By Caroline Helwick

San Francisco—Two studies presented at the 2013 American Society of Clinical Oncology Gastrointestinal Cancers Symposium demonstrated a role for molecular classification of tumor cells in the prognosis and treatment of patients with colorectal cancer (CRC).

“We have developed a diagnostic, single-sample predictor that allows the classification of CRC tumors of different intrinsic molecular subtypes,” said Josep Tabernero, MD, PhD, director of Vall d’Hebron Institute of Oncology in Barcelona, Spain, and lead investigator of one of the studies. “These subtypes could be clinically relevant as they differ in their underlying biology and clinical outcomes, and consequently require different treatment strategies.”

Molecular classification systems could be most helpful in managing patients with stage II CRC who have a variable risk for relapse that is hard to predict with current clinical and pathologic methods. The goal of subtyping is not only to identify patients who need aggressive treatment, but also to pair the right patient with the right drug.

Patterns of Gene Expression

Dr. Tabernero and his team used gene expression data taken from 188 patients with CRC (stages I-IV) to develop the classification system, which they subsequently validated in 543 patients with CRC (stages II-III).

Three distinct molecular patterns were identified: 21.5% of samples were categorized as subtype A (32 genes), 62% as subtype B (53 genes) and 16.5% as subtype C (102 genes). These subtypes differed according to three biological hallmarks of colorectal tumors: epithelial-to-mesenchymal transition (associated with aggressive tumors); deficiency in mismatch repair genes (associated with genetic alterations); and the rate of cellular proliferation—features that are known to independently affect outcomes, said Dr. Tabernero.

“The subtypes were significantly associated with prognosis and significantly correlated with benefit from adjuvant 5-FU-based treatment,” Dr. Tabernero said.

Patients with subtype A had a good prognosis, regardless of whether chemotherapy was given: 10-year survival rates were approximately 65% without chemotherapy and 80% with chemotherapy (P=0.183). Patients with subtype B had a pronounced benefit from chemotherapy, with 10-year survival rates of approximately 55% without chemotherapy versus 80% with chemotherapy (P=0.014). Patients with subtype C, in contrast, derived no benefit from adjuvant chemotherapy, with 10-year survival rates of approximately 65% without chemotherapy and 50% with chemotherapy (P=0.542). The five-year overall survival difference between types A and B, versus type C, was statistically significant (P=0.0166).

The findings suggest that patients with subtypes A and B would be treated differently: Subtype A patients would probably not require chemotherapy; subtype B patients would receive chemotherapy; and subtype C patients would require a more aggressive and perhaps novel treatment. The study researchers plan to evaluate these subtypes in patients who have received more current regimens, especially oxaliplatin.

Signaling Pathway Deregulation

Another study, presented by Joshua M. Uronis, PhD, found that deregulation of oncogenic signaling pathways can be used as a powerful tool in the classification of patients with CRC into molecular subgroups. The researchers, including David Hsu, MD, PhD, of the Duke Institute for Genome Sciences and Policy, Durham, N.C., have developed a preliminary set of biomarkers that are prognostic, predictive of treatment response and applicable to patients who have undergone resection for primary or metastatic tumors.

Dr. Uronis and his colleagues examined microarray data, based on the activity of 19 oncogenic pathways, from 850 patients with primary CRC and generated patterns of pathway deregulation for each patient’s tumor. A molecular profile of CRC was created that identified six molecular subgroups of CRC and showed that patients in each subgroup differed significantly in their risk for CRC recurrence. For example, although recurrence-free survival approached 100% for subgroup 4, it fell to about 40% for subgroup 3 (P=0.0004). The model was then applied to a data set of 133 tumor samples, of which 94 patients presented with metastatic disease and 34 with primary lesions. Again, the subgroups exhibited differences in recurrence-free survival rates (P=0.046), according to Dr. Uronis.

Additionally, researchers translated the unique patterns of pathway deregulation into gene expression signatures that were used to measure the probability of pathway activation in a panel of cell lines. With this approach, the model also was predictive of response to various drugs.

“We observed that molecular subgroups of CRC demonstrated differential sensitivity to specific targeted agents, such as molecular subgroup of CRC 1, 2 and 3 to inhibition of HER2 by lapatinib, and inhibition of the epidermal growth factor receptor by erlotinib [P<0.05],” Dr. Uronis said. “From this, we gather that we can make basic predictions using pathway signatures.”

The study was validated using a murine model of drug sensitivity that used patient-derived CRC explants implanted into mice. This model showed that subgroups with high mammalian target of rapamycin (mTOR) activity were highly sensitive to mTOR inhibitors and those with low mTOR expression were found to be resistant to these drugs. Researchers deduced that the combination of a genomic-based molecular profile of CRC and their preclinical murine model using patient-derived CRC explants can be used to develop a clinically relevant prognostic and predictive biomarker of CRC.

“There has been considerable interest in determining whether molecular alterations in primary CRC are more accurate prognostic indicators than a tumor’s pathologic stage, which is what is currently used,” noted William M. Grady, MD, associate professor of medicine and chief of the gastroenterology section at the University of Washington Medical Center, Seattle.

“[The researchers] have used a unique approach to identify molecular alterations that are predictive of recurrent cancer, not only for non-metastatic disease but also for metastatic disease,” Dr. Grady said of the study by Dr. Uronis. “If these results can be validated, these molecular alteration patterns have the potential to be used as markers to identify which patients should receive aggressive care after surgical resection of both primary and metastatic colorectal cancer and to direct the specific chemotherapeutic agents they should receive,” he said.

“The work by Uronis et al is critical for CRC care advancement, and symbolic of the new age of cancer care in general,” said Jennifer Tseng, MD, chief of surgical oncology and co-clinical director for surgery at Beth Israel Deaconess Medical Center, and associate professor at Harvard Medical School, both in Boston. “For example, we as surgeons can surgically remove tumors to the best of our oncologic ability, and enable our partners, the pathologists, to stage the tumors and help determine the best treatment; however, currently, there are serious limitations.

“At the extremes, there is less controversy,” she continued. “There, there is no question that chemotherapy does not (for stage I) or does (for stage III or IV) help the patient, overall. But the vast majority of patients fall into those shades of gray, stage II especially, where we are weighing potentially toxic chemotherapy risks and benefits versus disease recurrence risks and benefits. This type of study helps point us to personalizing cancer care for the individual tumor, and thus the individual patient,” Dr. Tseng concluded.