Analysis Identifies Immunogenomic Determinants of Response and Survival in RCC | OncLive

Analysis Identifies Immunogenomic Determinants of Response and Survival in RCC

September 9, 2020

David A. Braun, MD, PhD, discusses findings from a pooled analysis of 3 clinical trials evaluating the potential impact of immunophenotypes, somatic mutations, and chromosomal alterations on response to PD-1 inhibition and recurrence in patients with renal cell carcinoma.

During the 2020 ASCO Virtual Scientific Program, David A. Braun, MD, PhD, presented findings from a pooled analysis of 3 clinical trials evaluating the potential impact of immunophenotypes, somatic mutations, and chromosomal alterations on response to PD-1 inhibition and recurrence in patients with renal cell carcinoma (RCC).

“We are moving the needle forward with this research, but there is still a huge amount of work to do,” said Braun. “When we look at large genomic studies, we examine things in bulk. We take sections of a tumor, essentially grind it up in a blender to mix everything, and extract DNA and RNA to try to gain insight. However, these systems are complex and incredibly heterogeneous. Understanding what the individual cell tumor and immune cell components are that drive response and resistance will be important, but it will also require moving beyond conventional genomic tools into more single-cell exploration.”

In an interview with OncLive, Braun, a physician at Dana-Farber Cancer Institute, and an instructor of medicine at Harvard Medical School, discussed the findings from the analysis, the role of next-generation sequencing (NGS) in RCC, and explained how the field of kidney cancer may evolve in the coming years.

OncLive: What were some of the goals you wanted to accomplish with this analysis?

Braun: We know that immunotherapy, and PD-1 blockade specifically, have been transformative in the management of [patients with] advanced RCC. However, at the same time, we know that many patients do not achieve durable clinical benefit from these drugs. We don’t know what the determinants of response or resistance mechanisms are in this disease.

Although this has been explored in other cancer types, kidney cancer is different. As opposed to melanoma or non–small cell lung cancer, RCC has a modest mutation burden. Also, in contrast with other cancer types, RCC has a robust CD8 T-cell infiltration. This standard “hot” versus “cold” paradigm of response/resistance typically doesn’t apply to RCC.

Knowing there is a clinical need to understand these factors and knowing we can’t necessarily extrapolate from other tumors types, we endeavored to investigate the immune or genomic determinants of response or resistance to PD-1 blockade.

In order to do that, we worked as a large, collaborative group with a [substantial] cohort [of patients]. We took clinically annotated tumor specimens from patients who were enrolled in a phase 1, phase 2, and phase 3 trial that evaluated the anti–PD-1 agent nivolumab (Opdivo). The phase 1 trial was CheckMate-009, the phase 2 trial was Checkmate-010, and the randomized phase 3 trial was CheckMate-025. Half the patients [in CheckMate-025] were treated with the mTOR inhibitor everolimus (Afinitor).

In total, we had 592 tumor specimens that we could perform a mix of whole exome sequencing on to get mutations and copy number variants, as well as RNA sequencing to get gene expression patterns. We also performed an immunopathological analysis to visualize how many CD8 T cells were infiltrating the tumor, where they are spatially, and how this may impact response and resistance.

What were the findings of this analysis?

First, we looked to see what has been established in other tumor types to determine whether that paradigm works for clear cell RCC. We examined several factors. There is a common belief that a high total mutation burden is a positive factor for PD-1 blockade, as there are more targets for the immune system and more neoantigens that could be recognized. Therefore, we looked at those factors, which are better established in other cancer types. To our surprise, we found that total mutation burden, neoantigen load, or total number of copy number alterations did not have any impact on response or survival to PD-1 blockade. That did not seem to fit the picture for kidney cancer.

Next, we examined whether there were individual genomic features that may be associated with response and resistance. We searched through the genome for all possible truncating mutations or loss-of-function mutations that were significantly recurring and had an impact on survival and response. In this pooled analysis, we found only 1: loss-of-function mutations in PBRM1 were associated with improved response and survival to PD-1 blockade. We didn’t see that with mTOR inhibition.

A critical point is that these trials were all done in the post-antiangiogenic setting. All patients had advanced RCC, but they were all previously treated with something like a VEGF TKI, something that blocks angiogenesis. Then, after they progressed, they enrolled on this trial. In the context of other work on PBRM1, it appears that this effect of PBRM1 on response and survival does seem to be in the specific post-antiangiogenic setting. There is some interesting biology that probably underlies that.

After genomic features, we evaluated immune features; again, we looked at this paradigm of “hot” versus “cold” tumors. Moving beyond that, in other tumor types like bladder cancer there is a third immune phenotype that has been described: immune excluded. [This means that] the T cells have marched up to the border of the tumor margin, but they are not able to penetrate inside the tumor center to have an effect. Up to 50% of bladder tumors have the immune-excluded phenotype which accounts for a major source of resistance.

As part of this trial, we evaluated 219 tumors. We found that the immune-excluded phenotype, although prevalent in other cancer types, is not a prominent feature in kidney cancer. Only about 5% of those tumors were immune excluded, and in reality, the bulk of these tumors were very highly T-cell infiltrating. Nearly 75% [of tumors] were highly infiltrated by CD8 T cells.

There is [the belief] that many CD8 T cells will lead to better response; however, in our case, having a high number of CD8 T cells infiltrating [the tumor had no effect] on response or survival compared with non-infiltrated tumors. This was perplexing to us as this seemed to be a standard [understanding] of immune checkpoint inhibitors and response/resistance, but it doesn’t seem to be at play in RCC.

Next, we wanted to determine whether any of these features integrate together. Is there an interplay between genomic and immune features where we could miss the effect if we just look at 1 at a time? We looked at whether there were any genomic features between these “hot,” infiltrated tumors and the “cold,” non-infiltrated tumors.

To our surprise, we found that the clinically-favored PBRM1 mutations were highly enriched in the non-infiltrated tumors. In contrast, we found that the infiltrated tumors were enriched in a number of copy number variants. One in particular, deletion 9p21.3, was highly enriched in infiltrated tumors [and] is associated with a much worse survival in response to PD-1 blockade.

As such, we see this potential interplay of immune and genomic features where, in theory, it looks like the tumors with many T cells are poised to respond and the ones with low T cells aren’t. However, the ones with low T cells are being pulled up by having more of these clinically-favorable PBRM1 mutations. The infiltrated tumors are being dragged down by having more than 1 of these deleterious 9p21.3 mutations.

We don’t see an effect when we look at 1 feature at a time, which is an important finding. This needs to be validated in other clinical data sets, but I hope this provides some conceptual framework for how to approach these response/resistance questions that, in a complex system, may not make sense. We have to learn how to integrate multiple features together to see how they affect response/resistance.

You mentioned that these data need to be validated in other clinical datasets. Are there other settings in which you would like to evaluate immunogenomic determinants of response and survival?

All these trials were done in the post-antiangiogenic setting. The CheckMate-025 trial led to the approval of nivolumab in the second-line setting of RCC. We know that the field has moved at an incredible pace over the past few years, which is wonderful for our patients. Now, PD-1 blockade is in the first-line setting alone and in combination with other therapies. The standard of care is PD-1 blockade combined with either another checkpoint inhibitor such as ipilimumab (Yervoy), or PD-1/PD-L1 blockade together with an antiangiogenic drug. Obviously, those are different clinical situations. There is the difference of second-line therapy compared with first-line therapy, and the difference of monotherapy versus combination therapy.

This is an interesting finding that hints at some underlying biology and provides some framework; however, ultimately, I want this to be very useful to our patients. For that to be the

case, we have to explore these findings in the first-line setting where those drugs are now standard of care.

Should NGS be conducted in all patients with RCC?

It is a great question, and it touches on the notion of how this affects the individual patient. I want to be very up front because when I see a patient with kidney cancer I like to provide them with more information rather than less. As such, I tend to favor genomic sequencing.

However, there are some big caveats to that. One is that the number of targetable mutations in kidney cancer is almost nonexistent. There are specific situations where a mutation may lend itself to enrollment in a clinical trial in a later line of therapy, but not much that guides up-front therapy. As much as these explorations have helped move the needle forward in our understanding of response/resistance, none of these are usable biomarkers today. We are not in a situation where if a patient has a genomic feature, I know they should get therapy “X,” versus another patient who has a different genomic feature. That is a difficult but important area to develop; we are definitely not there yet in terms of biomarkers.

I am always up front with my patients. I say, “The chance of this having an enormous benefit for you individually is low, maybe not zero, but it is low.” However, I hope that by incorporating NGS [into the treatment continuum for] more patients, we will gain an understanding so that the patient who walks into my office in 5 years will have a better comprehensive understanding of the genomic features of response/resistance that will enable us to improve outcomes.

What are your thoughts on the emerging potential for triplet regimens in RCC?

It is incredibly exciting to see how fast the field has evolved. I’ve seen patients who had retired from medicine or urology. When they hear they have a diagnosis of kidney cancer, they are devastated because they remember nothing was available. Now, I can honestly say to my patients that this field is evolving so quickly that there has been an approval every year for the past few years. We saw nivolumab plus ipilimumab in 2018, we saw 2 PD-1/VEGF combinations in 2019, and it looks like we are on course to see something similar hopefully in 2020.

Triplet therapy, if it is tolerated, has the potential to move the needle even further. However, some big questions remain. One is that we are talking about intensifying up-front therapy. However, a subset of patients may respond well to PD-1 blockade alone; that group will have durable, long-term survival with basic immunotherapy and should be spared the toxicity of adding more therapies to their regimen.

Even as we add these therapies together and move the curve up, the patients who experience truly durable clinical benefit are still the minority.

Understanding those mechanisms of resistance, whether primary or innate resistance, or acquired resistance months or years later, will be critical as it will hopefully allow us to rationally design new therapies and combinations to overcome those resistance mechanisms.

Also, while this is incredibly exciting, we are using many of our best weapons in our arsenal up front. We would be using all of our immune-based therapy in a triplet with our antiangiogenic agent. What is next? What do we do for the majority of patients who progress on antiangiogenic therapy and PD-1 blockade? To this end, several interesting [approaches] are being explored. For example, hypoxia-inducible factor-2α is an important target, and inhibiting it is an critical area of drug development. Other novel combinations are under exploration as well, but what the appropriate treatment post–PD-1 treatment will be an important question.

What are some of the next steps you hope to take with this research?

Having a good tumor microenvironment requires an interaction between a T cell that can recognize a target antigen on a tumor. We still know very little about that interaction. What are the effective tumor antigens in RCC? What T cells are capable of recognizing them? Those are open questions in this disease.

Hopefully, as we begin to answer them, that insight will pave the way to new therapies and antigens. Rather than giving broad immunotherapy, perhaps we can steer the immune system to target a specific tumor antigen. In the coming years, my hope is that this will serve as the first step in understanding the genomic and immune features of RCC, and that we will continue to work toward a more comprehensive understanding [in that regard].

Reference:

Braun DA, Hou Y, Bakouny Z, et al. Immunogenomic characterization of advanced clear cell renal cell carcinoma treated with PD-1 blockade. J Clin Oncol. 2020;38(suppl 15):5010. doi:10.1200/JCO.2020.38.15_suppl.5010


x