Pathway activation model for personalized prediction of drug synergy

Quang Thinh Trac,Yue Huang,Tom Erkers,Päivi Östling,Anna Bohlin,Albin Osterroos,Mattias Vesterlund,Rozbeh Jafari,Ioannis Siavelis,Helena Backvall,Santeri Kiviluoto,Lukas Orre,Mattias Rantalainen,Janne Lehtiö,Soren Lehmann,Olli Kallioniemi,Yudi Pawitan,Trung Nghia Vu

DOI: https://doi.org/10.1101/2024.06.05.597583

2024-06-08

Abstract:Targeted monotherapies for cancer often fail due to inherent or acquired drug resistance. By aiming at multiple targets simultaneously, drug combinations can produce synergistic interactions that increase drug effectiveness and reduce resistance. Computational models based on the integration of omics data have been used to identify synergistic combinations, but predicting drug synergy remains a challenge. Here, we introduce DIPx, an algorithm for personalized prediction of drug synergy based on biologically motivated tumor- and drug-specific pathway activation scores (PASs). We trained and validated DIPx in the AstraZeneca-Sanger (AZS) DREAM Challenge dataset using two separate test sets: Test Set 1 comprised the combinations already present in the training set, while Test Set 2 contained combinations absent from the training set, thus indicating the model's ability to handle novel combinations. The Spearman correlation coefficients between predicted and observed drug synergy were 0.50 (95% CI: 0.47--0.53) in Test Set 1 and 0.26 (95% CI: 0.22--0.30) in Test Set 2, compared to 0.38 (95% CI: 0.34--0.42) and 0.18 (95% CI: 0.16--0.20), respectively, for the best performing method in the Challenge. We show evidence that higher synergy is associated with higher functional interaction between the drug targets, and this functional interaction information is captured by PAS. We illustrate the use of PAS to provide a potential biological explanation in terms of activated pathways that mediate the synergistic effects of combined drugs. In summary, DIPx can be a useful tool for personalized prediction of drug synergy and exploration of activated pathways related to the effects of combined drugs.

Bioinformatics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is **predicting the synergy of drug combinations**, especially in the context of personalized medicine. Specifically, the authors developed an algorithm named DIPx (Drug synergy Interaction Prediction), which predicts the synergy of drug combinations based on tumor - specific and drug - specific Pathway Activation Scores (PAS). ### Background and Motivation 1. **Limitations of targeted monotherapy**: - Although targeted monotherapy (such as specific inhibitors) has potential in cancer treatment, it often fails or only achieves temporary remission due to inherent or acquired resistance. - In theory, by acting on multiple targets simultaneously, drug combinations can produce synergy, improve drug efficacy and reduce resistance. 2. **Deficiencies of existing methods**: - Traditional drug combination discovery relies on expert knowledge and known biological mechanisms, and this method is limited in discovering new combinations. - Although computer models have shown potential in personalized prediction of drug synergy, challenges still exist, such as differences in drug screening protocols, and limited overlap of test drugs and cell lines. - "Black - box" machine learning methods are difficult to explore the underlying molecular mechanisms driving synergy. ### Innovations of DIPx 1. **Pathway Activation Scores (PAS)**: - DIPx uses biologically - motivated Pathway Activation Scores (PAS), which provide potentially relevant information and help explain the biological mechanisms of synergistic combinations. - PAS values combine information from upstream, downstream and driver genes to train a random forest regression model. 2. **Model validation**: - DIPx was trained and validated on the AstraZeneca - Sanger (AZS) DREAM challenge dataset and evaluated using two independent test sets. - Test set 1 contains combinations already in the training set, and test set 2 contains new combinations not present in the training set. - Compared with the best - performing method TAJI, DIPx outperformed TAJI on both test sets. 3. **External validation**: - DIPx was also externally validated on the ONeil dataset, further demonstrating its generalization ability. ### Results and Conclusions 1. **Predictive performance**: - In test set 1 of the AZS dataset, the Spearman correlation coefficient of DIPx was 0.50 (95% CI: 0.47–0.53), and in test set 2 it was 0.26 (95% CI: 0.22–0.30). - Compared with TAJI, DIPx showed better performance in predicting new combinations. 2. **Inference of biological mechanisms**: - DIPx can infer potential biological mechanisms through PAS values, for example, the association between the Capivasertib + Sapitinib combination and the ERBB2 signaling pathway. - By analyzing feature importance and PAS values, DIPx can identify the pathways that contribute the most to the synergy of drug combinations. In conclusion, DIPx is a useful tool that can not only predict drug synergy in a personalized manner but also provide biological explanations for activated pathways, thus providing new ideas for the design and optimization of drug combinations.

Pathway activation model for personalized prediction of drug synergy

Integrative Pharmacogenomics Analysis of Patient-Derived Xenografts

From cell lines to cancer patients: personalized drug synergy prediction

SYNPRED: prediction of drug combination effects in cancer using different synergy metrics and ensemble learning

DTF: Deep Tensor Factorization for predicting anticancer drug synergy

Prediction of Synergistic Anti-Cancer Drug Combinations Based on Drug Target Network and Drug Induced Gene Expression Profiles

Anti-cancer Drug Synergy Prediction in Understudied Tissues using Transfer Learning

Efficient Parameter Estimation Enables the Prediction of Drug Response Using a Mechanistic Pan-Cancer Pathway Model

Interpretable Drug Synergy Prediction with Graph Neural Networks for Human-AI Collaboration in Healthcare

Deep learning-based multi-drug synergy prediction model for individually tailored anti-cancer therapies

SynergyX: a multi-modality mutual attention network for interpretable drug synergy prediction

PAIRWISE: Deep Learning-based Prediction of Effective Personalized Drug Combinations in Cancer

An In Silico Method for Predicting Drug Synergy Based on Multitask Learning

DeepSynergy: predicting anti-cancer drug synergy with Deep Learning

DKPE-GraphSYN: a drug synergy prediction model based on joint dual kernel density estimation and positional encoding for graph representation

Synergistic Drug Combination Prediction by Integrating Multi-omics Data in Deep Learning Models

A systematic evaluation of deep learning methods for the prediction of drug synergy in cancer

Predicting drug synergy using a network propagation inspired machine learning framework

GraphSynergy: a network-inspired deep learning model for anticancer drug combination prediction

How to Predict Effective Drug Combinations - Moving beyond Synergy Scores

A network-based trans-omics approach for predicting synergistic drug combinations