CHICAGO (GenomeWeb) – Investigators in the National Cancer Institute's Molecular Analysis for Therapy Choice (MATCH) trial reported initial results for three treatment arms of the study at the American Society of Clinical Oncology's annual meeting here this week, finding that some patients with defined molecular tumor alterations responded to targeted drugs while others didn't.
MATCH, a Phase II signal-finding study, was launched in 2015. Using Thermo Fisher Scientific's 143-gene Oncomine assay, the study screened around 6,000 patients with advanced, heavily pretreated solid tumors, lymphomas, or myelomas who had stopped responding to standard therapies or lacked standard treatments. Based on their molecular profile, some of these patients were assigned to one of almost 40 treatment arms.
In arm W, led by Young Kwang Chae from Northwestern University, researchers evaluated cancer patients' response to AstraZeneca's investigational FGFR inhibitor AZD4547 according to different FGFR alterations they harbored — fusions, single nucleotide variants, and amplifications — and identified a potential response signal in patients with FGFR fusions.
In arm Q, led by Komal Jhaveri from Memorial Sloan Kettering Cancer Center, researchers investigated response to Genentech's Kadcyla (trastuzumab emtansine) in patients with HER2-amplified tumors other than breast and gastric cancers, where HER2-targeted treatments are already widely prescribed. The study failed to reach its primary endpoint of objective response rate, but identified promising responses in certain rare tumors types that might be worth further follow up.
In arm I, led by Ian Krop from Dana-Farber Cancer Institute, researchers looked at how patients with PIK3CA-mutated tumors responded to Roche's experimental PI3K inhibitor taselisib, and reported lackluster efficacy in patients with different kinds of PIK3CA mutations, co-occurring mutations, and across tumor types.
"MATCH is designed to find signals of drug activity," said Richard Schilsky, chief medical officer of ASCO. "It's inevitable that some of the drugs will not show a signal of activity in the tumor types and against the alterations in which they've been tested, and some will."
MATCH has been criticized for not being able to live up to its name by not being able to place the majority of genomically screened cancer patients into molecularly defined treatment arms. Only 19 percent of the 6,000 patients were matched to one of nearly 40 arms, and as of a May 31 update, only 15 arms have reached their accrual goal.
Because a number of arms haven't reached the minimum accrual goal of 35 patients, MATCH is in the process of qualifying additional laboratories so it can cast a wider net and identify more patients with these exceedingly rare biomarkers of interest.
Keith Flaherty, a medical oncologist at Massachusetts General Hospital and one of the study chairs, noted that reaching the enrollment goal for some MATCH arms has been difficult partly because the very nature of the enterprise — identifying and exploring knowledge gaps in precision oncology, and advancing new hypotheses — means studying small subsets of patients, or even a single patient, with rare tumors and rare tumor markers.
"This is a discovery trial. We're not trying to relearn things we already knew, but really map out the likelihood of response in areas that we didn't know," Flaherty told GenomeWeb. "That's what defines the treatment-eligible population [in MATCH]. That's a very different thing than just determining how many patients get a next-generation sequencing test and can have something useful done [in terms of treatment.]"
As an example, he noted that the combination of BRAF and MET inhibitors is already known to work in melanoma patients with BRAF V600 mutations and not to be effective in colorectal cancer patients. As such, these groups wouldn't be studied in a MATCH arm. "We went down this process over and over again, trying to hone, for every therapy and molecular feature in MATCH, where there is still discovery to be done," Flaherty explained. "Where can we still find things that haven't been found?"
For example, in arm W, researchers hypothesized that tumors harboring FGFR alterations may respond to an FGFR inhibitor like AZD4547 regardless of histology, because such mutations are found across tumors types, in up to 15 percent of cancers. However, patients with FGFR1-amplified gastric cancer and non-small cell lung cancer were excluded from enrolling in arm W due to prior evidence that the drug lacked efficacy in these subpopulations.
Out of 5,558 patients whose tumors were successfully profiled in MATCH, 70 had FGFR alterations and were assigned to arm W. Ultimately, 50 patients received AZD4547. Of these, 21 had an FGFR amplification, 20 an FGFR SNV, and nine a gene fusion.
Among those treated, common tumor types were breast, urothelial, cervical, and endometrial cancers. Breast cancers had more amplifications, cervical cancer had more fusions, while urothelial and endometrial cancers had more point mutations.
Four patients had partial responses, 20 had stable disease, and out of these 24 cases, five had progression-free survival of greater than 5.5 months. These five patients had intrahepatic cholangiocarcinoma, urothelial carcinoma of the renal pelvis, mucinous adenocarcinoma of the colon, epithelial-myoepithelial carcinoma of the submandibular gland, and transitional cell carcinoma of the renal pelvis. Eighteen patients experienced disease progression.
Median progression-free survival for all 50 patients was 3.6 months. However, the subset of patients with FGFR fusions fared best compared to patients with other alterations, with median progression-free survival of 5.1 months and a six-month progression-free survival rate of 37 percent.
Around half the 49 patients evaluable for adverse events experienced grade 3 adverse events. "AZD4547 demonstrated modest activity across various solid tumors with aberrations in the FGFR pathway with acceptable toxicities," Kwang Chae said. "Tumors harboring FGFR fusions seem to have superior progression-free survival outcomes compared to patients with FGFR amplifications or SNVs."
He recommended further exploration of the FGFR fusions in this setting, and noted that in arm K2 of MATCH, investigators will study response to Janssen's erdafitnib in patients with FGFR fusions and point mutations, and in arm K1, they will study the same drug in patients with FGFR amplifications.
A current challenge in precision oncology is figuring out which type of alteration — fusion, point mutation, and amplification — is targetable. Flaherty noted that based on signals from past studies, and drugs approved by the US Food and Drug Administration, fusions like BCR-ABL rank high on the list of targetable markers, point mutations are next, while amplifications rank the lowest.
HER2 amplifications were the focus in arm Q. Jhaveri and colleagues studied how patients with HER2-amplified tumors other than breast and gastric cancers responded to Kadcyla. The antibody drug conjugate is already an approved treatment for metastatic breast cancer patients who have had prior treatment with Herceptin (trastuzumab) and a taxane and overexpress HER2. And a study has shown that the Kadcyla didn't improve overall survival in advanced gastric cancer patients compared to taxane therapy.
HER2 is overexpressed in 15 percent of breast and ovarian cancer cases. However, between 2 percent and 7 percent of other solid tumors also harbor this biomarker, which is associated with poor survival and poor prognosis.
In arm Q, 37 eligible patients with HER2 amplification — as determined by Oncomine and orthogonally confirmed, and defined as a HER2 copy number gain of more than seven — were treated with Kadcyla. They had common malignancies, including colon and ovarian cancer, as well as rare cancers, like tumors in tubes that carry bile through the liver, cancer of salivary glands, and cancer of the uterus.
Researchers had aimed to show a 16 percent objective response rate. Although more than 50 percent of 29 evaluable patients saw some tumor shrinkage, only three had a confirmed partial response — one with salivary duct cancer of the parotid gland, one with squamous cell cancer of the parotid gland, and one with extramammary Paget's disease of the scrotum.
Twenty patients had a partial response or stable disease. The six-month progression-free survival rate was around 25 percent, and no new safety signals were seen in the study. "There was a trend for tumor shrinkage with greater HER2 copy number," Jhavery said during a meeting presentation.
The patient with squamous cell cancer of the parotid gland that experienced a partial response remains on the therapy at 16.5 months. Some ovarian, endometrial, and colorectal cancer patients experienced durable stable disease, and Jhaveri highlighted other single cases that had notable responses.
Although arm Q didn't meet the predefined overall response rate, Jhaveri and colleagues recommended further research into Kadcyla's efficacy in some of the rare HER2-amplified tumor types that saw some benefit. Researchers will also analyze the correlation of HER2 copy number gain, determined by the Oncomine assay, with HER expression determined by immunohistochemistry and fluorescence in situ hybridization.
Flaherty noted that studies investigating drugs targeting amplified genes have been disappointing in the past. "I'm not trying to disparage HER2 targeted therapy for HER2-amplified cancers," Flaherty said. "There's efficacy no doubt, but we've chased as a field multiple examples of amplifications as the indicator [of response] and been disappointed in terms of not seeing efficacy when targeting single amplified genes."
Some of this might be due to the fact that in genetically complex tumors, these gene amplifications may be playing a supporting role to other genetic features that are more important for the formation of the tumor. "When we go to the meta level [in MATCH] and look at all our patients across arms, I think we might be able to unpack some of that," he said.
Lastly, researchers led by Krop enrolled 65 patients with 36 different tumor histologies defined by PIK3CA mutations in arm I, and studied their response to the PI3K inhibitor taselisib. This study excluded patients who had breast cancer and squamous cell lung cancer because of competing ongoing studies, and patients with PTEN and KRAS mutations due to data suggesting that these co-occurring mutations are associated with resistance to PI3K inhibitors.
Although none of the patients experienced an objective response, and the arm didn't meet its primary endpoint, some had stable disease. Six-month progression-free survival rate was 24 percent and two patients have been on the study for more than a year. Overall median progression-free survival in the arm was 3.5 months.
Krop and colleagues delved deeper into this subset of patients to see if the type of PIK3CA mutation, other co-occurring mutations, or tumor histology impacted outcomes on taselisib. They noted that some patients with mutations in the kinase domain saw slightly greater tumor shrinkage than those with mutations in helical or other domains, but these differences "weren't striking," Krop said, because there were patients in all these groups who also experienced disease progression.
When researchers looked at response according to other co-occurring mutations, no specific patterns stood out in terms of taselisib's efficacy. "There was no clear relationship between these markers," Krop said.
Based on this data, the researchers concluded that taselisib had limited activity in cancer patients with PIK3CA mutations. "I think it's safe to say that PIK3CA mutations alone, without other data, don't appear to be a sufficient predictor of taselisib activity," Krop said.
Although there was preclinical and early study data that suggested patients might benefit from this drug, this was not the case in arm I, he suspected, because patients were heavily pretreated and by design, the study enrolled a heterogeneous population. "It may certainly be possible that PIK3CA [mutations] may not be the major driver in the particular tumors we looked at," he said, adding that other mutations may be helping cancer cells bypass taselisib's ability to block the PI3 kinase. "But we were not able to identify these other mutations."
In arm I, 12 percent of patients had grade 3 diarrhea, and while most of the toxicities were manageable, 9 percent discontinued the study due to taselisib-related adverse events. "The toxicity of targeting such ubiquitous signaling pathways, [which involve] PI3 kinase, may just inherently limit our ability to adequately inhibit PI3 kinase function," Krop said.
In reviewing the data from this arm, Fabrice Andre from the Institut Gustave Roussy in France noted that in some patients, PIK3CA mutations may not lead to the activation of the pathway and oncogene addiction, while in other cases, PIK3CA may be a codriver of the tumor.
Andre discussed the need for advancing triplet therapy combinations to target the driver mutation, PIK3CA mutation codriver, and then block cancer cells from escaping inhibition via another pathway. He also suggested the need to advance second-generation molecular screening strategies that go beyond tumor sequencing to identify single genetic drivers and integrate molecular pathology and assessment of phosphoproteins.
Currently, Roche is studying taselisib in solid tumors, metastatic hormone receptor-positive breast cancer, and metastatic HER2-negative breast cancer. At the meeting, researchers presented data from a randomized Phase III trial comparing taselisib plus hormone therapy against hormone therapy alone in advanced breast cancer patients.
Patients in the taselisib arm had a two-month progression-free survival advantage over patients who received just hormone therapy, but they also saw more serious side effects, more frequent diarrhea and hyperglycemia, and grade 3 toxicities. Based on the results, Genentech has decided to halt development of taselisib (see related story).
There are multiple isoforms of the PI3 kinase family that, when targeted by a drug, produce toxicity. Taselisib hits multiple PI3 kinases, which is likely why taselisib-treated patients had more adverse events. PI3K inhibitors that aren't so broadly active could have less toxicity, according to experts at the meeting.
Ultimately, MATCH is an efficient way to screen drugs for activity, Schilsky said. "Some will advance. Many will not advance," he noted. "Most drugs don't actually make it, but hopefully along the way, we learn something about disease biology and the effectiveness of the treatments."
Drugmakers haven't publicly stated whether the data from the three MATCH arms will impact their development programs, but according to Flaherty, they're deliberating over the findings. "What we're staring at in terms of signals, they're staring at as well," he said, noting that if the sponsor doesn't follow up on a lead, the MATCH investigators might. "We absolutely have an interest in chasing down every signal we've uncovered," he said.