A promising new biologic prognostic model in diffuse large B-cell lymphoma

In this issue of Blood, Perry et al show that immunohistochemical markers for the cell-of-origin and the microenvironment in diffuse large B-cell lymphoma delineate 2 groups with markedly different survival.¹ 

Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoma in all countries and its frequency varies from 35% to 60% of non-Hodgkin lymphomas. In the World Health Organization classification, DLBCL is clinically, biologically, and pathologically heterogeneous because it includes over 20 entities, subtypes, and variants.²  Nevertheless, the 2 most common (∼ 85%) subtypes of DLBCL are: (1) those with a germinal center B-cell (GCB) origin and (2) those with a non-GCB origin, most of which have activated B-cell (ABC) phenotype. The non-GCB subtype has a more aggressive clinical course than the GCB subtype.^3,4 

For optimal patient management, an accurate diagnosis is mandatory. For the next 5 or more years, there is agreement among senior academic lymphoma pathologists that immunohistochemistry-based markers will continue to be most valuable in making an accurate diagnosis. Immunostaining is widely available to most pathologists in the developed world, as well as in large hospitals in big cities in the developing world.

Several papers published in the past decade on the clinical utility of separating DLBCL of GCB subtype from the non-GCB/ABC subtype have yielded conflicting survival results, especially since the addition of rituximab to standard DLBCL therapy.^5,6  However, results from the Nebraska Lymphoma Study Group as well as the Lymphoma/Leukemia Molecular Profiling Project have been consistent.^3,4,6 

Recently, significant progress has been made in the use of immunohistochemical markers to predict survival of DLBCL patients, as shown in the compelling and highly-significant results in the article by Perry et al.¹  Over the past decade, the authors of this paper have designed, conducted, and published many original papers on DLBCL that identify the most important diagnostic, prognostic and biologic markers (immunohistochemistry, gene expression profiling, microRNA profiling, and so on) for use in the daily diagnostic and clinical practice.^3,4,6-9 

The current study by Perry et al is carefully designed and conducted by a seasoned team of highly experienced lymphoma investigators from North America and Western Europe who have incorporated their most important published information on DLBCL.^4,7-9  Based on the results presented, the authors conclude that their new biologic prognostic model (BPM) delineates 2 clinically distinct groups of patients: one with a low biologic score (0-1) and good survival, and the other with a high score (2-3) and inferior survival.¹  This new BPM could be used in the future with the International Prognostic Index (IPI) to stratify patients for novel or risk-adapted therapies. The conclusions reached in this paper are based on the immunohistochemical stains for the cell-of-origin of the malignant cells (GCET1, MUM1, FOXP1, CD10, and BCL6), as well as admixed–stromal cells and benign histiocytes expressing secreted protein acidic and rich in cysteine (SPARC), and the microvessel density (MVD) determined by analyzing digital images of the CD31-stained slides. Using the BPM, 1 subset of DLBCL had a short overall and event-free survival in multivariate analysis after adjusting for the IPI, as evidenced by the presence of the non-GCB phenotype, low SPARC expression of < 5%, and a high MVD score.¹  Importantly, their BPM based on immunohistochemistry is the first prognostic model in the rituximab era to combine the cell-of-origin of the malignant cells and stromal signatures of the benign component into one, integrated, numerical prognostic score that can be readily used in clinical practice.

Because the BPM model of Perry et al is newly introduced, it has not been tested by anyone else. Therefore, the authors emphasize that this BPM will need to be validated in future prospective clinical studies.¹  Because DLBCL is the most common lymphoma around the world and there is no similar model available, I also urge investigators to test this new BPM. Even though immunohistochemistry is widely available and easy to use, its reproducibility among different institutions has been variable due to differences in tissue processing, antibody clones used, and interobserver variability. Efforts should be made to standardize various aspects of immunohistochemistry and also use image analysis for quantification to promote better reproducibility so that the results from different studies can be compared in the future.

Furthermore, validation of this BPM model in formal clinical trials would be ideal. Toward this end, for new studies, participation of one of the coauthors of this study in the design and validation of this model could also be valuable for improving interobserver agreement, because reviewing of H&E-stained slides and extracting relevant histologic features and interpreting immunostains is subjective, and integrating all of this information with clinical and other information requires a high level of judgment that is surely dependent on the level of individual training, subspecialization, skills, and experience and, thus, making diagnosis is an art.^10,11  Beneath this art, there lies a solid foundation of scientific knowledge, but this new scientific information is highly complex, ever increasing, rapidly evolving, often uncertain, and continuously changing. This underlying complexity in each individual tumor accounts for at least, in part, the variability in the current and other predictive models. Future investigations may identify additional important markers that could be added to this model, not only to improve outcome prediction but also to guide target directed therapy.