RESUMEN
Immune Checkpoint Blockade (ICB) has revolutionized cancer treatment, however the mechanisms determining patient response remain poorly understood. Here, we used machine learning to predict ICB response from germline and somatic biomarkers and interpreted the learned model to uncover putative mechanisms driving superior outcomes. Patients with higher infiltration of T follicular helper cells had responses even in the presence of defects in the class-I Major Histocompatibility Complex (MHC-I). Further investigation uncovered different ICB responses in tumors when responses were reliant on MHC-I versus MHC-II neoantigens. Despite similar response rates, MHC-II reliant responses were associated with significantly longer durable clinical benefit (Discovery: Median OS=63.6 vs. 34.5 months P=0.0074; Validation: Median OS=37.5 vs. 33.1 months, P=0.040). Characteristics of the tumor immune microenvironment reflected MHC neoantigen reliance, and analysis of immune checkpoints revealed LAG3 as a potential target in MHC-II but not MHC-I reliant responses. This study highlights the value of interpretable machine learning models in elucidating the biological basis of therapy responses.
RESUMEN
Mitigation of enteric methane (CH4) emissions from ruminant livestock represents an opportunity to improve the sustainability, productivity, and profitability of beef and dairy production. Ruminal methanogenesis can be mitigated via two primary strategies: (1) alternative electron acceptors and (2) enzymatic inhibition of methanogenic pathways. The former utilizes the thermodynamic favorability of certain reactions such as nitrate/nitrite reduction to ammonia (NH3) while the latter targets specific enzymes using structural analogs of CH4 and methanogenic cofactors such as bromochloromethane (BCM). In this study, we investigated the effects of four additives and their combinations on CH4 production by rumen microbes in batch culture. Sodium nitrate (NaNO3), sodium sulfate (Na2SO4), and 3-nitro-1-propionate (3NPA) were included as thermodynamic inhibitors, whereas BCM was included as a enzymatic inhibitor. Individual additives were evaluated at three levels of inclusion in experiments 1 and 2. Highest level of each additive was used to determine the combined effect of NaNO3 + Na2SO4 (NS), NS + 3NPA (NSP), and NSP + BCM (NSPB) in experiments 3 and 4. Experimental diets were high, medium, and low forage diets (HF, MF, and LF, respectively) and consisted of alfalfa hay and a concentrate mix formulated to obtain the following forage to concentrate ratios: 70:30, 50:50, and 30:70, respectively. Diets with additives were placed in fermentation culture bottles and incubated in a water bath (39°C) for 6, 12, or 24h. Microbial DNA was extracted for 16S rRNA and ITS gene amplicon sequencing. In experiments 1 and 2, CH4 concentrations in control cultures decreased in the order of LF, MF, and HF diets, whereas in experiments 3 and 4, CH4 was highest in MF diet followed by HF and LF diets. Culture pH and NH3 in the control decreased in the order of HF, MF, to LF as expected. NaNO3 decreased (p < 0.001) CH4 and butyrate and increased acetate and propionate (p < 0.03 and 0.003, respectively). Cultures receiving NaNO3 had an enrichment of microorganisms capable of nitrate and nitrite reduction. 3NPA also decreased CH4 at 6h with no further decrease at 24 h (p < 0.001). BCM significantly inhibited methanogenesis regardless of inclusion levels as well as in the presence of the thermodynamic inhibitors (p < 0.001) while enriching succinate producers and assimilators as well as propionate producers (p adj < 0.05). However, individual inclusion of BCM decreased total short chain fatty acid (SCFA) concentrations (p < 0.002). Inhibition of methanogenesis with BCM individually and in combination with the other additives increased gaseous H2 concentrations (p < 0.001 individually and 0.028 in combination) while decreasing acetate to propionate ratio (p < 0.001). Only the cultures treated with BCM in combination with other additives significantly (padj < 0.05) decreased the abundance of Methanobrevibacter expressed as log fold change. Overall, the combination of thermodynamic and enzymatic inhibitors presented a promising effect on ruminal fermentation in-vitro, inhibiting methanogenesis while optimizing the other fermentation parameters such as pH, NH3, and SCFAs. Here, we provide a proof of concept that the combination of an electron acceptor and a methane analog may be exploited to improve microbial efficiency via methanogenesis inhibition.
RESUMEN
Immune Checkpoint Blockade (ICB) has revolutionized cancer treatment, however mechanisms determining patient response remain poorly understood. Here we used machine learning to predict ICB response from germline and somatic biomarkers and interpreted the learned model to uncover putative mechanisms driving superior outcomes. Patients with higher T follicular helper infiltrates were robust to defects in the class-I Major Histocompatibility Complex (MHC-I). Further investigation uncovered different ICB responses in MHC-I versus MHC-II neoantigen reliant tumors across patients. Despite similar response rates, MHC-II reliant responses were associated with significantly longer durable clinical benefit (Discovery: Median OS=63.6 vs. 34.5 months P=0.0074; Validation: Median OS=37.5 vs. 33.1 months, P=0.040). Characteristics of the tumor immune microenvironment reflected MHC neoantigen reliance, and analysis of immune checkpoints revealed LAG3 as a potential target in MHC-II but not MHC-I reliant responses. This study highlights the value of interpretable machine learning models in elucidating the biological basis of therapy responses.