to validate an advanced molecular assay for routine clinical use as it is to report outcomes after receipt of experimental treatment regimens. It may be that an advanced molecular panel for clinical use could become more widely available in a way that is analogous to how commercial laboratories responded to the increased use of FISH to evaluate for the presence of DHL by offering aggressive B-cell lymphoma FISH panels.^21,22 But regardless of the mechanism, we hope that such panels would be made available in all clinical settings.

Although much effort has been made over the past 10 years to accurately identify and optimally treat patients with DHL, the first widely recognized molecularly defined high-risk aggressive B-cell lymphoma, we feel that this effort is just the beginning of a broader recognition of the importance of molecular risk stratification in these diseases. It is becoming increasingly clear that our widely used pathologic tests may not provide adequate information to optimally inform first-line treatment decisions for patients and that we are on the verge of having more robust molecular assays available for clinical use. If they are validated in clinical trials, performing genomic sequencing and GEP may ultimately become the standard practice for risk stratification and treatment selection for patients with DLBCL or HGBL. However, for such assays to meet their potential and improve outcomes for all patients with DLBCL or HGBL, we strongly feel that they must be available for use in all clinical settings and that their results must be reported within a clinically relevant timeframe.

AUTHOR CONTRIBUTIONS

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Molecular Risk Stratification in Aggressive B-Cell Lymphomas

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Daniel J. Landsburg

Consulting or Advisory Role: Celgene, Curis, MorphoSys

Speakers' Bureau: Seattle Genetics

Research Funding: Takeda (Inst), Triphase Accelerator (Inst), Curis (Inst)

No other potential conflicts of interest were reported.