A new computational tool that incorporates viral evolution and epidemiology predicts an above-average outbreak of influenza A (H3N2) in the 2017 to 2018 season, but not as severe as last year’s.
The model, reported in an article published in the October 25 issue of Science Translational Medicine accurately predicted the severity of last year’s influenza season before the season started, researchers report.
The H3N2 strain is responsible for most influenza-associated morbidity and mortality in the United States. Outbreak sizes vary from year to year because of interactions between viral evolution, including changes in the hemagglutinin (HA) protein sequence and epidemiological characteristics, such as the percentage of a population that has immunity against circulating viral variants.
Therefore, Xiangjun Du, PhD, a postdoctoral researcher from the Department of Ecology and Evolution at the University of Chicago Medical Center in Illinois, and colleagues developed a computational model that combines surveillance data with new amino acid changes to predict influenza outbreak severity sooner than is currently possible using epidemiological factors alone.
“We predict interannual disease risk rather than finer-scale outbreak timing during the season; that is, we forecast whether the upcoming season will be anomalously large or small,” the researchers write. Their approach enables overall prediction in the summer, rather than early fall.
The researchers’ “novel evolutionary index” compared the amino acid sequences for the H3N2 HA epitopes, the viral protein that changes, using the sequence data from 2002 to 2011 to “train” the model and data from 2011 to 2016 to test the accuracy of its predictions. They developed both continuous (small genetic changes that do not warrant a new vaccine formulation) and cluster (larger genetic changes) models. Both approaches enabled the investigators to predict last year’s high number of cases.
To test the strategy, the researchers considered the geographic region encompassing the District of Columbia, Maryland, Delaware, Pennsylvania, Virginia, and West Virginia. It is a relatively small area with a high population and moderate climate.
“Forecasts based on the best cluster and continuous models capture both the interannual variation of the outbreaks and disease risk for this U.S. region,” the researchers write.
Earlier forecasts, such as during the summer, can help healthcare providers prepare for vaccination programs and an influx of cases.
The researchers conclude, “Overall, the fact that incorporation of pathogen evolution into epidemiological models increases forecasting skill should embolden future efforts to further improve on the model presented here.”
Limitations of the study include not considering age or social structure, focusing only on the HA protein, and not other viral parts, and on the H3N2 strain, as well as restricting geographic region to the United States.
In an interview the journal conducted, Mercedes Pascual, PhD, professor of ecology and evolution at the University of Chicago and senior author of the study, said, “We can produce our incidence forecast for the US in the summer before the transmission season begins. Our method is skillful for prediction of overall season incidence rather than for the peak timing of epidemics…. [W]e have produced a prediction for H3N2 influenza this coming season. We now have to wait and see how well we do.”
The authors have disclosed no relevant financial relationships.
Sci Transl Med. 2017;9:eaan5325. Abstract
For more news, join us on Facebook and Twitter
Tidak ada komentar:
Posting Komentar