BACKGROUND
Chemotherapy-induced cognitive impairment is highly heterogeneous, and as such, i) patients cannot be given accurate predictions about the neurocognitive risk of their treatment, and ii) cognitive symptoms are managed reactively.
Here, we compared the capability of multi-omics data to predict patients at risk of cognitive impairment and to understand the omic drivers of individual treatment response.
METHODS
Pre-chemotherapy saliva (N=20) and stool (N=20) samples of newly diagnosed patients with breast cancer, were subjected to multi-omic profiling of:
i) gut microbiota using 16S rRNA gene sequencing, and
ii) oral metabolome using untargeted metabolite acquisition in GC-MS
Multi-omic profiles were assessed using established in-house pipelines to compare patient cohorts, aggregated based on their development of cognitive impairment (FACT-Cog <106.6), following chemotherapy.
RESULTS
Beta diversity of the pre-chemotherapy gut microbiota as well as enrichment of the genus Sutterella, were associated with the development of cognitive impairment. Salivary metabolomic profiling also revealed strong predictive potential, with several molecules including 6-hydroxynicotinic acid, hypoxanthine and glycylproline decreased, and 5′-Methylthioadenosine increased in patients who later developed cognitive impairment (p<0.05, FDR corrected). Combining the top-ranked features of pre-therapy gut microbiota and salivary metabolome, each with strong predictive performance for cognitive impairment, achieved an AUC of 100% which, whilst unlikely to be sustained in larger, more heterogenous cohorts, it demonstrates the predictive potential of integrating multi-omic layers.
CONCLUSIONS
Profiling of pre-chemotherapy omic measures, such as the gut microbiota and salivary metabolome, presents as a minimally invasive tool to predict a patient’s risk of cognitive impairment. Integration of these data within a multi-parameter risk prediction model of cognitive impairment enhances accuracy. Confirmation of these results in a larger, more heterogenous cohort represents an important next step in being able to provide patients with clarity on their risk of neurocognitive symptoms, enabling the provision of precise and targeted supportive care.