SAN FRANCISCO-Routine KRAS testing in colon cancer patients received a major boost in January when ASCO issued a provisional clinical statement in favor of pretreatment genetic screening in metastatic disease. But KRAS is only the first step toward personalized medicine in colorectal cancer, according to reports presented at the 2009 Gastrointestinal Cancers Symposium. In fact, colon cancer treatment is headed in the same direction as breast cancer, with other genetic polymorphisms emerging as both predictive and prognostic.
ABSTRACT: Predictive and prognostic polymorphisms, such as transporter genes and BRAF mutations, make headway as personalized treatment options for colorectal cancer patients.
SAN FRANCISCO-Routine KRAS testing in colon cancer patients received a major boost in January when ASCO issued a provisional clinical statement in favor of pretreatment genetic screening in metastatic disease. But KRAS is only the first step toward personalized medicine in colorectal cancer, according to reports presented at the 2009 Gastrointestinal Cancers Symposium. In fact, colon cancer treatment is headed in the same direction as breast cancer, with other genetic polymorphisms emerging as both predictive and prognostic. At the symposium, Alan P. Venook, MD, offered an overview on how genetic markers will truly guide treatment planning. In addition, several researchers presented their latest work in the field.
In breast cancer, recurrence scores and pretreatment tests are calculated from tissue samples, said Dr. Venook, professor of clinical medicine at the University of California, San Francisco.
“Can we do the same in colorectal cancer? Would this help determine which stage II patients need chemotherapy?” Dr. Venook asked. “Not only would this optimize patient care but it would help us afford care in an era of escalating costs.”
Fifteen years ago, researchers established that stage II patients with 18q deletions “behaved more like stage III patients” while those with intact 18q almost never succumbed to their disease. Now this information is being used in a clinical trial: ECOG 5202 is stratifying patients (n = 3,125) according to presence or absence of 18q and microsatellite instability, and randomizing them to FOLFOX with or without bevacizumab (Avastin) or to observation.
“I emphasize that this is 15 years from the initial observation that deletion of 18q could be an important marker,” Dr. Venook noted. The 18q deletion will be but one factor in a predictive or prognostic genetic signature. “But we are not far from where our breast cancer colleagues are,” he said.
The importance of the KRAS oncogene, which is downstream of EGFR and is mutated in some 30% or more of colorectal tumors, has been confirmed as being predictive of response to EGFR antibodies (wild-type KRAS) or lack of response (mutated KRAS). Based on a strong body of data, the CALGB/SWOG 80405 trial that Dr. Venook is leading was amended to exclude patients with mutated KRAS, and future trials will follow this lead.
While KRAS is predictive, it does not appear to be prognostic; that is, it is not associated with outcomes independent of treatment. On the other hand, BRAF, the serine-threonine kinase that is the principal effector of KRAS, may be prognostic (see below, abstract 293).
Italian investigators recently reported that V600E mutations in BRAF preclude a response to panitumumab (Vectibix) or cetuximab (Erbitux) in patients with normal KRAS. In their study of 113 subjects, BRAF-mutated patients did not respond to treatment and had significantly shorter progression-free survival (PFS) and overall survival (OS) than patients with wild-type tumors. In other words, to obtain benefit from EGFR antibodies, both wild-type KRAS and wild-type BRAF may be necessary.
“Double-hit therapies aimed at simultaneous inhibition of EGFR and BRAF warrant exploration in colorectal cancer patients carrying the V600E oncogenic mutation,” the authors commented (J Clin Oncol 26:5668-5670, 2008).
Looking even further downstream, loss of the PTEN gene appears to protect cells from apoptosis and may someday serve as a predictive or prognostic marker, as could PI3 kinase. It has been estimated that loss of PTEN and mutations in PI3K may account for 10% of nonresponses to EGFR antibodies.
Beyond the tumor genotypes, the genomic DNA of the patient will also play a role in individualizing treatment, as it is related to treatment response and toxicity. “Pharmacogenetics interacts with pharmacokinetics and dosing, making for a very complex algorithm,” Dr. Venook said.
In the CALGB 80203 trial, benefit from oxaliplatin (Eloxatin) could be predicted based on certain transporter genes involved in oxaliplatin and irinotecan clearance. The problem, however, is that the most predictive gene, ABCG2 34 G>A, is rare, found in only four patients in the study, all of whom responded to treatment.
With irinotecan, a UGT1A1 promoter polymorphism has long been recognized as predicting for toxicity and lack of activity. “This information is not very useful either, however, as the data pertained to a regimen that is no longer used,” he pointed out.
One of the limitations of the field is that what applies to one dosing regimen does not apply to another, Dr. Venook said.
Current investigations are evaluating host polymorphisms that may predict for response to VEGF antibodies, and FCγR polymorphisms that may relate to cetuximab efficacy.
“Clearly, this is all just scratching the surface, and many other genetic polymorphisms or markers promise to be identified and incorporated into treatment planning,” he concluded.
In another report from the GI Symposium, the predictive and prognostic values of common point mutations were evaluated with respect to treatment response and clinical outcomes, and BRAF mutations were found to herald worse outcomes (abstract 293).
Th e cohort was 168 patients treated for metastatic colorectal cancer first-line with 5-FU-based regimens, primarily oxaliplatin and irinotecan, as well as bevacizumab in 58%.
Assays revealed that KRAS, BRAF, and PIK3CA mutations were present in 37%, 8%, and 15% of primary tumors, respectively. In a multivariate analysis, the BRAF V600E mutation (compared with wild-type BRAF) was an independent prognostic factor for decreased survival after first-line therapy, reported John Sougklakos, MD, PhD, of University Hospital, Heraklion, Greece. The study was done in collaboration with Dana Farber Cancer Institute.
In patients receiving oxaliplatin-based therapy, median PFS approached 12 months for patients with wild-type BRAF but was just 5 months for those with BRAF mutations. With an irinotecanbased regimen, PFS was 12.8 months and 3.5 months, and in regimens containing bevacizumab, 12.5 and 4.2 months, Dr. Sougklakos reported.
Median PFS was similar between patients with mutant or wild-type KRAS and PIK3CA. BRAF mutations were also associated with worse outcomes aft er salvage cetuximab.
“BRAF mutation in the primary tumor identifies patients who carry an especially poor prognosis, regardless of the specific treatment regimen,” Dr. Sougklakos said. “If our conclusions are independently confirmed, patients with BRAF V600E mutation might be justified in foregoing approved treatments in favor of investigational therapy.”
Testing for BRAF should be available for clinical use in about two years, according to Dr. Venook.
Understanding KRAS mutations
“Signaling downstream of KRAS is not as linear as we thought. There are clearly subgroups that handle KRAS mutations differently,” added Wendy De Roock, MD, who presented data from a team of Belgian investigators (abstract 289).
Their study evaluated dual specificity phosphatases (DUSPs) as markers of MEK/Erk activation in primary tumors. DUSPs inactivate P-Erk and are activated upon RTK signaling. Investigators studied their role in predicting outcome to cetuximab in chemorefractory patients, asking whether DUSPs contribute additional information beyond KRAS status. KRAS-mutated tumors were more likely than wild-type tumors to have increased levels of DUSP4 and 6a. Increased DUSP4 expression was correlated with improved OS and PFS in KRAS-mutated tumors; however, in wild-type tumors OS (but not PFS) was better in low DUSP4 expressors.V “DUSP4 level may be an indicator of differential levels of Erk activation in KRAS mutations. Th is seems to have a prognostic effect within the KRAS mutation and may be used to select patients for therapy with MEK/Erk inhibitors,” said Dr. De Roock of the University Hospital Gasthulsberg and Center for Human Genetics, Leuven.
Molecular markers differ by stage Adding to the complexity, an international group reported that molecular markers have stage-specific prognostic value. Their study included 1,400 patients with stage II or III colorectal cancer, yielding 36,000 slides or “half a ton of material,” said Arnaud Roth, MD, of Geneva University Hospital, Switzerland (abstract 288).
In an analysis of eight markers, only KRAS and BRAF did not vary in frequency per stage. Highly variable by stage, in terms of level of expression and in some cases prognostic value, were high microsatellite instability, p53, and thymidylate synthase expression. The findings suggest that “pooled analysis of stage II and III patients might be confounded by this heterogeneity,” Dr. Roth said.
Predicting drug-specific response
Finally, investigators from Duke University in Durham, N.C., used gene expression analysis to identify three gene clusters that were associated with differential responses to certain antitumor drugs (abstract 294).
Gene cluster 1 did not appear to have a distinct pattern of oncogenic pathway deregulation. Cluster 2 was characterized by increased deregulation of the Myc and PI3K pathways and cluster 3 by increased deregulation of the beta-catenin and Src pathways.
Cluster 2 was likely to be resistant to oxaliplatin and sensitive to irinotecan while Cluster 3 was more likely to be sensitive to oxaliplatin and resistant to irinotecan. A positive correlation was found between predicted oxaliplatin sensitivity and Src deregulation in Cluster 3.
This suggests that patients who are sensitive to oxaliplatin are also likely to respond to a Src inhibitor, and that this combination might be synergistic, according to David Hsu, MD, PhD.
“More importantly, this shows that gene expression profiling can be used to identify novel therapeutic combinations and bring us a step closer to personalized medicine.”
The symposium was jointly sponsored by ASCO, the American Society for Radiation Oncology (ASTRO), the Society for Surgical Oncology, and the American Gastroenterological Association.