Skip to main content
Premium Trial:

Request an Annual Quote

Molecular Tumor Board Therapy Matching Predicts Immunotherapy Best Responders


NEW YORK – New results from a small but specific study of Molecular Tumor Board-guided cancer care have provided new evidence that taking a broad approach to precision immunotherapy may be the best way to identify those likely to respond.

The study, led by investigators from the University of California, San Diego and the Center for Personalized Cancer Therapy at Moores Cancer Center, analyzed a subset of patients in their clinical cohort who received immunotherapy either as a single agent, combined with additional immune-targeted drugs, or alongside a molecularly targeted therapy.

"We thought it was worth analyzing the specific immunotherapy subset, and the reason for that is, despite the talk about precision oncology, a lot of immunotherapy is given to everybody without thinking about the biomarkers," said Razelle Kurzrock, the study's senior author, now associate director of clinical research at the Medical College of Wisconsin's Cancer Center and Genomic Sciences and Precision Medicine Center. "And, yes, we're all really excited about immunotherapy," Kurzrock added. "I really think it's some of the best therapy that's ever been created for cancer, but it certainly doesn't work for everybody, and it does have side effects that are very significant to a subset of patients."

The study, published in NPJ Precision Oncology, intentionally included patients regardless of their tumor type. Pan-cancer indications for immunotherapy have generated much excitement in the field in recent years, but are also controversial.

Two studies published last year, for example, provided evidence that only some patients with high tumor mutational burden (TMB), as defined by the universal cutoff of 10 mutations per megabase, see improved survival with checkpoint inhibition compared to their TMB-low counterparts.

Kurzrock has been a vocal defender of TMB's pan-cancer relevance in the face of these challenges. She said she recognizes that all current immunotherapy biomarkers are imperfect predictors, but argued that this imperfection doesn't render them useless.

"I think that with immunotherapy, the lesson of genomics has been lost … which is that while biomarkers may not be perfect, giving therapy without biomarkers is not very useful. We would not consider at this point giving an EGFR inhibitor if a patient didn't have an EGFR alteration. But at the beginning, that's exactly what we did," she said.

Some immunotherapy drugs, like anti-PD-1 agents, have been effective enough that they could get approved for certain indications without a biomarker. That hasn't been true in a pan-cancer setting, where checkpoint inhibitors aren't effective in enough patients to be prescribed to all comers. Moreover, Kurzrock said, newer molecules like IDO inhibitors have faltered when studied without a biomarker.

Meanwhile, researchers have theorized from the early days of immunotherapy that doctors would likely need to consider networks of different biomarkers to best identify the most treatment-sensitive patients, something Kurzrock and her colleagues' new data seems to bear out.

"Ultimately, what our study suggests is that the use of immunotherapy agents is best guided by a comprehensive review of tumor characteristics, and a multifactorial approach to developing treatment plans for patients," they wrote.

The study, led by UCSD oncologists Bryan Louie and Shumei Kato, focused on 80 patients from the cancer center's molecular tumor board program who had received some form of immunotherapy.

Overall, 60 of the 80 total MTB immunotherapy patients had a high degree of matching between their tumor molecular characteristics and the therapy ultimately administered. Patients with high versus low matching score experienced significantly longer median progression-free survival (6.4 versus 3 months) as well as longer median overall survival (15.3 versus 4.7 months, respectively).

"The essence of what we found is that the most important thing to predict outcome was how well the patients were matched," Kurzrock said. In other words, regardless of individual biomarkers, the fact of having some biomarker or combination of markers and being treated accordingly confers better outcomes.

Although she thinks individual biomarkers are better than nothing, Kurzrock said it was clear that matching score was the only independently significant variable in the team's multivariate analysis.

If the matching score were discounted from the multivariate analysis, TMB should have remained significant and MSI would just miss, likely due to a too-small number of patients, "but once you put in the composite, nothing else is significant," Kurzrock said.

Whether patients received a single drug or multiple drugs was also not a significant predictor of outcome in the study.

Kurzrock said that when she presented earlier versions of this data in lectures, one of the first questions she'd get was whether she and her team were sure that the improved outcomes for patients with high matching scores wasn't just a feature of them getting more drugs.

"When we looked at number of drugs, it made no difference whatsoever," she said. "It wasn't just about giving more drugs. It was about giving more drugs that matched to the patients."

Kurzrock has been working since her move to Wisconsin to set up a new precision medicine and rare cancer center, largely modeled on the protocols she helped develop at UCSD, but hopefully even more expansive.

"We are learning about biomarkers at a very quick rate, and we know things that we didn't know when we opened up our protocols at UCSD in 2015. We can also now access way beyond genomics. We can access transcriptomics, proteomics, a lot of these things that were really not available five, seven years ago, so I think that we're really at the beginning of the journey — just touching the tip of the iceberg in that," she said.

In their study, Kurzrock and her coauthors cited a number of caveats to their findings. The study was not randomized, and as such could suffer from unknown confounders. It was also small and included a limited number and distribution of cancer types. Finally, matching score, although independently predictive, was not a perfect indicator of response. Some patients with higher matching scores didn't respond to immune checkpoint blockade.

"You'll notice that there's very little in the way of randomized trials in the precision medicine space," Kurzrock said, adding that this isn't because the precision oncology community doesn't believe in randomized trials. Rather, it's because "it is almost impossible to operationalize them."

"The drugs are not that freely available, so you can randomize people to the treatment group, but they may not get treatment because you cannot get the drugs for them," she added.

At the same time, the field is evolving so quickly that randomization can rapidly become an ethical problem, or at least a problem for patient recruitment. "Now that we know that RET inhibitors work well and they are FDA approved, can you really randomize a patient with a RET fusion to get unmatched chemotherapy? Personally, I can't do that. I don't feel it's the right thing to do," Kurzrock said.

"I think in this new day, despite the fact that randomized controlled trials are the gold standard, they're not the only standard, and so that's what we've been trying to do with things like matching score and so forth," she added.