“Oh no, not again!” That was the reaction of many observers of gene therapy last month, after researchers in Paris announced that a child given a pioneering treatment for severe combined immunodeficiency disease (SCID) had succumbed to a leukaemia-like disease (see Nature 419, 545; 200210.1038/419545a).

Just three years ago, the death of 18-year-old Jesse Gelsinger, in a trial of a gene therapy for a liver defect that causes a dangerous build-up of ammonia, threw the field into crisis — with good reason. Gelsinger died of an inflammatory reaction to the viral vector used to deliver the corrective gene, and serious faults were soon found in the way that the trial, at the University of Pennsylvania in Philadelphia, had been conducted. Patients had been inadequately informed of the potential risks, despite evidence of problems at high doses of the vector in animal experiments. And when signs of liver stress emerged in some patients, regulators weren't informed.

As these facts sunk in, questions began to be asked about the reporting of adverse events in other trials, and gene therapists embarked on a bout of soul-searching. Perhaps, some acknowledged, the field had been too eager to rush into the clinic.

The present case, in which a retroviral vector's site of insertion into the genome seems to have activated a cancer-causing gene, has come as a further blow. And given recent history, it's understandable that authorities in France, directly responsible for the affected trial, and the United States, where the Gelsinger debacle has left painful memories, should move quickly to suspend such trials.

However, a closer examination of the French case reveals important differences to the Gelsinger case that should convince regulators to proceed once more with SCID trials, albeit more cautiously.

First, there has been no suggestion that the investigators are at fault. Cancer triggered by 'insertional mutagenesis' was always recognized as a risk of gene therapy using retroviral vectors, and the parents of patients enrolled in the SCID trials were informed of this possibility. The researchers involved are now pursuing studies to investigate the risks facing the other patients they have treated (see page 116). And although a definitive risk assessment isn't possible, the results should be promptly communicated to regulators and the patients' parents.

Perhaps the most important difference to the Gelsinger case is that, whereas SCID gene therapy is a potential life-saver for children who otherwise face an extremely bleak outlook, the Gelsinger trial was a safety study in adults of a treatment designed for young children. Gelsinger and the other volunteers did not stand to gain any therapeutic benefit. In trials of drugs for life-threatening conditions, on the other hand, severe adverse events do sometimes occur, and may be tolerated if the benefits outweigh the risks. It is in this light that the current SCID setback should be viewed.

This is not to say that changes are unnecessary. Procedures for obtaining informed consent from the parents of future SCID gene-therapy patients must be adjusted to stress that cancer caused by insertional mutagenesis is now a tangible, rather than a theoretical, risk. Regulators must also reconsider the wisdom of using retroviral vectors, particularly where genes are introduced to mark populations of cells for study, rather than for their intrinsic therapeutic effects.

The challenge for gene therapists and regulators is to show that the field can respond appropriately to a serious adverse event in an otherwise successful clinical trial. It is unlikely to be the last: such setbacks are inherent to the development of new medical treatments.