Promise, problems for molecular medicine
From the January/February ACP Observer, copyright © 2004 by the American College of Physicians.
By Margie Patlak
Recently, a patient of family physician Peter Rabinowitz, MD, MPH, shared an unusual insight with him: She had learned through gene testing that her acetylator enzymes did a poor job of metabolizing different drugs and toxins.
The patient was convinced that she was more sensitive to chemicals in her environment than the average person. What, she wanted to know, should she do about the indoor air pollution problems in her apartment building?
Fortunately, Dr. Rabinowitz was familiar with acetylator enzymes, which metabolize environmental pollutants as well as common prescription medications, including hydralazine and the antiarrhythmic drug procainamide. But he was surprised that she had been tested—and that he was actually having a conversation with a patient about her acetylator status.
Such conversations, however, may soon become much more common. "I think there are going to be more and more patients who come up to you and say, 'I'm a slow acetylator' " said Dr. Rabinowitz, an assistant professor of medicine at Yale University in New Haven, Conn., who specializes in occupational and environmental medicine. "We'll need to know what that means."
Tests that will allow you to tailor dosages based on patients' genetic makeup are just one way that "molecular medicine" may make inroads into clinical practice.
A first-of-its-kind test of protein patterns for early detection of ovarian cancer in patients at high risk for the disease is about to hit the market. Other tests may soon dramatically improve the detection of other cancers and distinguish indolent cancers from lethal ones.
Additional tests in the pipeline may one day provide detailed profiles of patients' heart disease risk or forecast the effects of diet or pesticides on the health of individual patients.
Together, these breakthrough tests may help take the guesswork out of prescribing medications. They may also critically enhance physicians' ability to prevent, treat and screen for disease.
'The science is coming at us like a train roaring down the tracks, and it's going to roar right through clinical medicine.'
—Richard M. Weinshilboum, MD
"We're on the cusp of significant changes in the way we practice medicine," said Richard M. Weinshilboum, MD, an internist and clinical pharmacologist at Mayo Clinic in Rochester, Minn. Dr. Weinshilboum is an expert in pharmacogenomics, the study of how genomic differences affect individual variation in drug effect. "The science is coming at us like a train roaring down the tracks, and it's going to roar right through clinical medicine."
That "roaring" science, however, may hit some major roadblocks on the way to clinical practice. Although genomic and proteomic tests have produced dazzling preliminary results, they have yet to be put through vigorous clinical trials.
And some of the most promising applications—the ability to design drug treatments for individual patients—are on a collision course with the pharmaceutical industry, which thrives on blockbusters, not tailored treatments. Despite the scientific promise of genomics and proteomics, physicians' ability to adopt this brave new world to patient care will face some familiar, real-world obstacles.
Many of these new tests being developed are made possible by revolutionary new "microarray" technologies. (A microarray is made up of thousands of different genetic or protein sequences spotted onto discrete sites on a glass microscopic slide or silicon chip.) Microarray technologies allow automated machines to detect thousands of genes or proteins in a blood sample in less time than it takes to read this sentence.
The resulting "genomic" or "proteomic" fingerprints can reveal clinically inconspicuous diseases. With these sophisticated signatures, researchers say they are no longer forced to limit themselves to single-gene diseases that doctors rarely encounter. Instead, they can now venture into the new territory of common but genetically complex diseases, like cancer and heart disease.
The National Heart, Lung, and Blood Institute has made proteomics a major new research funding priority, creating 10 special centers in 2002 for research on proteomics and cardiovascular disease. While no tests are expected in the immediate future, proteomic studies have already identified more than 40 proteins linked to various cardiovascular diseases in animal models or human tissue studies.
In oncology, genomic and proteomic research is much more advanced. Physicians can already order a test that seeks out gene mutations in stool samples to screen for colorectal cancer. While these tests are less sensitive than colonoscopy, they are more sensitive than fecal occult blood testing, giving patients another non-invasive—though more expensive—screening alternative.
In prostate and breast cancers, proteomic tests may be particularly useful in distinguishing common benign cell overgrowths from similar malignant tumors. A research lab at the Eastern Virginia Medical School in Norfolk, Va., for instance, recently developed a proteomic test that, when fed serum samples from almost 400 men, identified all of those who had prostate cancer. The test's generated sensitivity was slightly better than that of the prostate-specific antigen (PSA) test, which some studies have shown has a sensitivity rate of 95%.
Even more impressively, the proteomic test yielded no false positives for men with normal prostates or benign prostatic hyperplasia. That 100% specificity far outstrips the 25% specificity of conventional PSA testing.
Researchers hope that a proteomic test combined with PSA could substantially reduce the number of men subjected to biopsies each year. Similarly, proteomic tests could whittle away the number of women who undergo breast biopsies because of suspicious mammograms. In initial studies, for instance, a serum proteomic test developed by researchers at Baltimore's Johns Hopkins University distinguished between women who had breast cancer and either normal controls or benign breast disease.
Once cancers develop, genomic or proteomic tests may be able to accurately pinpoint patients who need aggressive treatment, a distinction that isn't yet possible through standard pathological findings such as tumor size and grade. Eventually, researchers claim, genomic and proteomic testing may identify which women with breast cancer don't need chemotherapy, and which men with indolent prostate cancer can forgo the complications of surgery, radiation or hormonal therapy.
Ready for prime time?
Screening and diagnostic tests that target prostate and breast cancer are still years away. But one of the first proteomic tests to be offered outside the research setting is about to make its debut. The test—which detects early ovarian cancer and should be available this winter—will be offered by two national laboratories, Quest Diagnostics Inc. and Laboratory Corporation of America (LabCorp).
The test uses a high-resolution mass spectrometer to detect a protein fingerprint of ovarian cancer. Initial studies on a few hundred samples found the test had a remarkable 100% sensitivity and specificity.
Not only could it detect stage 1 ovarian cancer, but it could also distinguish ovarian cancer from more benign conditions, such as ovarian cysts. Further, it didn't confuse ovarian cancer with other types of cancers.
Sound too good to be true? It may be. The test has yet to undergo the scrutiny of large-scale clinical trials, although those trials are currently underway.
In the meantime, LabCorp and Quest Diagnostics are releasing their versions via a FDA "home brew" loophole that allows companies and institutions to offer screening or diagnostic tests to patients before FDA approval and extensive clinical trials. The only catch is that the companies can't sell the tests to other labs or clinics.
The home brew loophole applies to tests custom-made by individual laboratories. The lab must meet federal quality standards before the test can be used clinically, but it doesn't have to be shown safe and effective prior to use.
Most genetic tests, such as the BRCA tests used to determine women's susceptibility to breast cancer, are considered home brew tests. Genbank, an NIH-contracted resource for genetic tests, offers more than 1,000 genetic tests, only six of which have been brought to the FDA for approval.
Although the ovarian cancer detection test will be marketed to women at high risk for ovarian cancer, it's not clear whether insurers will pay for it before the FDA gives its blessing. Analysts point out that most insurers pay for the PSA screening test even though the FDA approved it primarily as a monitoring test.
However, many experts, such as oncologist Robert C. Bast Jr., FACP, of Houston's University of Texas M. D. Anderson Cancer Center, are already skeptical of the proteomic ovarian cancer test because it hasn't yet been extensively tested.
Dr. Bast was one of the researchers who developed a diagnostic test for CA [cancer antigen] 125, which—like the test about to come to market—seemed highly sensitive and specific for early detection of ovarian cancer in initial studies. The test proved, however, to be clinically inadequate for this purpose after additional larger trials were done.
A recent article in the Nov. 1, 2003, Lancet pointed out that small sample sizes are a major shortcoming in most genetic microarray research on classifying cancers. The use of such samples leads to "inflated, over-promising results," the authors wrote.
"It does give one pause to say, 'What are we concluding here?' " noted Emmanuel Petricoin, PhD, lead microbiologist at the FDA. (Working with researchers from the National Cancer Institute, Dr. Petricoin helped develop the new ovarian cancer test.) "With our own proteomic work, I always say it's not the first 50 samples that concern me, it's the next 50,000."
Another significant problem with much proteomic research, Dr. Petricoin added, is that results are highly sensitive to minor variations between the mass spectrometers that are used to detect protein patterns. (Test results can also be affected by minor changes in the way serum samples are prepared.)
Consequently, findings usually aren't duplicated when conducted on a different machine—or even over time on the same machine.
Dr. Petricoin's research group has tried to solve that problem by using a higher resolution mass spectrometer, which yields results that don't appear to fluctuate with variations in machines or sample preparation. As with all research, he noted, reproducible results are the key to the clinical utility of proteomic tests.
"It's not enough for us to get it to work in our lab," he said. "It has to work in every lab in the United States."
Drug therapy's promise, roadblocks
Pharmacogenomics is another genetic application that has the potential to revolutionize patient care, particularly for internists.
Preliminary results, for example, suggest that subtle but common genetic variations (polymorphisms) can accurately predict patients' responsiveness to statins or antihypertensives. Genomic screens may also identify which types of blood pressure drugs would be most beneficial to specific patients, as well as how individual patients might respond to certain chemotherapy drugs.
And genetic screens for drug metabolism might be particularly useful when treating patients for depression. With as many as 40% of patients not responding to initial treatment, experts believe that close to one-third of nonresponders to tricyclic antidepressants, as well as a sizable number of patients taking selective serotonin reuptake inhibitors, could be identified through currently available genetic blood tests.
One test product being developed by Roche Molecular Diagnostics may help predict, from a single small blood sample, how patients will respond to one-quarter of all therapeutic drugs.
Despite genotyping's potential, researchers have yet to validate its usefulness with prospective clinical studies. Once again, the companies and academic institutions marketing pharmacogenomic tests are using the "home brew" loophole.
Advocates for "pre-prescription genotyping" hope that future prospective trials will eventually pave the way for much wider use of—and coverage for—pharmacogenomics. Genotyping could help avert serious drug reactions, advocates say, and help contain soaring drug costs by identifying optimal treatments.
In fact, the FDA issued a guidance last November encouraging drug developers to conduct pharmacogenomic tests during drug development. "Using genomic testing to guide drug therapy," Janet Woodcock, MD, FDA's director of the Center for Drug Evaluation and Research, was quoted in an FDA press release, "will constitute a significant shift from the current practice of population-based treatment towards 'fine-tuning' individual therapy."
But the FDA guidance falls far short of mandating submission of pharmacogenomic data, unless a drug label specifically states how such data determine dosing or patient selection criteria. Genetic profiling advocates point out that drug companies have very little incentive to submit data that could potentially limit the number of people taking their drugs.
"The idea of stratifying patients based on genetic responsiveness to drugs represents a significant threat to the blockbuster drug mentality of the pharmaceutical industry," said Mayo Clinic's Dr. Weinshilboum.
It will also likely make physicians think twice about how to treat many patients. "You aren't going to be able to look a patient in the eye and just give him a drug when you know you can have a test that can distinguish whether he'll respond or not, or whether he'll develop an idiosyncratic toxicity or not," he explained.
Proponents of genotyping point to another major advantage of genetic profiles: Most of the drug-metabolizing enzymes featured in pharmacogenomic tests also metabolize environmental toxins. Patients who know their drug-metabolizing status will also know how they are likely to respond to certain pollutants or substances in their diet. That information could have profound implications for preventing disease.
"The environment and what you can do about it is more modifiable than the genome," pointed out Yale University's Dr. Rabinowitz. "As we become more fascinated by ways we can test patients, doctors will have to get more involved at looking at the occupational, environmental and dietary exposures of their patients."
Indeed, if molecular medicine becomes more mainstream, researchers say, preventive medicine is going to mean much more than just monitoring patients' cholesterol levels and advising them to quit smoking. It may involve counseling patients to use extra caution at their jobs because tests show their bodies can't safely process the toxins they encounter there, for example. Or you may be able to relieve some patients of the burden of adhering to a low-cholesterol diet because tests show their bodies process cholesterol in a benign fashion.
"We'll be able to do things that we never thought we'd be able to do," said Dr. Weinshilboum. "It's going to be fun."
Margie Patlak is a freelance science writer in Elkins Park, Pa.
Internist Archives Quick Links
What will you learn from your Annals Virtual Patient?
Annals Virtual Patients is a unique patient care simulator that mirrors real patient care decisions and consequences. CME Credit and MOC Points are available. Start off with a FREE sample case. Start your journey now.
Internal Medicine Meeting 2015 Live Simulcast!
Unable to attend the meeting this year? On Saturday, May 2, seven sessions will be streamed live from the meeting. Register for the simulcast and earn CME credit after watching each session. Watch it live or download for later viewing.