Gene therapy used to treat children with no immune system could be far riskier than previously thought, a US study has suggested.
Rhys Evans underwent gene therapy in the UK in April 2002
The treatment is used for children with X-SCID - commonly known as "bubble boy" syndrome.
The latest animal study in Nature found the treatment itself can cause cancer.
But experts from Great Ormond Street, where nine children were successfully treated, said the study had used unnaturally high doses of the gene.
They said other studies showed the cancer risk was not present when lower doses were given.
The gene therapy does have a chequered past. A major study was halted after three children developed leukaemia.
And US research found the treatment could cause cancer through genes inserted into cells to replace faulty copies spreading and damaging healthy genes.
X-SCID is caused by mutations in the IL2RG gene, which governs the behaviour of a protein involved in the development of a number of immune system cells.
Without the protein, the cells cannot develop normally, and are unable to protect the body.
The gene therapy works by replacing a defective gene.
A sample of bone marrow cells is taken from the patient deliberately infecting them with a non-infectious retrovirus modified to carry the corrective gene.
The retrovirus is then used as a "vehicle" to insert the therapeutic DNA randomly into the cells' chromosomes.
Hopefully, when the cells are returned to the patient, they are able to initiate the creation of a functioning immune system.
In this latest research, a team from the Salk Institute for Biological Studies in La Jolla, California, looked at the long-term effect of infecting the IL2RG gene into mice.
They were injected with the gene - the same gene that is given in the treatment of X-SCID.
They studied for an average of 18 months - three times longer than in previous animal studies
A third of the animals developed lymphoma, with most doing so when they were about 10 months old.
Dr Niels-Bjarne Woods, one of the scientists who carried out the research, said: "We were surprised by the strength of the association.
"These results suggest that curing X-SCID by replacing IL2RG in the manner it is currently being done puts patients at an increased risk of developing cancer."
A earlier trial of the gene therapy in France had found nine out of 10 children were successfully cured of their condition.
But three were diagnosed with T-cell leukaemia, and the study was halted in 2002.
Two developed the disease because IL2RG inserted itself into the cellular genome next to a known cancer-causing gene and activated it, but the cause of the third cancer case had not been found.
The Salk Institute researchers say their work could provide the explanation.
Professor Inder Verma, who led the research, said: "The bottom line here is that if you replace a gene that has multiple effects, you have to know more about its regulation and its ability to affect other genes, and that requires extensive preclinical work and a much more careful analysis."
The gene therapy team at Great Ormond Street Hospital was highly critical of the study.
A spokeswoman said: "The paper in Nature is in our view, not helpful."
Professor Adrian Thrasher, who led the GOSH work, added: "This is a very preliminary study published in an incomplete form.
"The researchers have taken artificially high doses of these genes and given them to animals.
"I'm not sure how it is relevant to human treatment.
"We know already that, in lower doses, the gene therapy does not have that effect."