The precision oncology landscape is in full effect, it has held great promise, and it is constantly changing. It is also very confusing. On many levels. In the current era, a more-than-superficial grasp of basic molecular oncology is required (or at least should be) for routine medical care of many patients with solid tumor malignancies.
The evolving era of precision oncology has changed the way we look at some tumor types and subgroups of tumors-literally-as shifts from histologic to molecular classifications. Gastric cancer and medulloblastoma brain tumors are two examples of cancers whose subtypes have been better identified using molecular classifications rather than histologic.
But beyond reclassification and learning names to designate tumors, what is the legacy and message from the first era of molecular oncology? Let us begin by discussing the fruit of new technology that promised a new era of precision medicine in cancer care.
What's in a Target, Anyway?
The word "target"-as in, molecular target-has been in vogue for nearly 2 decades. It is a loaded term. Why so? Because its definition differs on one's perspectives, and sometimes in relation to its prevalence and identification of a drug that fits, taken in combination with demonstration that drug development leads to equitable access to drugs that effectively demonstrate meaningful (as well as significant) improvements in survival and quality of life of treated patients. I will use recent advances in molecular profiling and reports of increased identification of molecular targets in biliary tract cancers (cholangiocarcinomas) as an example.
There is no question that molecular alterations are identifiable in cholangiocarcinoma patients; with advances in next-generation sequencing, that has actually become the easy part. Definitions of what constitutes enrichment and targets may vary. In my mind, I consider the following points that go well beyond identification of alterations before considering them to be true "targets."
Labeling of testable and identifiable mutations as readily targetable can mistakenly presuppose a one target-one drug model for likelihood of success. In the world of GI oncology, we have seen this before: BRAF mutations are prevalent in melanomas, and druggable with inhibitors. Therefore, the same tactic should have worked in BRAF-mt CRC. But it didn't. The tumors upregulated an alternate pathway that led to no improvement in overall survival (OS).
What did we learn from this? A multipronged and more rational trial design that would truly mark BRAF as a viable target was required. Enter examples like the BEACON trial, which presented a combination treatment strategy that gained FDA breakthrough status and eventual approval for treatment of BRAF-mutant CRC in second-line settings and beyond.
Biliary tract cancers have caught up in terms of identification of alterations using next-generation sequencing (NGS). In 2019, the ClarIDHy trial was presented and introduced us to updates regarding the trial utilizing ivosidenib in IDH1-mutant cholangiocarcinoma. Is its use based on a rational premise when treating patients whose tumors harbor this mutation? Of course! Was there an improvement seen in progression-free survival as compared to use of placebo? Yes. Was there an observed improvement in OS? Not as of yet.
The trial reported a "trend toward favorable OS compared to placebo." We can take two approaches when attempting to objectively answer the question of whether, as a result, IDH represents a true target. On the one side, taking into account the concern that biliary tract cancers is a relatively rare cancer for which there has been only one tried-and-true standard-of-care regimen for unresected case (gemcitabine and cisplatin), any progress in a seemingly desperate corner of oncology can be perceived as monumental. As stated during the presentation at ESMO 2019 Annual Meeting (Abstract LBA10_PR), PFS improvement using ivosidenib was essentially better than nothing for "a rare cancer having few effective therapies."
However, let us also consider measuring progress and definition using a different yardstick, and regardless of prevalence of the cancer and this mutation, let's layer the perspective with understanding of perception and also access: until mid-2020, I could not find ivosidenib on my hospital's formulary off-trial. If I am a molecular GI oncologist at an NCI-designated comprehensive cancer center and have this challenge, how much harder would it be for a non-specialist in general oncology practice? If we're allowed to be philosophical: If a drug can fit a molecular alteration and increases PFS (not necessarily OS), but I cannot prescribe the drug, is it truly a target?
This point is not meant to be semantic, but rather to bring realistic points to the conversation, not unlike recent points made online regarding the ADAURA trial (N Engl J Med 2020;383(18):1711-1723) and baseline brain MRI versus CT, or availability of PET/CT scans, on social media. Here it was pointed out during the September ESMO 2020 Annual Meeting that many oncologists around the world do not readily have access to this level of scans. If we want to take away that argument, decrying the point that access to equitable care is not the same as a biologic rationale for discussing druggable targets, here are further points regarding access to drugs.
If we claim that biliary tract cancers are enriched in targets, and these targets have drugs that do not increase OS and are not widely accessible across the country, what we should say is that there are identifiable molecular targets for which drugs exist that for the most part achieve stable disease, and that is much better than what we have for these difficult-to-treat cancers. Perspective on this issue may come down to a difference in expectations, but I offer consideration that it is the combination of molecular testing, randomized controlled clinical trials that prove efficacy that includes increased OS, and also reasonably widespread access to these drugs as criteria for designating any cancer as "target-rich." Utilizing this term excessively risks misinforming both patients as well as oncologists of outcomes that have not yet been achieved regarding these alterations.
To paraphrase a famous saying, some potential targets are more equal than others. One of the biggest white whales that has traditionally escaped drug targeting is the RAS family of proteins, which in their mutated form drive ~19 percent of all malignancies (Cancer Res 2020;80:2969-2974). Lack of efficacious targeting has not been for lack of trying and extensive research. The notion that KRAS is "undruggable" makes for a good headline or attractive statement in grants, but it's not that KRAS, or any molecular player, is absolutely "undruggable"-but it is very difficult to target, especially due to its biostructural composition and conformation. Despite its prevalence across cancers-or perhaps because of it-for good reason targeting this difficult-to-medicate class has remained a Holy Grail of molecular oncology.
A very recent example is publication and presentation at the 2020 ESMO Annual Meeting of inhibition of the G12C isoform of KRAS using the novel drug sotorasib (N Engl J Med 2020;383:1207-1217). In this Phase I trial, there were 129 patients with advanced solid tumors; the most common tumors in these patients were NSCLC (59), colorectal cancers (42), and various others. In NSCLC, objective response was reported in 32.2 percent; in CRC cases, this number was only 7.1 percent. Responses were reportedly also seen in patients with pancreatic, endometrial, appendiceal cancers, and melanoma (N Engl J Med 2020;383:1207-1217).
The formulation of such a small molecule inhibitor, at the biochemical level, is even more of a major achievement than any of the responses reported in this early trial. Yet the relatively low response rate-compared to what one would hope, or expect, for a major molecular driver-should not be exceedingly surprising. Strategically, blocking KRAS using a single-agent approach was not likely to be a valid method without blockage of one, or a number of, potential resistance mechanisms from this or corollary molecular pathways, or in concert with cytotoxic chemotherapies. The success of sotorasib and other KRAS-targeting small molecular inhibitors will be determined not by their ability to succeed in many types of cancers, but rather their ability to achieve a deep level of success demonstrated by biologic response rate accompanied by significant improvements in OS of patients with those types of cancers.
Genomics vs. Tumor Microenvironment
I will never pass up an opportunity to mention the role of cell biology and the tumor microenvironment that not only co-exists with genomics-driven biology of tumors, but also how these aspects are synergistic and far from mutually exclusive. Understanding cellular architecture and three-dimensional space of tumors in the complex and constantly changing matrix is a crucial yet underexplored component of discussions of tumor genomics.
Intra- and inter-tumoral heterogeneity affect accuracy of NGS. Spatially, gene expression varies throughout a tumor, at least in part due to stage of tumor evolution (e.g., earliest stages of carcinogenesis, versus localized, versus more highly invasive and post-treatment), and results of NGS are not yet being routinely interpreted through that lens. The field of spatial genomics is shedding new light on approaches through in situ analysis of tumors in their native microenvironment and promises to yield a higher level of accuracy than we have seen to date.
I speculate that more such studies will in some ways contradict results published to date in some tumor types and will call earlier forays into large-scale NGS into question. A prime example of a tumor type that will benefit most from this approach is pancreatic adenocarcinoma, in which the role of molecular targeting is questionable and not yet meaningfully achieved, in part, because of the role of dense desmoplastic stroma, questions of how it may be driven at the genomic level, and some remaining controversy about whether stroma-depleting strategies block tumor growth or actually enhance it (J Clin Invest 2020;130:4704-4709, Cancer Lett 2017;391:38-49).
Next Generation of Molecular Oncologists
My own career and understanding of molecular oncology have paralleled the rise of the field overall. As an internal medicine resident, I learned about care of patients with lung cancer with chemotherapy combination regimens in the mid-2000s. A few years later, I had a front-row seat as a fellow at Memorial Sloan Kettering Cancer Center during a time when it was one of the epicenters of the emergence of understandings of EGFR and the first generations of inhibitors of this target.
What has transpired in NSCLC alone has been an explosion of knowledge and many additional generations of TKI and immunotherapeutic approaches to treating NSCLC, which have now given hope and improved quality of life and OS to thousands of patients. This paradigm informs us that we can no longer practice oncology in isolation and without at least a basic understanding of the underlying molecular biology that drives cancer genesis and the evolution of drug resistance. Many of the terms previously used in cancer research labs are now well integrated into clinical jargon, and we now have a generation of clinical cancer biologists whose job description entails grasping, if not mastering, biologic principles as they apply directly to patient care.
In this context, as we teach ourselves to become the effective clinical cancer biologists that our profession now demands, how well are we teaching up-and-coming oncologists to practice in this brave new world (and yes, I did use that oft-used phrase, and yes, I do consider it relatively new considering it will guide our profession for the remainder of this century, and beyond)? Are our trainees in the oncologic sciences, both clinically and in the lab, learning the nuances of interpretation of NGS correctly, or is their expectation that NGS is the end goal, rather than a means to uncovering targets for individually tailored treatment options? And are we conveying the fact that identification of a putative target does not absolutely equate to a corresponding drug working effectively?
This is one of the great challenges that our field faces now and in the years to come: understanding what precision oncology is and what it is not, how accurate analysis is performed and most importantly, when to use the results to help our patients in daily practice. In this era of big data, better distribution of and access to genomic and transcriptomic data are key to making conclusions that are not biased even more by limited pools of tumors that may not be representative of the disease as a whole on a population level. Few institutions have both the patient volume and the means to generate big data on their own; thus cross-institutional collaboration is crucial.
My institution, the University of Minnesota, joined one such collaboration-the Caris Life Sciences Precision Oncology Alliance-in 2019. It is a collaborative network of leading cancer centers that demonstrate a commitment to precision medicine. As members, our institutions work together to support and create evidence-based data that establishes molecular testing as a key component in precision cancer care.
Participation has allowed our researchers to work with other researchers at over 40 cancer centers to access a combined dataset of over 200,000 tumors profiled using NGS and, more recently, in-depth transcriptomics. This collaboration amongst leading institutions and access to big data provides us with a better ability to identify biological patterns and true targets in a more real-world setting, which in turn should be leveraged to better design future clinical trials.
Where Are We Going?
We can ask where we should be going and what we should do differently based on current understanding of molecular targets and pathways. How can we better anticipate resistance mechanisms ahead of real time, leading to optimal design of trials that can then be destined to be the "game changers" that we all want to see?
We can be encouraged if we regard the past 1-2 decades as just the first stage of a long era that will likely deliver on the promise of precision oncology, but not always to the extent expected and hoped for. Key factors will be improved, like acceptance of the role of the tumor microenvironment and non-genomic factors in tumor biology that will affect response to targeted molecular therapies and also better translational understanding of targeted molecular pathways and putative resistance mechanisms, rather than considering targets in isolation. The future is promising, and it requires diligence, respect for the heterogeneity of cancer, and scientific discipline.
EMIL LOU, MD, PHD, FACP, is Associate Professor of Medicine at the University of Minnesota and Medical Director in the Clinical Trials Office-Solid Tumor Unit for the Masonic Cancer Center Division of Hematology, Oncology and Transplantation.