Identifying Predictive Biomarkers in Non-Hodgkin Lymphoma

The focus in non-Hodgkin lymphoma (NHL) now needs to shift to predictive biomarkers, according to Randy Gascoyne, MD.

Randy Gascoyne, MD

The focus in non-Hodgkin lymphoma (NHL) now needs to shift to predictive biomarkers, according to Randy Gascoyne, MD, research director, Centre for Lymphoid Cancers at the BC Cancer Agency.

Gascoyne is investigating predictive biomarkers in follicular and diffuse large B-cell lymphoma (DLBCL). In an interview withTargeted Oncology, Gascoyne explains how predictive biomarkers can help oncologists select the right therapies for their patients and why they are the critical next step to improving outcomes in NHL.

Targeted Oncology:What do researchers need to do to identify predictive biomarkers in NHL?

Dr Gascoyne:You need smart people who are very creative. You have to do it in the context of uniform treatment and you have to look for markers, usually molecular or genetic markers, which actually inform an outcome. If they are also related to a pathway perturbation that could possibly be targeted, then you are off to the races. Now you have a biomarker that not only predicts outcome, but also could predict or inform a specific therapy.

Which studies are looking at predictive biomarkers?

We are conducting a study on biomarkers in follicular lymphoma right now that has been submitted but not yet published. In this study, we took the clinical variables that used the Follicular Lymphoma International Prognostic Index (FLIPI), a set of clinical variables used to inform outcomes. It works quite well in almost all treatment regimens for which it has been tested.

However, it does not help one to plan a therapy; it just provides an expectation of survival or outcome. We actually sequenced at a pretty good depth of 74 genes. We did this in a discovery cohort and then validated it in an independent cohort. The study showed that approximately 25% to 28% of patients were identified with a markedly inferior survival that was not predicted by the clinical FLIPI. Suddenly, we have a tool that could be applied using routinely available paraffin blocks. The original diagnostic blocks were what was used to extract DNA. When we did that, we distilled down to seven genes that when combined with the FLIPI, showed a markedly inferior subset.

This immediately provides a tool that people can use going forward in the context of planning clinical trials, and could even be used to enroll. You would be able to identify those patients who are destined to do very poorly very quickly. That would allow us to do trials where we are not waiting 10 to 15 years to see differences in outcome. They would occur quite quickly.

That is an example of a tool in the new molecular era that could actually penetrate clinical practice. It sounds complicated (in the original work we actually sequenced a lot of genes) but in the end, we distilled down to seven. Some were associated with favorable outcome, and some were associated with inferior outcome. Together, in a model with the FLIPI, they were highly predictive of survival.

Could a similar model be applied to multiple malignancies?

That model is purely restricted to follicular lymphoma, but in my own research I am also looking at a similar thing for DLBCL. While we know a fair bit about the mutational landscape with that disease, we don’t know the relevant genes that impact the clinical outcome. We only know about two of them. One we have known about for 20 years; the other one was discovered in 2013. Beyond that, we know nothing about the recurrent mutations and their impact on outcome.

Do you see the discovery of more biomarkers for NHL on the horizon?

I think it will be a significant area of growth. However, we have to move away from these endless publications, which I can admit I have contributed to, and all of these publications talk about prognostic biomarkers. There is some value in that because if you hit on the right marker you can learn some insights about the biology.

But at the end of the day, being able to show that a marker is prognostic is not hugely relevant. We have to move away from the idea of a prognostic marker into a marker that is predictive. By that, I mean a marker that informs the choice of treatment or an alternative therapy in either the upfront or relapse setting. It is already happening in follicular lymphoma and, if we are successful, also in DLBCL. You could make an argument that we should be able to do that using these recurrent mutations and other genetic alterations.

CD30 is another example of a predictive marker, because it is being used in the setting of anaplastic large-cell lymphoma (ALCL) and other T-cell lymphomas to inform a treatment choice. By definition it becomes a predictive marker.

Can you talk a little bit more about the potential of CD30?

In ALCL, this biomarker is universally and strongly expressed. It is used as a diagnostic marker, but it also now informs a therapy that specifically targets that molecule. That has an impact on both ALK-positive and ALK-negative, in particular the latter because the outcomes there are not as good.

What we also know is that you look across the spectrum of the other peripheral T-cell lymphoma, particularly peripheral T-cell lymphoma and angioimmunoblastic T-cell lymphoma, and a number of diseases where that antigen is expressed in a subset. That begs the question: are those patients who express logical candidates appropriate for that targeted therapy? One could make the argument that they are. Some of those studies are actually underway.