Cancer's cure may be found by using the right diagnostic tree

The recent publication of a very promising phase 3 clinical trial of a small molecule, vemurafenib, for the treatment of metastatic melanoma ushers in the one of the first new drug treatments for this condition since the introduction of interferon-based therapy decades earlier (N Engl J Med. 2011;364:2507-16). Though probably not as effective as imatinib (Gleevec) for chronic myelogenous leukemia, this new drug provides a solid treatment foothold on a condition that has been relentlessly difficult to treat.

The vemurafenib example adds to a growing list of targeted cancer therapeutics with higher efficacies and often much lower toxicities than currently used medications. These advances have depended on a clear understanding of the molecular pathogenesis of cancer, rooted in knowledge of the genetic derangements underpinning each tumor type. This leads to the question of whether we should classify cancers by their tissue of origin, or their mutation of origin. Which diagnostic tree is the better example?

In this case, the molecular defect that vemurafenib targets is a mutated form of BRAF, a serine-threonine protein kinase that plays a role in cell signaling. From 40% to 60% of melanomas have defects in this pathway, and about 90% of those with defects harbor a mutation that results in a single amino acid substitution (V600E). Inhibition of the function of the mutant protein blocks cancer cell proliferation in vitro, and enhances overall survival and progression-free survival at six months over standard care in humans. It seems highly likely that testing for the BRAF mutation in cases of melanoma will become part of routine care.

Analogous to the story of imatinib, which has proven to be a treatment for not only chronic myelogenous leukemia but gastrointestinal stromal tumors, inhibition of BRAF V600E may be important for the treatment of other types of tumors; about 7% to 8% of other cancers harbor BRAF variations. A remarkable finding recently showed that 47 of 47 cases of hairy-cell leukemia (HCL) tested harbored the BRAF V600E mutation, and that vemurafenib inhibited kinase activity in HCL cells in vitro (N Engl J Med. 2011;364:2305-15). It seems likely that clinical trials are already being planned or are underway. A host of other cancers, such as papillary thyroid cancer, are known to harbor BRAF mutations; very likely this story will have additional chapters.

The BRAF V600E experience to date provides powerful support for the value of learning as much as possible about a diversity of cancer genomes as rapidly as possible. Efforts like the Cancer Genome Atlas are striving to do just that. Very likely, there will be other examples of novel therapeutics that can be brought to bear where few good treatment options have previously existed for tumors with similar mutational spectrums. Comprehensive catalogs of the mutations present in different tumors can be leveraged to ensure that the maximal mileage from costly drug development programs is achieved.

It is very interesting to consider how these types of discoveries may upend thinking about cancer classification. Does tissue of origin matter less than spectrum of mutation to prognosis and treatment?

Traditionally, cancers have been thought of by system or organ of origin (e.g., he has leukemia, she has breast cancer). This approach dates back to the Egyptians' and Greeks' first pictographic or written descriptions of tumors and where they arose on or in the human body. Little progress in classifying tumors was made beyond physical exam and gross anatomic observation until the end of the 19th century. At that time the cellular basis of cancer was firmly established, and the tumor origin could be refined by cell type. It became possible to determine that a liver lesion represented a metastatic melanoma rather than a hepatocellular carcinoma. Special staining in the form of immunohistochemistry came into use in the mid 20th century, further refining the ability to determine tissue of origin. Later, international guidelines for tumor classification were developed by the World Health Organization. Over the last several decades a succession of increasingly sophisticated molecular techniques such as expression profiling, including microRNA expression, have provided ever keener tools to refine tissue of origin. World Health Organization guidelines have been modified to reflect these advances, but remain organized around tissue of origin.

However, consider the extant overlap in BRAF mutations between melanoma and hairy-cell leukemia. It seems unlikely that, a priori, pathologists would have grouped these tumors together using any previous anatomic classification scheme. Perhaps the approach of classifying tumors by tissue of origin that's been taught to medical students over the last two millennium should be replaced by classification by mutation of origin. Such upending of classification schemes by DNA sequence has already rippled through organismal biology; perhaps biomedicine is next. The genome can be a very humbling teacher indeed.