SAN FRANCISCO – Gene expression profiling of patients with colorectal cancer and pharmacogenomic modeling of patients with pancreatic cancer could help identify which patients will respond to particular chemotherapies, preliminary evidence from two small studies suggests.
Spanish investigators developed a molecular classification system using gene expression data from 188 patients with colorectal cancer to identify three subtypes of the cancer, which they called subtypes A, B and C. They then performed a multicenter validation study using tumor samples from 543 patients with stage II and III colorectal cancer, and classified 35% of samples as subtype A, 50% as subtype B, and 15% as subtype C.
They found that the different subtypes were associated with different clinical and biological characteristics and helped predict response to adjuvant chemotherapy regimens based on 5-fluorouracil (5-FU) when compared with 10 years of patient records, Dr. Josep Tabernero and his associates reported at the American Society of Clinical Oncology’s Gastrointestinal Cancers Symposium.
Patients with subtype C colorectal cancer had a mesenchymal gene expression phenotype and did not respond to 5-FU–based chemotherapy, suggesting that new, targeted treatments need to be found for these tumors, said Dr. Tabernero, director of clinical research at Vall d’Hebron Institute of Oncology, Barcelona.
Patients with subtype A colorectal cancer had the best prognosis – "some of them don’t even need adjuvant therapy," he said in a separate press briefing – and did respond to 5-FU–based chemotherapy. Patients with subtype B had a poor prognosis but showed an even more dramatic response to adjuvant chemotherapy.
Subtypes A and B had more proliferative and epithelial phenotypes than did subtype C. Tumors in subtype B showed a low frequency of overall kinase mutation, compared with subtype A or C.
Dr. Tabernero’s gene expression research follows previous work by others who successfully used gene expression patterns in breast cancer to create molecular subtypes that have been associated with prognosis and used to help create treatment plans. In colorectal cancer, previously only KRAS gene status has been shown to be a predictor of response to treatment focusing on epidermal growth factor receptor activity in advanced disease, he added.
A separate study is underway to validate the molecular subtype classification system in patients with stage IV colorectal cancer.
Dr. Kenneth H. Yu and his associates reported preliminary results of an ongoing, separate study that suggests pharmacogenomic profiling of 20 patients with nonresectable pancreatic adenocarcinomas predicted who would, or would not, respond to first-line chemotherapy.
The study so far has enrolled 50 patients with stage II-IV pancreatic adenocarcinoma who get blood samples taken before starting 1 of 12 different drug combinations chosen by their physician for first-line chemotherapy. If the tumor progresses on first-line treatment, another blood sample is taken before starting second-line chemotherapy. The blood is analyzed for pharmacogenomic characteristics in circulating tumor cells and invasive cells that might predict response or resistance to treatment.
The scientists analyzing the blood samples are blinded to treatment choices, and the treating physicians are blinded to the pharmacogenomic profile results of the blood samples, noted Dr. Yu of Memorial Sloan-Kettering Cancer Center, New York.
Twenty patients so far have had the two blood samples. Using a pharmacogenomic model created in laboratory work by CellPath Therapeutics, the investigators predicted that tumors in six patients would be sensitive to the first-line treatment received, six other patients in an intermediate group were predicted to have some response to the first-line treatment, and tumors in eight patients were predicted to be resistant to the first-line treatments received.
The median time to tumor progression supported these classifications – 7.3 months in the sensitive group, 5.3 months in the intermediate group, and 3.7 months in the resistant group, Dr. Yu reported.
Analyses of gene pathways seem to have identified other predictors of treatment response. "Genes do not exist in isolation. The proteins these genes encode interact with each other in pathways – you can think of them as networks of genes and gene products. Changes in one gene along the pathway can affect other genes," he said in a separate press briefing.
A longer response to treatment was seen in patients with disruptions in the pathway called E2F1, which is known for regulating cell division. A shorter response to treatment was seen in patients who had disruptions in the NFkB pathway, which has been implicated in regulating cell division and cell survival, particularly in cancer. The investigators also found disruptions in the phospholipase C and retinoblastoma 1 pathways when tumors became resistant to treatment.