CC BY
8МЕДИЦИНСКИЕ НАУКИ (MEDICAL SCIENCES)
UDC 615.45
Soham Mukherjee
2nd Year, Mbbs Perm State Medical University (Perm, Russia)
Himanshu Bhagbole
3rd Year, Pharmacy Srmist College of Pharmacy (Chennai, India)
Suraj Kumar
3rd Year, Pharmacy Srmist College of Pharmacy (Chennai, India)
Krishna Kumar
3rd Year, Pharmacy Srmist College of Pharmacy (Chennai, India)
PERSONALISED DOSING OF MEDICINES FOR CHILDREN
Abstract: Doses for most drugs are determinedfrom population-level information, resulting in a standard ? one- size-fits-all' dose range for all individuals. This review explores how doses can be personalized through the use of the individuals' pharmacokinetic (PK)-pharmacodynamic (PD) profile, its particular application in children, and therapy areas where such approaches have made inroads.
Key words: personalised dosing, medicine, health care
Key findings
The Bayesian forecasting approach, based on population PK/PD models that account for variability in exposure and response, is a potent method for personalising drug therapy. Its potential utility is even greater in young children where additional sources of variability are observed such as maturation of eliminating enzymes and organs. The benefits of personalised dosing are most easily demonstrated for drugs with narrow therapeutic ranges such as antibiotics and cytotoxics and limited studies have shown improved outcomes. However, for a variety of reasons the approach has struggled to make more widespread impact at the bedside: complex dosing algorithms, high level of technical skills required, lack of randomised controlled clinical trials and the need for regulatory approval.
Summary
Personalised dosing will be a necessary corollary of the new precision medicine initiative. However, it faces a number of challenges that need to be overcome before such an approach to dosing in children becomes the norm.
Introduction
For the majority of drugs, therapeutic doses are proposed based on population-level information, focussing on the typical patient and recommending a standard, 'one-size- fits-all' fixed dose range. However, this approach to dosing does not to a great extent account for the prevalence of between-patient variability in systemic exposure
(pharmacokinetics (PK)) and the consequential biological response (pharmacodynamics (PD)). Demographic, genetic, clinical and environmental factors have been shown to contribute considerably to this population variability, and hence, individual patients can differ substantially in their response to drug therapy or their susceptibility to adverse drug reactions. This is of particular importance for drugs with narrow therapeutic ranges where the variability increases the likelihood of serious toxicity or otherwise treatment failure. Drug dosing in children has traditionally been extrapolated linearly from adult doses with adjustments based on age, body weight or body surface area. This method is easy, simple and does not require the use of complex dosing algorithms. However, the relationship between dose and age is not linear as children are in a continuous state of
development and maturation and which can have a significant impact on both the PK and the PD of many drugs.
Age-related changes in absorption, distribution, metabolism and excretion and response to drugs have been demonstrated in young children as a result of the ontogeny process. Absorption, and hence the bioavailability, of orally administered drugs usually approach adult values by approximately 5 years of age because of alterations in gastric pH and gastrointestinal motility as well as the maturation process for efflux transporters and intestinal metabolism. The distribution of drugs is also affected in neonates and infants because of the increased total body water-to-body fat ratio and the decreased amount
and affinity of plasma proteins, albumin and a1-acid glycoprotein. In addition, hepatic metabolism and renal excretion of drugs are decreased in the first year of life, whereas the enzymatic activity of specific hepatic cytochrome P450 isoenzymes exceeds the adult values in the age range 1-12 years. The development process can also affect drug efficacy response or sensitivity to adverse effects. For instance, it has been shown that children are more sensitive to the anticoagulant effects of warfarin as compared to
adults due to significantly lower levels of prothrombin and vitamin K-dependant clotting factors.
Therefore, a simple linear extrapolation of drug doses from adults to children may result
in systemic exposures and/or clinical responses that are not equivalent in the two populations.
Moreover, in terms of the challenges to optimising drug therapy, developmental effects
should be considered an additional source of variability, increasing the complexity of drug
treatment in children, particularly the neonatal and infant population. The risk-benefit profile
for all drugs is intrinsically linked to the doses administered and even more so for potent
drugs with a narrow therapeutic range. From this perspective, it could be argued that where
doses can be personalised for patients through an understanding of a drug's PK and PD
response, and the factors
contributing to their between- and within-patient variability, the potential for improving
the risk benefit profile is likely to be even greater in children.
Personalising dosing in children through Bayesian forecasting
Personalised dosing is a concept that recognises each individual has unique PK and PD
characteristics, governing the time course of drug effect, and pivotal to optimising therapy.
Knowledge of the individual's PK/PD parameters is therefore key to individualising drug
120
doses and improving treatment response.
The concept of Bayesian forecasting, a proactive approach to dose individualisation of
drugs with narrow therapeutic ranges, was first introduced by Sheiner et al. in 1979 and was
first applied on a microcomputer by Peck et al. The method utilises population PK/PD
models, incorporating clinically significant covariates that explain the between- and within-
patient variability, to prospectively identify individual's PK/PD parameters and hence
personalise dosing. The Bayesian approach has been shown to be a major advance on
traditional therapeutic drug monitoring (TDM), which is a reactive attempt at dosage
individualisation. Bayesian forecasting has several advantages over traditional TDM; it can
be used during complicated drug dosage regimens, at non-steady state conditions, and when
only a limited number (1 or 2) of serum drug data are available.
The latter is of particular importance and a major advantage in the paediatric population
where decreasing the aggressiveness of interventions is always preferable.
To illustrate the difference between the two approaches, the example of aminoglycoside
antibiotics will be discussed(Figure 1). The traditional method of monitoring
aminoglycosides to ensure the target therapeutic range is achieved is to give an initial
standard dose and then to measure the peak and trough concentrations at steady state.
The observed concentrations are then used to adjust the dose, through the use of a
nomogram or individualisation by estimating PK parameters, assuming a one- compartment
model with clearance and volume of distribution and linear kinetics. However, often more
complex models than one-compartment may be required to avoid bias in parameters and the
steady state condition is not always attainable, particularly in premature neonates and
critically ill children with variable renal function that can alter the drug disposition. In
contrast, in the Bayesian approach population PK models are implemented that incorporate
not only the typical PK parameters describing
aminoglycoside disposition but also the between- and within-patient variability, and the
covariates that explain this variability. An initial a priori estimation of the individual patient's
PK parameters, and hence, dose can be predicted with greater certainty using the population
parameters of the PK model (knowing that the model is representative of other patients in
the population) and the individual patient covariates (age, weight, renal function etc). The
121
parameters can subsequently be refined by considering the individual's observed drug concentrations (taken at any time with no need to attain the steady state). The Bayesian individualised PK parameter estimates are then used to predict the subsequent drug dose (a posteriori) to attain the predefined target concentration. After the first few observations, the individualised parameter estimates become patient data driven with less influence from the population model parameters.
This is expected given the nature of the Bayesian objective function. In the future as models become more mechanistic, accommodating more patient and disease-specific factors, further improvement in individualised dose will be realised. Indeed, the highly detailed, mechanistic, physiologically based PK (PBPK) models potentially offer enhanced prediction capabilities and also extrapolation capacities. Individualised,physiological (system specific) parameters derived from a PBPK model for a given drug should be transferable to other drugs administered to the same individual as informative priors.
Figure 1
The difference between the traditional therapeutic drug monitoring (TDM) approaches
and the Bayesian approach. (a) The traditional TDM approach involves administering a
standard dose to the patient; frequent blood samples taken at steady state are required for
individual parameter estimation and subsequent dose adjustment. (b) The Bayesian approach
involves developing a population PK/PD model using population PK/PD data. A priori
(initial) dose for a new patient is estimated using the mean population PK/PD parameters
and the individual patient's covariates (age, weight, etc.). The parameters can subsequently
122
be refined using the individual's drug blood concentrations taken at non-steady state for a posteriori dose estimation (dose adjustment).
Most Bayesian forecasting algorithms utilise the maximum a posteriori probability (MAP) method to estimate the individual PK parameters. The approach involves incorporating the mean and the standard deviation (SD) of the population PK parameters with the observed drug concentration data to forecast the individual PK parameters. Alternative, non-parametric approaches such as the 'multiple model'
Bayesian forecasting make no assumption about the shape of the parameter distribution. Instead the non-parametric prior includes one set of PK parameter values with associated probabilities for each subject in the population. As Bayesian feedback from serum concentrations is obtained, each set of parameter values in the non-parametric prior has its probability recomputed and in this way provide maximum precision in achieving the target concentrations.
The aim of this review was to highlight the clinical utility of personalising drug dosing in the paediatric population using the Bayesian approach and some examples of therapy areas where such an approach has been evaluated. The opportunities and challenges to implementing personalised drug therapy for children in clinical practice will also be discussed.
The Academic Search Premier (EBSCO), CINHAL, MEDLINE, PubMed and EMBASE databases were searched for relevant references. The search terms used included 'Bayesian forecasting', 'dosage individualisation', 'personalised medicine', 'children', 'developmental', 'pharmacokinetics', 'pharmacodynamics', 'models',
'antibiotics', 'anticancer', 'chemotherapy', 'oral anticoagulants' and 'dosing software'. Additionally, the bibliographies of the retrieved articles were used to identify additional relevant references.
Anticancer drugs
Optimising dosing regimens of anticancer drugs for children present a major challenge in the clinical oncology setting. This is particularly evident in infants and the very young children due to the considerable developmental physiological changes occurring in this age
group. Traditionally, dosing of anticancer drugs in children has been based on body surface area with dose reductions in the very young children, due to their narrow therapeutic range. However, anticancer drugs are generally associated with large between-patient variability in PK and PD in addition to the considerable effect of pharmacogenetics. The use of standard fixed dosing in anticancer treatment has been shown to cause twofold to fivefold variability in drug concentration between patients.
Moreover, ignoring the variability in drug disposition and efficacy/toxicity leads to increased risk of serious toxicity and treatment failure in this population.
Population PK/PD models of for anticancer drugs have been developed to identify and quantify the complex PK and the relationship between PK and PD including the influence of pharmacogenetics. Modelling can assist in optimising dosing schedules for both single-agent and combination regimens and identifying possible drug interactions with anticancer agents. In the paediatric setting, modelling has been used to describe the wide variability in this population and to identify the covariates that contribute to this variability and hence assisted in optimising dosing regimens to avoid toxicity and treatment failure.
Busulfan, an alkylating agent used as a part of the conditioning treatment prior to haematopoietic stem cell transplantation, has a narrow therapeutic range, requiring the area under the curve (AUC) to be targeted in the range 4-6 ^g h/ml, to reduce the risk of liver toxicity (veno-occlusive disease, VOD) and graft rejection.Busulfan is associated with high between-patient variability which has been attributed to various factors which makes it very difficult to achieve the target AUC with a standard dose. The use of intravenous busulfan has alleviated some of this variability, yet there is still considerable variability in response and toxicity which necessitates dose individualisation.
Personalising oral busulfan dosing in children using the Bayesian forecasting approach has been proven to improve the clinical outcomes, reduced doses in 69% of patients, lower incidence of VOD compared to the conventional approach (3.4% vs 24.1%) and successful engraftment in all patients. The approach has also been shown to improve the clinical outcome in a combination regimen with cyclosporine; 90-day VOD-free survival was 97% compared to 76% using the conventional approach, and the graft-
versus-host disease outcome was successful in the majority of the patients who received
124
the individualised doses.
The Bayesian dose individualisation approach has also been shown to improve the outcome of intravenous busulfan treatment with a higher proportion (72%) of personalised doses achieving a therapeutic steady state concentration range as compared to 57% of the FDA approved doses and 70% of the European Medicines Agency approved doses.
Carboplatin, a platinum compound widely used in solid paediatric cancers such as neuroblastoma and brain tumors, is a narrow therapeutic index drug whose systemic exposure correlates with both response and toxicity.The standard formula used for carboplatin dosing depends on the estimation of the drug's AUC which usually requires frequent blood sampling to be obtained and is particularly burdensome for children, costly and time-consuming.
Estimation of carboplatin AUC using limited sampling strategy and Bayesian forecasting has been shown to be more reliable and convenient than the conventional methods that use renal function or body surface area. The method has been used for adjusting high-dose carboplatin treatment in children and was shown to achieve carboplatin exposure within 80-126% of the target AUC values as compared to 65- 213% of the target values without dose adjustment.
Methotrexate (MTX), an antifolate drug used in high doses for the treatment of various paediatric cancers such as acute lymphoblastic leukaemia and acute myelocytic
leukaemia, exhibits large between-patient PK variability that has been shown to affect clinical outcomes. The use of a standard fixed dose can result in up to sevenfold
difference in plasma concentration.Individualising MTX dosing in children with B-cell leukaemia based on the individual's capacity to clear the drug resulted in better outcomes than standard doses, with the mean rate of continuous complete remission at 5 years of 76% vs 66%, respectively.
Methotrexate forecasting algorithms in the clinical setting can also provide clinicians with valuable information such as the patient's capacity to clear MTX as expected, the probability of developing MTX dose-limiting toxicity (mucositis), and whether more aggressive leucovorin rescue therapy is warranted to reduce the risk such toxicity.
125
Antimicrobials
Perhaps the most obvious therapy area and patient population for which to implement
personalising dosing is antibiotics for critically ill patients. Such patients commonly have
severely altered and variable PK are infected by less susceptible organisms and have a
reduced capacity to fight the infection as a consequence of an overwhelmed and/or
suppressed immune system. Infection in critically ill patients is associated with excessive
morbidity, mortality, length of hospital stay and healthcare costs. Early and appropriate
antimicrobial therapy as well as optimal dosing is associated with improved clinical
outcomes. Moreover, the standard dosing regimens for antimicrobials are usually derived
from studies in non-critically ill, ambulatory patients, so that when applied to critically ill
patients, risks suboptimal drug exposure. 43 Marked between- patient variability in PK is
characteristic of critically ill patients and may result from
alterations in cardiac output, tissue perfusion, end-organ failure, increased capillary
permeability, hypoalbuminemia and use of extracorporeal circuits.The amplified PK
variability increases the probability of clinical failure (to clear infection) and emergence of
antimicrobial resistance through low systemic exposures and increases the likelihood of
toxicity through high systemic exposure. Optimisation of antimicrobial dosing in the
critically ill patients therefore needs an individualised approach, which takes into
account the antibiotic-specific minimum inhibitory concentration for the infecting
pathogen, and selects a dosing regimen that has the highest probability of obtaining the target
PK/PD ratio predictive of successful treatment. Antibacterials such as aminoglycosides and
glycopeptides have a narrow therapeutic range and have been subject to traditional TDM
methods to optimise doses.However, in addition to the large between-patient variability,
critically ill patients also display significant intra-individual variability in PK, sometimes
from dose to dose, and coupled with long delays in feedback of plasma levels can result in
inappropriate dose adjustment.
Bayesian dose individualisation of vancomycin in children with malignant
haematological disease, based on the patient's weight, creatinine clearance and
susceptibility of the pathogens involved, has been investigated. The approach was
shown to attain the target therapeutic range significantly better than the fixed dosing
126
approach.Bayesian forecasting has also been shown to aid in predicting the pharmacokinetics of tobramycin in children with cystic fibrosis and hence dose individualisation. Additionally, a retrospective evaluation of the predictive performance of a Bayesian feedback programme for tobramycin dose individualisation has been shown to result in a significantly improved proportion of patients with therapeutic concentrations. Furthermore, personalising doses of aminoglycoside antibiotics have shown not only to improve treatment outcomes but also to reduce costs.Bayesian models can also be used to predict trough levels from opportunistic blood samples taken any time after dosing, obviating the need for time constrained trough levels.
Voriconazole, the triazole antifungal used in the treatment of invasive fungal infections, also has a narrow therapeutic range and shows highly variable non-linear pharmacokinetics such that small dose changes are associated with disproportionately large changes in plasma concentrations. Inappropriate dosing can result in either treatment failure or serious adverse events such as hepatotoxicity and central nervous system toxicity. Personalising voriconazole treatment using a Bayesian adaptive control algorithm was capable of accurately managing therapy in children independently of steady state conditions.
The antiviral drug, valganciclovir, used for the prophylaxis or treatment of cytomegalovirus infection in paediatric solid transplant patients,exhibits widely variable pharmacokinetics especially in children. Inappropriate dosing can result in toxicity or treatment failure and antiviral resistance which have a major impact on the patients' morbidity and mortality. Individualising valganciclovir treatment, based on individual AUC, is providing a promising tool for optimising valganciclovir treatment in paediatric kidney transplant patients.
Oral anticoagulants
Congenital heart disease accounts for about one-third of all major congenital abnormalities with a worldwide prevalence of nine per 1000 live births. Substantial improvements in surgical procedures and medical treatment have resulted in increased survival rates of children with this disorder. However, there is an associated increased risk of thrombosis in children undergoing heart surgery with a prevalence of 11% which can lead to
life-threatening complications like arterial stroke (5.3%), pulmonary embolism (2.9%), cardiopulmonary arrest (4.1%) and death (5.3%). Hence, the use of oral anticoagulants for long-term thrombotic prophylaxis has risen to reduce the associated morbidity and mortality. Warfarin is the most widely used oral anticoagulant,
but its dosing has been challenging to clinicians due to considerable between- and within-individual variability in its PK/PD and the effect of genetic polymorphisms. The current evidence available from the largest cohort study in children shows that the proportion of International Normalized Ratio (INR) values within the target therapeutic range was found to be 47% for the range of 2.0-3.0 and 61% for the range of 2.5-3.5.
Warfarin acts by inhibiting the enzyme vitamin K epoxide reductase (VKOR), and the gene for vitamin K epoxide reductase is found in complex subunit 1 (VKORC1).
Mutations in this gene lead to altered VKORC activity and therefore sensitivity to warfarin, manifesting as significant between-patient variability in anticoagulant dose
requirement.Metabolism of warfarin is influenced by the genetic polymorphism of the enzyme cytochrome P450 2C9 (CYP2C9), where the variant alleles, CYP2C9*2 and CYP2C9*3, are associated with reduced enzyme activity, as compared with the wildtype allele, CYP2C9* 1. Patients with the variant alleles of the enzyme CYP2C9 require lower doses of warfarin, have a greater difficulty in obtaining optimum anticoagulation and have a higher probability of developing bleeding episodes and elevated INR levels particularly during the initiation period.In addition, variability in warfarin PK in young children is further complicated by the ontogeny of the metabolising enzymes. CYP2C9 activity does not reach mature adult values until between 6 months and 1 year of age.
Population PK/PD models of warfarin incorporating pharmacogenetic variables have been developed and adopted as a tool for Bayesian forecasting. These models can address variability in both rate and extent of response and hence account for the time delay between warfarin exposure and increase in INR and can be extrapolated from one population to another, and they can be used for the prediction of initial doses as well as maintenance doses.
Lee et al. developed a PK/PD model based on the Bayesian approach to optimise warfarin dosing in adults. The model involved a starting dose nomogram based on CYP2C9
and VKORC1 genotypes and a titration scheme to adjust the maintenance dose according to
128
the observed INR values. This model was used as a prior for derivation of
a paediatric PK/PD model and taking into account the effect of body size on clearance, the established maturation pattern of the drug metabolising enzymes and the mechanism of action of warfarin. The dosing approach includes a starting dose nomogram and a titration scheme. Warfarin starting dose can be derived from the nomogram by matching CYP2C9 genotype with VKORC1 genotype and body weight. Dose revision can be accomplished using the titration scheme depending on the INR value.
The Bayesian approach was also used by Hamberg et al. to develop a model for individualising warfarin dosing in adults. The covariates included in the model were age. and CYP2C9 and VKORC1 genotypes; age caused a decrease in the dose requirement by about 6% per decade whilst CYP2C9 and VKORC1 genotype explained up to 4.2- fold and 2.1fold difference in warfarin maintenance doses, respectively. This model was bridged to children by allometric weight scaling of the PK parameters, clearance and volume of distribution, and the addition of a function to account for maturation of the warfarin metabolising enzymes. The children's model has been implemented in a user-friendly Java-based decision support tool. The tool uses the patient's age, weight, baseline and target INR, CYP2C9 genotype and VKORC1 genotype to predict warfarin dose. It can be used for both a priori (initial) and a posteriori (maintenance) dose prediction. An evaluation of the predictive performance of the model showed that the model was able to predict ideal doses (within ±20% of the observed doses) in 41% of cases, increasing to 70% when three or more INRs were available.The model was also retrospectively evaluated using a cohort of postoperative cardiac children and has been shown to predict ideal maintenance doses in 70% of patients.
Personalised medicine software
Many population PK/PD models incorporating the Bayesian forecasting concept have
been developed in children, but as yet there is not a single example of widespread
implementation of personalised dosing of a drug in clinical practice. A primary reason for
this is that the Bayesian forecasting approach has been hampered by the lack of
easy-to-use software. Commonly used population PK/PD modelling software
possessing Bayesian forecasting functionality are difficult to use in a clinical or healthcare
129
setting and require a high level of technical expertise. Routine clinical activities require rapid drug dosing decisions particularly in patients who are critically ill, and therefore, the personalised dosing software should enable easy and quick use during daily care activities even by non-experienced users. The software should have a graphical user interface to enhance navigation through windows or menus. They should also have the ability to run on computers as well as smart phones and personal digital assistants as their use has increased dramatically during the past few years.
Additionally, the software should be able to interface with laboratory information management systems and hospital information systems to facilitate the collection of
laboratory results and relevant patients' administrative and clinical data, and they should have the ability to store patients' databases and ensure confidentiality. Moreover, the software should be able to generate comprehensive reports in a timely manner which can then be transmitted to physicians for consultation.
Recently, with the enormous strides in computer processor speeds and performance, user-friendly personalised dosing software that utilises cloud-based processing systemsthat can run on smart phones and tablets has been introduced to the healthcare market.Nevertheless, presently available software is still sufficiently complex and requires training to enable rapid use at the bedside by healthcare professionals. Indeed, users still require basic knowledge about the PK/PD (model) for proper interpretation of the software output and subsequent dosage adjustment decision.
Necessary steps for the clinical implementation of personalised dosing of medicines Clinical effectiveness trials and regulatory approval
It is apparent that many of the scientific achievements over the past few decades in
pharmacometrics need to be transformed into clinical benefits for patients in the real world
setting. The need for conducting randomised clinical trials to prospectively evaluate the
clinical effectiveness and utility, as well as economic utility, of Bayesian forecasting methods
is now pivotal in order to provide the basis for their clinical use, to convince healthcare
providers to invest and adopt these methods as a powerful intervention for improving patient
outcomes. Unfortunately, it has generally proven very difficult to obtain funding from
governmental or charitable organisations for this type of translational work. Clearly, for
130
novel drugs, industry will need to incorporate such studies into the development programme. Although in recent years there have been a number of examples of personalised therapies (on the basis of genetics or biomarkers) approved by regulatory agencies, there are presently no examples of approvals incorporating personalised dosing. To satisfy regulatory authorities, the effect of personalised dosing on the risk/benefit profile will doubtless need to be demonstrated in randomised controlled trials and validated. Observational studies/data or modelling studies alone may not be adequate to provide a basis for a regulatory action such as recommendation for personalised dosing information in the product literature (SPC/label). Furthermore, it is to be expected the companion personalised dosing software will need to be approved too. Presently, there is no legislation or guidance on this specific subject to refer to.
Integration into existing and future clinical decision support systems
Healthcare systems increasingly use electronic medical records and electronic prescribing in primary and secondary care. They enable access to vast amount of patient-related data, history, investigations, laboratory results and permit accurate calculation of doses based on protocols and guidelines, and generate reminders for monitoring and warnings of potential drug interactions. Although some of this data could usefully feed into the Bayesian forecasting model, the integration of such
applications can be technically complex and not without administrative challenges local and governmental (approvals from public health agencies), as well as requiring acceptance by medical personnel.
Conclusion
Pharmacokinetic-pharmacodynamic model-based clinical decision-making has been the goal of clinical pharmacologists for the last few decades, ever since the power of Bayesian forecasting methodology was first described and demonstrated. Its ability to optimise dosing is likely to be even greater in young children due to the additional variability in PK and PD observed as a consequence of maturation of eliminating enzymes and organs. The utility of personalised dosing is most easily discerned and compelling for drugs with narrow therapeutic ranges such as antibiotics, cytotoxics, immunosuppressants and warfarin, and
indeed, studies have shown improved clinical outcomes through the use of Bayesian forecasting methodology. Many other drugs also have complex dose adjustments such as monoclonal antibodies, where fine tuning of doses and dosing intervals can avoid intermittent exposure and hence the development of loss of response through antidrug antibodies.
Unfortunately, aside from a limited number of sporadic examples, such model informed dosing decisions have struggled to make significant headway into bed-side clinical practice. In the past, output from the complex algorithms was slow; the technical skills required were high and in general not practical for every day clinical use. More recent advances in computer technology and parallel advances in precision medicine strategies have however renewed interest in personalised dosing of drugs. As Bayesian forecasting methods become accessible to non-PK/PD experts, the reality of personalised dosing should not be too far behind.
However, for personalised dosing to become more widely applicable to a multitude of drugs, there is a requirement for similar, parallel, innovative breakthroughs in microsampling technologies and POC testing devices for drugs, biomarkers and
pharmacogenomics, that are accurate, user-friendly and low-cost and that can provide rapid feedback to clinicians for real-time decision-making. It will also require integrating into existing or future electronic medical recording and prescribing systems.
Precision medicine is an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment and lifestyle. The goal of precision medicine is to improve the way we anticipate, prevent, diagnose and treat
diseases and is driven by 'big data' biological databases (human genome, proteomics and metabolomics) and the computational tools to be able to analyse such large data
sets. Clearly, there is a long way to go before personalised dosing is the norm, and there is a risk that recent developments in the field are yet another false dawn.
REFERENCE:
Elibrary.ru
