Historically, these cost have increased over time above the rate of inflation, but there aren't more recent data available for pharmaceuticals, says Joseph DiMasi, a study author who is director of economic analysis at the Tufts Center for the Study of Drug Development.

Many drug makers say they routinely evaluate why a formulation didn't work, and genetic information offers another way of sleuthing.

"Does genomic information help us understand the cause of failure and give us any information about whether we can resurrect the project?" says Liam Ratcliffe, head of clinical quantitative and innovative medicine at Pfizer Inc. "We do this routinely."

Ratcliffe says this approach hasn't yet helped salvage any Pfizer products, though — an assessment echoed by some other big pharmaceutical companies.

Some industry experts say it is too time-consuming for big pharmaceutical makers to try to salvage drugs. They say it might be more promising to form a partnership with or sell development rights to smaller biotech or specialty pharmaceutical companies.

Such alliances have grown sharply in recent years. About $19 billion was spent on securing such partnerships in both 2006 and 2007, compared with less than $10 billion annually from 1999-2005, although the increase can't be attributed solely to targeted treatments, according to the Ernst & Young Global Biotechnology Center.

When the antidepressant vilazodone failed to show a significant benefit in a mid-stage trial its originator, Merck KGaA, decided to take its portfolio in a different direction.

The German company — which isn't affiliated with New Jersey-based Merck & Co. — sold the rights to Genaissance Pharmaceuticals, a company since acquired by Clinical Data Inc. of Newton, Mass. The drug had shown promise in some patients and had a good safety profile, and Clinical Data wanted to identify the patient population for whom it works best, according to Dr. Carol Reed, the company's chief medical officer.

It was a matter of figuring why certain patients responded and others didn't, Reed says. In September, Clinical Data said it found potential genetic markers that enhance the drug's efficacy in some patients, and announced this month that it would begin clinical trials for the second of its two critical late-stage studies. The company is "confident" it will be able to develop the biomarker test to identify the population segment most likely to respond, Chief Executive Drew Fromkin says.

Another drug close to being resurrected is bucindolol, which had been sidelined by other pharmaceutical companies when research on similar cardiovascular medicines appeared to show more promise. Rather than let the drug die, a company research scientist, Dr. Michael Bristow, bought the rights and formed a new company to develop it.