Session Item

The search for factors beyond radiation dose that could improve the identification of prostate cancer patients (PCa) more at risk of toxicity is of extreme interest for personalized RT aiming at ameliorating quality-of-life of all survivors.
To investigate the role that the intestinal microbiota (MB) might play in the development of RT-induced toxicity in the bowel, the MicroLearner study collected faecal samples from 136 (discovery) and 79 (validation) consecutive PCa pts treated with conventional (78Gy @2Gy/fr) or hypofractionated (65Gy @2.6Gy/fr) VMAT+IGRT in 5 fr/week before and after RT.

Gastrointestinal (GI) toxicity was assessed weakly during RT using CTCAE. Average grade > 1.3 over time points was used to identify a group of 16/136 (discovery) and 8/79 (validation) pts that suffered from acute form of GI toxicity. Operational Taxonomic Units abundances were extracted using 16S rRNA amplicon sequencing and Metagenomic 16S Thermo Fisher bioinformatics pipeline. A core MB at the genus level was extracted in the discovery cohort by selecting taxa found with a relative abundance greater or equal to 2% in at least 10% of the samples. Using the core MB, we developed a clinical decision-tree that evaluates the relative abundance of specific genera and tested its ability to predict the risk of toxicity both in internal and external (MARS population, Ferreira 2019) validation cohorts.

Hierarchical clustering of the core profiles (Fig. 1a) revealed 8 clusters of pts in the discovery cohort displaying toxicity rates ranging from 0 to 60% (Fig. 1b). The cluster with the highest rate (10 pts, 7% cohort, 38% toxicity population) was significantly enriched for toxicity (Fisher’s exact test P<0.005). The high-risk MB composition was translated in the decision tree (Fig. 2 top), which can assess the MB-related risk of a patient to develop toxicity. We validated the model using the internal validation cohort and the external MARS early cohort (Fig. 2 bottom). A higher rate of toxicity is observed in all cohorts for pts with high-risk MB.

We highlighted that the predisposition to GI toxicity requires an understanding of the gut MB both at the ecological and at the functional level. Our analysis suggests that an altered ion homeostasis at the level of bacterial communities might predispose pts to the onset of GI toxicity. The predictive features identified are the relative abundance of genera Faecalibacterium, Bacteroides, Parabacteroides, Alistipes, Prevotella and Phascolarctobaterium and seem to generalize prediction beyond sequencing technology and platforms and overcome potential bias associated with the ethnic origin of patients (at least for what concern western European countries).
The study opens up concrete prospects for the use of baseline faecal samples in the clinical practice to improve pre-RT toxicity risk assessment and RT planning optimization.
MicroLearner was funded under the Call for the Promotion of Institutional Research INT-year 2016, 5 × 1000 Italian Ministry of Health. MicBioRadio was funded by FRRB, ID 2721017