Skip to content

GenomeWeb – Tumor Mutation Burden Predicts Immunotherapy Benefit Across Cancers, Though Cutoffs Differ

GenomeWeb – Tumor Mutation Burden Predicts Immunotherapy Benefit Across Cancers, Though Cutoffs Differ

Cancer patients with high tumor mutational burden who were treated with immunotherapy tended to live longer than those with fewer mutations, researchers at Memorial Sloan Kettering Cancer Center reported in Nature Genetics this week.

 

The study — involving cancer patients with a variety of tumor types, including almost 1,700 who were treated with a range of anti-CTLA4 and anti-PD-1/PD-L1 drugs and almost 5,400 patients who weren’t — is the largest evaluation of TMB to date. All patients received testing with the MSK-IMPACT 468-gene panel. The 20 percent of patients with the highest TMB in each type of cancer had improved survival.

However, although for most tumor types patients with higher TMB were less likely to die than those with lower scores after immunotherapy treatment, the cutoff defining high TMB differed between histologies. The paper is a step forward in the field’s understanding of TMB (defined as the total number of somatic mutations per megabase), according to Luc Morris, associate director of MSKCC’s immunogenomics and precision oncology platform and the senior author of the study, because it is the largest study to demonstrate the potential predictive value of the biomarker in a pan-cancer fashion. At the same time, the paper shows that how TMB-high and -low patients are defined is not universal across histologies, he said.

 

While the results need further exploration in additional studies, the MSK study is worth paying attention to, Morris said, because a number of labs are reporting TMB status using a universal cutoff in reports from next-generation sequencing cancer panels. “Our data suggests that cutoff is likely to be very different from one cancer to another,” he said. 

 

The use of immunotherapies has taken off in recent years because some cancer patients have experienced astonishing, durable responses. Unfortunately, such sustained benefit has only been experienced by a minority of cancer patients. Since these immunotherapies are also costly and can cause significant adverse events, researchers have been trying to identify predictive markers that will help identify which patients are likely to benefit, and TMB has generated significant interest.

 

It has been easy for the field of oncology to accept TMB as a predictive marker, Morris noted, because it’s based on the biologically plausible premise that a high number of somatic mutations in a tumor would correlate with the level of neoantigens in that tumor and increase the immune system’s ability to recognize it as foreign. But that acceptance was likely premature, in his view, because to date the association between TMB levels and immunotherapy response has been demonstrated largely in small studies involving non-small cell lung cancer, melanoma, and bladder cancer patients.

 

“Up until our paper, there had been very limited data that TMB is relevant pan-cancer, or that TMB would be relevant outside of a clinical trial setting,” he said.

 

In the study, the majority of patients had advanced, metastatic disease. Of these, 1,256 received anti-PD-1/PD-L1 drugs (Tecentriq (atezolizumab), Bavencio (avelumab), Imfinzi (durvalumab), Opdivo (nivolumab), or Keytruda (pembrolizumab)), 146 received an anti-CTLA-4 drug (Yervoy (ipilimumab) or tremelimumab), and 260 received a combination of anti-PD-1/PD-L1 and anti-CTLA-4 drugs. Researchers tracked overall survival from the date of treatment to death or the most recent follow-up, with a median follow-up of 19 months.

 

Morris and colleagues wrote that because tumor mutational load varied in range and median in each histology, a universal cutoff for “high TMB” would be skewed in favor of cancers that tend to accumulate more mutations. To account for this, they defined TMB cutoffs by percentile for the various cancer types and showed that the range for high-TMB status varied from the top 10 percent to 50 percent of scores, but that there was a significant association between high-TMB status and overall survival.

 

Longer overall survival was associated with higher TMB when researchers considered the top 20 percent of patients in each histology, though the TMB cutoff delineating the top 20 percent of patients varied by histology. The researchers also noted that higher TMB was similarly linked to higher rates of objective response to immunotherapy or progression-free survival in patients with NSCLC, melanoma, esophagogastric cancer, head and neck cancer, and renal cell cancer.

 

The association between high TMB and improved survival didn’t reach statistical significance in some tumor types with small sample sizes, such as estrogen receptor-positive breast cancer and cancer of unknown primary. Also, in estrogen receptor-negative breast cancer and gliomas, higher mutational burden was associated with worse survival.

 

However, the overall trend suggests that higher TMB is likely to be predictive of response to immunotherapy in most histologies, the researchers wrote. This association remained when they excluded data from NSCLC and melanoma patients to ensure that the finding wasn’t being driven by these cohorts.

 

Morris and colleagues also looked at TMB in approximately 5,400 patients who didn’t receive immunotherapy to assess whether those with higher mutational load generally had a better prognosis. They found no association between higher TMB and improved overall survival across the whole group and in histology-specific subgroups.

 

“The paper is important but provides preliminary information in terms of the actual clinical application of TMB,” noted Morris. He and his colleagues wrote that while this work demonstrates that TMB cutoffs associated with longer survival differ by histology, prospective studies are needed to establish the specific cutoffs in each tumor type.

 

A number of labs report TMB within NGS tumor panels, and each presents this information differently. After Foundation Medicine garnered FDA approval for its FoundationOne CDx, it began reporting TMB quantitatively on the first page of its report (the FDA-approved portion), without indicating whether those numbers are associated with higher or lower likelihood of immunotherapy benefit. Foundation said it is currently transitioning to reporting a TMB range in the following pages of the report that will indicate whether a patient has high, intermediate, or low TMB status, based on more discrete cutoffs for cancer-specific subtypes. It also plans to indicate drugs and clinical trials that might be appropriate, based on this information.

 

With FDA approval of FoundationOne CDx, the firm also had to shift how it presented results in the report so that markers the agency deemed to be a companion diagnostic are distinguished from those that aren’t, such as TMB. In the non-FDA version of FoundationOne, the company previously reported TMB quantitatively (as mutations per megabase) and qualitatively in terms of high (20 or more mutations per megabase), intermediate (between six and 19 mutations per megabase), and low (five or fewer mutations per megabase).

 

Foundation is working on garnering an FDA-approved TMB companion diagnostics indication, however. Last year, Bristol-Myers Squibb reported data from the CheckMate-227 study, which demonstrated that NSCLC patients with high TMB had significantly longer progression-free survival on the combination of Opdivo (nivolumab) and low-dose Yervoy (ipilimumab) than on chemotherapy, while among patients with lower TMB levels, there was no difference in PFS between the two regimens. However, BMS defined high TMB status as 10 or more mutations per megabase in this study.

 

“We have always understood that a singular TMB cutoff will likely identify differential levels of response to immune-oncology therapy according to each indication,” said David Fabrizio, VP of product development and cancer immunotherapy leader at Foundation Medicine. “Some factors which may influence this could include the varying levels of immune suppression associated with each disease type, which could influence the need to pursue higher cutoffs in some diseases versus others.”

 

Prospective studies will help define TMB cutoffs that are associated with immune-oncology treatments in specific tumor types, Fabrizio noted, but he said he still sees value for a universal cutoff in certain settings. “We anticipate that certain immune-oncology indications will likely be pursued using individual TMB cutoffs, especially in settings that include early lines of therapy,” he said. “However, the need to identify a relevant pan-tumor TMB cutoff still exists for very late lines of therapy, and the decision of how to define that cutoff will need to evaluate the performance of TMB more broadly across multiple disease types.”

 

Meanwhile, the MSK-IMPACT test reports TMB numerically, indicating where a patient’s score falls in terms of percentile and median for a particular histology. However, the report doesn’t say what those numbers mean in terms of a patient’s chance of deriving benefit from immunotherapy. “We now need to collect data prospectively to define the specific numbers that mean a patient is more likely to benefit,” Morris said.

 

However, the different ways in which labs are presenting TMB may confuse oncologists. Recognizing this, the cancer advocacy organization Friends of Cancer Research has convened a workgroup, involving cancer institutions, diagnostics providers, drug makers, and the FDA, to standardize the way TMB is calculated and reported. Foundation and MSK are both involved.

 

In the first phase of this effort, stakeholders used TCGA data to look at the way labs are calculating TMB using NGS panels and exome sequencing and found a strong correlation between the two. “We also found, similar to what the latest clinical data are pointing to, that the association of the panel-based TMB and exome-determined TMB differs by cancer type,” said FOCR CEO Jeff Allen.

 

FOCR plans to publish these findings soon. “We did see that there was some variation in TMB quantified across the different panels,” he said. “That supports the concept of trying to understand how each panel relates back to a common reference standard.”

 

This leads into the second phase of the project, where the focus is on creating a universal reference standard using exome sequencing, and identifying variability among TMB scores from NGS panels after they are aligned to the reference standard. The NCI will develop the exome sequencing-based reference standard using human cell lines and 15 labs are partaking in this phase.

 

In the third phase, the work group will come up with standards for applying TMB to clinical practice. This will be a validation stage for the harmonization algorithms identified in the first two phases.

 

“What we hope to be able to contribute at the end of our pilot project is a better understanding of the extent to which variability exists between different platforms, and ultimately strategies around how different tests for TMB could harmonize toward each other and how they relate to each other going back to a reference standard,” Allen said. The aim is to wrap up the harmonization project in the first half of the year, he added, since TMB is used more often by doctors as more clinical data is emerging and labs are reporting it.

 

As an offshoot from these harmonization efforts, FOCR is also discussing with stakeholders the feasibility of developing a tumor agnostic cut-off for TMB, which wouldn’t supersede data emerging in specific drug and cancer settings, but could still be a useful resource in rare cancer indications where trials aren’t being conducted or are difficult to do. “Our goal for the add on discussions … is to try and reach agreement between several major biopharma companies on developing a path forward to further define a cut off for tumor agnostic TMB use,” Allen said. “We believe that defining a path forward with numerous stakeholders will help to ensure that data is being collected similarly across different trials and drugs in order make the outcomes as useful and applicable as possible.”

 

While the evidence supporting the use of TMB is evolving, oncologists are using this biomarker in the same way they use PD-L1 expression, and are not using it to definitively give or deny immunotherapy. “If you have a patient for whom you think immunotherapy is the next best treatment option and their PD-L1 staining is low, you’re not going to disqualify them from immunotherapy based on that finding,” Morris reflected. “You’re going to consider that in the larger context of your decision making, knowing that the probability of responding to immunotherapy is a bit lower than if that number had been higher.”

 

“TMB is very similar in that it’s not going to be the make-or-break factor for what your patient’s next therapy is,” he said, “but it’s going to be one of many factors that you’re going to consider.”

 

In addition to TMB and PD-L1, there is a growing body of data around the predictive capabilities of T-cell infiltration levels, T-cell clonality, gene expression signatures, and blood markers for informing immunotherapy treatment. A number of studies have suggested that predictive models combining PD-L1, TMB, tumor, and immune markers could improve response predictions.

 

For example, a recent evaluation in the Journal of Clinical Oncology involving patients with 20 different kinds of cancer in the KEYNOTE-028 trial showed that patients were most likely to respond to Keytruda (pembrolizumab) if they had both high TMB and high levels of inflammation, measured by either gene expression profiling or PD-L1 status.

 

“Early on, there was some thought that TMB might be an independent indicator [of immunotherapy benefit],” Allen said. “More recent data, over the last year or two, have shown that TMB is probably one part of a larger story.”

 

MSK is also involved in such research. “We are actively involved in refining these models, using additional information you can get from the DNA and RNA levels that allow you to capture both the genome and tumor microenvironment,” Morris said.

 

https://www.genomeweb.com/sequencing/tumor-mutation-burden-predicts-imm…