Within the last week, biotech investors were served a reminder that extreme volatility cuts both ways and that stocks don't only pop upwards over 100% as we saw with OSI Pharmaceuticals (NASDAQ:OSIP).

Genta (NASDAQ:GNTA) and Allos Therapeutics (NASDAQ:ALTH) have both seen staggering declines in their stock prices, 70% and 50% respectively, after having their drugs reviewed by the FDA's Oncologic Drugs Advisory Committee (ODAC) early last week. This is a board of experts that evaluates oncology drugs and makes recommendations to the FDA. The ODAC saw significant problems with the clinical data from these drug programs and as a result these drugs are not likely to be approved without additional clinical trials.

Successful biotech investing is not just a matter of picking the winners, but also dodging the landmines such as these. In Zeke Ashton's article Value Investing's 10 Commandments, he states that "Warren Buffett is fond of saying that the first rule of investing is 'Don't lose money.'" I am quoting Buffett in a biotech article to make value investors everywhere writhe in agony, but also because that simple premise of not losing money is critical to being a successful biotech investor. It will be very difficult to generate a successful track record by holding a number of biotech stocks that get cut in half even if the winners in the portfolio do exceptionally well.

Since it is so important to avoid high-risk situations where the probability of permanent capital loss is significant, I will use Genta and Allos as case studies to identify factors which can be used as warning signs that can be applied to other biotech investments.

Data dredging: a big no-no
Allos' drug RSR13 was examined in a phase 3 trial to determine if the drug, in combination with whole brain radiation, would have clinical benefits over radiation alone. Patients that have cancers originating in the breast or lung frequently have the tumors metastasize to the brain and these brain metastases are currently treated with radiation. The idea behind the development of RSR13 is to enhance the effectiveness of radiation therapy.

The phase 3 trial enrolled primarily patients with non-small cell lung (NSCLC), breast, and to a far lesser extent, other cancers with metastases in the brain. For all of the patients in the study, there was not a statistically significant increase in survival when RSR13 was added to radiation therapy. When NSCLC and breast cancer patients were considered as a combined group, there was still not a statistically significant increase in survival. It was only when just breast cancer patients were analyzed that a statistically significant increase in survival was seen with the use of RSR13.

Allos argues that RSR13 has a clear benefit in the treatment of brain metastases in patients with breast cancer as median survival was 8.7 months compared to 4.6 months in the patients that received radiation therapy alone. ODAC vigorously disputed this finding due to the possibility that this result is a false positive resulting from a subgroup analysis in a population that was not predefined before the study began. The briefing information and presentation slides prepared by Allos and ODAC are highly informative and it was interesting to see the vigor with which each side made their case as to whether or not this was a preplanned analysis.

After reading these documents, it should be very clear that data dredging is an issue that ODAC takes very seriously. This presentation on subgroup analyses in clinical trials makes it very clear that:

1.) Overall analyses are of primary interest with subgroups predefined in advance.

2.) Subgroup analyses are of secondary interest as "hypotheses generating" techniques for future studies. Unplanned subgroup analyses are considered exploratory. Data dredging is a concern.

3.) When unplanned, they will not ordinarily provide an adequate basis for definitive conclusions.

4.) "Any conclusion of treatment efficacy based solely on exploratory subgroup analyses are unlikely to be accepted."

Ultimately, according to ODAC, this was the problem with the Allos submission. The data looked like there was a benefit in breast cancer patients, but because of the way the analysis was done, the effect is not convincing and a confirmatory trial is required to support the efficacy claim.

The take home message to biotech investors is that no matter how good the data looks within a subpopulation, that data by itself is probably not going to be sufficient for approval. Extreme caution is warranted when a biotech company is touting a benefit in a specific patient group that was not observed in the overall population of patients in the trial.

Response rate not always enough
Genta's Genasense is an antisense drug that the company filed for approval in the treatment of metastatic melanoma. Metastatic melanoma is a cancer type with a poor prognosis and represents an unmet medical need. Dacarbazine (DTIC) is commonly used in the treatment of metastatic melanoma though the drug has limited efficacy and significant toxicity.

Genta filed for approval of Genasense in the treatment of metastatic melanoma based on data from a phase 3 clinical trial which showed that Genasense plus DTIC had a statistically significant improvement in tumor response rate (shrinkage) vs. DTIC alone. No improvement in overall survival was observed.

While improvement in survival remains the gold standard for approval of cancer drugs, according to the FDA,"a significant objective response rate with tolerable treatment toxicity may be adequate to support approval..." (emphasis mine.) Drugs such as AstraZeneca's (NYSE:AZN) Iressa and ImClone Systems' (NASDAQ:IMCL) Erbitux were approved in such a manner.

Given the response rate improvement with Genasense in combination with DTIC, the approval of other cancer drugs on response rate data, and the poor prognosis of patients with metastatic melanoma, there were reasons for believing that Genasense would get a positive recommendation from ODAC.

As it turned out, ODAC had several reservations about the Genasense data, including: complete responses reported by Genta that were not independently verified, increased toxicity in the Genasense arm without an increase in survival, and prognostic factors that were skewed in favor of the Genasense arm.

As is the case with Genta, a company can report clinical data that is suggestive a drug may have some benefit. However, the key point that can be learned here is that the data a company releases to the public cannot be taken at face value. Management should be pressed to answer tough questions regarding the balance of prognostic factors between trial arms, whether or not the responses have been independently verified, and if the methods in which the data was collected could bias the results.

Where do we go from here?
The Allos and Genta cases serve as reminders that extreme caution is warranted when investing in biotech. However, I firmly believe that biotech is a high-growth sector where carefully selected investments can do very well. While it would be nearly impossible to invest heavily in biotech without taking a few lumps from time to time, a few basic rules of thumb can be used to avoid the majority of catastrophic investments.

A very reasonable solution is to simply not hold a biotech stock going into meetings of FDA advisory committees or PDUFA dates. Even if the drug looks good going in, it's simply not worth the risk as many biotech stock blowups are tied to these events.

Another approach is to avoid small biotech companies that are going to sink or swim on the fate of a single drug. While these companies typically offer the largest rewards, the risks are also greatest. Investments in large biotech or pharmaceutical companies may not offer the return potential of the small caps but they are less volatile and less risky.

Share your thoughts on these -- and any other biotech issue -- on our Biotechnology discussion board.

Fool contributor Charly Travers does not own shares of any of the companies mentioned in this article. The Motley Fool is investors writing for investors.