Opinion paper on the systematic application of integrated bioinformatic tools to actuate routine precision medicine in poly-treated patients

Introduction

The bulk of knowledge brought by omics sciences and the increasing availability of cost-effective molecular diagnostics are finally allowing to realize Precision Medicine (PM), replacing the concept of the “average patient” with the concept of “patient-sized treatment”. Presently, we are going throughout a transitional era where PM strategies are fully accepted and applied in critical medical areas, mainly oncology [1], [2], [3], [4], whereas are almost ignored in several different clinical settings, including primary care settings. Here, we would highlight the unseen PM potential to break down the current socio-economic burden of inappropriate drug prescription, which represents a concern for all medical fields and affects patients’ health as well as healthcare systems resources.

We mean, PM is ready to improve medication management of the millions of people who daily get many drugs to control comorbid, chronic diseased states. These poly-treated patients have increased rates of inefficacy, adverse drug reactions (ADRs), non-compliance and non-adherence, which in turn lead to health worsening, impaired social functions, pharmaceutical waste. The functional link among taking an increasing number of drugs, ADRs and non-adherence is well described and had paramount importance in the elderly [5], [6], [7], [8]. The impact of non-appropriate prescription of poly-therapies could be roughly estimated considering that the number of people aged 65 years or more in the European Union (EU) is expected to grow to around 141 million by 2050 and will account for 28.7% of the EU population by 2080 [9].

The following sections will briefly recap how the basic knowledge about the mechanism of drugs’ action can now easily be matched, by bioinformatics, with each specific “patient-disease system”, driving informed drug selection and prescription.

Factors affecting inter-individual variation in drug response

The biological variation in drug response is a multi-factorial phenomenon. In each subject, the extent of drug’s effect is determined by the rate of drug adsorption, distribution, metabolism and elimination by the body (e.g. the drug’s pharmacokinetics, PKs) and by the mechanism of action at the molecular target (e.g. the drug’s pharmacodynamics, PDs). PK and PD profiles are affected by general factors as diet, age, gender, and by “personal” factors as renal and hepatic function, body composition, comorbidities, and genomic profile [10], [11], [12].

At the molecular level, PKs and PDs are determined by the interaction of the drug with a plethora of protein partners acting as drug’s membrane transporters, drugs’ metabolizing enzymes (DMEs) and pharmacological targets. DMEs, including the superfamily of Cytochrome P450 (CYP450) enzymes, represent a wide and variegate group of proteins aimed to transform xenobiotics into hydrophilic metabolites, more easily eliminated by the body [13], [14], [15].

The expression and activity level of the above-mentioned groups of proteins is a personal matter, since depends by the combinatorial effect resulting by the interaction among taken drugs, individual genomic profile and the “biochemical environment”, as briefly explained below.

The effect of genomic profile is exemplified by the CYP450s. Genes encoding important CYP450s members are highly polymorphic, and different alleles produce proteins with markedly different enzymatic features, ranging from null activity to deeply enhanced activity [16], [17], [18]. In subjects carrying such alleles, the PKs of a substrate drug is changed, and the risk of toxicity or inefficacy is increased, since the drug concentration is below, or above, the therapeutic window. Similarly, DNA sequence variations in drug transporters and drug targets may alter drug distribution and action, respectively. The recognition of genomic markers predictive of drug response opened the way to Pharmacogenomics (PGXs), that is the possibility to preemptively assess genotype to predict individual drug response and drive patient-sized drug prescription. Systematization of the pharmacogenomic knowledge has been fronted by different scientific consortium, as the Clinical Pharmacogenetics Implementation Consortium (CPIC) and the Dutch Pharmacogenetics Working Group (DPWG) [19], [20], [21].

The effect of the “biochemical environment” consists in the presence of molecules acting as inducers or inhibitors of the protein’s activity, thus affecting the rate of drug biotransformation. Many DMEs are modulated by molecules contained in foods, drinks [22], and by several drugs themselves [23], determining the phenomenon of drug-food interactions and drug-drug interactions (DDIs).

Commonly, DDIs happen when two or more drugs, administered together, act as a substrate, inhibitor or inducer of the same metabolizing enzyme, drug transporter or drug target. Thus, one drug can impair or enhance the PKs and PDs of the other drug [24]. The more are the drugs in a polypharmacy regimen, the more is likely the development of negative effects caused by DDIs [5, 6]. Notably, DDIs are well described for nearly all available drugs and many bio-informatics tools are available to check interactions before prescribing two or more drugs [25, 26].

Bioinformatic tools: user-friendly translators of a complex biological code

As we are increasingly aware of many of the factors affecting individual drug response, we are increasingly enabled to preemptively recognize such factors in each patient. Indeed, personalized drug prescription is an old concept implied in drug labels and recommendations. The novelty of Precision Medicine is that we have hundreds of markers to drive drug prescription, but they seem too difficult to be summarized, producing a usable information. We would highlight the actionable progresses made in the field and their present limits.

True therapy personalization would require integration of clinical and diagnostic evaluation with DDIs prediction and pharmacogenomic screening and interpretation.

DDIs analysis is currently accessible to anyone, since many commercial or free web-based tools have been developed [27], [28], [29], [30]. So, the main limit to the systematic evaluation of DDIs before drug prescription seems to be cultural. Systematic training and education programs could be easily carried out by scientific and academic communities to disseminate more and more this information among doctors, particularly general practitioners. The latter represent the first healthcare line and have the precious advantage to have the all-round vision of the patient’s condition, allowing prompt identification of subjects at high risk for DDIs (patients with multi-morbidities). In our opinion, general adoption of the easy, accessible and free-of-charge DDIs analysis has the potential to dramatically cut the burden of inefficacy and ADRs in poly-treated subjects.

More efforts would be needed to achieve broader application of PGX testing in these patients. First, its clinical utility and appropriateness should be finally accepted and codified into clinical guidelines. PGXs testing accessibility is growing thanks to increasing cost-effectiveness of DNA sequencing technologies. The limit of PGXs testing in poly-treated patients is rather the functional interpretation of many genomic markers upon many medications at once. To address the issue, a last generation bioinformatic tools has been developed, recently, to optimize poly-therapy according to both DDIs and patient specific characteristics, including pharmacogenomic markers, gender, age, renal and hepatic function [31, 32]. This software allows an in-silico evaluation of a proposed polytherapy, producing a risk score, that is a warning system highlighting potential deleterious effects (in terms of both safety and efficacy); when the patient-polytherapy pair produces a high therapy-related score, the system suggests a ranked lists of alternative drugs for each therapeutic target. By selecting one or more of the proposed alternative drugs, the system recalculates the therapy-related score, so the doctor can rapidly re-design (optimize) the drug cocktail to minimize the risk of toxicity/inefficacy.

Such a pipeline for drugs prescription is transversal to all medical specialties, and its clinical utility has already been showed by various reports [33, 34]. It is expected that many similar algorithms would proliferate in the next future, making more and more accessible the approach to the medical community.

However, even if the field of PGXs bioinformatics is growing, main issues should be faced by healthcare stakeholders. Ethical and legal implications should be carefully evaluated and regulated to ensure protection of personal (including genomic) data and medical responsibilities should be accurately defined. It should be kept in mind that patient-sized drug prescription is based on a “phenotype-prediction”, based on a certain number of data. So, the precision of the prediction varies according to the number of analysed data and according to the algorithm used to interpret data. Thus, different evaluation pipelines may generate different recommendations for the same patients, who should be adequately informed and give consent about the procedure. It is particularly important to inform the patient about the limits of PGXs prediction: albeit most genotype-phenotype associations are unambiguous, some uncertainty areas exists, e.g. the prediction of CYP2D6 phenotype: this is affected by both single nucleotide polymorphisms and by gene copy number variations (CNV), but often CNV cannot be definitely established by common PGXs tests [35, 36]. Development of adequate genotypic and phenotypic assays to control this variable is ongoing [35], [36], [37], [38].

Although general ethical and regulative guidelines in the field can be shared, country-specific aspects (as reimbursement system and data protection) should be developed by national legislative systems [39, 40].

In summary, bioinformatics is essential and almost ready to improve the appropriateness of drug prescription. The level of accessibility to the general population of such novel tools will depend upon the level of awareness of healthcare managers and decision makers in the private and public sectors, which could determine the extent of public funding for development and spreading of free-of-charge or low-cost prescription support tools. This decision should be supported by cost-effectiveness considerations. Since the benefits of the evaluation/optimization strategy are proportional to the number of drugs in a poly-therapy, as well as to the fragility of the treated patient, the elderly population is supposed to be its elective target.

Adoption of a prescribing method reducing ADRs and increasing efficacy is expected to produce a virtuous cascade: from a patient’s point of view, engagement in a medical process specifically targeted on our own unique characteristics, increases adherence and compliance; from the healthcare stakeholders point of view, it means cost containment in terms of drug consumption, emergency care to manage ADRs, additional care to manage disease’s progression.

Although poly-therapy evaluation by integration of multi-dimensional data as DDIs, patient’s genomic data, clinical data etc., is presently scarcely applied, we are firmly convinced that a rapid diffusion of appositely developed informatic-tools, facilitated by the commercial competition among producing companies, and hopefully by the engagement of the public sector, will soon allow system-level actuation of patient-sized drug prescription. Dissemination of the concepts underlying the methodology of poly-therapy evaluation/optimization is crucial to the whole community (patients, healthcare operators and healthcare managers), to be ready to use innovative medical decision support tools.

Conclusions

Poly-therapy regards a huge number of fragile patients who have higher rates of DDIs, in turn associated with adverse outcomes, ADRs, increased access to emergencies and increased rate of hospitalization. The systematic adoption of computer-assisted tools for the preemptive evaluation and optimization of poly-therapies in daily clinical practice is expected to unlock the huge potential of Precision Medicine to improve disease treatment, in all medical fields.