TCS Daily

It's the Treatment, Stupid

By Charles Murtaugh - February 25, 2002 12:00 AM

Not long ago, many biologists could get by on two rules of thumb. On the one hand, it was said, "the first time your experiment works, it's a mistake; the second time, it's an artifact; the third time, it's a finding." On the other hand, "if you need to use statistics to prove something, it's not worth proving."

The latter rule is a holdover from the heroic era of molecular biology, when the tools of physics and chemistry were brought to bear on the components of the cell. As the lens of biology pulls back, however, to take in entire tissues and organisms, the complexity of life itself blurs the clarity normally found in biochemistry, and statistics has assumed increasing importance in the life sciences.

Nowhere is this more true than in clinical research, where proving the efficacy of a new diagnostic tool or a new treatment often requires extraordinary mathematical rigor. Even techniques considered mainstream for decades can have their credibility undermined by statistical analysis, as illustrated by the current furor over routine mammograms. Although millions of women have for years undergone annual X-ray screening, recent studies raise the chilling possibility that these tests do not statistically reduce the risk of fatal breast cancer. Those studies have in turn been called into question, leaving this critical question unsettlingly, well, unsettled.

Medical research was not always thus. Medicine enjoyed its own heroic era, in the first half of the twentieth century, when previously fatal diseases could be treated and prevented for the first time. What particularly stands out about discoveries like penicillin and insulin is how obvious their effects were in "clinical trials," and how quickly they entered the medical mainstream. The latter case was particularly dramatic: Frederick Banting began experiments to isolate the hormone in April of 1921; in May of the following year, human tests of the newly-purified factor began; by 1923, when he shared the Nobel Prize in Medicine for his discovery, insulin treatment for diabetes "[had] come into use in practically all countries and with favourable results."

Comparing such breakthroughs to our current slow struggle with diseases like breast cancer, Parkinson's and Alzheimer's, it's hard to avoid being impatient. It's also easy to forget just how long and frustrating the road was even to these earlier cures and treatments.

For most of its history, medicine has been unable to affect the course of many diseases that it diagnoses. Diagnosis has been a tool to separate patients into those who have a hope of recovery and those who do not. Diabetes was diagnosable in classical Greece, and in the 17th century the idea first arose that it involved sugar metabolism. But even the realization, in the 19th century, that diabetes had some connection to the pancreas, had no immediate impact on the suffering and death of patients. Of course, this finding proved of utmost importance for the discovery of insulin some decades later, which has saved millions of lives; even insulin, however, remains only a treatment for diabetes, rather than a cure.

Nonetheless, even without the hope of a cure, patients will always want to know the nature of their affliction, if only to prepare for the worst. In the current controversy over mammographies, it is hard to disagree with oncologist Vincent Rajkumar, who told the New York Times, "The bottom line is that if you're still not sure whether it's good or not, it can't be that good. It can't be phenomenally effective." But even if a routine mammogram will not give a woman a statistical edge over her breast cancer, it does provide her with, for lack of a less overused term, empowerment -- perhaps not power to overcome the cancer, but at least to face up to the possibility of death.

No matter how the mammogram debate is resolved, it may prove to be a moot point, an artifact of outdated technology. On its own, mammography may not be "phenomenally effective," but new and better diagnostic approaches are continually entering the pipeline that promise to augment it. For instance, transcriptional profiling of a biopsy sample, taken after a positive mammogram result, may authoritatively distinguish an aggressive tumor from a more benign one, as well as suggest the best course of treatment, therefore giving the decades-old technique a new life-saving edge.

In a way, these newer approaches hearken back to the oldest function of medicine, to discern the treatable from the untreatable. Gene chip analysis of a breast tumor may indicate that it will respond to the anticancer drug tamoxifen, but it may also indicate that tamoxifen will be useless. Surely one factor in the mammogram debate is that we don't have a "magic bullet" for breast cancer, a safe and effective treatment that will cure all of the disease's manifestations. Even a poor diagnostic tool can be useful if it is paired with a robust cure.

If today's debate proves to have a silver lining, it will be in a renewed focus on treatment. Finding a treatment for a disease can be much harder than developing a diagnostic technique. Like diabetes, cancer was first diagnosed in antiquity, but cancer patients have nothing like insulin today. Biomedical researchers must ask for their forbearance for what promises to be, in the scheme of things, only a little while longer.

TCS Daily Archives