No single diet works for everyone. Some people can slurp cabbage soup for a week and lose only a few ounces, while others on the same spartan regimen lose 10 pounds. But what if you could measure your metabolism and get a prescription for a customized diet?

Metabonomics may do just that. It is one of the latest offshoots of the "-omics" revolution--after genomics (genes) and proteomics (proteins). With the understanding that some diseases such as obesity are metabolic syndromes in which multiple biochemical pathways interact to cause complex symptoms, metabolic testing offers a way to gauge health over a lifetime. What is more, metabonomic technology might identify disorders before they produce symptoms. Such testing could help people choose diet and exercise regimens that are tailored to their individual metabolic states.

Alan J. Higgins of Icoria, a company based in Research Triangle Park, N.C., that is applying its metabonomic technology to human health, explains that "metabono-mics gives you the functional component" that is not always evident from a genetic or protein analysis. Changes in gene expression do not necessarily affect health, because the body's homeostatic mechanisms may compensate. Moreover, genes and proteins interact and only sometimes cause net changes in metabolic pathways. Metabonomics attempts to unify genomics and proteomics by examining an organism as a system. "We pick up those changes on the downstream end," Higgins adds.

At any given moment, the human body excretes thousands of metabolites that can be measured in urine, plasma and various body tissues. Conventional technology such as mass spectrometry and nuclear magnetic resonance can measure such components--that is how biochemists test the toxicity of drugs or environmental pollutants on human cells. The challenge, however, has been interpreting the reams of data generated. The bioinformatics boom has helped solve that problem. Scientists can now analyze metabolites in greater detail and also conduct more informative comparative studies. For example, three years ago London-based Metabometrix demonstrated that high-frequency radio waves bounced off a blood sample could identify atherosclerosis. The radio waves measure the sample's magnetic properties, and computer software generates a telltale pattern.

Profiling metabolic disease before symptoms appear may also be possible. Researchers at BG Medicine, based in Waltham, Mass., examined mice genetically engineered to develop atherosclerosis if placed on a high-fat diet. Scientists fed the mice a moderate-fat diet and after nine weeks measured lipid molecules in their livers and plasma. Compared with levels in a control group, certain lipid metabolites were elevated in the transgenic mice, even though they appeared perfectly healthy.

Of course, biochemical markers that flag disease, such as high cholesterol, already exist, but they are not sufficient. "A single biomarker gives information," says Jan van der Greef of BG Medicine, "but typically biomarker patterns are necessary to tell the complete story." Van der Greef suspects that many diseases have metabolic signatures that technology can detect even before a marker such as cholesterol would be elevated--the challenge is to identify the patterns. That is no small task: there is not yet a clear understanding of normal human metabolism, let alone abnormal metabolism.

Relatively speaking, "gene sequencing is so easy," says José M. Ordovas, director of the Nutrition and Genomics Laboratory at Tufts University. He notes that sequencers have to cope with only four components (A, C, T and G), whereas "in metabonomics you have different [technology] platforms that measure things in different ways. We are talking about thousands of components."

To move forward, scientists would like to see a human metabonome established--an equivalent of the human genome for metabolism. But the field lacks coordination and money, says Ordovas, who estimates that it might take analyses of half a million people or more to accomplish the task.

Ordovas is, however, inching his way ahead. In a joint project with Metabometrix, he is examining a few thousand subjects, some who have severe cases of metabolic diseases such as obesity and some who are healthy, to identify how extremes differ from the norm. He will also investigate whether diet and exercise tailored to a person's unique metabolic profile can bring weight down to normal and prevent a premature slide into bad health. In the future, instead of scanning food labels for calories, fats and carbs, we might be matching labels to our metabolic type.