Knowing that a risk exists is one thing. But knowing the odds is something very different.

Take gene therapy. In 2000, 11 baby "bubble boys" with x-linked severe combined immunodeficiency disorder (X-SCID) were treated by gene therapy in a clinical trial at the pediatric Hospital Necker Enfants Malade in Paris. The boys were treated with blood stem cells from their own bone marrow, which were modified ex vivo to contain the protein they were missing: a functioning interleukin-2 receptor gamma chain. At the time, the intervention was hailed as gene therapy's first stunning success. (see BioWorld Today, May 1, 2000.)

But three of the patients in the Necker trial developed leukemia. (See BioWorld Today, March 3, 2003, and March 7, 2005.)

It was clear before the trial that there was some risk of the gene therapy leading to the integration of the gamma-chain gene in such a way that normal gene functioning would be disrupted, including a risk of cancer. But the exact magnitude of the risk was vastly underrated.

"People had done calculations," Brian Sorrentino told BioWorld Today, "and arrived at the conclusion that the risk of the virus integrating into an oncogene was 1 in 100,000."

The Necker debacle left basic scientists with a number of questions - among them, why the boys developed cancer at such an unexpectedly high rate in the first place, whether the high cancer rate is unique to patients with X-SCID and how to better predict the safety of gene therapy approaches. In a paper published in the Aug. 1, 2006, issue of the Proceedings of the National Academy of Sciences, Sorrentino, who is the director of the division experimental hematology at St. Jude's Children's Research Hospital in Memphis, Tenn., and his colleagues presented a new mouse model that can help answer some of those questions.

The animals presented in PNAS lack the tumor suppressor gene Arf, as well as the gamma chain gene that is missing in X-SCID patients. When they were treated with gene therapy to reconstitute the gamma chain, they showed a high rate of leukemia.

Their clinical symptoms are similar to what was seen in the Necker trial, though the oncogenic insertions were seen in different places than had been the case in the patients. Sorrentino said the difference in insertion could be due to a number of factors, including the use of young adult unpurified bone marrow in the animal experiments vs. purified neonatal cells in the clinical trial, but that the difference should not affect the basic conclusions of the paper.

In the PNAS paper, the researchers used their mice to investigate whether treating immunodeficient animals with gene therapy carries special risks compared to other diseases that are good candidates for gene therapy treatment.

The researchers found that for a high rate of leukemia to occur after gamma chain gene therapy, mice needed to lack both the Arf tumor suppressor and the gamma chain genes - in other words, animals that did not have X-SCID in the first place were far less susceptible to the cancer-causing effects of gamma chain gene therapy.

Sorrentino said that his team concluded from the data that "there's something special about putting the gamma chain back when you have X-SCID. Our findings predict that the risk of getting tumors is far lower than what we've seen in X-SCID" for other diseases, such as sickle-cell anemia or beta thalassemia.

Why X-SCID should predispose to the development of cancer after gene therapy is not entirely clear yet; there are several possibilities, including the fact that a nonexistent immune system cannot carry out antitumor surveillance. But Sorrentino and his team noticed that there is an expanded pool of bone marrow progenitors. Sorrentino said that the accumulation of such primitive cells might be "an important factor" in the eventual development of leukemia.

The mechanisms are not entirely clear, but might be a simple numbers game: "If there are 10-fold more cells that can be transduced with the vector and repopulate the animal, there is a 10-fold increase of having that animal exposed to an insertion in an oncogene," Sorrentino said.

Aside from understanding exactly what went wrong in the Necker X-SCID trial, the team hopes that the new model will be useful in giving better risk estimates for different gene therapy methods.

The work "provides the field with an animal model that can test for safer vectors," Sorrentino said. Since the Necker trial, researchers have been working on a number of methods that might make gene therapy safer, including a move to different viral vectors, or even to do away with viral vectors altogether. "All of these ideas are attractive," Sorrentino said. "But if you don't have a model to test them, they remain theoretical [and] our results are really going to allow us to move forward in a practical sense."