Skip to main content
Premium Trial:

Request an Annual Quote

Boolean-Based Computational Approach Identifies Colon Cancer Prognostic Marker


NEW YORK (GenomeWeb) – Using a boolean logic-based approach, researchers from the University of California, San Diego School of Medicine, Columbia University, and Stanford University found that CDX2, a diagnostic marker for colon cancer, could also mark high-risk patients and predict who would most likely benefit from chemotherapy.

According to the results of a retrospective study published last week in the New England Journal of Medicine, the researchers found that colon cancer patients with CDX2-negative tumors were less likely to remain disease-free up to five years post-surgery. They also found that stage II colon cancer patients with CDX2-negative tumors would most likely benefit from receiving adjuvant chemotherapy after surgery, perhaps the most significant finding of the study, Debashis Sahoo, an assistant professor of pediatrics, and computer science and engineering at UCSD and first author on the paper, told GenomeWeb.

Currently, stage II colon cancer patients do not receive chemotherapy after surgical removal of their tumors and often die because their tumors are very aggressive, he said. This study shows that there is a subset of patients — around 4 percent to 10 percent of the whole patient population — who would benefit from follow-up treatment. With CDX2 — short for caudal-type homeobox transcription factor 2 — "we now have a biomarker to tell the difference," Sahoo said.

Also, because CDX2 is used in the clinic, it has already surmounted at least one of the barriers that has stopped some known stem cell markers, hopefully making the path to clinical prognostic use a little less arduous. "It's really hard," Sahoo said. "There [are] already some markers that people have published [that] didn't make it to the clinic because when [tested] in large settings it didn't work or [because of] practical issues like ... the technology is too expensive."

To identify CDX2 as a potential prognostic marker, the researchers used a computational tool called BooleanNet to explore differences in gene expression patterns in samples from thousands of colon cancer patients. The basis for the study was the idea that tumors with lots of aggressive stem and progenitor-like cells result in bad patient outcomes, so identifying markers in these cells might indicate which patients were at higher risk of a poor outcome no matter what stage of the disease they were in, Sahoo said.

BooleanNet is described in a paper published in Genome Biology in 2008. The method, which was developed by Sahoo, then at Stanford University, and other colleagues at the institution, works by making Boolean implications — if-then relationships — from large quantities of data. As explained in Genome Biology, researchers first classify the expression level of each gene as low or high compared to a threshold set for each gene. They then make Boolean implications between pairs of genes — an example from the paper runs thus "if gene A's expression level is high, then gene B's expression level is almost always low." 

Sahoo and collaborators have used the method in at least one study aimed at finding markers of stem cell differentiation. In addition to the current NEJM study involving colon cancer, they've also applied it to gene expression data from other cancer types including bladder cancer, he said. But the method can be used to explore other kinds of data; for example, Sahoo is using it to explore mutations in genomic datasets as well as to study viral evolution, he said.

For the NEJM study, the researchers used BooleanNet to search for an unknown gene X that, if its expression was low, then the expression of a second gene called the activated leukocyte-cell adhesion molecule (ALCAM/CD166) — a cancer stem cell marker — is high. Specifically, they searched a database of information on over 2,300 human colon gene-expression array experiments from the National Center for Biotechnology Information's Gene Expression Omnibus. They turned up a list of 16 candidate genes including CDX2 ranked based on their expression levels. Compared to the other genes in the list, the low-high relationship between CDX2 and ALCAM showed up in almost all of the datasets that the researchers looked at, Sahoo said. Coupled with the fact that a clinical test already existed for CDX2, it made sense to start with that particular marker, he said. "But certainly, as people develop [new] antibodies and we have something as good as CDX2, we can test them."

Among other things, CDX2 regulates intestinal development and oncogenesis. Colon cancers that don't express CDX2 are associated with aggressive features such as advanced stage, poor differentiation, vascular invasion, and so on, according to the paper. "We reasoned that colon tumors lacking CDX2 would likely contain a higher number of stem-like cells, and would therefore be more aggressive than CDX2-positive tumors," Piero Dalerba, an assistant professor of pathology and cell biology at Columbia University and co-first author on NEJM paper, explained in a statement.

To explore the relationship between CDX2 expression and disease-free survival, the researchers analyzed data from 466 patients at varying stages of colon cancer — these patients were from the NCBI-GEO dataset. Their analysis showed that patients with CDX2-negative tumors were associated with poorer prognosis. Specifically, they found that the rate of five-year disease-free survival was lower — 41 percent — in the 32-patient cohort with CDX2-negative tumors than it was in the remaining 434-patient with CDX2-positive tumors, 74 percent of which survived five years disease-free.

To validate their findings, the researchers turned to an independent dataset of 366 colon cancer patients drawn this time from the National Cancer Institute's Cancer Diagnosis Program — 48 patients here were CDX2-negative while 318 patients were CDX2-positive. As in their previous analysis of the NCBI-GEO set, CDX2-negative tumors were associated with a worse five-year prognosis than CDX2-positive tumors — 48 percent of CDX2-negative patients survived longer than five years versus 71 percent of CDX2-positive cases. CDX2-negative patients also had worse overall survival and disease-specific survival rates compared to CDX2-positive patients — 33 percent versus 59 percent and 45 percent versus 72 percent, respectively.

The researchers also found evidence to show that treating stage II and III colon cancer patients with CDX2-negative tumors with chemotherapy following surgery could improve their survival. They analyzed data from a cohort of about 1,800 patients with stage II and III colon cancers, finding that 91 percent of CDX2-negative stage II colon cancer patients survived five years disease-free when they were treated with chemotherapy after surgery. In contrast, only 56 percent of CDX2-negative Stage II colon cancer patients who did not receive chemotherapy after surgery survived five years disease-free. The stage III subgroup also showed higher rates of survival for CDX2-negative cases; 74 percent of patients treated with chemotherapy post-surgery versus 37 percent with no chemotherapy.

The next step would be to test CDX2's prognostic efficacy in prospective clinical trials. That's not a task that Sahoo plans to take on himself — he's a computer scientist by training not a clinician — but he hopes that the NEJM paper spurs an independent group to action. "I think a prospective study should be independent ... and they need expert physician scientists who can dedicate their time to conduct the trial," he said. "I think there [will be] a lot of interest in using this because this will save a lot of lives."