Diffuse Large B-Cell Lymphoma Cell of Origin Determination from Formalin-Fixed Paraffin-Embedded Tissues

Diffuse large B-cell lymphoma (DLBCL) consists of two distinct subtypes, Germinal Center B-cell (GCB) and Activated B-Cell (ABC), each with distinct prognostic and biological profiles. With the advent of new targeted therapeutic approaches for DLBCL treatment, molecular sub-typing of the patient's tumor has been proposed as an important step in therapeutic selection and recommended by 2016 WHO classification guideline.

Gene expression-profiling (GEP) based subtyping is considered the gold-standard method for DLBCL molecular sub-typing, but this method was established using fresh frozen / non-fixed tissues on microarray platforms. Since fresh tissue is not routinely collected during patient diagnosis, it severely limits the ability to utilize this assay in routine clinical practice.

Multiple immunohistochemical (IHC) approaches have been developed to approximate the GEP methods, but these techniques generally suffer from lower than desired agreement rates with GEP classifications, and require staining of 4 to 8 sections of limited biopsy material. Interpretation of the slides can also be variable, leading to low inter-observer reproducibility.

We have developed a 12 gene GEP-based DLBCL cell-of-origin (COO) assay using the HTG EdgeSeq System specifically designed to use a minimal amount of FFPE tissue. To build this system, we profiled a total of 107 samples previously subtyped using the HG-U133 Plus 2.0 Affymetrix microarrray and algorithm (Visco C et al Leukemia 2013); this algorithm was then validated in an additional 58 samples. The methodology we have developed produces 92% concordance with the microarray-based approach.

Briefly, 107 DLBCL cases, of which 58 were previously sub-typed as ABC and 49 cases as GCB, were used as the training cohort. After classifier training and cross-validation, a separate cohort of 58 cases were used to verify the performance of the assay/classification system. Approximately 5 mm² of 5 µm thick FFPE tissue was used to generate the data set for each of the cases.

In addition to the DLBCL COO classification, the assay also contains additional genes including potential drug targets, T-cell, B-cell, and macrophage biomarkers, and housekeeping/normalization genes. These markers could be used to further understand the nature of the tumor and potentially help identify the characteristics of atypical tumors and immune infiltrates in the microenvironment.