Is the paradigm of cancer treatment bound to change? At a time when new drugs are registered for use on a specific molecular target instead of a specific tumor, experts raise some warnings: this new model might be hard to sustain financially, and might spark too much optimism, if one doesn’t take into account tumor cells interactions with the microenvironment and genetic intratumor variability
For 20 years, since trastuzumab was approved for breast cancer in 1998 and imatinib for chronic myeloid leukemia in 2001, we have been told that the paradigm of cancer treatment is bound to change. We are entering a new era of targeted, precision, personalised oncology, it is being said. The improved knowledge of the molecular secrets of tumours will allow us to tackle the disease by zeroing in on specific alterations and pathways, eventually treating any single tumour, in any single patient, in any specific phase of the disease course. Cancers will not be classified any longer based on their primary location, as a brain, lung or colon cancer, or on their histopathological features, for how its cells look under the microscope, but rather based on specific molecular tags that could be targeted by tailored drugs, regardless of the cancer tissue of origin. No matter where the tumour is, within the body. That’s where the Achilles’ heel of cancer is: in its cells. Regardless of where they are. It seemed quite easy; the obvious consequence of the advancement of our understanding.
The first one only in 2017
In fact, the road to this different approach was, and still is, full of hurdles. The first approval of a drug based not on tumour location, but on its genetic specific features, came only last year, in 2017, by the US Food and Drug Administration. It refers to pembrolizumab, a well-known monoclonal antibody, already in use for immunotherapy in several tumours, such as advanced melanoma, bladder and non-small cell lung cancer. For the new approval, anyway, location and type of tumour is irrelevant. The drug can be used against any solid tumour with two genetic markers showing that cancer cells have a flaw in their capacity to repair DNA: microsatellite instability-high (MSI-H) and mismatch repair deficiency (dMMR). These defects make them more recognizable by our immune system, when pembrolizumab, like other checkpoint inhibitors, release the brake that prevent our immune T-cells from identifying cancer cells as foreigner and attacking them.
In these cases, the drug is so efficacious that, in the study published on Science that hastened FDA’s decision, some disease control (with partial or complete response and stable disease) was achieved in 66 (77%) of the 86 patients with 12 different types of advanced tumours that had resisted every standard treatment. More than half of the 86 patients saw their tumours shrinking, and 18 achieved a complete radiographic response.
“Other targeted therapies based on genetic features, irrespective of tumour’s location or histological type, are on the way” says Francesco Perrone, head of Clinical Trials at the Istituto Nazionale Tumori, Fondazione Pascale, in Naples, Italy. “For nivolumab, another checkpoint inhibitor like pembroluzimab, FDA granted an accelerated approval, and an application has already been presented for another kind of drug, larotrectenib. In a recent trial published on the New England Journal of Medicine, larotrectenib gave an impressive 78% rate of response across several paediatric and adult type of cancer, mostly rare, which only had in common a genetic signature, fusions involving one of three tropomyosin receptor kinases (TRK)”.
The tumour’s address still has its role
“The cancer’s site cannot be fully ignored, anyway” Pier Paolo Di Fiore, director of Molecular Carcinogenesis and Stem Cell Biology research unit at Istituto Europeo di Oncologia (IEO) in Milan, Italy, says. “Even in the presence of the same mutation or genetic marker, responses can be different”. In the recent SUMMIT study, for example, 141 patients with 21 different types of tumours that had mutations in HER2 and HER3 genes were treated with a targeted drug called neratinib: while the treatment was active in patients with breast, cervical and biliary cancers, it wasn’t in those with bladder or colorectal tumours.
“It has been well established that the tumour has a strict relationship with the microenvironment where it develops and grows” he explains. “Its genetic is important, of course, but we shouldn’t overlook the role of epigenetic factors. The same mutation can have a very different impact in different organs”.
And this is not all: “Most cancers do not depend on one single driver gene and addressing one single mutation could therefore be disappointing”.
This proved to be true with PI3KCA, another gene mutated in many cancers. The clinical results of PI3KCA inhibitors appear to be less favourable than thought. The extensive intratumour genetic heterogeneity that we are increasingly aware of makes it difficult to tackle the disease with a single, or a simple combination, of targeted agents.
A needle in the haystack
So, where cancer develops is still relevant, and it will be for a long time. The gene-based approach regards only a minority of cases and, even when it can be tried, the tumour’s location is anyway important to improve our chance to find patients carrying one target or the other. Each marker, in fact, is present only in a very small part of cancer patients. To find patients that carry one of those, and that are consequently eligible for each treatment, is a hard task. Since some can be more frequent in some type of cancer than in others, the tumour’s site is still important to be considered in order to improve the sensitivity of screening.
It is like looking for a needle in the haystack: one needs to screen many patients, spending a lot of money and feeding countless false hopes, to find those few who could (in part) respond. The flaws in DNA repair addressed by pembrolizumab are present only in about 15% of sporadic, non-familial colorectal cancers, while 90% of colorectal cancers in patients with Lynch syndrome, a genetic condition, have it. In other types of cancer, the prevalence is much lower. “We could therefore start focusing on colorectal cancers, which represented nearly half of the tumours included in the study on DNA repair and pembroluzimab” Perrone suggests, “but when we will have more and more markers, how can we afford testing all patients for each mutation or characteristic, that may be present only in 2-5% of them? Feasibility and sustainability are crucial issues. Even if a more widespread use is usually accompanied by a decline in prices, a genome-wide testing of all patients for all possible mutations will not be easy to perform in the short term”.
Many patients are undergoing wide gene testing offered by private companies. “When they go to their oncologist with the results, they often get disappointed: this information is usually completely useless, since most mutations don’t have a targeted therapy to address them”, Perrone says.
Are these drugs “magic bullets”?
In 2017, FDA also approved F1CDx, a 324-gene panel offered by Foundation Medicine, for any solid tumour type, with enrolment in an observational registry. In doing so, the FDA praised the ability of F1CDx to pair patients with targeted treatment options. But in the recent MOSCATO-1 trial, on 1,035 patients originally intended to undergo next-generation sequencing, only 199 (19%) were paired with a genome-informed therapy and just 22 patients, or 2.1%, of the original cohort, were able to achieve an objective response.
Another cold shower came from a trial recently published in JAMA Oncology, where 15.4% of 610,000 US patients with metastatic cancer were found eligible for an FDA-approved, genome based, or guided, drug. But only 6.6% likely had benefits from the treatment and many patients relapse after a couple of years on the drugs.
“We also estimated that ‘real-world’ response rates would be 10% or 20% lower than those reported in clinical trials” Vinay Prasad, one of the author and a young, renowned oncologist from Oregon Health & Science University, Portland, Oregon told us. “We calculated that not more than 10% of patients would benefit from genome-targeted and genome-informed therapies in 2018”.
Basket, umbrellas and other new trials
Looking at cancer from this new perspective also means that under the same definition of a common cancer such as lung cancer there could be many and many rare different tumours, with many different molecular characteristics, and available (or more often) potential treatments accordingly.
The traditional design of clinical trials sounds inadequate to face the complexity of the new approach, with so many possible variations in the subtype of tumours, regardless of their site. New master protocols have therefore been conceived. The most common are called “basket” and “umbrella” trials: basket trials gather patients with cancer of different types but one or few similar mutations; umbrella trials, on the contrary, enrol patient with the same kind of tumour, whom are then tested for different mutations, in order to be channelled in a specific arm, toward a specific, targeted drugs.
These new designs of trials are expected to make it easier to gather a statistically sufficient number of patients and to move them from one arm to the other when requested.
They are not always successful. The SHIVA trial, the first reported ‘basket’ trial of this type, which compared standard of care against treating a variety of solid tumours based purely on their genomic profile, showed that this approach produced more toxicities than standard treatment. The off-label use of drugs, based on molecular targets, without evidence obtained by clinical trials for specific tumours, should therefore be discouraged, the authors think.
Feasibility and sustainability of this revolution
On the field, in oncology hospital wards, how can doctors orienteer in such a complex landscape, facing also patients’ expectations towards this, somehow maybe hyped, personalized medicine? “We have known for 20 years that this revolution would have come, but we were not prepared to its consequences in clinical practice,” Perrone adds. “All our knowledge of cancer, our system of prevention, diagnosis, care and payment depends on a classification based on site. We don’t have an epidemiology of genetic characteristics of cancer, neither have any idea of how we could manage this change. Until few months ago, most clinical oncologists had hardly heard about TRK fusions. We don’t have diagnostic kits to find them in patients. Genetic test for DNA instability start being available, but how can we be sure they are well validated and standardized? These are not trivial issues, since the decision about which patients should be tested, and then treated, has a huge economic impact”.
This approach, in fact, requires a strict selection of eligible patients to be treated. Pembrolizumab costs about $156,000 a year in USA and €135,000 in Europe. Independent trials assessing the feasibility, safety and effectiveness of reducing dosage and times of administration of new targeted drugs are a possibility. “They are ongoing” told us Mark Ratain, director of the Center for Personalized Therapeutics at the Comprehensive Cancer Center of the University of Chicago, USA, “and are highly appropriate in the current environment of many similar drugs (some with marginal efficacy) and high costs”.
“Patients and their families need studies that aim to reduce costs. High costs lead to decreased access for effective drugs, but are also a burden on healthy individuals, as they fund these costs indirectly through taxes and insurance premiums. Corporations that provide health care coverage in some companies must fund the costs of these drugs in lieu of providing higher compensation to its employees”.
This is all but impossible. “Many – if not most – modern oncology drugs can be administered at a significantly lower cost” Ratain continues. “For those companies that choose not to optimize the dosage, this may result in financial consequences when others do so. On the other hand, optimizing the dose is important to patients – even if there are no financial savings – since this will result in decreased toxicity, as well as potentially greater convenience”.
Is it the only way?
“Targeting individual genetic alterations in a personalized manner will bring increasing benefits as we get to know cancer better and better, but this will need to be integrated with other approaches that take into consideration the interaction of the tumour with the host” Di Fiore concludes. “The case of pembrolizumab is particularly interesting because the two approaches are synthetized in one drug, which is given based on the characteristics of the tumour but acts on the immune system of the host”.
This and other strategies, not strictly personal, won’t be less important than individualized medicine in the future treatment of cancer. Closing in on it, by several sides, it will be easier to achieve better results.