A few weeks apart, Mr. Green and Mrs. Brown presented to their primary care doctors with intermittent rectal bleeding. Both were referred for colonoscopies, and each was found to have a colonic mass. With great trepidation, they awaited the results of the pathology and the CT scans that followed. Ultimately, both were diagnosed with stage-II colon cancers, confined to the wall of the colon without evidence of systemic spread. Filled with anxiety, they awaited their physicians’ recommendations for treatment.
Determining treatment plans for these two patients with stage-II colon cancer is increasingly complicated. Unlike in patients with stage-I cancers, for whom surgery is sufficient, or those with stage-III cancers, which have spread to regional lymph nodes and clearly benefit from systemic chemotherapy, patients with stage-II cancers face uncertainty: some seem to do well with surgery alone, but a subset develops recurrent disease that might have benefited from adjuvant chemotherapy. So far, we have lacked tools to differentiate these subtypes, and researchers are increasingly asking, can the genetics of individual tumors predict prognosis and guide treatment?
Now published in NEJM, a study by Dalerba and colleagues tackles this very question. The investigators describe a novel approach used to identify a tumor marker called CDX2 that might predict prognosis and determine which patients with stage-II colon cancer might benefit from adjuvant chemotherapy, rather than surgery alone.
Investigators reasoned that tumor cells that closely resemble mature colonic epithelial tissue would be less aggressive; by contrast, those tumors that have characteristics of more immature undifferentiated colonic stem cells might behave more aggressively and result in worse outcomes. So they set out to find markers of immature colonic tissue present in tumors. Using a novel bioinformatics approach, the researchers mined a database of more than 2,000 human colon gene-expression array experiments and identified a number of candidate genes, ultimately selecting the transcription factor CDX2. CDX2 is expressed in more mature cells and is not expressed in more primitive colon precursor cells. With this candidate biomarker identified, the investigators conducted a retrospective analysis using about 2200 colon cancer tissue microarrays from different databases to determine if there was an association between CDX2 expression and either survival outcomes or response to chemotherapy.
The results: the data demonstrate that the absence of CDX2 portends worse outcomes. In the discovery dataset, for patients with stage-II colon cancer with CDX2-negative tumors, the 5-year disease-free survival was 49%, as compared with 87% for those CDX2-positive tumors (p=0.003). The validation dataset yielded similar results. A separate analysis of those with CDX2-negative tumors suggests that treatment with adjuvant chemotherapy was associated with improved disease-free survival.
In an accompanying editorial, Drs. Boland and Goel of Baylor University remind us of the limitations of small, retrospective analyses, and they call for prospective trials to confirm these results. Still, they describe the fundamental importance of this study: “This work provides an opportunity for oncologists to move beyond what has been an inadequate method of selecting Stage II colon cancer patients for adjuvant chemotherapy.”
NEJM Deputy Editor Dr. Dan Longo agrees: “While these data might prove important for the treatment of patients with Stage II colon cancer, the implications of this study are broader – as they suggest novel ways that we might harness our improving understanding of tumor biology to personalize treatment and cure disease.”
So, for Mr. Green and Mrs. Brown, further testing to determine whether their tumors express CDX2 might help guide treatment options and might result in very different treatment plans for what was thought to be the same disease. To be sure, more data are needed, but this study points to the importance of this rapidly developing field of inquiry. As Boland and Goel conclude, “it is likely that a combination of genetic and epigenetic panels will soon help refine and advance the field of predictive oncology.”