A major collaboration between multiple NIH-funded centers in the U.S. has identified and validated a gene transcriptional signature that can predict whether or not any one individual is likely to mount an effective antibody response to an annual influenza (flu) vaccination. The consortium’s analysis of data from multiple cohorts found that high baseline (i.e., before vaccination) expression levels of a signature of nine genes were associated with high antibody responses to flu vaccination, whereas low expression levels correlated with low antibody responses to vaccination. The positive correlation was relevant to young adult individuals who were under 35 years of age, whereas in individuals aged 60 years and older, genes that correlated positively with the antibody response in younger subjects appeared to suggest a poorer response.
Biological Mechanisms Underpinning Vaccination Repsonse
The results, which implicate the involvement of previously unreported genes and pathways, could help to identify molecular mechanisms that underpin successful responses to flu vaccination, as well as provide clues about how such biological mechanisms might change with age, the researchers suggest. “Understanding the mechanisms underlying these signatures will be the next step for the vaccine research community,” commented Purvesh Khatri, Ph.D., assistant professor at Stanford University’s Stanford Institute for Immunology, Transplantation and Infection, speaking with GEN. Prof. Khatri is a co-corresponding author for the researchers’ published report, which is released today in Science Immunology, in a paper entitled “Multicohort Analysis Reveals Baseline Transcriptional Predictors of Influenza Vaccination Responses.” The data, including all source code used to analyze the data, are being made freely available through ImmPort and also through ImmuneSpace. “With the availability of the code, we expect the community to further analyze these and other data as they become available in future,” Prof, Khatri stressed.
Flu is a major global health concern, and annual flu vaccinations are recommended for everyone over the age of six months. However, vaccination is nowhere near 100% effective. The U.S. CDC cited figures suggesting that the adjusted overall flu vaccine effectiveness (VE) was actually as low as 19% in the 2014–2015 season. Figures for the 2015–2016 and 2016–2017 seasons indicated an overall VE of 47%, and 42%, respectively.
Lack of flu vaccine effectiveness is due in part to the fact that not everyone who is vaccinated will mount an effective antibody response. Efficacy is particularly low among the over 65 age group, who are 20% less likely to seroconvert than young adults and sometimes don’t generate the necessary protective neutralizing antibodies. Also, we don’t yet have a universal flu vaccine, so each seasonal vaccine can only protect against a limited number of virus strains that are predicted to be the most prevalent during that season.
Previous attempts to identify gene expression signatures that might help to predict who will mount an effective antibody response to flu vaccination have been reported, but have been carried out in relatively small numbers of individuals. “Typical studies are a single cohort study where the discovery and validation samples are from the same cohort,” Prof. Khatri explained.
For the latest work the researchers examined data on pre- and post-vaccination antibody titers and gene expression data covering more than 32,000 genes from four Human Immunology Project Consortium (HIPC) cohorts. In addition, successful validation of the young signatures was performed using an independent cohort from the Trans-NIH Center for Human Immunology (CHI), an intramural human immunology research program within the NIH. In contrast with previous studies, the discovery and validation cohorts spanned a wide geographical spread across the U.S. and encompassed five consecutive vaccination seasons. “Our study used a multicohort analysis framework, which has been shown to successfully identify diagnostic and prognostic signature in a broad range of diseases,” Prof. Khatri commented to GEN. “The advantage of this multicohort analysis framework is that it is better able to account for the real-world biological and technical heterogeneity that typical single cohorts cannot.”
Collaborators from the HIPC-CHI Signatures Project Team and HIPC-I Consortium divided the cohorts into individuals who were under 35 years of age and those who were over 60 years of age. Their analyses showed that among the under 35 group, high pre-vaccination expression levels of the nine-gene transcriptional signature were predictive of higher vaccination responses. The lower the pre-vaccination expression of the gene signature, the lower the post-vaccination antibody response.
“In addition to analyzing each gene separately across multiple cohorts, we also analyzed groups of co-expressed genes (called modules) across these cohorts,” Prof. Khatri explained. “Because genes do not function in isolation, we explored predefined gene modules as predictors of vaccine response.” This analysis also confirmed the correlation between expression of the three module signatures—BCR signaling, platelet activation and inflammatory response, and antibody responses to flu vaccine—but again, only among the under 35 group.
Inverse Correlation Among Over-60s
Interestingly, while the researchers didn't find a significant link between gene transcription and low or high antibody responses among the older individuals, the nine-gene transcription signature did correlate somewhat with poorer antibody responses among the older individuals. “Across all cohorts, we found that baseline differences, either at the transcriptome level or module level, were inversely correlated between young and older participants,” Prof. Khatri added.
In other words, “genes that were positively associated with higher vaccination responses in young individuals tended to be negatively associated with higher vaccination responses in older individuals,” the authors write. “Thus, increased expression levels of our gene signature before vaccination were associated with better antibody responses in the young, but were inversely correlated with those in older individuals.… Although the presence of an inflammatory gene signature, for example, was associated with better antibody responses in young individuals, it was associated with worse responses in older individuals.”
So, could the transcriptional signature be used to screen individuals before vaccination for likely antibody response? “In theory, yes,” Dr. Khatri indicated to GEN. “But they have to be translated into a robust point-of-care assay that needs to be further validated. There is also a need to increase accuracy of the signature, because it has high sensitivity, but low specificity.”
Boosting Vaccination Response
The researchers do suggest that their findings might lead to new approaches to boosting vaccination responses among individuals predicted to respond poorly. “The discovery of these gene signatures offers the possibility of improving the vaccination response by modulating an individual’s immune state prior to vaccination,” said Steven Kleinstein, Ph.D., who is also a co-corresponding author for the published report. “Future research should be focused on developing ways so that potential nonresponders' transcriptome can be modulated to be in the state as defined by our signature prior to vaccination,” Prof. Khatri suggested.
John Tsang, Ph.D., co-corresponding author of the paper and chief of the Systems Genomics and Bioinformatics Unit, Laboratory of Systems Biology at the NIH, and associate director of the Trans-NIH Center for Human Immunology (CHI), pointed out that a causal association between the transcriptional signature and vaccine response does still need to be confirmed. “In this study we found blood trancriptomic states that correlate with a good antibody response,” he noted. “It remains to be determined whether these correlates are also causal, and if so, then finding ways to modulate/manipulate their levels/states could confer better vaccine responsiveness. However, if they are correlates that merely report on the status of other causal factors that are yet to be identified, then manipulating on the correlate itself would not lead to improvements in vaccination outcomes.”