NEW YORK (GenomeWeb) – Researchers at Children's Mercy Kansas City found that more than half of 35 critically ill infants undergoing a rapid whole-genome sequencing test received a genetic diagnosis, and almost two-thirds of those diagnosed saw a change in their clinical management, suggesting clinical utility of the test for selected patients.
The results of the retrospective study, published online in The Lancet Respiratory Medicine yesterday, helped the scientists design a larger, ongoing prospective randomized study of their test, called STAT-seq, in infants in the intensive care unit. In addition, the scientists are currently assessing the assay's cost-effectiveness.
"What surprised me the most was the perception by clinicians of the impact that the diagnosis had on their clinical management," said Laurel Willig, assistant medical director of the Center for Pediatric Genomic Medicine at Children's Mercy and the lead author of the study.
STAT-seq employs rapid whole-genome sequencing and can be completed in as little as 50 hours. Three years ago, Children's Mercy published a pilot study of the test, and a year ago the researchers presented some early results of the current study at a conference.
For the study, the researchers recruited 35 patients younger than four months of age that were treated at Children's Mercy's neonatal or pediatric intensive care units between the fall of 2011 and 2014, as well as their parents. All children had a suspected monogenic disorder and STAT-seq was deemed likely to lead to a diagnosis. Patients and their parents had their genomes sequenced, and 32 of the patients also received standard clinical genetic testing, such as array CGH, FISH, or gene panel sequencing, as ordered by their physicians.
STAT-seq was performed in the hospital's CLIA-certified and CAP-accredited laboratory using the Illumina HiSeq 2500, either in rapid run mode (a 50-hour protocol) or in high-output mode (a 7-day protocol), depending on the severity of the patient's illness.
Variants were annotated using the institution's in-house RUNES software, and primarily interpreted using its VIKING software, which allows for the input of patients' clinical features. Likely causative variants were further interpreted by experts and confirmed by Sanger sequencing.
While the study was ongoing, the US Food and Drug Administration granted it non-significant risk status for reporting a provisional STAT-seq diagnosis prior to Sanger confirmation in exceptional cases, when such a result could change clinical management and the patient was likely to die quickly. The researchers made use of this exception for a single case, reporting a preliminary result after three days.
STAT-seq provided a diagnosis for 20 of the 35 infants, or 57 percent, and two of these diagnoses could also be made using standard genetic testing results.
Standard testing, on the other hand, only led to diagnoses in three out of 32 babies, or 9 percent. STAT-seq missed one of these, a structural variant that was discovered by a microarray-based test. According to Willig, the variant was actually present in the sequencing data but not called by the algorithm, and improving the ability of STAT-seq to identify large structural changes is a future goal.
One reason for the dismal performance of standard genetic testing might be that clinicians, knowing that STAT-seq was also being performed, did not order as many other tests as they would have done otherwise, according to the paper.
The high diagnostic yield of STAT-seq was likely related to the fact that the patients were highly selected, Willig said, and the yield might go down in the ongoing prospective study that involves a less selected patient population.
Overall, turnaround time for STAT-seq was quick, but nowhere near the 50 hours suggested in previous reports: the median time between enrollment and start of STAT-seq variant analysis was 5 days, and the median time to a confirmed diagnosis was 23 days.
According to Willig, Sanger confirmation alone added about a week to the turnaround time and might not always be necessary in the future when whole-genome sequencing becomes more established. Also, she said, the time required for data analysis and interpretation varied significantly between cases.
According to patients' doctors, 13 out of the 20 STAT-seq diagnoses, or 65 percent, were useful for their clinical management, for a variety of reasons that included changes in medication, changes in genetic counseling, and the decision to start palliative treatment.
An interesting finding of the study was that the mortality rate of infants who received a genetic diagnosis – by STAT-seq or standard testing – was higher than in babies with no diagnosis: 57 percent of those diagnosed died within 120 days, compared to 14 percent of those without a diagnosis. According to Willig, this result "highlights the need for a rapid diagnosis for some patients," as there seems to be only a narrow time window for interventions. "If we are to even have a chance, you have to make the diagnosis before they pass away," she said.
A large proportion of the deaths might not be preventable, with or without a diagnosis, because the patients have a lethal genetic condition. But after a diagnosis has been made, those patients can switch to palliative care and suffer less than they might otherwise, she said. In other cases, there might be treatment options available if a diagnosis is made quickly enough.
The study did not address the cost-effectiveness of STAT-seq, but the researchers are currently working on such an analysis for different patient groups, which they plan to publish separately, Willig said. The cost of the test was on the order of $30,000 per patient-parent trio, and it was important to sequence family trios because many of the diagnostic variants were de novo mutations.
Given the high cost of care in the ICU, it may be cost-effective to run such an expensive test because based on the diagnosis, patient may be able to leave the ICU earlier, either because it affects their treatment and makes them improve sooner, or because it suggests palliative treatment. "We expect a subset of patients for whom rapid diagnoses are important, and a subset of patients for whom a slower test would be as appropriate," Willig said.
The published results helped the researchers design their ongoing prospective randomized study, which is part of the National Institutes of Health-funded Newborn Sequencing In Genomic Medicine and Public Health, or NSIGHT, program.
The goal of that study is to assess STAT-seq in a broader group of critically sick infants, and it has broader inclusion criteria than the published study. Over four years, it plans to enroll 500 patients for STAT-seq and 500 in a control arm. About 30 families have been signed up so far, and diagnoses made for a few STAT-seq patients suggest a lower diagnostic rate than in the published study, Willig said.
Unlike the published study, the NSIGHT trial will also look at the psychosocial effects of diagnoses on parents and healthcare providers, she said.