The media and stock market are again atwitter with news of another supposed cancer breakthrough. Avastin (search), a drug developed by biotech giant Genentech (search), reportedly extended the median survival time of terminally ill colon cancer patients in a clinical trial by 4.7 months.

The news was formally released at last weekend’s meeting of the American Society of Clinical Oncology (search) (ASCO). The price of Genentech stock has risen by about 65 percent since mid-May when the news began leaking.

One biotech analyst said annual sales of Avastin could reach $2 billion as almost 150,000 Americans are diagnosed with colon cancer and 57,000 die from it every year.

And based on headlines such as the Los Angeles Times' “New Drug Combinations Effective on Colon Cancer,” the Avastin claims sound exciting.

But since cancer breakthrough news is usually more smoke than fire, a closer look is warranted -- especially since Avastin wasn’t effective in an earlier breast cancer trial.

It’s too bad a closer look isn’t possible.

Detailed information in standard study form about the Avastin trial isn’t available -- not from the Duke University Medical Center, whose news release claimed Avastin’s efficacy was “proved,” and not from Genentech, whose market value increased by $15 billion on the news.

Both were happy to be contacted about the study, no doubt expecting more giddy and gullible reporting. But no detailed write-up was available. The Genentech spokesperson couldn’t even say when a study might be published.

Only a brief study summary or abstract was available, one omitting or glossing over key information and basic questions about the trial.

The trial involved 925 patients. About 800 patients were either given Avastin plus a standard chemotherapy or the chemotherapy alone. Another 100 patients were given Avastin in combination with another standard chemotherapy.

The abstract only contains results for the 800-patient treatment group. What happened in the 100-patient treatment group? Was Avastin effective there, too? If so, why not report it?

Avastin reportedly increased survival time by almost five months. But that claim relies on the major assumption that at the beginning of the trial, patients in the Avastin group, on average, had a similar expected survival time as patients in the chemotherapy-alone group.

Individual study subjects, though, likely had different types of colon cancer and were at different stages in the progression of their colon cancers. If the Avastin treatment group was, on average, at an earlier stage of colon cancer or had less aggressive colon cancers and metastases, it wouldn’t be surprising that their survival time is longer.

The researchers apparently hoped that random assignment of subjects to the Avastin and non-Avastin treatment groups would equalize the expected survival times of the treatment groups at the trial’s beginning.

This may have worked, but we just don’t know. Without some information about, and validation of the assumption, the touted results are based on a huge leap of faith.

The trial was multicenter in nature, meaning that patients were treated at several locations around the country. Such decentralization may give rise to a phenomenon known as “multicenter bias,” where the results from one study center may be skewed because of some systematic difference in the conduct of its part of the trial.

We can’t tell whether multicenter bias occurred in the Avastin trial because the data weren’t reported by a study center.

I asked lead researcher Herbert Hurwitz whether the study was peer-reviewed. He said it was reviewed by a committee of ASCO -- the group putting on the medical conference where the results were announced.

But how could the committee perform a credible review with only the superficial abstract? No reputable journal would publish results without more. We must also wonder if the committee was truly objective since ASCO may have been eager to have such headline-grabbing claims announced at its annual meeting.

One final reason for requiring something more than “science by press conference” is that the biological mechanism by which Avastin is supposed to work, the blocking of blood vessels in tumors (anti-angiogenesis), hasn’t really panned out yet.

Based on studies in laboratory animals, anti-angiogenesis drugs were first touted several years ago in a front-page New York Times article that caused dramatic speculation in biotech stocks. But subsequent studies in humans have disappointed and biotech stock prices collapsed.

Given the huckster-ish history of anti-angiogenesis drug claims, Genentech and Hurwitz should realize more than a vague abstract is needed to show Avastin works. A detailed study would be a start, followed by more clinical trials.

Proven effective or not, terminally-ill cancer patients shouldn’t be denied Avastin or any other potentially helpful drug they are willing to try on their own dime. But cancer treatments shouldn’t be hailed until they are actually proven to work.

Steven Milloy is the publisher of JunkScience.com, an adjunct scholar at the Cato Institute and the author of Junk Science Judo: Self-defense Against Health Scares and Scams (Cato Institute, 2001).

Respond to the Writer