Share this article on:

Gordon Mills: Progress in Molecular Diagnostics Depends on Academia-Industry Cooperation

Carlson, Robert H.

doi: 10.1097/


DENVER—The AACR's 4th International Conference on Molecular Diagnostics in Cancer Therapeutic Development here brought together a heterogeneous mix of about 200 basic and clinical researchers from industry and academia, biotech CEOs and department chairs, movers, shakers, and others in the arena of molecular medicine. (And even some reporters.)

An overview and key points from the meeting might best come from the words of conference chair Gordon B. Mills, MD, PhD, Chair of the Department of Systems Biology at the University of Texas MD Anderson Cancer Center.

AACR Molecular Diagnostics Conference

AACR Molecular Diagnostics Conference

To say Dr. Mills is plugged into this world is a vast understatement. His keynote lecture (prepared at the last minute when the scheduled speaker could not attend) detailed a score of challenges, opportunities, ongoing research, and, to keep it honest, failures of molecular diagnostics since the days of the first tests for hormone receptors in breast cancer.

Here are some of his lecture highlights, plus excerpts from an interview with OT before the meeting.

Back to Top | Article Outline

Conflicts of Interest

Dr. Mills said that in the past, organizers of basic science meetings were hesitant to invite industry representatives to present their data. “We now realize that they are a very important part of the process,” he said. In fact, more than half of the registrants for this meeting were from the biotech and pharmaceutical industries.

“The way forward with molecular diagnostics—and I'm happy to say this—is forging effective academic, public, industrial collaborations that are going to work together to help our patients. These are important, but most of the time people look at these as problems.

“I actually look at them as what we have to do, that we must figure out how to work with industry and control the conflict,” Dr. Mills said. “And trust me—I spend more time with our conflict-of-interest committee than I do on some of our research projects.”

Back to Top | Article Outline

Biomarkers: Prognostic vs Predictive

It's important to remember that the term “biomarker” has no intrinsic meaning—it should always have a modifier, Dr. Mills said. That is critical both in the way they are identified, and the way in which they are validated and implemented.

There are biomarkers for exposure, risk detection, prediction, etc., he said, and each would take a completely different process to develop.

Predisposition markers identify people who are at increased risk, or potentially decreased risk, for the development of cancer. “They certainly could direct patients to screening, or chemoprevention or lifestyle changes,” he said. All are incredibly important he said, emphasizing, though, that all are discovered and validated by different processes.

Prognostic markers represent the natural behavior of untreated disease. “These are virtually impossible to study today because we don't have samples from patients with untreated disease. It's also very difficult to distinguish prediction from prognosis—do they change the behavior without treatment, or do they change the behavior with our current treatments including surgery, radiation, and chemotherapy?”

Predictive markers identify patients likely to respond, but it is important to remember that these are a combination of sensitivity and resistance, and resistance tends to be dominant over sensitivity.

“So if you're thinking of how you would use a marker, or how you would design a drug trial, think about resistance first, and then markers of sensitivity,” he said.

Pharmacodynamic markers indicate whether the drug is hitting the target in the tumor, and could be used to personalize dose and schedule.

“Is a pharmacodynamic marker an early response marker?” he speculated. “Is the drug having the desired functional effect, not just changing a biomarker but really altering the behavior of the cells? This would be great to determine whether we continue or change a drug.”

The overall goal of molecular diagnostics and targeted therapies is to capitalize on the vulnerabilities that cancer cells attain, he said.

Back to Top | Article Outline

High Failure Rate

Despite the incredible enthusiasm that he and many others have had about molecular diagnostics, “we have had a far higher rate of failure than we would like in this particular field,” Dr. Mills said.

He noted that another speaker at the meeting had said that there are more than 800 drugs targeting important signaling molecules in trials or about to enter trials.

“We have an incredible tool at our hands, if we can determine the population of patients likely to benefit, and design appropriate trials to link the patient and the drug together. Because even when we have active drugs, they are only active in a subset of patients.

“Indeed, most patients, even with the marker, do not benefit,” Dr. Mills continued. “And further, in most cases, these therapies have so far been relatively short-term in their efficacy, so we have to do a better job if we are to have benefit from any of these targeted therapy drugs.”

Back to Top | Article Outline

New Orphan Diseases

Personalized medicine is making progress in diagnosing and treating tumors based on specific collections of molecular biomarkers. As more and more tumors are identified by discrete molecular signatures, cancer types will be divided into increasingly smaller subtypes.

It follows that some common cancers could be divided into “orphan” diseases. One result might be that pharmaceutical companies become less interested in developing drugs for those niche markets.

For example, Dr. Mills said, it's becoming clear there are many classes of breast cancer and each one of them is an independent disease.

“We must embrace that concept or we are doomed to failure. It is as unreasonable to combine luminal breast cancer and basal breast cancer in the same clinical trial as it is to combine ovarian cancer and breast cancer, yet we still do that.”

Indeed, he said, there is more in common between basal breast cancer and ovarian cancer than there is between basal breast cancer and luminal breast cancer.

The fact that these are truly different diseases comes with some problems, he said.

“I think breast cancer is now becoming an orphan disease,” he said. He pointed out that today, luminal A is treated with hormonal manipulation; luminal B with chemotherapy and hormonal manipulation; HER2 ER+ and HER2 basal are very different diseases in their behavior and probably in their therapy; basal can be split into BRCA1 and 2; and non-BRCA 1 and 2 can be treated with chemotherapy and PARP inhibitors having activity.

“Can there be too many classes? And how do we do a clinical trial if we start running out of patients to do the studies?” Dr. Mills said.

A way out of this challenge is to focus on the combination of biomarkers located in a specific pathway—what he called the “pathway-ness” of a tumor.

“If you look at a single member in a pathway, one molecule at a time as we used to do, I think we're doomed to failure in the same way as if we don't embrace the concept that many of these cancers are different diseases,” he said.

© 2010 Lippincott Williams & Wilkins, Inc.
Home  Clinical Resource Center
Current Issue       Search OT
Archives Get OT Enews
Blogs Email us!