Molecular monitoring of adenovirus reactivation in faeces after haematopoietic stem-cell transplantation to predict systemic infection: a retrospective cohort study.

BACKGROUND: Faecal shedding of adenovirus following allogeneic haematopoietic stem-cell transplantation (HSCT) is an early sign of loss of immune control over adenovirus, but there is no consensus on the role of monitoring of faecal adenoviral load by serial testing. We investigated whether serial faecal PCR monitoring could predict the risk of adenoviraemia and survival outcomes after HSCT.

METHODS: We did a retrospective cohort study at the Royal Manchester Children's Hospital, Manchester, UK, of patients who had received their first allogeneic HSCT between Feb 1, 2003, and Sept 1, 2016, and adenovirus infection recorded in their medical records. We excluded patients who had received second or third transplants or autologous HSCT transplants. We obtained characteristics of patients and transplants, including mortality and adenoviral reactivation, from medical records and the hospital database. All patients had blood samples tested weekly for adenovirus by PCR until immunosuppression was stopped and CD3 T-cell count recovered to greater than 0·3 × 109 /L. Faecal PCR was done before transplantation in all patients, and after transplantation in patients who had diarrhoea, at the onset of symptoms and weekly thereafter until diarrhoea resolved. We analysed all samples available before and after HSCT. We did subgroup analyses for patients undergoing HSCT for cancer versus non-malignant conditions. We also assessed whether 5 log10 copies per g faeces was a suitable predictive threshold for adenoviraemia.

FINDINGS: We included 341 patients who had undergone a first allogeneic HSCT (median age 4·6 years, IQR 1·5-8·0, range 0-20·0). After HSCT, PCR was done in 4116 faecal samples from 293 (86%) patients who had diarrhoea and in 10 649 blood samples from 341 patients. Follow-up ended on July 14, 2017. 173 (59%) of 293 patients had adenovirus in faecal samples and 63 (18%) of 341 had adenovirus in blood samples. Maximum faecal viral load before adenoviraemia correlated significantly with maximum blood viral load (r=0·51, 95% CI 0·38-0·61, p<0·0001). Faecal adenoviral viral load greater than 5 log10 copies per g faeces was predictive of adenoviraemia (odds ratio 10·2, 95% CI 4·9-21·6, p<0·0001) with sensitivity 75·9% and specificity 74·8%. These values were increased further in patients with cancer, to 86·4% and 87·5%, respectively. Among the 28 patients who had positive faecal and blood samples and who had undergone serial faecal PCR monitoring after HSCT, the median time between reaching the faecal viral load threshold and onset of adenoviraemia was 8·0 days (IQR 2·3-21·8, 95% CI 4·0-16·0). Non-relapse mortality was not associated with adenovirus reactivation in faeces alone (9·2%, 95% CI 5·4-14·3 in patients without reactivation vs 7·8%, 3·8-13·7 in those with positive faeces only), but was significantly increased in patients who developed adenoviraemia (27·0%, 95% CI 16·7-38·4, p<0·0001).

INTERPRETATION: We identified a threshold faecal viral load that can predict the risk of adenoviraemia. Our findings support proliferation of adenovirus in the gastrointestinal tract before viraemia develops. Faecal PCR is suitable for early detection of children and young adults at risk of adenoviraemia, and its use might help reduce non-relapse mortality in allogeneic HSCT recipients.

FUNDING: None.