This is the second article in a two-part series on tissue-agnostic drugs in cancer. The first part focuses on the emerging biomarkers researchers are evaluating and the second part focuses on the regulatory and financial considerations for drug developers.
NEW YORK (GenomeWeb) – Merck and Loxo Oncology's experience developing and commercializing cancer drugs for pan-cancer indications has piqued the interest of other pharmaceutical companies in this new paradigm.
Following the approvals of Merck's pembrolizumab (Keytruda) and Bayer and Loxo's larotrectinib (Vitrakvi), drugmakers are increasingly exploring biomarkers that cut across different histologies in order to better understand disease biology and the landscape of rare genomic alterations. Some pharmaceutical companies are even advancing drugs with tissue-agnostic indications in clinical trials and aiming for a regulatory filing.
Pan-cancer indications have their own scientific issues due to the rarity of the markers of interest and because patients with different kinds of cancers may have other molecular and non-molecular factors that impact their ability to respond to drugs. These issues may give some drugmakers pause as to whether it is financially worthwhile to take a drug through a new, risky paradigm that, if successful, will likely never be a blockbuster drug.
Ahead of the American Society of Clinical Oncology's annual meeting, researchers presented data on the activity of Roche's entrectinib in pediatric patients with a variety of solid tumors, including cancers in the brain and spine, characterized by NTRK, ROS1, and ALK fusions. The study showed that only patients with these rare alterations saw their tumors shrink, but not those who didn't harbor these alterations.
Giles Robinson, a pediatric neuro-oncologist at St. Jude Children's Research Hospital, highlighted that patients in this study had some very rare tumors, such as neuroblastoma, pediatric high-grade gliomas, and infantile fibrosarcoma, a cancer that forms on the limbs of young children and can be deforming. "I predict that all these [cancers] would be universally fatal if they didn't respond to this medicine," he said.
While researchers are now able to identify these rarest of rare cancer patients due to the increased availability of next-generation sequencing and basket trials like the one Robinson’s group performed, the smaller payoff for biopharma firms in targeting these molecularly defined indications might discourage investment.
Even so, with the FDA’s support for tissue-agnostic approaches, some drugmakers are taking on the challenges. For example, NTRK fusions show up in less than 0.5 percent of solid tumors. Yet, Roche subsidiary Genentech said that it has submitted data to the FDA, the European Medicines Agency, and the Japanese Health Authorities as part of its bid for regulatory approval for entrectinib in advanced adult and pediatric solid tumors with NTRK fusions. The FDA is slated to make its decision in mid-August.
"Six or seven years ago there was an enormous amount of risk" in pursuing pan-cancer indications, but the availability of pembrolizumab and larotrectinib has "tremendously lowered" the risk, said Razelle Kurzrock, director for the Center for Personalized Cancer Therapy and Clinical Trials Office at the University of California, San Diego. "I think companies will be much more willing to do this."
Luis Diaz, head of the solid tumor oncology division at Memorial Sloan Kettering Cancer Center, led the research that provided the first prospective clinical data showing that patients who were MSI-high could derive benefit from pembrolizumab regardless of the type of cancer they had. The pembrolizumab and microsatellite instability (MSI) story "has really opened up the FDA's, physicians' and drug developers' eyes that this is a very real possibility," he said.
A pathway emerges
The FDA plans to issue guidance on the topic of tissue-agnostic drug development and held a public workshop last year to discuss the ways in which the orphan drug designation program can incentivize pharmaceutical firms to advance products for rare biomarker-defined indications. "Being able to confront disease based on how it develops and not simply how it appears at the time of diagnosis is a key evolution in how we approach scientific research and medicine," former FDA Commissioner Scott Gottlieb said at the workshop. "These approaches aren't just theoretical anymore. FDA is opening a door to these methods and embracing new regulatory constructs like basket trials and master protocols that make these approaches easier and more efficient."
In working with Merck and Loxo to approve their pan-cancer indications, the FDA recognized the difficulty of conducting traditional randomized-controlled trials in rare biomarker indications. As such, the agency was flexible in terms of the evidence it was willing to accept.
For example, in order to gain accelerated approval for pembrolizumab, which had already been marketed and shown to be safe and effective in a variety of tumor types, Merck submitted data to the FDA from five single-armed studies involving 149 patients with 15 types of solid tumors characterized as MSI-high or mismatch repair deficiency (dMMR). The FDA didn't restrict biomarker testing to a single test or lab, but allowed patients to have their MSI-high and dMMR status determined prospectively via PCR and immunohistochemistry in local labs, as well as accepting retrospective testing in a central lab using PCR.
Larotrectinib, meanwhile, was a new drug and unlike pembrolizumab hadn't been extensively studied in a variety of cancers. Yet, it was approved with a pan-cancer indication from the outset, likely because the drug targets NTRK fusions, which don't occur frequently enough in any one cancer type. Loxo was able to garner approval for larotrectinib in a relatively rapid timeline — around four years from starting clinical studies — and submitted data from 55 patients with solid tumors enrolled in three single-arm trials, one of which was a Phase II basket study.
In a presentation describing the regulatory considerations for pembrolizumab's pan-cancer indication, Steven Lemery, associate director at FDA's Division of Oncology Products 2, has explained that there is no one-size-fits-all answer to tissue-agnostic drug approvals. For example, there is no set number of tumor types that need to be studied. The agency has said it will consider whether it can approve a drug based on the totality of evidence submitted, if the treatment effect is generally consistent across tumor types studied, as well as the scientific basis of the approach.
"Absolute certainty regarding drug effect will not exist for every biomarker-tumor-drug combination," Lemery said in the presentation, pointing out that "absolute certainty also does not exist in tumor-specific approvals."
Still, studying a drug in patients with one kind of tumor and who are similar in other ways, allows for a more controlled research environment. When trying to establish the safety and efficacy of a drug in patients with all different types of cancers, there may be more confounding factors to deal with.
"For these pan-cancer indications, it's harder to establish the medical and the therapeutic effect, because you are dealing with the individual aspects of all these diseases," said Michael Korn, chief medical officer of Caris Life Sciences, a company that provides molecular profiling services. "You are a priori starting with a more heterogeneous group of patients."
Cancers are successful at evading treatment because they are so varied. The cells that make up a breast tumor in a single patient can have different molecular drivers. This heterogeneity within a patient’s own tumor and between patients makes cancer difficult to study and defeat it with drugs. Tissue-agnostic drug development could add to this complexity.
This is a reason for drugmakers to be wary of combining different cancer types in therapeutic studies, agreed Sameek Roychowdhury, a medical oncologist and physician scientist who specializes in precision oncology at the Ohio State University Comprehensive Cancer Center. The same biomarker may occur in patients with different kinds of cancer, but these patients may respond very differently to the same drug due to a myriad of other factors. Some markers, while they show up in lots of tumors, may not be driving the disease in a particular cancer type and may co-occur with other resistance markers, limiting a drug's ability to attack and destroy cancer cells.
Roychowdhury recalled the experience with BRAF mutations. Melanoma patients with certain BRAF mutations have strong responses to BRAF inhibitors, but colorectal cancer patients with these mutations don't respond to these treatments as single agents. "In general, you don't want to make assumptions about the same genetic alterations across diseases," he said. "But, sometimes, we have to when [the biomarker] is just not common enough to do those studies in the different diseases by themselves."
For this reason, when a biomarker occurs frequently enough in a specific cancer type, drugmakers are studying the drugs in that setting first before venturing into the tissue-agnostic path. "The rationale for pan-cancer is really a numbers thing. If [the biomarker] was in half of breast cancers, then it would just be a breast cancer study," observed Roychowdhury. "Once it becomes something that's in 5 percent or less of the patients in a cancer population, then it becomes much harder to do."
For example, since 15 percent of advanced colorectal cancer patients are MSI-high or dMMR, drugmakers have focused on getting immunotherapy indications in this subset of patients. Bristol-Myers Squibb's nivolumab (Opdivo) and the combination of nivolumab and ipilimumab (Yervoy) have FDA approval in this setting. BMS declined to comment on its plans to seek a tissue-agnostic indication for nivolumab, but notably, the NCI's MATCH basket trial has reported out some preliminary data that suggests the drug may have activity in solid tumors with dMMR solid tumors beyond colorectal cancer.
Similarly, FGFR alterations also show up across tumor types but more frequently in certain histologies: fusions in this gene occur in around 15 percent of cholangiocarcinomas, while fusions and mutations show up in approximately 20 percent of metastatic urothelial cancer patients. The FDA last month approved Janssen Pharmaceuticals’ metastatic bladder cancer drug erdafitinib (Balversa) for patients with FGFR3 gene mutations or FGFR gene fusions. During a recent business review presentation, Peter Lebowitz, head of J&J's global oncology drug development, highlighted that because genetic changes in this gene occur in a variety of malignancies, the company plans to initiate tissue-agnostic studies later this year.
Incyte intends to seek FDA approval for its FGFR inhibitor pemigatinib later this year in FGFR2 translocated cholangiocarcinoma and pursue approval for the drug in FGFR3 mutated bladder cancer next year. Incyte Chief Medical Officer Steven Stein noted during a quarterly earnings call in April that the company will also open a registration-directed Phase II study in the tumor-agnostic setting now that the regulatory pathway is better validated by two commercialized pan-cancer treatment indications.
A financial gamble?
Despite the scientific challenges of cancer drug development, when the tissue-agnostic approach works, "you do see pretty impressive responses in patients who wouldn't have had any options whatsoever," UCSD's Kurzrock said. For example, in the trials that led to the approval of larotrectinib in solid tumors with NTRK fusions, 75 percent of patients saw their tumors shrink, with 73 percent responding to treatment for at least six months and 39 percent responding for a year or more at the time results were analyzed.
"Everybody, whether you work for a [drug] company or not, knows someone with cancer," she said, "and when you see these fantastic responses, that's gratifying."
Still, tissue-agnostic drugs present a difficult financial prospect for pharmaceutical companies, because the rarity of the target biomarkers in the population makes it hard to do studies and ensure a return on investment after commercialization. "The prevalence [of these biomarkers] starts to become a concern because if they're exceptionally rare, it just becomes so challenging to identify them," said Ken Carson, senior medical director at Roche subsidiary Flatiron Health, a health technology firm with a clinico-genomics database that can be used as a source of real-world data on how patients with certain biomarkers respond to treatment.
"But, in an era of wide acceptance of NGS testing, maybe we can identify those patients who are needles in haystacks, so you can demonstrate the efficacy of these agents," he said.
Some drugmakers are investing in these indications, however, as a way to grow the revenue opportunities for an already approved, successful therapy and to diversify their cancer drug pipelines.
For example, pembrolizumab is already a blockbuster for Merck and netted around $4.15 billion in the US last year, primarily driven by the drug's first-line lung cancer indication. However, Chief Commercial Officer Frank Clyburn said during the firm's fiscal year 2018 earnings call that the newer pan-cancer indication in MSI-high solid tumors also demonstrated "good growth," accounting for around 5 percent of sales, or around $208 million.
Meanwhile, Lilly paid $8 billion to snap up Loxo earlier this year for its suite of cancer compounds, including larotrectinib, the follow-on pan-cancer TRK inhibitor Loxo-195, and Loxo-292, a drug that targets RET fusions and alterations that occur in a variety of tumor types. The acquisition triggered Bayer to exercise an option it had under a 2017 deal with Loxo to gain full rights to larotrectinib and Loxo-195, which means that Bayer will now pay royalties to Lilly on the sales of these two drugs.
Industry observers have said the purchase of a tiny startup with a pipeline of orphan drugs is a financial gamble for Lilly. However, the move may signal its competitors to not ignore tissue-agnostic indications just because the intended patient population will be small.
Roche is certainly pushing ahead with entrectinib in NTRK fusion-positive advanced solid tumors. The pre-ASCO data presented recently on the activity of the drug in pediatric patients with solid tumors characterized by fusions in NTRK, ROS1, and ALK, suggests that combining rare biomarker subpopulations may be a strategy of bolstering the market for these orphan drugs.
"The NTRK story has taught a great lesson to the pharmaceutical industry," said Caris' Korn. "We see clearly much broader interest in working with us on identifying the right patient populations [for drugs] … and I think there'll be a number of new companion diagnostics for these kinds of drugs coming to market in the next few years."
Caris recently launched a whole-transcriptome test that is designed to detect gene fusions in cancer patients, so doctors can use that data to identify drugs that target those alterations. The company is seeking regulatory approval for the MI Transcriptome assay as a companion diagnostic, which recently received breakthrough device designation from the FDA. Korn said that one of the CDx indications the company is seeking approval for involves a pan-cancer biomarker, though he didn't disclose further details.
The emergence of pan-cancer indications also highlights the importance of continuing to expand cancer patients' access to advanced molecular tests that can gauge many biomarkers at once, a lesson Kurzrock learned in 2007, when she was at MD Anderson Cancer Center. Back then, sequencing a person's genome cost around $10 million, and next-generation sequencing panels weren't available, but at MD Anderson oncologists had access to a PCR test that gauged 12 genes one by one. In most cases, however, there was only enough tumor tissue to analyze three genes, so doctors would prioritize the top-three genes they wanted tested for every patient and wait a couple of weeks for the results for each gene.
Kurzrock and her colleagues learned all too quickly that this paradigm wasn't viable. Indeed, today, because there are many more drugs targeting increasingly rare tumor alterations, oncologists are quickly realizing the drawbacks of sequential biomarker testing. "You run out of tissue, you run out of time, and there is no way in the world a doctor can guess which gene will be abnormal in a patient," she said. "You have to do a panel."