January 23, 2020 – A Croatian group of scientists from the St. Catherine Special Hospital participated in the publication of yet another remarkable scientific paper, this time explaining the concept of pharmacogenomics testing based on the principles of artificial intelligence.
“Pharmacogenomics” is one of the world’s leading scientific publications in the field of pharmacogenomics (PGx) and their latest issue included an article titled “Pharmacogenomics at the center of precision medicine: challenges and perspective in an era of Big Data“.
The authors are a group of the Croatian and American scientists, led by Professor Dragan Primorac, who propose the model of the systematic introduction of PGx testing into clinical practice. Along with that, they propose the implementation of the concept into the health systems of various countries, using Artificial Intelligence (AI) models, as well as some sub-systems within the AI framework, such as so-called Machine Learning.
Through the specific algorithms analysing the data, the patients’ data is compared with all the data already deposited in large databases (using the Big Data approach), with the goal of optimising diagnostic procedures, the prevention of disease on time, and personalised treatment. Unlike the typical model of machine learning where the algorithms are defined by certain parameters based on expert knowledge, the concept of AI primarily uses the neural networks, continually evaluating a large amount of data and processing it in a similar manner to human thinking.
Pharmacogenomics analyses a whole series of genes, or even the entire genome, and then studies the connections between the genetic predisposition of an individual and their reaction to a drug. It helps understand why some people respond to some medicines while others don’t, why some people need to have the doses of their medicines adjusted to get the perfect therapeutic response, and it can even warn you if a patient won’t respond to therapy at all or even when someone will experience toxic side effects.
The model proposed in the paper is based on the experiences by the St. Catherine Hospital and the PGx testing it has been performing in cooperation with OneOme American company (a spin-off company of the famous Minnesotan Mayo Clinic).
The testing uses the RightMed system and analyses 25 genes at the same time (CYP1A2, CYP2B6, CYP2C9, CYP2C19, the CYP2C cluster, CYP2D6, CYP3A4, CYP3A5, CYP4F2, COMT, DPYD, DRD2, GRIK4, HLA-A, HLA-B, HTR2A, HTR2C, IFNL4, NUDT15, OPRM1, SLC6A4, SLCO1B1, TPMT, UGT1A1, VKORC1), which are responsible for the synthesis of the enzymes important for the drug’s metabolism (especially the genes of the enzymatic system of cytochrome P450), transport proteins, receptors, other proteins important for the functioning of drugs, as well as those from the HLA system, which is important for the reactions of oversensitivity to medicines.
The system allows for the prediction of the response of each patient for over 300 of the most frequently used medicines, and so the patients are given the possibility to find the one which will help them best. All of the algorithms used in the system related to the use of the genetic information and the selection of the drug and its dosage are following the Clinical Pharmacogenetics Implementation Consortium
The algorithm of the analysis of genes responsible for the drug metabolism of each patient will sort them into five categories of metabolizers: slow, intermediary, normal, fast or very fast. The paper also includes a very detailed SWOT (Strength, Weakness, Opportunity, Threat) analysis of the proposed strategy, which can lead to a significant new step in the development of modern medical sciences.
The importance of the introduction of PGx methods into routine clinical practice is best confirmed by the information recently published in the leading American medical sciences journal, JAMA, in which it was said that in the US, more than 2 million hospitalized patients have serious side-effects from the drugs they were given annually, and over 100.000 of them die. Some estimate that the number is even higher today. Today, side-effects from drugs constitute the fourth cause of mortality in all populations. In the US, the health system spends 136 billion dollars a year to mitigate the damage done by the side-effects of drugs. European data shows that between 7 and 13 percent of patients get admitted into hospitals because of the side-effects of drugs, and 30 to 50 percent of patients do not respond to therapy at all.
Professor Dragan Primorac said that the “right therapy for the right patient at the right time” is the key phrase of personalized medicine, however that can’t be achieved without an insight into the molecular status of the patient.
Our goal is to reduce the morbidity caused by the side-effects of drugs to the lowest possible level, as well as to integrate pharmacogenomics through the concept of AI with all the other diagnostic procedures into an integrated system which will lead to the optimisation of diagnostic and therapeutical procedures.
The proposed concept of the integration of PGx methods into clinical practice, developed by the scientists from the St. Catherine Special Hospital and the OneOme company, has attracted huge interest on the world’s health market, and the first implementation of the model outside of Croatia is soon to start in German health institutions.