University of California—San Diego School of Medicine, La Jolla, CA 92093; E-mail: tfriedmann@ ucsd. edu doi: 10.1016/mthe. 2003.1022 patient, treated in Australia by Ian Alexander with the same vector and protocol employed in the French study, has also shown a partial immune reconstitution. Sadly, in September 2002, approximately thirty months after treatment, one of the first children treated by the French group began to develop signs of T-cell leukemia. Although there was a family history of cancer—two close relatives had died in childhood of medulloblastoma—it was difficult to avoid the presumption that the genetic manipulation was the immediate trigger for the leukemia. Fischer and his colleagues quickly found that the leukemic cells consisted of a single, clonally expanded population of γδ T cells that contained a single copy of the retroviral provirus integrated in reverse transcriptional orientation in the first intron of the LMO-2 oncogene. LMO-2 is known to be associated with some forms of childhood T-cell leukemia. In this case, the integrated provirus had apparently introduced the strong retroviral long terminal repeat enhancer element sufficiently close to the LMO-2 gene to activate it to a level and time inappropriate for its stage in T-cell development. By mechanisms that are still not entirely understood, the combination of a potential growth-promoting function (γc) with an activated oncogene (LMO-2) in one of thousands of transduced cells may have provided that cell with a growth or survival advantage that led it, after almost three years, to outgrow all other T cells in the patient’s hematopoietic system and produce a clinically evident leukemia. To many observers, this terrible outcome seemed to confirm the long-held expectation that randomly integrating vectors could by chance hit upon or activate an oncogene and initiate tumorigenesis. Integration of retroviruses is generally thought to occur by stochastic mechanisms into fairly random or quasi-random sites. If that is so, a very high efficiency of initial transduction of a large enough population of target cells could virtually guarantee an integration event within enhancer-activation proximity to any one of the hundreds of known oncogenes or proto-oncogenes in the human genome. Furthermore, the integration event into a known oncogene in the first patient could have compounded the potential cancer predisposition in the patient’s genome, thereby representing a “second hit” in a multistep process of leukemogenesis. Because the precise mechanisms of leukemia induction in this child were still not understood, the French investigators and their regulatory agency suspended the study until more information could be gathered. In the United States, the FDA placed two X-SCID stud-