Skip to main content

Precision Medicine Tool Sorts Diffuse Large B-Cell Lymphomas into Seven Unique Subtypes

NEW YORK – Researchers have developed an algorithmic tool to classify diffuse B-cell lymphoma (DLBCL) into seven distinct genetic subtypes, according to a study published today in the journal Cancer Cell.

Study authors said the tool, called LymphGen, could offer insights into individual responses to targeted therapies. They expressed hope that the tool would prove valuable in future precision medicine trials, and have made it available on a publicly accessible server.

The LymphGen algorithm calculates the probability that a tumor belongs to a specific DLBCL subtype based on the genetic features that the researchers, led by first author George Wright of the National Cancer Institute, have associated with these different groupings. The researchers defined "genetic subtype" as a group of tumors that is enriched for aberrations in a set of subtype predictor genes.

"These subtype predictor genes are identified by considering each possible combination of genetic aberrations (i.e., mutations, copy-number alterations, or fusions) as a separate genetic 'feature' and scoring a tumor as positive for a feature if one or more of its genetic aberrations is observed," Wright and colleagues explained in the paper. "LymphGen uses the presence or absence of each subtype predictor feature to provide a probability that a tumor belongs to the subtype."

Before they could implement the LymphGen algorithm tool in DLBCL, however, researchers had to define the genetic subtypes into which of the cancers could be classified. They began with four subtypes that were previously defined by the GenClass algorithm: MCD, which includes MYD88L265P and CD79B mutations; BN2, which includes BCl6 translocations and NOTCH2 mutations; N1, which includes NOTCH1 mutations; and EZB, which includes EZH2 mutations and BCL2 translocations.

For the remaining DLBCL cases in their NCI cohort that did not fall into one of these genetic "seed" subtypes, the researchers identified two additional subtypes: A53, based on the aneuploidy with TP53 inactivation, and ST2, based on the presence of SGK1 and TET2 mutations. They also further divided the previously defined EZB group into two subtypes, EZB-MYC-positive and EZB-MYC-negative, due to their findings that several key abnormalities were only present in some of the genetically distinct lymphomas.

Having defined the seven genetic subtypes — MCD, BN2, N1, A53, ST2, EZB-MYC-positive, and EZB-MYC-negative — the researchers applied the LymphGen tool to DLBCL samples from the 574-patient NCI cohort. Altogether, the tool was able to classify more than 63 percent of the cases into one of the subtypes. This percentage, the study authors pointed out, was substantially higher than previous tools had been able to classify.

Among the remaining cases for which the LymphGen tool could not assign one of these subtypes, the researchers cited three potential reasons: one, that the tumors had a few features from one or more subtype, but not enough to be classified; two, that the tumors had unique features that were not recurrent in DLBCL; and three, that tumors had very few genetic features at all.

To determine whether these results were reproducible, the researchers ran the LymphGen tool on tumors from two additional validation cohorts with more than 300 cases each. Importantly, because each of the additional cohorts presented different data types, LymphGen's design was further refined to be able to calculate subtypes using various combinations of mutational data, including from whole-exome or gene panel re-sequencing, copy number data, including regional or whole genome data, and rearrangement data for BCL2 and BCL6.

After these validation studies showed LymphGen to have "robust performance," the researchers implemented the tool for general research use.

As a part of their study, the researchers also assessed each DLBCL subtype group for five-year overall survival and sensitivity to precision drugs that go after a specific target. This work is ongoing, however, and the researchers hope that by making the tool available to researchers generally, it will advance the field of precision medicine in DLBCL.

"Our combined genetic, phenotypic, functional, and clinical data demonstrate that the DLBCL genetic subtypes differ strikingly in their response to standard immunochemotherapy and may also respond differentially to targeted therapies," wrote the authors, later adding, "we feel that the LymphGen algorithm will be a useful tool in DLBCL clinical trials that extend the utility of gene expression-based assays."

The researchers also speculated that the LymphGen classification will find initial utility in retrospectively analyzing clinical trials.