Artificial intelligence techniques for prediction of drug synergy in malignant diseases: Past, present, and future
Introduction
The uncontrolled growth of abnormal cells in the body is called malignancy. Malignancy cells will often break away from the original cell stack, relocate via lymph and blood systems and occupy other organs where the uncontrolled growth cycle can repeat. A single-drug approach to treat malignant diseases is inefficient. Combination therapy is a promising medication choice that incorporates many treatments. It have greater clinical effect than the sum of their individual effects [1].
When compared to single-drug treatment, administering drug synergy instead of monotherapy can result in greater efficacy [2,3]. The adverse side effects and host toxicity are minimized with drug synergy as the doses of the formulated drug are usually smaller than the doses of single-agent [4,5]. Drug resistance can be reduced or even eliminated by drug synergy [[6], [7], [8], [9]]. As a result, drug synergy becoming an effective strategy for treating malignant diseases. Previously, the drug synergy for malignant diseases was determined by clinical trials. But, this method proves to be costly and time-consuming. High-throughput screening (HTS) is another technique to classify the synergistic drug combinations [10]. But, it is rigorous to measure the entire combinatorial space through HTS [11,12].
With the advancements in technology, artificial intelligence (AI) techniques are used to develop the drugs. MYCIN, a computer-based consultation system was an early AI application designed for drug treatment [13]. MYCIN is a rule-based expert system with 600 rules. The purpose of this system was to give therapeutic recommendations for patients in case of bacterial infections [13]. A fuzzy logic function was incorporated in MYCIN's reasoning assessment procedures for integrating uncertain statements inside each rule [13]. While MYCIN was never utilized in practice, it did obtain a 69% success rate in determining approved medication, which was better than that of infectious disease specialists using the same criteria [13]. An expert system was also employed in AI-based application for drug therapy selection, however, in this case, a knowledge base of just roughly 100 rules were used to identify optimal antimicrobial therapy [14]. Despite this, the system proved valuable for medical practitioners, despite its small information base [14]. Machine learning models are utilized the unimodal data for drug synergy prediction. Nowadays, researchers are working on multimodal data for the effective drug synergy computation [15]. Recently, Wang et al. [16] created Therapeutic Target Database (TTD) that consists of information about patented agents, target-interacting proteins, target-regulating microRNAs, and transcription factors. They provided targets for computing the drug synergy prediction. Li et al. [17] analysed the eight-nine targets to predict the clinical success of target drugs. They found that the trail-speed differential features are closely related to the human targets. Beside this, drug-target interactions (DTIs) are also used in the development of effective drugs. DTIs describe interaction regions between drug molecules and protein targets. The main goal of DTIs is to discover novel ligands for the specific protein targets [18]. Artificial Intelligence (AI) techniques utilized the concept of DTIs in the computation of drug synergy. Initiative to enhance earlier AI applications by improving physician skill has been made by merging expert systems, in which information contained in the system is internally defined by frames and rules, with artificial neural networks (ANNs) [19]. As an example, multilayer ANNs were utilized in conjunction with expert systems to create a machine learning (ML) system for diagnosis and treatment of hypertension [19]. The selection of hypertension drug combinations in the system was achieved by training the system with blood pressure time series [13]. The training of system was performed with a learning set of data that contain 300 healthy subjects and testing data with 85 hypertensive subjects [19]. The multilayer ANN appears to extract differentiating characteristics from a learning set of data and detect these patterns even with noisy input data sets [19]. The ANN, which had more than three layers was one of the earliest deep learning (DL) approaches utilized for drug combination selection [13]. This review paper explains various AI techniques that are used for drug synergy.
The purpose of this paper is three folds. The drug synergy prediction for malignant disease using artificial intelligence techniques are presented in the first fold. Second, it elaborates the applications of deep learning in various domains like biological imaging analysis, De novo design, estimation of reactions, and drug-target interaction prediction. Last, the challenges associated with drug synergy and future research approaches are explored.
The main contributions of this paper are as follows:
- 1.
The malignant diseases and their treatment methods are discussed.
- 2.
The role of artificial intelligence techniques in drug synergy is presented.
- 3.
The reference models for computation of deep learning techniques are discussed.
- 4.
The challenges and future research directions associated with drug synergy are investigated.
The rest of this paper is structured as follows: Section 2 presents preliminary concepts of drug synergy and reference additivity models. In Section 3, various machine learning and deep learning architectures for drug synergy are discussed. In Section 4, the datasets used for drug synergy are presented. Section 5 presents the deep learning applications for drug development. In Section 6, drug resistance towards malignant diseases are discussed. The current challenges and research directions are presented in Section 7. The concluding remarks are drawn in Section 8.
Malignant diseases refer to the process through which genetic mutations develop in normal cells, causing them to expand uncontrolled. Almost every cell in the body has the potential to accumulate mutations and form a tumor. Carcinogenesis or tumorigenesis refers to the process through which a normal cell transforms into a tumor [20]. Malignant tumors are those that can spread from their initial site of origin. Virulent, cancer, and malevolent are medical synonyms for malignant. The antonyms of malignant are harmless, noncancerous, and benign. Benign is a type of tumor, that is not dangerous to health.
In the simplest possible way, a malignant tumor is caused by deoxyribonucleic acid (DNA) damage in a normal cell that results in growth and survival [[21], [22], [23]]. A DNA mutation is occur at a location where it does not cause cell death [20].
There are billions of cells in the human body. For example, there are cell of blood, connective tissue cells, lymphatic systems, and epithelial tissue cells. Table 1 depicts several forms of malignant tumors. Eighty-five percent of malignant tumors are the tumor because of the epithelial cells and are known as carcinomas [20].
Mechanical, physical, chemical, and biological treatments are the major approaches to treat patients with malignant diseases. Fig. 1 depicts different approaches for the treatment of malignant diseases.
Mechanical approach of tumor removal is surgery. Surgery is the most often utilized option, and this method has the highest potential benefit. It is a medical approach that involves removal, examination, and tissue repairing. It is a local therapy that is only effective for solid tumors that are confined in a single location. It is not utilized to treat liquid tumors such as leukemia. It may not always eliminate all malignant cells from the patient, thus the tumor may reappear after some time.
Physical approach of tumor removal destroys tumor with photodynamic therapy, radiation therapy, hyperthermia, and heat therapy.
Chemical treatment of malignant diseases uses chemical agents and targeted medications to kill the tumor. Chemical therapies are the type of systemic therapy that is administered through the bloodstream and may be used to treat malignant diseases at any anatomic site in the body. It is the sole choice for the treatment of metastatic illness when medication must be administered systemically throughout the body.
Biological treatment of malignant diseases includes therapies that consist of viruses, recombinant proteins, and the use of antibodies.
Synergy is derived from the Greek word syn and ergon. Syn means together and ergon means work. Synergy is generally described as the impact of two or more compounds acting together that is higher than their predicted additive effect. It can be categorized based on pharmacology and medicine.
When one drug activity is enhanced by another, the drugs are said to be synergistic [24]. The diseases that require combination drug therapy are hypertension, cancer, diabetics, tuberculosis, heart diseases, anti-depression, etc [25]. Drug Synergy can be defined in two types such as additive synergy and supraadditive (potentiation) synergy.
The impact of two medicines acts in the same direction and simply adds up in additive synergy [24]. The additive synergy is given as:
The side effects of components of an additive pair may be differ. The combination is considered safer than the larger dose of one constituent [24]. For example, the combination of aspirin and paracetamol increases their impact than the individual effects of components in supraadditive [24]. The supraadditive synergy is given below:
For example, the combination of fluorouracil and leucovorin increases the activity of 5-fluorouracil by increasing the intracellular concentration of reduced folates in case of gastric cancer.
Section snippets
Drug synergy for malignant diseases
Implementing drug synergy rather than monotherapy will result in greater efficacy in the treatment of malignancy disease. Drug combination is a possible solution to combat the complicated diseases including inflammation, cancer, and type 2 diabetes. The drugs are combined in unpredictable ways and provide a wide range of outcomes. The formulations of drug synergy have been found to be more effective and clinically specific. Drug antagonism, on the other hand, is often unfavorable. However, it
Artificial intelligence in drug synergy for malignant diseases
Drug synergy become a feasible method for the treatment of malignant diseases. Both time and resources, methodologies based on experiments are extremely costly. Several methods have been suggested to solve these problems by starting the combinations of known drugs. The main concern associated with drug synergy is production cost and time consumption. Further, inappropriate dosage and inefficiency are barriers to the process of drug synergy [42]. High-throughput screening (HTS) is one of the
Database utilized in drug synergy
This section discusses the datasets, which were used for prediction of drug synergy.
The largest known drug combination public database is DrugComb [95]. DrugComb database consists of 739,964 drug combinations screening result for 8397 drugs with 2320 cancer cell lines. Thirty-three types of tissues are utilized for this dataset. Some of these issues are skin, bone, lung, breast, ovary, prostate, kidney, etc. This database curates data from NCI Almanac dataset [96], Forcina dataset [97], Cloud
Applications
In this section, the deep learning applications for drug development are described followed by the applications of pharmacological interaction in drug synergy.
Drug resistance towards malignant diseases
Resistance to anticancer treatment is a complicated process that results from changes in the drug targets. Drug resistance can arise through a variety of mechanisms including cell death inhibition, multi-drug resistance, epigenetic, enhancing DNA repair, and alterations in drug metabolism. Proteomics technology and advances in DNA microarray provide the new strategies to overcome the drug resistance. In the field of drug resistance, the treatment of malignant diseases can be formed with
Current challenges and possible research directions
The main challenges of deep learning in drug discovery are addressed followed by the possible research directions.
Conclusion
The relevance of knowing the importance of treating malignant diseases has become crucial as it is one of the leading causes of deaths in recent decades. The application of artificial intelligence to various drug programmes has shown promising results. This paper presents an overview of the need of drug synergy prediction for malignant diseases using artificial intelligence approaches. Artificial Intelligence has shown promising results in expediting drug formulation and exploring drug synergy
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (143)
- et al.
Structure and dynamics of molecular networks: a novel paradigm of drug discovery: a comprehensive review
Pharmacol. Therapeut.
(2013) - et al.
A model of inexact reasoning in medicine
Math. Biosci.
(1975) - et al.
An artificial intelligence program to advise physicians regarding antimicrobial therapy
Comput. Biomed. Res.
(1973) - et al.
Modality adaptation in multimodal data
Expert Syst. Appl.
(2021) Cancer causation: the darwinian downside of past success?
Lancet Oncol.
(2002)Hallmarks of cancer: a 2012 perspective
Ann. Oncol.
(2012)- et al.
Prediction of aquatic toxicity of chemical mixtures by the qsar approach using 2d structural descriptors
J. Hazard Mater.
(2021) - et al.
Qsar modeling of algal low level toxicity values of different phenol and aniline derivatives using 2d descriptors
Aquat. Toxicol.
(2020) - et al.
Predict effective drug combination by deep belief network and ontology fingerprints
J. Biomed. Inf.
(2018) - et al.
Current trends in multidrug optimization: an alley of future successful treatment of complex disorders
SLAS TECHNOLOGY: Translating Life Sciences Innovation
(2017)
Pdc-sgb: prediction of effective drug combinations using a stochastic gradient boosting algorithm
J. Theor. Biol.
Genomic signatures for paclitaxel and gemcitabine resistance in breast cancer derived by machine learning
Molecular oncology
Combining genomic and network characteristics for extended capability in predicting synergistic drugs for cancer
Nat. Commun.
Mechanisms of drug combinations: interaction and network perspectives
Nat. Rev. Drug Discov.
Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies
Pharmacol. Rev.
An unbiased oncology compound screen to identify novel combination strategies
Mol. Cancer Therapeut.
Fulvestrant reverses doxorubicin resistance in multidrug-resistant breast cell lines independent of estrogen receptor expression
Oncol. Rep.
Increased oral bioavailability of topotecan in combination with the breast cancer resistance protein and p-glycoprotein inhibitor gf120918
J. Clin. Oncol.
Bexarotene (lgd1069, targretin), a selective retinoid x receptor agonist, prevents and reverses gemcitabine resistance in nsclc cells by modulating gene amplification
Cancer Res.
Combinatorial drug therapy for cancer in the post-genomic era
Nat. Biotechnol.
Methods for high-throughput drug combination screening and synergy scoring
A new drug combinatory effect prediction algorithm on the cancer cell based on gene expression and dose–response curve
CPT Pharmacometrics Syst. Pharmacol.
Systematic analysis of quantitative logic model ensembles predicts drug combination effects on cell signaling networks
CPT Pharmacometrics Syst. Pharmacol.
Therapeutic target database 2020: enriched resource for facilitating research and early development of targeted therapeutics
Nucleic Acids Res.
Clinical trials, progression-speed differentiating features and swiftness rule of the innovative targets of first-in-class drugs
Briefings Bioinf.
Deep learning in drug target interaction prediction: current and future perspectives
Curr. Med. Chem.
A neural network expert system for diagnosing and treating hypertension
Computer
The Biology and Treatment of Cancer: Understanding Cancer. Plus 0.5em Minus 0
The clonal evolution of tumor cell populations
Science
Essentials Of Medical Pharmacology. Plus 0.5em Minus 0
Present and Future Prospect of Combination Drugs Therapy
The search for synergy: a critical review from a response surface perspective
Pharmacol. Rev.
Drug synergism: its detection and applications
J. Pharmacol. Exp. Therapeut.
Synergy, additivism and antagonism in immunosuppression. a critical review
Clin. Exp. Immunol.
Understanding synergy
Am. J. Physiol. Endocrinol. Metabol.
Quantifying synergy: a systematic review of mixture toxicity studies within environmental toxicology
PLoS One
Synergy, additivism and antagonism in immunosuppression. a critical review
Clin. Exp. Immunol.
Quantitation of the synergistic interaction of edatrexate and cisplatin in vitro
Cancer Chemother. Pharmacol.
Biomol: a computer-assisted biological modeling tool for complex chemical mixtures and biological processes at the molecular level
Environ. Health Perspect.
Safe uses of hill's model: an exact comparison with the adair-klotz model
Theor. Biol. Med. Model.
Beyond ic50s: towards robust statistical methods for in vitro association studies
J. Pharmacogenomics Pharmacoproteomics
The hill equation: a review of its capabilities in pharmacological modelling
Fund. Clin. Pharmacol.
An introduction to terminology and methodology of chemical synergy—perspectives from across disciplines
Front. Pharmacol.
Systems biology approaches for advancing the discovery of effective drug combinations
J. Cheminf.
Systems biology and combination therapy in the quest for clinical efficacy
Nat. Chem. Biol.
Effect of combinations: mathematical basis of problem
Arch. Exp. Pathol. Pharmakol.
Statistical determination of synergy based on bliss definition of drugs independence
PLoS One
Artificial Intelligence to Deep Learning: Machine Intelligence Approach for Drug Discovery
Machine learning approaches for drug combination therapies
Briefings Bioinf.
In silico drug combination discovery for personalized cancer therapy
BMC Syst. Biol.
Cited by (4)
Strategies of Artificial intelligence tools in the domain of nanomedicine
2024, Journal of Drug Delivery Science and TechnologyArtificial intelligence representation model for drug–target interaction with contemporary knowledge and development
2023, Deep Learning in Personalized Healthcare and Decision SupportArtificial Intelligence and Anticancer Drug Development—Keep a Cool Head
2024, Pharmaceutics