I watched a good friend die a strange death from prostate cancer. He was diagnosed with the cancer in his late 60s, and already the cancer was throughout his body. In the hope that the cancer was hormone sensitive, he was treated by castration. And it worked dramatically: starved of testosterone, the cancer melted away. Unfortunately his spine was being held together by the cancer, and the disappearance of the cancer meant that his neck became completely unstable. Surgeons thought that there was some chance that they could stabilise his neck with a complex operation. My friend knew that he might not survive the operation, and he didn't.

I tell this story to make two points. First, I've been touched personally by the pain of losing a friend to prostate cancer. Second, a drug that caused the cancer to melt away would not have saved my friend.

So I thought of my friend as I read the front-page headline in yesterday's Times: "Cancer drug could save the lives of 10,000 a year". My immediate reaction was to think that this was appalling hype, and as I dug into the details I decided I was right.

The story follows on from publication of a very small study in the Journal of Clinical Oncology of testing the drug, abiraterone, in 21 patients. Nobody would have noticed this study, but Dr Johann S de Bono, one of the authors of the study and a consultant to Cougar Biotechnology, the manufacturers of the drug, held a "briefing". Journalists are very grateful for stories that allow them to fill Monday's papers, and the Times had enough material to fill two pages, including a story from a patient who had taken the drug and was photographed behind what looked like lilies.

Dr de Bono told journalists that the drug was "spectacularly active ... we believe we have made a major step forward in treating patients who have failed all other treatments". Recognising the need for some human interest, he added: "Within three months I have had men stop their morphine and say I'm going to see my daughter living in Australia."

But all that we have apart from De Bono's understandable enthusiasm is what's called a "phase 1 trial". Such a trial is designed simply to find out if patients can tolerate the drug, whether there are any side-effects that would stop patients from taking the drug, and what dose of the drug should be used in the large, controlled, randomised trials that are needed to determine for real whether a drug is effective and to get it onto the market so that it can be prescribed. These phase 1 trials are not designed to evaluate whether the drug is effective, but they do give some information on what happens to the patients.

The 21 patients were admitted into the trial between December 2005 and February 2007. The patients all had prostate cancer that did not respond to castration or treatment with anti-androgens. (So my friend could not have entered this study as he did respond to castration – rather too dramatically.) Patients who had other severe diseases – like heart failure – could not enter the trial, which is important because most patients with prostate cancer are elderly men and many do have other conditions.

Several of the men developed side-effects to the drugs – high blood pressure, low blood potassium, and swelling of the legs – but these side-effects could mostly be controlled with other drugs. One man developed a severe headache. Although these phase 1 trials look at safety, they are most unlikely to detect side-effects that may be very severe but occur only in, say, one in 500 patients. Such side-effects are often discovered later – meaning that a drug that looks promising in a phase 1 study never reaches the market.

Of the 21 patients, five are still being treated with abiraterone alone and seven with the drug plus dexamethasone, a strong steroid. So nine are presumably either dead or have stopped the treatment for some reason. Half of the patients had a 50% decline in their prostate specific antigen, a marker in the blood of the size and activity of the cancer. What this means to patients in reality is impossible to know: "I'm not interested in what some chemical is doing in my blood – I want to know whether I'm going to live or die and be in pain." Some of the patients showed some shrinkage of their cancer on X-ray examination, and of the 11 patients who had pain, eight were able to reduce the amount of painkillers they took or stop them altogether.

So we might legitimately conclude that there is some evidence that the drug will act against the cancer, but we are a very long way from being able to conclude that it will save 10,000 lives a year. The drug may never reach the market, and even if it does it could take many years and may prove impossibly expensive.

Years ago I used to be the BBC Breakfast Time doctor, and I had to struggle every week with how to present the results of these sorts of studies to an audience only half awake. My conclusion was that I did much more harm by hyping treatments than I did by being over cautious – because new treatments famously go through phases of being "a major breakthrough" to being "terrible poison" to ending up as "a treatment of limited effectiveness that has some use in some patients".