Conversation with Scott Kopetz: RAS in Colorectal Cancer, Initiation and Resistance

September 28, 2018, by Jim Hartley

How is colorectal cancer treated today? How is it diagnosed, what sorts of people do you see in your clinic, and what are the decision points when you make treatment decisions?

About half of patients, either at the time of diagnosis or during follow-up after initial surgical resection will develop metastatic disease.  The needs for colorectal cancer are better avenues to detect the tumor early so that surgery will have higher rate of cure, or failing that to have more effective therapies for established disease.  Our current treatments range from cytoxics (including the now 50-year-old 5-fluorouracil, along with oxaliplatin and irinotecan), to targeted therapies, mostly targeting the VEGF and EGFR pathway.  There are rare subsets that have a very different treatment approach.  In the 2-3% of patients with microsatellite instability-high [MSI-H] tumors, we utilize immunotherapy with inhibitors of PD1 +/- CTLA4.  For the 5% of patients with BRAF^V600E-mutated tumors, we use BRAF inhibitors combined with EGFR inhibitors which intercept adaptive feedback mechanisms.  And emerging data suggests reproducible benefit from dual HER2 pathway targeting in the setting of the 4% of patients with HER2 amplifications, although these are not yet on the clinical guidelines or FDA approved.  The remaining patients are defined by the presence of KRAS and NRAS mutations, which we collectively call RAS-mutated in the clinical literature. The big distinction in management between RAS-wild type and RAS mutated tumors is that we use EGFR inhibitors when the tumors are wild type RAS.  Recent data has also indicated that tumor location in the colorectum is also highly predictive of benefit to EGFR inhibition, with limited to no activity of EGFR inhibition in right-sided tumors despite lack of RAS or BRAF mutations.

How would you summarize the role of mutant RAS genes in colorectal cancer?

In early disease, meaning stages I - III, the presence of a KRAS or NRAS mutation has a modest negative prognostic effect.  However the magnitude of that effect is not sufficiently strong to define a subset that is clinically useful, so it is not usually tested or incorporated into our decision-making.  We have evaluated the utility of EGFR inhibitors in the treatment of stage III CRC, and have not shown a clinical benefit, regardless of the RAS status.  In a metastatic setting the guidelines now suggest that all tumors should be tested for KRAS, NRAS, and BRAF, as well as microsatellite instability.  The current KRAS and NRAS codons that are recommended to be tested are 12, 13, 59, 61, 117, and 146. But the reality in the clinic right now is that most patients are being tested as part of a next-generation sequencing panel, so the issue is less about insufficient codon coverage but management of patients with with the occasional atypical RAS mutations outside of well defined codons.

One other point I would make is that one of the key questions from a clinical oncologist's perspective is not only how do we develop novel therapies for tumors that have innate resistance to therapies because of RAS mutations, but also interest in the large number of patients who develop RAS mutations in secondary resistance (Strickler, et al., 2018, PMID 29196463).  We see this both with EGFR inhibitors as well as BRAF therapies.  That it is a very interesting and common phenomenon. It's intriguing that in many patients it's not just a single mutation in one codon of KRAS that appears, but we see a large number of different RAS mutations occurring in the same patient, where they develop multiple different resistant clones in parallel.  So a key questions from my perspective is whether these acquired RAS mutant clones are signaling in the same way as a tumor that developed a RAS mutation early in carcinogenesis, and are there things about the biological context of these mutations that may require different strategies to target?

What's new in clinical trials for colorectal cancer?

There has been a recognition in the field that the yield of genomic biomarkers will be lower in the future, meaning that the most common subgroups have increasingly been defined.  As an example, tumors with fusions are a subgroup of interest, including ROS, NTRK, ALK, and FGFR members.  However, these subgroups are very rare with total prevalence of <1%.  Thus there is increased interest in evaluating convergent biology through transcriptomic profiling.  Transcriptomic efforts to consolidate the various classifications systems resulted in the colorectal cancer "Consensus Molecular Subtypes" (Guinney, et al., 2015, PMID  26457759), which are now being used to select patients for clinical trials.

While adjuvant studies have been difficult to do in the past, another trend is the integration of circulating tumor DNA to define minimal residual disease in patients who have no evidence of disease by CT scans.  This promises to substantially accelerate drug development in early stage CRC, and may allow novel therapeutic targets to be effectively explored without the need for 1,000 patient studies as we have done in the past.

Finally, immunotherapy for colorectal cancer is an area of substantial interest like most other solid tumors, with over 2/3^rd of the current studies in refractory CRC utilizing an immunotherapy agent. Many of these studies are being done without strong preliminary data due to the poor representation of available immunocompetent preclinical models.  However, outside of the rare MSI-H tumors, activity has been minimal in CRC to date as highlighted by the recent failure of a Phase 3 study exploring dual MEK and PD1 inhibition (IMblaze370 trial).  Nevertheless, interest in the integration of targeted and immunotherapy agents is high, and a number of promising leads are being explored in the clinic.

Why are MSI-high colon cancers rare?  I thought that colon cancers had a pretty high mutational burden.

It's interesting that the mutational burden for MSI-high colorectal cancer is 100 fold higher than many other kinds of high mutation burden cancers, including MSI-stable colorectal cancer.  A high mutation burden is a double-edged sword, as it theoretically generates a higher rate of clones and therefore may have higher propensity to metastasize or be treatment resistant, but this same high mutation rate also generates a large number of neoantigens which may be targeted by the immune system.  We see a much higher rate of MSI in earlier stage cancers (15% to 20%), compared to metastatic disease where it is 2-3 %.  Our thinking is that this microsatellite instability probably causes a fairly robust immune surveillance in earlier stage cancers, resulting in immune clearance of any micrometastatic disease and low representation of MSI-H in a clinically evident metastatic population.  However, once these tumor have evaded the immune system and have established metastatic disease, the tumors are more aggressive and resistant to standard chemotherapy as would be predicted.

But most of the patients don't have a definite genomic signature that you can target?

Right.  While the field is continuing to find therapies that are beneficial, they tend to be for smaller and smaller populations.  RAS-mutant tumors are the “elephant in the room” for genomically defined tumors with the greatest clinical need. Because the RAS-mutant tumors tend to have a slightly worse prognosis and are resistant to EGFR inhibition, they tend to have shorter time on therapies and fewer treatment options.  The population of colorectal cancer patients with a RAS mutation is probably the largest population of patients going on phase I trials currently. This is a measure of the fact that these are patients who have exhausted all other standard therapies but who are still in good physical shape and are looking for novel treatments. By that measure, RAS-mutated CRC patients are probably one of the greatest unmet needs by pure patient numbers out there right now.

We've also recognized that even within a population with a given RAS mutation there is not homogeneous biology on the molecular level.  When we look at RNA profiling as one way to measure tumor biology, we see, for example, the BRAF mutation tumors are very homogenous, the microsatellite instability subgroup is very homogeneous, but KRAS and NRAS tumors tend to have a very wide pattern of biology.  So I think we're also recognizing that it's probably a little optimistic to view RAS-mutant tumors as a homogeneous basket.

Is there a consensus about why the RAS tumors are more heterogeneous? 

Some of this interpatient heterogeneity may be due to the underlying biology of the tumor, both other co-occurring mutations but also the epigenetic environment.  For example, there is a subset of KRAS-mutated tumors that are occurring with a hypermethylation phenotype on the right side of the colon which may have distinct biology.  Wonderful basic science studies are starting to illuminate this heterogeneity, for example highlighting subsets of RAS mutant tumors that remain dependent on RAS for proliferation once established and others where RAS may be dispensable.  I’m encouraged that these insights will help us dissect this heterogeneity in the clinic.  We would welcome opportunities to bridge the clinical and preclinical divide in this area.

Another interesting phenomenon is the intrapatient heterogeneity, especially in patients that acquire RAS mutations after progression on EGFR inhibitors.  Intriguingly, it appears from our clinical data that cells expressing mutant RAS seem to be less fit than cells that are wild type, because once the selective pressure of the EGFR inhibitor is removed, these mutant RAS clones regress as a proportion of the total tumor burden (Morelli, et al., 2015, PMID 25628445). Why are those RAS clones not as fit once you stop inhibiting EGFR signaling and how do we target this vulnerability? There are a lot of implications to those questions, and I think even if we don't have a RAS inhibitor ready for the clinic, there are ways we can better understand those RAS dynamics and design therapies that improve duration of benefit.

How do you organize your work week?

As a physician-scientist, my work is divided between the lab and the clinic. I help coordinate the clinical trial efforts for the NCI for colon cancer, as the chair of the Colon Cancer Task Force, which adds an administrative role but provides satisfaction in seeing great science being translated into the clinic.  My clinical responsibilities are in the management of patients with colorectal cancer in the clinic two days a week, which incorporates treating patients and getting them on clinical trials.  I also run a heavily translational lab focused on colorectal cancer therapeutics and biomarkers, which allows interrogation of patient samples from clinical trials and development of strategies for future clinical trials.

Is it difficult being an oncologist?

It's a very rewarding profession but it does require a balance between the hope for the future and the realities of the best treatments we have available now. We're buoyed by the hope and promise of what's coming.