Blog | Monday, June 11, 2012

10 newly defined molecular types of breast cancer, and a dream

Breast cancer is not one disease. We've understood this for decades. Still, and with few exceptions, knowledge of breast cancer genetics, information on tumor-driving DNA mutations within the malignant cells, has been lacking.

Most patients today get essentially primitive treatments like surgical hacking, or carving, traditional chemotherapy and radiation. Some doctors consider hormone therapy as targeted, and thereby modern and less toxic. I don't.

Until there's a way to prevent breast cancer, we need better ways to treat it. Which is why, upon reading the new paper in Nature on genetic patterns in breast cancer, I stayed up late, genuinely excited--as in thrilled, optimistic.

The research defined 10 molecular breast cancer subgroups. The distinct mutations and gene expression patterns confirm and suggest new targets for future, better therapy.

The work is an exquisite application of science in medicine. Nature lists 31 individuals and one multinational research group, METABRIC (Molecular Taxonomy of Breast Cancer International Consortium), as authors. The two correspondents, Drs. Carlos Caldas and Samuel Aparicio, are based at the University of Cambridge, in England, and the University of British Columbia in Vancouver, Canada. Given the vastness of the supporting data, such a roster seems appropriate, needed. The paper, strangely and for all its worth, didn't get much press.

Just to keep this in perspective, we're talking about human breast cancer. No mice.

The researchers examined nearly 2,000 breast cancer specimens for genetic aberrations, in two parts. First, they looked at inherited and acquired mutations in DNA extracted from tumors and, when available, from nearby, normal cells, in 997 cancer specimens, the "discovery set." They checked to see how the genetic changes (SNPs, CNAs and/or CNVs) correlated with gene expression "landscapes" by probing for nearly 29,000 RNAs. They found that both inherited and acquired mutations can influence breast cancer gene expression. Some effects of "driver" mutations take place on distant chromosomal elements, in what's called a trans effect; others happen nearby (cis).

Next, they honed in on 45 regions of DNA associated with outlying gene expression. This led the investigators to discover putative cancer-causing mutations (accessible in supplementary Tables 22 to 24, available here). The list includes genes that someone like me, who's been out of the research field for 10 years, might recall: PTEN, MYC, CDK3 and –4, and others. They discovered that 3 genes, PPP2R2A, MTAP and MAP2K4 are deleted in some breast cancer cases and may be causative. In particular, they suggest that loss of PPP2R2A may contribute to luminal BM breast cancer pathology. They find deletion of MAP2K4 in ER positive tumors, indicative of a possible tumor suppressor function for this gene in breast cancer.

The investigators looked for genetic "hotspots." They show these in Manhattan plots, among other cool graphs and hard figures, on abnormal gene copy numbers (CNAs) linked to big changes in gene expression.

Of interest to tumor immunologists (and everyone else, surely!), they located two regions in the T-cell receptor genes that might relate to immune responses in breast cancer. They delineated a part of chromosome 5, where deletions in basal-like tumors marked for changes in cell cycle, DNA repair and cell death-related genes.

They performed integrative cluster analyses and defined 10 distinct molecular breast cancer subtypes. The new categories of the disease, memorably labeled "IntClust 1-10," cross older pathology classifications (open-access: Supplementary Figure 31) and, it turns out, offer prognostic information based on long-term Kaplan-Meier analyses (Figure 5A in the paper: Supplementary Fig 34 and 35). Of note, here, and a bit scary for readers like me, is identification of an ER-positive group, "IntClust 2" with 11q13/14 mutations. This breast cancer genotype appears to carry a much lesser prognosis than most ER-positive cases.

Finally, in what's tantamount to a second report, the researchers probed a "validation set" of 995 additional breast cancer specimens. In a partially-shortened method, they checked to see if the same 10 molecular subtypes would emerge upon a clustering analysis of paired DNA mutations with expression profiles. What's more, the prognostic (survival) information held up in the confirmatory evaluation. Based on the mutations and gene expression patterns in each subgroup, there are implications for therapy. Wow!

I won't review the features of each type here for several reasons. These are preliminary findings, in the sense that it's a new report, albeit a model of what's a non-incremental published set of observations and analysis; it's early for patients, but not for investigators, to act on these findings. (Hopefully, this will not be the case in 2015, or sooner, preferably, for testing some pertinent drugs in at least a subset of the subgroups identified.)

Also, some of the methods these authors used came out in the past decade, after I stopped doing research. It would be hard for most doctors to fully appreciate the nuances, strengths and weaknesses of the study.

Most readers can't know how skeptical I was in the 1990s, when grant reviewers at the NCI seemed to believe that genetic info would be the cure-all for most and possibly all cancers. I don't think that's true, nor due most people involved with the Human Genome Project, anymore.

The Cancer Genome Atlas and Project should help in this regard, but they're young projects, larger in scope than this work, and don't necessarily integrate DNA changes with gene expression as do the investigators in this report.

What's clear, now, is that some cancers do respond, dramatically, to drugs that target specific mutations. Recently-incurable malignancies, like advanced melanoma and GI stromal tumors, can be treated now with pills, often with terrific responses.

Last night I wondered if, in a few years, some breast cancers might be treated without surgery. If we could do a biopsy, check for the molecular subtype, and give patients the right breast cancer tablets. Maybe we'd just give just a tad of chemo, later, to "mop up" any few remaining or residual or resistant cells. The primary chemotherapy might be a cocktail of drugs, by mouth. It might be like treating hepatitis C, or tuberculosis or AIDS. (Not that any of those are so easy.) But there'd be no lost breasts, no reconstruction, no lymphedema. Can you imagine?

Even if just 1 or 2 of these investigators' subgroups pans out and leads to effective, Gleevec-like drugs for breast cancer, that would be a dream. This can't happen soon enough.

With innovative trial strategies like I-SPY, it's possible that for patients with particular molecular subgroups could be directed to trials of small drugs targeting some of the pathways implicated already. The pace of clinical trials has been impossibly slow in this disease. We (and by this I mean pharmaceutical companies, and oncologists who run clinical trials, and maybe some of the breast cancer agencies with funds to spend) should be thinking fast, way ahead of this post.

And given that this is a blog, and not an ordinary medical publication or newspaper, I might say this: thank you, authors, for your work.

This post originally appeared at Medical Lessons, written by Elaine Schattner, ACP Member, a nonpracticing hematologist and oncologist who teaches at Weill Cornell Medical College, where she is a Clinical Associate Professor of Medicine. She shares her ideas on education, ethics in medicine, health care news and culture. Her views on medicine are informed by her past experiences in caring for patients, as a researcher in cancer immunology and as a patient who's had breast cancer.