NEW YORK (GenomeWeb) – More than a decade after the first studies demonstrating that diffuse large B-cell lymphomas are actually made up of two molecularly distinct subtypes with different prognoses and drug sensitivities, the incorporation of molecular testing into the treatment of DLCBL has remained largely theoretical and experimental.
But this is poised to change, according to clinicians at this year's annual meeting of the American Society of Clinical Oncology, who described a field at the cusp of significant shifts as more practical and accurate molecular diagnostics approach the clinic alongside novel targeted drug regimens.
Speaking in a session at the meeting earlier this month, the British Columbia Cancer Agency's David Scott explained the long road the field has taken in translating the knowledge of DLBCL's molecular heterogeneity into something that is just now becoming clinically actionable.
After the first evidence that DLBCL is not a single disease, but a conglomeration of at least two main subtypes — germinal center B-cell-like (GCB) and activated B-cell-like (ABC) — researchers quickly developed methods to distinguish these types using gene expression microarrays.
This gold standard for distinguishing ABC from GCB DLBCL, the two "cell-of-origin" subtypes of the disease, developed by National Cancer Institute researchers, uses microarrays to determine the expression of a handful of different genes and collapses this data into a linear prediction score.
Based on this score, cancers with over a 90 percent chance of being an ABC are designated ABC, and those with less than ten percent chance of being an ABC are called a GCB. The approach also produces an "unclassified" group — roughly 10 to 15 percent — in which the tumor cannot be assigned to either GCB or ABC with sufficient confidence.
"Using the gold standard has been very useful… and has provided a scaffold on which we have hung and arranged our knowledge of the pathogenesis of DLBCL," Scott said.
For example, over the last decade research has solidified the notion that the different cell-of-origin subtypes have significantly different prognoses when treated with standard-of-care therapy, a chemo combination called R-CHOP. ABC cancers are distinctively more aggressive and more likely to relapse after R-CHOP than GCB cancers.
Even more recently, Scott explained, it's become clear that ABC or GCB status can also be a predictive tool.
Greg Nowakowski, a lymphoma physician at the Mayo Clinic Rochester, also spoke during the ASCO session, highlighting the potential of new targeted therapies to shift a one-size-fits-all treatment modality that has lingered even more than a decade after the first evidence of the molecular heterogeneity of DLBCL.
"None of this is new," he said. "The technology and classification has actually been around for 15 years. Yet, in 2015, if you approach a patient in the clinic, the standard of care is still R-CHOP for the majority of patients."
"But things have changed. Most importantly, we have a number of agents now that target different subtypes, particularly the bad ABC subtype." Nowakowski explained.
The advancement of such drugs has in turn spurred the need for and the development of more accurate and reproducible molecular diagnostics to distinguish ABC from GCB lymphomas.
"Cell of origin has moved from being a prognostic biomarker in the R-CHOP era to being a predictive biomarker," Scott said. "In this setting, we require accurate and reproducible assays ... because the gold standard was never going to work in clinical practice."
The main problem with the initial gold standard array approach, Scott said, is that it requires fresh frozen tissue, which is impractical for routine use in clinical trials let alone in clinical practice.
To open up testing to FFPE samples, teams first began using immunohistochemistry methods that matched up as closely as possible with the outcome of the gold standard gene-expression microarray approach.
However, Scott said, while IHC looked promising in initial studies, the misclassification rates compared to gold standard array analysis have been as high as 14 percent in further validation, suggesting that the technology may be too variable to support clinical decision making.
In light of this, alternative gene expression-based methods, either using novel array approaches that work with FFPE or taking advantage of alternative technologies like Nanostring's nCounter, have also emerged.
Scott highlighted two assays in particular that are being used to stratify patients and guide therapy in Phase III trials of R-CHOP combined with novel targeted drugs.
In one, researchers have harnessed Illumina's DASL platform and created a method using 20 genes to distinguish the cell-of-origin subtypes. This test is being used in a trial called REMoDL-B, to provide equal proportions of ABC and GCB patients for randomization with either R-CHOP alone, or R-CHOP plus the drug bortezomib (Takeda's Velcade).
A second Phase III trial is using a different test, initially developed by the Leukemia Lymphoma Molecular Profiling Project (of which Scott is a member) which relies on Nanostring's nCounter This trial, ROBUST, is assessing the efficacy of adding the Celgene drug lenalidomide (Revlimid) to R-CHOP in patients with ABC DLBCL.
Nanostring announced last year its agreement with Celgene to develop a companion diagnostic test for Revlimid and its intention to seek regulatory approval of the test following completion of the Revlimid DLBCL trial.
According to Scott, data from testing of the Nanostring assay, called Lymph2Cx, has shown a low failure rate, high concordance across different laboratories, and substantial separation of outcomes following R-CHOP between the ABC and GCB groups.
Scott reported at the ASCO meeting that the BC Cancer Agency currently subtypes patients with DLBCL using IHC-based assays, as part of standard practice. Now, as current trials of ABC-specific drugs like Revlimid — and hopefully GCB-specific drugs as well — move forward, it's likely the newer gene expression methods being used in such trials will usurp IHC.
However, Scott said, several issues must be resolved before such tests can be widely distributed and applied to patients in practice. First, he highlighted a need for accuracy testing by external and independent groups. In addition, he said that a trend away from excisional biopsies toward core needle biopsies raises a question about how the accuracy of various assays will hold up using these smaller samples.
Finally, he said, the community as a whole will have to decide if it will embrace centralized, lab-based testing, or if it wants US Food and Drug Administration-approved test kits.
In an email to GenomeWeb this week, Scott said that he is not aware of any centers currently offering Nanostring or Illumina DASL array-based gene expression testing for DLBCL. However, he said many are planning to adopt the Nanostring approach in the near future.
A number of other technologies have also been harnessed for DLBCL subtyping. Recently, a team of French researchers published an alternative method, which relies on reverse transcriptase multiplex ligation-dependent probe amplification (RT-MLPA), capillary electrophoresis, and freely available software.
Cleveland Clinic researchers have also developed an RT-PCR method using Primera Dx's ICEPlex system, now Qiagens' Modaplex. The group reported in the British Journal of Hematology last year that the Modaplex-based assay "has the potential to provide rapid and accurate subclassification for DLBCL patients for prognostic implication as well as clinical trial patient selection in a [CLIA]-certified laboratory environment."
Cleveland Clinic physician and molecular pathologist James Cook told GenomeWeb that Modaplex-based testing for differentiation between GCB and ABC DLBCL subtypes is currently available through the clinic's reference lab as a clinical test.