Gut Microbiota Role in Response to Checkpoint Inhibitor Treatment in Patients with Relapsed/Refractory B-Cell Hodgkin Lymphoma: The MICRO-Linf Study

Single-agent monoclonal antibodies targeting the immune checkpoint PD1 (programmed death 1) are an efficient and safe therapeutic option in patients with relapsed/refractory B-cell lymphoma. However, many patients progress or lose response to anti-PD1. Recent studies have highlighted the role of the gut microbiota (GM) in influencing the response to chemo-immunotherapeutic agents. Here we hypothesize that the GM dynamics in B-cell lymphoma patients during anti-PD1 therapy correlate with treatment response.

From December 2017 to December 2020, we enrolled 17 patients (12 with classical Hodgkin lymphoma [cHL] and 5 with primary mediastinal B-cell lymphoma [PMBCL]) treated with anti-PD1 due to relapsed/refractory disease. Feces were collected at baseline, before each therapy cycle, at response assessment (both during therapeutic course and at the end of treatment) and for grade >2 adverse events. All the samples were profiled through Illumina sequencing. At each time point, patients compiled a 7-day weighted food intake record that was analyzed by MètaDieta (METEDA).

We report the results of the first 6 patients enrolled, all affected by cHL. Of the six patients, five were females. The median age was 31 years (range 26-71). Five patients were refractory to the last therapy, with a median of previous treatments of 3 (range 3-5). All the six patients discontinued the chemo-immunotherapy. In particular, three patients were discontinued due to disease progression, two achieved a complete remission and consolidated the response with autologous stem cell transplantation and the last one discontinued due to a grade 3 adverse event, despite partial remission. The median number of anti-PD1 cycles was 15 (range 7-18).

The baseline GM was found to be distinct from that of age-/gender-matched healthy controls (HC). In particular, the Bray-Curtis dissimilarity index showed significant segregation between the two study groups (p value < 1×10 ^-4, PERMANOVA), as well as greater dispersion in the patient group. Regarding intra-individual diversity, although no significant differences were found with both metrics used (Inverse Simpson and Shannon index, Wilcoxon test), a slight decrease in patients compared to HC was observed. By analyzing the genus-level composition, compared to HC, the microbial ecosystem of patients was enriched in the pathobiont Collinsella while depleted of health-associated taxa, e.g., Faecalibacterium, Ruminococcus, Coprococcus and Roseburia (p value < 0.05; Figure 1A). When focusing the analysis on the GM trajectories along the checkpoint inhibitor treatment (Figure 1B), we found that intra-individual variability underwent cyclical fluctuations in responders, while values remained nearly constant in non-responders. Furthermore, significant differences were found between the GM structures of the two patient groups (responders vs non-responders), both at the level of dominant (weighted UniFrac, p value = 0.02) and subdominant (unweighted UniFrac, p value = 0.01) microbial components. The analysis of the questionnaires on eating habits filled in by the patients at each time point made it possible to estimate the daily consumption of the main macronutrients. Despite comparable energy intake and protein and fiber consumption, a greater lipid consumption and lower carbohydrate consumption were observed in responders than in non-responders.

In conclusion, the GM of patients affected by B-cell Hodgkin lymphoma showed some peculiarities compared to the HC microbiota, with a depletion of health-promoting microbial components, including producers of short-chain fatty acids (i.e., Faecalibacterium, Roseburia, Coprococcus and Ruminococcus). During therapy, a peculiar trend of GM response emerged in patients with different therapeutic outcomes. Beyond the different structures in terms of dominant and subdominant microbial components, responders showed greater biodiversity plasticity than non-responders. This difference possibly suggests a lower resilience of the disease state. Furthermore, the responder group showed a greater consumption of lipids and a lower intake of carbohydrates during therapy, opening interesting perspectives towards the development of integrated intervention strategies in this peculiar setting.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal