NEW ORLEANS – Researchers at Memorial Sloan Kettering are recruiting colorectal cancer patients with immunologically "cold" tumors into a clinical trial in the hopes that they can be made "hot" and responsive to immune checkpoint inhibitors by giving them temozolomide and cisplatin chemotherapy.
At the American Association for Cancer Research's annual meeting on Tuesday, Luis Diaz, head of the solid tumor oncology division at MSK, described a study that his team is conducting in collaboration with Bristol Myers Squibb to explore whether advanced colorectal cancer patients with mismatch repair-proficient tumors will respond to the PD-1 inhibitor Opdivo (nivolumab) when given in combination with temozolomide and cisplatin.
Researchers aim to enroll 35 colorectal cancer patients who are microsatellite stable or mismatch repair-proficient and have BRAF and POLE wild-type tumors. Patients must provide blood and tumor tissue samples for further biomarker research. Investigators will track patients' response to the Opdivo-temozolomide-cisplatin combination for one year.
"Knowing that mismatch repair-deficient tumors are incredibly sensitive to checkpoint blockade and understanding some of the fundamental, albeit rudimentary, features of these tumor types, [such as] high tumor mutation burden … could we pharmacologically engineer immunity in tumors that are not immunogenic?" Diaz proposed during a presentation at the meeting. "In other words, [could we] turn … a cold tumor into a hot tumor, or, in this case, turn a mismatch proficient tumor into a mismatch deficient tumor?"
The accumulation of tumor mutations is a key feature of cancer development and progression, and tumor mutation burden has become a key measure for predicting whether a patient will respond to immune checkpoint inhibitors. Opdivo was the first anti-PD-1 inhibitor approved in the US in 2014 for advanced melanoma and then approved the following year for non-small cell lung cancer. The first approvals were in these two tumor types, Diaz said, because they are mutagen-associated tumors that responded particularly well to checkpoint blockade.
In 2017, the US Food and Drug Administration granted accelerated approval to Merck's PD-1 inhibitor Keytruda (pembrolizumab) for any refractory, advanced solid tumor that was mismatch repair-deficient or had high microsatellite instability. The biological rationale for this tissue-agnostic approval centered on the idea that because mismatch repair-deficient or microsatellite unstable cancers tend to accumulate a large number of mutations, there's a better chance that the immune system will recognize these mutations as foreign antigens and a PD-1 inhibitor can further unleash this anti-tumor response.
Tumors that are mismatch repair-deficient cannot repair mutations that occur during replication and therefore accumulate a lot of single-base mismatches and indels. "This leads to the accumulation of microsatellites across the genome, and you can see microsatellite instability and ... an extraordinarily high tumor mutation burden," Diaz said, adding that these tumors tend to be inflamed and infiltrated with lymphocytes. "Pathologists could look at a colon cancer, see that lymphocytic infiltrate, and say that, 'This is likely mismatch repair-deficient.'"
After that first tissue-agnostic approval for Keytruda, the FDA approved Opdivo for mismatch repair-deficient metastatic colorectal cancer that has progressed on other treatments; Keytruda as a first-line option for mismatch repair-deficient metastatic colorectal cancer; and finally, yet another tissue agnostic indication for Keytruda for refractory solid tumors with high tumor mutation burden defined as at least 10 mutations/mega base.
Despite the growing use of tumor mutational burden and similar biomarkers like mismatch repair deficiency, microsatellite instability, and POL-E or -D mutations to predict immunotherapy response, physician-scientists like Diaz have wondered whether it is really the overabundance of tumor mutations that's driving the responses to checkpoint inhibitors, or if there are other biological mechanisms involved. Moreover, they wanted to know if there was something unique about the types of mutations tumors acquired that made them especially sensitive to checkpoint blockade.
When MSK researchers looked across tumor types that had high TMB (at least 10 mutations/mega base) that was not due to mismatch repair-deficiency or POL-D mutations, they found that only patients with non-small cell lung cancer, head and neck cancer, and melanoma saw improvements in overall survival with checkpoint inhibitors compared to TMB-low patients. A commonality between the tumor types is that they often have environmental causes: UV light for melanoma and tobacco use for lung and head and neck cancer.
There may be "a dichotomy in what the etiology of the hyper-mutagenesis might be and that tobacco and UV light, external exposure, might be driving this responsiveness to immune checkpoint blockade," Diaz observed.
Embarking on a 'crazy' adventure
This is the idea that Diaz's group is testing out in the trial that's underway in mismatch repair-proficient colorectal cancer. His group considered whether they could give "other exogenous agents," like chemotherapy, "to induce an immunogenic phenotype in tumors that are not immunogenic," Diaz said.
There are several potential barriers to using drugs to induce tumor mutations, he cautioned, because the treatments might induce mutations in tumor and normal tissues that end up harming the patient, the agents could be otherwise cytotoxic, or they may have suboptimal activity. And since prior research suggests that tumors must have strong clonal antigens in order to respond to anti-PD-1 treatment, the attempted pharmacological strategy could induce a "subclonal architecture," Diaz said.
Despite these caveats, he said his MSK colleagues Benoit Rousseau and Mitesh Patel are "embarking on this maybe crazy adventure … to [see] can we screen for drugs that would induce immune phenotype?" Based on prior research, researchers sought to rationally pick drugs that could induce high TMB, especially frameshift mutations.
Among several pharmacological strategies, the combination of temozolomide and cisplatin fit the bill. Temozolomide is known to induce tumor mutation burden though not so much indels, while cisplatin has been shown to induce a lot of indels, many of which are likely to be frameshift mutations. Researchers tested out the hypothesis first in non-immunogenic mouse cell lines cultured for eight weeks with these treatments, both individually and in combination.
They then looked at the genetic landscape in the cultured cells using whole-exome sequencing. The cells unexposed to any drugs had unremarkable tumor mutation burden and MSI and harbored some missense and frameshift mutations. The cells cultured with cisplatin had slightly increased TMB and some increases in frameshift mutations. With just temozolomide, the cultured cells had a lot more mutations, but it was really the combination that induced the immunogenic profile that researchers were looking for.
The temozolomide-cisplatin combo increased the TMB in cultured cells by 124 mutations per megabase, the number of frameshift mutations by more than 1,600, and the number of missense mutations by 12,000. The MSI score was also significantly increased. This showed that "we could at least recapitulate the genomic architecture of a tumor with these two drugs," Diaz said, adding that the combination treatment also recapitulated the mutational signatures associated with mismatch repair-deficient tumors.
Next, researchers injected these cultured cancer cells into mice and exposed some of them to anti-PD-1 treatment. The mice with cancers that hadn't received any treatment progressed, as did the mice given just anti-PD-1 treatment. The mice that had cancer cells previously exposed to single-agent temozolomide or cisplatin also progressed with or without anti-PD-1 treatment.
But in mice injected with cells exposed to the temozolomide-cisplatin combo, which Diaz said were "ultra mutagenized" and had a mutational signature consistent with mismatch repair-deficiency, not only was tumor growth delayed, but when anti-PD-1 treatment was added, the tumors "were completely eradicated," he said.
This pre-clinical experiment showed that "we can pharmacologically, in vitro change these tumors from non-immunogenic into immunogenic by introducing mutations at a very accelerated rate," Diaz said.
Early clinical trial experience
Encouraged by these findings, with the help of Neil Segal, another oncologist at MSK, Diaz and colleagues began the clinical trial underway now in mismatch repair-proficient advanced colorectal cancer patients. Although it's still early days in this study, using circulating tumor DNA testing researchers are tracking the accumulation of mutations in patients as they received the temozolomide-cisplatin-Opdivo cocktail. What they've seen so far, according to Diaz, is that "those patients that accumulate the most mutations tend to be achieving stable disease, whereas those that are not appear to be progressing."
With three drugs, there will be dosing issues to consider, Diaz acknowledged. Still, what the experiments so far indicate to Diaz is that there is a strong association between these tumor intrinsic features, such as mismatch repair deficiency, POLE mutations, and high TMB, and that they are highly predictive of response to checkpoint blockade.
"Is that proof? No, but there is a very strong correlation," Diaz said. "The recapitulation of the mismatch repair deficient genotype to convert cold tumors into immunogenic tumors provides further evidence that these tumor intrinsic features, these mutational alterations, are critical elements of tumor immunogenicity."