Erbitux, for those who have managed to miss the headlines, is ImClone Systems Inc.'s medicine to fight colon cancer. By late last year, the company had advanced the drug to late Phase III clinical trials, the exquisite moment before a drug is approved for people — or tossed in the dumpster. However, the FDA said that last year's clinical trial was so poorly implemented, it could not establish the merits of Erbitux compared with other colon cancer treatments.

Information about all clinical trials is in a "black box." It has to be: Statisticians do not want the data tainted by hopeful doctors who may have a personal stake in the outcome. Months and even years may go by with no clear indication of the effectiveness of any drug during clinical trials. Sifting through data recorded largely on paper is a tedious process susceptible to many loose ends. So an Erbitux — which Bristol-Myers Squibb (BMS) paid $2 billion to share with ImClone — could happen to just about any pharmaceutical company.

The bad news for the company is that congressional and Securities and Exchange Commission (SEC) investigators have been poring over documents suggesting that ImClone and BMS may have had a few hints about missing or faulty data. Specifically, BMS knew there might not be enough clinical data on 11 of the 27 patients who had responded favorably to Erbitux. Both companies knew that excluding data from those 11 patients could have disappointed doctors and investors.

Controversy aside, one important question is whether a better system for managing clinical trial data could have proved conclusively — one way or the other — whether Erbitux worked. The stakes are obviously high. BMS shelled out at least $735 million on Erbitux R&D and did not get a clear signal about whether the compound worked. ImClone and BMS are now funding another clinical trial that is already under way.

—Mark D. Uehling