In other trials, Herceptin improved survival rates after three years, from 91.7 per cent to 94.3 per cent. So the reduction in the death rate is indeed the headline 33 per cent (from 8.3 per cent to 5.7 per cent). But it can also be expressed like this: Herceptin keeps about three people out of every 100 alive after three years-i.e., just 3 per cent.

Of course, that 3 per cent only applies to patients who have HER2 breast cancer. However, we know that only roughly a quarter of all breast cancer cases are HER2-positive. In other words, Herceptin is completely useless for 75 per cent of women with breast cancer.

On a more global analysis, therefore, the drug will help one quarter of 3 per cent of sufferers-i.e., less than one in a hundred women. As broadcaster, writer and breast cancer sufferer, Lisa Jardine, recently observed: "For women like myself, the new drug seems to promise a smallish reduction of an already lowish chance" of the disease recurring.

Earlier this year, Professor Peter Littlejohns, Clinical Director of NICE, analyzed the trial data, and calculated that 18 patients would have to be treated in order to prolong one life. "For every 100 suitable patients prescribed Herceptin," he wrote, "94 will have been exposed to the side effects without any benefit, at a cost of £400,000 per recurrence pre-vented" (Lancet Onc, 2006, 7: 22-3).

But, for other experts, even that puts too favourable gloss on Herceptin. In November 2005, Canadian biostatisticians delivered a bombshell. They too re-analyzed the Herceptin figures, and came to the conclusion they were phony. In a paper entitled "Randomized trials stopped early for benefit", their major criticism was that the original trials were just "interim analyses", and thus likely to show "implausibly large treatment effects". They heavily criticized the decision to stop the trials early, and recommended "clinicians should view the results of such trials with scepticism" (JAMA, 2005: 294: 2203-9).

Even The Lancet, a normally conservative medical journal, has been driven to condemn the whole Herceptin hype. "The best that can be said about Herceptin's efficacy and safety for the treatment of early breast cancer is that the available evidence is insufficient to make reliable judgments. It is profoundly misleading to suggest, even rhetorically, that the published data may be indicative of a cure for breast cancer", said a stinging editorial (Lancet 2005; 366: 1673).

Hype, hoopla and licensing
But surely Herceptin must be good for something, otherwise how would it have been licensed in the first place?

Let's look back at how it got onto the market. The drug was originally developed in the late 1990s for advanced breast cancer, when initial tests were begun with a few hundred patients-organized by the US manufacturers Genentech. It claimed significant "response rates" to Herceptin when the drug was combined with standard chemotherapy.

Two small-scale studies followed, in the first of which nine out of 37 patients with advanced breast cancer experienced "responses", although only for an average of about 8 months (Semin Oncol, 1999; 26S: 89-95). The second study used Herceptin alone, and found that five out of 43 patients "responded", although again the benefit was short-lived-about 5 months (Semin Oncol, 1999; 26S: 78-83).

Despite such modest results, press releases quoted the studies' authors as describing Herceptin as "a significant medical break-through".