New Tumor Classification Could Change Therapy for Many Patients

Alexander M. Castellino, PhD

August 07, 2014

It might be more appropriate to classify tumors on the basis of molecular signatures than on the basis of the tissue in which they originate, researchers from The Cancer Genome Atlas (TCGA) suggest.

This reclassification of tumors could change the approach to treatment, because molecular signatures are more indicative of the treatment tumors are likely to respond to, explain senior researchers Joshua M. Stuart, PhD, professor of biomolecular engineering at the University of Santa Cruz, California, and Charles Perou, MD, from the University of North Carolina at Chapel Hill, and colleagues.

Their research was published online August 7 in Cell.

Using their new system, the team reclassified 1 of 10 tumors in "clinically meaningful" ways, which "matters if you're one of those patients," Drs. Stuart and Perou told Medscape Medical News.

In fact, the 10% reclassification rate might be an underestimation because tumors from all tissues were not represented in the study, and some tissues were underrepresented. It is possible that up to 50% of tumors will be reclassified as more and more get analyzed, Dr. Stuart said.

The researchers told Medscape Medical News that they are in the process of expanding their dataset to include 22 different tumor types.

The Origins for Molecular Reclassification

The idea of classifying tumors on the basis of molecular features came from the TCGA Pan-Cancer Initiative. "In the TCGA project that is being undertaken across tumor types, we were struck with commonalities and differences that we saw," Dr. Stuart explained. "This led to the idea of doing a comprehensive analysis of molecular data across tumor types."

The researchers analyzed molecular data from more than 3500 tumors included in the Pan-Cancer 12 set — the first 12 tumor types profiles by the TCGA consortium. Data were generated in 6 different "platforms", including genomic platforms for DNA and RNA sequencing and a platform for protein expression.

Molecular Data Do Not Always Define Tissue of Origin

The data were analyzed on the each platform, and then on a combination of the 6 platforms. "The data defining the reclassification were based on an analysis of the molecular profiles across all the platforms," lead researcher Katherine A. Hoadley, PhD, from the University of North Carolina at Chapel Hill, told Medscape Medical News.

The molecular data generated defined commonalities and differences, regardless of whether they were analyzed on 1 platform or a combination of platforms.

The researchers note that for many tumors, molecular signatures or subtypes reflected the tissue from which they originated. However, "some tumors resisted categorization based on the tissue of origin," Dr. Stuart reported.

Bladder cancer was found to be very diverse, and could be molecularly classified as bladder-specific cancer, squamous cancer (such as that seen in the head, neck, and lung), or adenocarcinoma (such as that seen in some lung cancers).

In addition, the majority of squamous cell head and neck cancers, squamous cell lung cancers, and some bladder cancers converge on the basis of their molecular data.

The researchers believe that the divergence and convergence might be better explained by the cell of origin than by the tissue of origin, and that the classification of tumors on the basis of molecular data has ramifications for treatment decisions.

Implications for Treatment

Most current protocols for treating tumors are based on the tissue of origin. Perhaps when tumors have a molecular diversity, such as bladder cancer, treatment should be based on molecular analysis in addition to tissue of origin, Dr. Stuart suggested. "If you look at survival rates, the bladder cancers that clustered with other tumor types had a worse prognosis," he said in a press release.

Basal-like breast cancer was also found to be an anomaly, and looked much like squamous cell cancer or ovarian cancer.

"Although basal-like breast cancer originates in the breast, at the molecular level, it looks much like squamous carcinoma of the head and neck when examined on one platform, or like ovarian cancer when analyzed on another platform," said study researcher Christina Yau, PhD, from the Buck Institute of Research on Aging and the University of California at San Francisco. On the basis of these observations, basal-like breast cancer could be treated in much the same way as ovarian or squamous cell cancers, Drs. Hoadley and Yau told Medscape Medical News.

Follow-up studies are needed to validate these findings, but the study has laid the foundation for defining tumors on the basis of molecular data. Molecular subtyping is poised to offer patients treatment options.

The molecular reclassification of tumors will help redefine clinical trial design, several of the researchers told Medscape Medical News. In the near future, it is conceivable that patients will be enrolled in clinical studies on the basis of the molecular signature of their tumors, rather than the tissue of origin.

"We can now say what the telltale signatures of the subtypes are, so you can classify a patient's tumor just based on the gene-expression data or just based on mutation data, if that's what you have," Dr. Stuart said in a statement. Having a molecular map like this could help get a patient into the right clinical trial, he added.

That sentiment was echoed by senior researcher Christopher Benz, MD, from the Buck Institute. "Future iterations of it will fuel better clinical trial designs, whereby patients become eligible for novel therapeutics based on this type of genomic reclassification of tumors," he said.

The Future of Molecular Reclassification of Tumors

Although tumors arise in specific tissues, several factors contribute to tumor progression, such as cell-specific origin, tissue microenvironment, and immune response components. Future work will define how the molecular subtypes defined in this study will fit into data generated by others.

Data generated in the Pan-Cancer Initiative are available on the Synapse Web site. In the future, researchers can do similar analyses and compare their datasets with data from this study. "It will be insightful to know how other cancer types will converge with or diverge from the molecular subtypes defined in this study," Dr. Yau said.

The current study was restricted to primary tumors. Future work might define stage-specific and metastatic categories not assessed in this study. In addition, molecular classes defined from this analysis might converge or diverge when tumor metastasis is analyzed molecularly. This approach will provide a deeper understanding of the cells that populate metastatic sites.

For tissue types that have a high reclassification rate, such as in bladder cancer, it will be important to retrospectively re-examine clinical trial data and analyze outcomes using this new classification system, said Dr. Perou.

In the future, molecular profiling will be used to personalize tumor treatments. "It will ultimately provide the biologic foundation for the era of personalized cancer treatment that patients and clinicians eagerly await," according to Dr. Benz.

The authors have disclosed no relevant financial relationships.

Cell. Published online August 7, 2014. Abstract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.