OPTIMIZATION AND INDIVIDUALIZATION OF MEDICATION SELECTION AND DOSING
The invention provides systems and methods for producing a ranked list of drugs for administration to a patient in need thereof, the ranked list produced by calculating for each drug a score that is the product of a patient vector, a weighting vector, and the drug's column value in a disease matrix.
The present application claims the benefit of U.S. Provisional Patent Application Ser. No. 60/740,430, filed Nov. 29, 2005 and of U.S. Provisional Patent Application Ser. No. 60/783,118, filed Mar. 16, 2006, the disclosures of which are incorporated herein by reference.
FIELD OF THE INVENTIONThis invention relates to methods for combining a patient's genetic information, a patient's non-heritable host factors and candidate medication characteristics to optimize and individualize medication dosage and compound selection.
BACKGROUND OF THE INVENTIONOne of the most important but unresolved problems in therapy with potent and often toxic drugs has been the lack of our ability to describe, understand, and quantify the important mechanistic relationships and variability between drug doses, concentrations in blood, concentrations of metabolites in other body compartments, and the therapeutic and toxic drug effects. For the most part, defining drug action and inter-patient variability has been limited to simplistic, less informative descriptions of average maximum and minimum drug dose requirements that do not permit true individualization of therapy for each patient.
For some drugs over 90% of the measurable variation in selected pharmacokinetic parameters has been shown to be heritable. Traditionally in pharmacokinetic (PK) analysis a series of concentrations over time is measured. A structural model is defined and fit to the data in order to obtain estimates of the desired parameters such as clearance and volume of distribution. The model is fitted to the individual data by using a least squares algorithm that minimizes the difference between observed and the model predicted concentrations. For reasons of simplicity the assumption is made that differences between the observed and predicted concentrations are caused by random error. With this traditional type of analysis, a model is defined for each subject and the individual parameters are then summarized across individuals. However, imprecision in the sample mean and sample standard deviation frequently are greater than expected, while estimates of variability in these parameters are not well characterized.
The FDA is recognizing the importance of the genetic contribution to the inter-individual variation in response to therapy. There has been a significant increase in the number of new drug applications sent to the FDA containing pharmacogenetic information (Wendy Chou, Ph.D/FDA Apr. 3, 2003). Two package inserts reflect this trend. Thioridazine (Mellaril) which is used for neuropsychiatric conditions is contraindicated in patients who are CYP2D6 poor metabolizers; this warning is specifically stated in two places in the insert. Similarly in multiple places in the package insert for Atomoxetine (Strattera, a medication used for ADHD), the association between genetic polymorphisms in drug metabolism and adverse drug reactions is stated.
In certain ethnic groups as many as 10% of the adolescent population have a CYP2D6 haplotype that is associated with poor metabolism of many antidepressant medications. See Wong et al. (2001) Ann. Acad. Med. Singapore 29:401-406. Clinical genomic testing of these individuals has clear implications for their treatment and prognosis. In extreme cases, children who were poor metabolizers and who were not identified have had tragic outcomes. These negative case reports have included a reported death of a nine-year-old boy who was not recognized to be a poor CYP2D6 metabolizer. The treatment of this child with fluoxetine continued despite the development of multiple symptoms because these symptoms were not recognized as being related to his extremely high serum levels of fluoxetine. Sallee et al. (2000) J. ChildAdol. Psychiatry 10(1):27-34.
Adverse drug reactions occur in 28% percent of hospitalized patients and in 17% percent of hospitalized children. In a report by Phillips in JAMA, 27 drugs were most frequently sited in adverse drug reaction reports. 59 percent (16/27) of these drugs were metabolized by at least one enzyme having a poor metabolizer genotype. 37 percent (11/27) were metabolized by CYP2D6, specifically drugs acting on the central nervous system. The annual cost of the morbidity and mortality associated with adverse drug reaction is $177,000,000 dollars (Year 2000 dollars). Clearly drug toxicity is a major health issue with 100,000 deaths a year and 2,000,000 persons suffering permanent disability or prolonged hospitalizations as a result of direct medication adverse reactions.
Although significant inter-individual variability exists in the response to most medications, medication selection and titration is usually empiric rather than individualized. The main reason that physicians do not incorporate genetic and non-heritable host factors responsible for this inter-individual variability into treatment plans is the lack of applicable, easy to use algorithms that translate the patient's characteristics into clinical recommendations. Thus there is a need in the art for a pharmacokinetic dose individualization technique that is informative, cost saving, and effective.
SUMMARY OF THE INVENTIONThe present invention is concerned generally with the field of identifying appropriate medications and treatment regimens for a disease based upon genotype in mammals, particularly in humans. It is further concerned with the genetic basis of inter-patient variation in response to therapy, including drug therapy. Specifically, the invention describes the use of gene sequence variances for optimizing efficacy and safety of drug therapy. The invention relates to computerized methods and/or computer-assisted methods for identifying patient population subsets that respond to drug therapy similarly.
The invention provides computerized methods and/or computer-assisted methods of targeting drug therapy, particularly dosing regimens and compound selection to an individual subject or patient. The methods incorporate genetic and non-heritable factors into drug selection and titration. The invention provides computational algorithms for recommending a dosing regimen for a particular patient utilizing population models, genotype information, and clinical information. The methods of the invention allow iterative integration of patient information and clinical data. The methods of the invention provide timely, easy to understand, and easy to implement recommendations. Further the invention provides proactive identification of patients potentially requiring more in depth assessment by a clinical pharmacology specialist.
It is therefore a first aspect of the present invention to provide a computerized method and/or computer-assisted method of selecting a dosing regimen for a patient the method that includes the steps of: (a) integrating patient data with patient associated genotype information; (b) generating a drug concentration profile for the patient; (c) integrating the drug concentration profile and the target drug concentration profile; and (d) providing a dosing regimen for a first compound likely to result in the target drug concentration profile in the subject. In a more detailed embodiment, the method further includes the steps of (x) providing a biological sample; (y) monitoring a biomarker in the biological sample; and (z) integrating the biomarker value with the drug concentration profile information. Alternatively or in addition, the patient data may comprise patient demographic data and clinical data. Alternatively or in addition, the clinical data may include information regarding a second compound, where the second compound may modulate metabolism of the first compound. Alternatively or in addition, the first compound may be a neuropsychiatric medication. Alternatively or in addition, the method may further comprise the step of determining the genotype of a patient at one or more loci of interest.
It is a second object of the present invention to provide a computerized method and/or computer-assisted method for selecting a dosing regimen for a patient, where the method includes the steps of (a) obtaining patient data; (b) obtaining patient associated genotype information; (c) integrating the patient data with the patient associated genotype information; (d) generating a drug concentration profile for the patient; (e) integrating the drug concentration profile and a target drug concentration profile; (f) providing a dosing regimen for the compound likely to result in the target drug concentration profile in the subject; (g) providing a biological sample from the patient; (h) monitoring a biomarker in the biological sample; (i) integrating the biomarker value with the drug concentration profile information; (j) generating a second drug concentration profile for the patient; (k) supplying a second target drug concentration profile; (l) providing a second dosing regiment for the compound likely to result in the second target drug concentration profile. In addition, the method may further include the step of performing the processes of (f) through (l) at least a second time. Alternatively or in addition, the method may further include the step of selecting a population model for the patient. Alternatively or in addition, the method may further include the step of generating a probability value for a designated response by the patient.
It is a third aspect of the present invention to provide a computerized method and/or computer-assisted method of selecting a dosing regimen for a patient, where the method includes the steps of: (a) generating statistical population models of drug interactions for a plurality of genotypes; (b) obtaining patient associated genotype information; and (c) establishing a dosing regimen by applying the genotype information against the population models. In addition, the step of generating population models may include the use of Bayesian algorithms. Alternatively or in addition, the population models of drug interactions may be defined for a combination of genotypes and non-genetic information.
It is a fourth aspect of the present invention to provide a computerized method and/or computer-assisted method for selecting one or more drugs for a patient that includes the steps of: identifying the phenotype; providing a first plurality of possible medications based upon the identified phenotype; and calculating a ranked list or a predictive index of medications from the first plurality of medications based upon, at least in part, patient specific genetic factors, non-heritable patient factors and drug specific factors. In addition, the calculating step may further consider one or more preclinical toxicity variables, one or more pharmacokinetic variables, one or more clinical efficacy variables, one or more clinical toxicity variables, one or more clinical safety issues, and/or one or more ease of use/adherence variables. In addition, in the calculating step, one or more of the following variables could contribute linearly: TI (therapeutic index—the ratio of (50% lethal dose/50% therapeutic dose)=measure of the drug's inherent toxicity); F (Bioavailability=fraction of the dose which reaches the systemic circulation as intact drug); fu (the extent to which a drug is bound in plasma or blood is called the fraction unbound=[unbound drug concentration/[total drug concentration]); f-BIND-T (fraction of drug that is a substrate for a drug-specific efflux transporter “T”); MET-L (drug with linear metabolism); f-MET-E (fraction of drug that is metabolized by drug metabolizing enzyme “E”); PEX (percentage of drug metabolizing enzyme “E” with functional polymorphism “X”); CLcr (creatinine clearance=the volume of blood cleared of creatinine per unit time=(liters/hour)); IDR (rate of idiosyncratic reactions); FORM (formulation); FREQ (frequency of daily drug administration); MAT ED (maternal education level); SES (socio-economic class); and TRANS (method of transportation to/from clinic). Alternatively or in addition, in the calculating step, one or more of the following variables could contribute exponentially: ATA (number of functional non-wild type transporter polymorphisms for the specific patient); MET-NonL (drug with non-linear metabolism); AEA (number of functional non-wild type drug metabolizing enzyme polymorphisms for the specific patient); MED-IND (concurrent use of medications that induce metabolizing eanzymes); MED-INH (concurrent use of medications that inhibit metabolizing enzymes); DIET-IND (concurrent use of dietary supplements that induce metabolizing enzymes); DIET-INH (concurrent use of dietary supplements that inhibit metabolizing enzymes); NNT-EFF (number need to treat=number of patients who need to be treated to reach 1 desired outcome); META-EEF (results from an efficacy meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder); NNT-TOX (number need to treat=number of patients who need to be treated to have a 1 toxicity outcome); and META-TOX (results from toxicity meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder).
In another alternative detailed embodiment of the fourth aspect of the present invention, the calculating step may involve linear algebra computational science to integrate disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology principles, and/or patient specific environmental and genetic factors to produce a ranking of potential medications. In addition, or alternatively, the calculating step may assign, for each potential medication, computational values corresponding to a favorability of utilizing the potential medication for a corresponding plurality of factors. In addition, the plurality of factors may include factors from a plurality of the following categories: disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology principles, patient specific environmental and patient specific genetic factors. Alternatively or in addition, the plurality of computational values may include positive values for favorable factors and negative values for unfavorable factors, and the calculating step involves adding the computational values to determine a score. Alternatively or in addition, the plurality of computational values may include positive values for favorable factors and negative values for unfavorable factors and weights corresponding to the relative importance of such factors, and the calculating step involves adding the weighted computational values to determine a score.
In yet another alternate detailed embodiment of the invention, the computerized method may further comprise a step of generating an adherence score corresponding to a predicted likelihood that the patient will adhere to a scheduled therapy or prescription.
It is a fifth aspect of the present invention to provide a computerized method and/or computer-assisted method for selecting a starting dose of a medication for a patient that includes the steps of: for a given medication, determining if the patient is an extensive metabolizer for the medication, an intermediate metabolizer for the medication, or a poor metabolizer for the medication; calculating the starting dose based upon, at least in part, a usual drug dose for a given population (Dpop), the frequency of extensive metabolizers in the given population (fEM), the frequency of intermediate metabolizers in the given population (fIM) and/or the frequency of poor metabolizers in the general population (fPM); and determining a minimal dose adjustment unit for the medication based, at least in part, upon the patient's genetic information. In addition, the step of determining if the patient may be an extensive metabolizer for the medication, an intermediate metabolizer for the medication, or a poor metabolizer for the medication is based, at least in part, upon the patient's genetic information. Alternatively or in addition, (a) the percent of the usual drug dose Dpop for an extensive metabolizer DEM is
DEM=100/(fEM+fIM·S+fPM·R)
where S is the Area Under the Time Concentration Curve for extensive metabolizor subpopulation divided by the Area Under the Time Concentration Curve for intermediate metabolozier subpopulation, and where R is the Area Under the Time Concentration Curve for extensive metabolizer subpopulation divided by the Area Under the Time Concentration Curve for poor metabolizer subpopulation; (b) the percent of the usual drug dose Dpop for a poor metabolizer DPM is
DPM=R·DEM; and
(c) the percent of the usual drug dose Dpop for an intermediate metabolizer DIM is
DIM=S·DEM.
Alternatively or in addition, the minimal dose adjustment unit for the medication may be based, at least in part, upon a number of non-functional alleles, DEM, DIM, and/or DPM.
It is a sixth aspect of the present invention to provide a computerized method and/or computer-assisted method for selecting one or more drugs for a patient that includes the steps of: identifying the phenotype; providing a first plurality of possible medications based upon the patient's diagnosis; and calculating a ranked list or a predictive index of medications from the first plurality of medications based upon, at least in part, patient specific genetic factors, non-heritable patient factors and drug specific factors. In addition, the calculating step may involve linear algebra computational science to integrate disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology principles, and/or patient specific environmental and genetic factors to produce a ranking of potential medications. Alternatively or in addition, the calculating step assigns, for each potential medication, computational values corresponding to a favorability of utilizing the potential medication for a corresponding plurality of factors, where the plurality of factors may include factors from a plurality of the following categories: disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology principles, patient specific environmental and patient specific genetic factors. Alternatively or in addition, the plurality of computational values include positive values for favorable factors and negative values for unfavorable factors, and the calculating step involves adding the computational values to determine a score, where the plurality of computational values may include positive values for favorable factors and negative values for unfavorable factors and weights corresponding to the relative importance of such factors, and the calculating step involves adding the weighted computational values to determine a score.
In another detailed embodiment of the sixth aspect of the present invention, the method may include a step of generating an adherence score corresponding to a predicted likelihood that the patient will adhere to a scheduled therapy or prescription.
It is a seventh aspect of the present invention to provide a computer, a computer system or a computerized tool designed and programmed to perform any or all of the above computer implemented methods. In addition, the computer, computer system or computerized tool may provide a graphical user interface to provide for the collection of appropriate data from users, such as any of the above-discussed factors. Alternatively, or in addition, the computer, computer system or computerized tool may provide a graphical user interface (or any other known computer output, such as a printout) to provide the report, analysis, recommendation or any other output resulting from any of the above-discussed methods.
Defining and describing the often complex relationships of drug action and inter-patient variability has historically been very difficult. Developing pharmacokinetic (PK) and pharmacodynamic (PD) models of these variables provides a method of defining and describing the relationships between drug action and patient variability. Further drug or compound actions (effects) are directly related to the drug concentration at the site(s) of action. There is usually a better relationship between the effect of a given drug and its concentration in the blood than between the dose of the drug given and the effect.
The invention provides population models for various compounds that incorporate pharmacokinetic and pharmacodynamic models of drug action and interpatient variability. Further the invention provides computerized methods and/or computer-assisted methods (including software algorithms) that utilize the one or more population models of the invention to predict a dosing regimen for a particular compound or to predict patient response to a compound. The computerized methods and/or computer-assisted methods (including software algorithms) of the invention generate a prediction regarding a subject's ability to metabolize a compound of interest. The computerized methods and/or computer-assisted methods (including software algorithms) of the invention provide for iterative evaluation of a patient's response to a dosing regimen or compound incorporating data obtained from monitoring at least one suitable biomarker. Often subjects receive more than one medication. These additional medications may affect the subject's ability to metabolize a compound of interest. Thus, in an embodiment computerized methods and/or computer-assisted methods (including software algorithms) of the invention provide a means of integrating information regarding such an additional compound or compounds and the effects of such an additional compound on the subject's ability to metabolize a compound of interest.
A “compound” comprises, but is not limited to, a drug, medication, agent, therapeutically effective agent, neuropsychiatric medications, neurotransmitter inhibitors, neurotransmitter receptor modulators, O-proteins, O-protein receptor inhibitors, ACE inhibitors, hormone receptor modulators, alcohols, reverse transcriptase inhibitors, nucleic acid molecules, aldosterone antagonists, polypeptides, peptides, peptidomimetics, glycoproteins, transcription factors, small molecules, chemokine receptors, antisense nucleotide sequences, chemokine receptor ligands, lipids, antibodies, receptor inhibitors, ligands, sterols, steroids, hormones, chemokine receptor agonists, chemokine receptor antagonists, agonists, antagonists, ion-channel modulators, diuretics, enzymes, enzyme inhibitors, carbohydrates, deaminases, deaminase inhibitors, hormones, phosphatases, lactones, and vasodilators. A compound may additionally comprise a pharmaceutically acceptable carrier.
Neuropsychiatric medications include, but are not limited to, antidepressants, mood elevating agents, norepinephrine-reuptake inhibitors, tertiary amine tricyclics, amitriptyline, clomipramine, doxepin, imipramine, secondary amine tricyclics amoxapine, desipramine, maprotiline, protriptyline, nortriptyline, selective serotonin-reuptake inhibitors (SSRIs), fluoxetine, fluvoxamine, paroxetine, sertraline, citalopram, escitalopram, venlafaxine, atypical antidepressants, bupropion, nefazodone, trazodone; noradrenergic and specific serotonergio antidepressants, mirtazapine, monoamine oxidase inhibitors, phenelzine, tranylcypromine, selegiline; antipsychotic agents, tricyclic phenothiazines, chlorpromazine, triflupromazine, thioridazine, mesoridazine, fluphenazine, trifluoperazine, thioxanthenes, chlorprothixene, clopenthixol, flupenthixol, piflutixol, thiothixene, dibnzepines, loxapine, clozapine, clothiapine, metiapine, zotapine, fluperlapine, olanzapine, butyrophenones, haloperidol, diphenylbutylpiperidines, fluspirilene, penfluridol, pimozide, haloperidol decanoate, indolones, neuroleptics, anti-anxiety/sedative agents, benzodiazepines, chlordiazepoxide, diazepam, oxazepam, clorazepate, lorazepam, prazepam, alprazolam, and halazepam; mood stabilizing agents, lithium salts, valproic acid; attention deficit hyperactivity disorder agents, dextroamphetamine, methylphenidate, pemoline, and atomoxetine; anticonvulsants, phenobarbital, phenytoin, carbamazepine, valproic acid, felbamate, gabapentin, tiagabine, lamotrigine, topiramate, zonisamide, oxcarbazepine, levetiracetam, pregabalin, ethotoin, and peganone; headache medications, ibuprofen, aspirin/acetometaphen/caffeine, diclofenac, ketoprofen, ketorolac, flurbiprofen, meclofenamate, naproxen, ergotamine tartrate, dihydroergotamine, ergotamine, acetometaphen/isometheptene mucate/dichloralphenazone, sumatriptan succinate, zolmitriptan, rizatriptan, naratriptan hydrochloride, ahnotriptan, frovatriptan, eletriptan, diclofenac, fenoprofen, flurbiprofen, kepaprofen, naproxen sodium, amitriptyline, desipramine, doxepin, imipramine, nortriptyline, fluoxetine, paroxetine, sertraline, venlafaxine, trazodone, bupropion, atenolol, metoprolol, nadolol, propranolol, timolol, diltiazem, nicardipine, nifedipine, nimodipine, verapamil, divalproex sodium, gabapentin, valproic acid, and topiramate; and dementia medications, tacrine, donepezil, galantamine, galanthamine, rivastigmine, and memantine.
By “drug” is intended a chemical entity, biological product, or combination of chemical entities or biological products administered to a person to treat, prevent, or control a disease or condition. The term “drug” may include, without limitation, agents that are approved for sale as pharmaceutical products by government regulatory agencies such as the U.S. Food and Drug Administration, European Medicines Evaluation Agency, agents that do not require approval by a government regulatory agency, food additives or supplements including agents commonly characterized as vitamins, natural products, and completely or incompletely characterized mixtures of chemical entities including natural agents or purified or partially purified natural products. It is understood that the methods of the invention are suitable for use with any of the drugs or compounds in the 2005 Physicians Desk Reference, Thomson Healthcare 59th. ed., herein incorporated by reference in its entirety.
The computerized methods and/or computer-assisted methods (including software algorithms) of the invention utilize subject or patient associated genotype information. The term “genotype” refers to the alleles present in genomic DNA from a subject or patient where an allele can be defined by the particular nucleotide(s) present in a nucleic acid sequence at a particular sites(s). Often a genotype is the nucleotide(s) present at a single polymorphic site known to vary in the human population. By “genotype information” is intended information pertaining to variances or alterations in the genetic structure of a gene or locus of interest. Genotype information may indicate the presence or absence of a predetermined allele.
A “loci of interest” may be a gene, allele, or polymorphism of interest. Genes or loci of interest include genes that encode a) medication specific metabolizing enzymes, b) medication specific transporters, c) medication specific receptors, d) enzymes, transporters or receptors affecting other drugs that interact with the medication in question or e) body functions that affect that activities of the medication in question. In an embodiment of the invention loci of interest include, but are not limited to, five cytochrome P450 genes, the serotonin transporter gene, the dopamine transporter gene, and the dopamine receptor genes. The five cytochrome P450 genes can encode CYP2D6, CYPIA2, CYP2C19, CYP2C9 and CYP2E1. Alleles of particular interest include, but are not limited to, the CYPIA2*1A or 1A2*3 allele, the CYP2C19*1A, 2C19*1B, or 2C19*2A allele, and the CYP2D6*1A, 2D6*2, 2D6*2N, 2D6*3, 2D6*4, 2D6*5, 2D6*6, 2D6*7, 2D6*8, 2D6*10, 2D6*12, or 2D6*17 allele. The serotonin receptor genes encode serotonin receptors 1A, 1B, 1D, 2A, or 2C and the dopamine receptor genes encode dopamine receptors D1, D2, D3, D4, D1, and D6. The serotonin transported gene is also an important part of the genotype. Additional genes, alleles, polymorphisms, and loci of interest are presented in Tables 1 and 2.
In an embodiment of the invention, the computerized methods and/or computer-assisted methods (including software algorithms) are utilized to select a dosing regimen for a patient in need of a neuropsychiatric medication. A major gene in the neuropsychiatric panel is CYP2D6. Substrates of CYP2D6 typically are weak bases with the cationic binding site located away from the carbon atom to be oxidized. In particular, substrates of CYP2D6 include amitriptyline, nortriptyline, haloperidol, and desipramine. Some individuals have altered CYP2D6 gene sequences that result in synthesis of enzymes devoid of catalytic activity or in enzymes with diminished catalytic activity. These individuals metabolize SSRIs and tricyclic antidepressants (TCAs) poorly. Duplication/multiplication of the functional CYP2D6 gene also has been observed and results in ultrarapid metabolism of SSRIs and other drugs. Individuals without inactivating polymorphisms, deletions, or duplications have the phenotype of an extensive drug metabolizer and are designated as CYF2D6*1. The CYP2D6*3 and *4 alleles account for nearly 70% of the total deficiencies that result in the poor metabolizer phenotype. The polymorphism responsible for CYP2D6*3 (2549A>del) produces a frame-shift in the mRNA. A polymorphism involved with the CYP2D6*4 allele (18460>A) disrupts mRNA splicing. These changes produce truncated forms of CYP2D6 devoid of catalytic activity. Other poor metabolizers are CYP2D6*5, *10, and *17. CYP2D6*5 is due to complete gene deletion. The polymorphisms in CYF2D6*10 and *17 produce amino acid substitutions in the CYP2D6 enzyme which have decreased enzyme activity. All of these polymorphisms are autosomal co-dominant traits. Only individuals who are homozygous or who are compound heterozygous for these polymorphisms are poor metabolizers. Individuals who are heterozygous, with one normal gene and one polymorphic gene, will have metabolism intermediate between the extensive (normal) and poor metabolizers. Individuals who are heterozygous for duplication/multiplication alleles are ultra-rapid metabolizers.
CYP IA2 metabolizes many aromatic and heterocyclic anilines including clozapine and imipraniline. The CYP1A2 * IF allele can result in a product with higher inducibility or increased activity. (See Sachse et al. (1999) Br. J. Clin. Pharmacol. 47: 445-449). CYP2C19 also metabolizes many substrates including imipramine, citalopram, and diazepam. The CYP2C19*2A, *2B, *3, *4, *5A, *5B, *6, *7, and ‘:’8 alleles encode products with little or no activity. See Theanu at al. (1999) J. Pharmacol. Exp. Th 290: 635-640.
CYP1A1 can be associated with toxic or allergic reactions by extra-hepatic generation of reactive metabolites. CYP3A4 metabolizes a variety of substrates including alprazolam. CYP1B1 can be associated with toxic or allergic reactions by extra-hepatic generation of reactive metabolites and also metabolizes steroid hormones (e.g., 17-estradiol). Substrates for CYP2A6 and CYP2B6 include valproic acid and bupropion, respectively. Substrates for CYP2C9 include Tylenol and antabuse (disulfuram). Substrates for CYP2E1 include phenytoin and carbamazepine. Decreases in activity in one or more of the cytochrome P450 enzymes can impact one or more of the other cytochrome P450 enzymes.
Methods of determining genotype information are known in the art. Genotype information obtained by any method of determining genotype known in the art may be employed in the practice of the invention. Any means of determining genotype known in the art may be used in the methods of the invention.
Generally genomic DNA is used to determine genotype, although mRNA analysis has been used as a screening method in some cases. Routine, commercially available methods can be used to extract genomic DNA from a blood or tissue sample such as the QIAamp@ Tissue Kit (Qiagen, Chatsworth, Calif.), Wizard Genomic DNApurificationIdt (Promega) and the A.S.A.P.™ Genomic DNA isolation Iat (Boehringer Mannheim, Indianapolis, Ind.).
Typically before the genotype is determined, enzymatic amplification of the DNA segment containing the loci of interest is performed. A common type of enzymatic amplification is the polymerase chain reaction (PCR). Known methods of PCR include, but are not limited to, methods using paired primers, nested primers, single specific primers, degenerate primers, gene-specific primers, vector-specific primers, partially-mismatched primers, and the like. Known methods of PCR include, but are not limited to, methods using DNA polymerases from extremophiles, engineered DNA polymerases, and long-range PCR. It is recognized that it is preferable to use high fidelity PCR reaction conditions in the methods of the invention. See also Innis et al., eds. (1990) PCR Protocols: A Guide to Methods and Applications (Academic Press, New York); Innis and Gelfand, eds. (1995) PCR Strategies (Academic Press, New York); Innis and Gelfand, eds. (1999) PCR Methods Manual (Academic Press, New York); and PCR Primer: A Laboratory Manual. Ed. by Dieffenbach, C. and Dveksler, G., Cold Spring Harbor Laboratory Press, 1995. Long range PCR amplification methods include methods such as those described in the TaKaRa LA PCR guide, Takara Shuzo Co., Ltd.
When using RNA as a source of template, reverse transcriptase can be used to synthesize complementary DNA (cDNA) strands. Ligase chain reaction, strand displacement amplification, self-sustained sequence replication or nucleic acid sequence-based amplification also can be used to obtain isolated nucleic acids. See, for example, Lewis (1992) Genetic Engineering News 12(9):1; Guatelli et al. (1990) Proc. Natl. Acad. Sci. USA 87:1874-1878; and Weiss (1991) Science 254:12921293.
Methods of determining genotype include, but are not limited to, direct nucleotide sequencing, dye primer sequencing, allele specific hybridization, allele specific restriction digests, mismatch cleavage reactions, MS-PCR, allele-specific PCR, and commercially available kits such as those for the detection of cytochrome P450 variants (TAG-ITTM kits are available from Tm Biosciences Corporation (Toronto, Ontario). See, Stoneking et al., 1991, Am. J. Hmn. Genet. 48:370-382; Prince et al, 2001, Genome Res. 11(1):152-162; and Myakishev et al., 2001, Genome 11(1):163-169.
Additional methods of determining genotype include, but are not limited to, methods involving contacting a nucleic acid sequence corresponding to one of the loci of interest or a product of such a locus with a probe. The probe is able to distinguish a particular form of the gene or the gene product, or the presence of a particular variance or variances for example by differential binding or hybridization. Thus, exemplary probes include nucleic acid hybridization probes, peptide nucleic acid probes, nucleotide-containing probes that also contain at least one nucleotide analog, and antibodies, such as monoclonal antibodies, and other probes. Those skilled in the art are familiar with the preparation of probes with particular specificities. One of skill in the art will recognize that a variety of variables can be adjusted to optimize the discrimination between variant forms of a gene including changes in salt concentration, pH, temperature, and addition of various agents that affect the differential affinity of base pairing (see Ausubel et al., eds. (1995) Current Protocols in Molecular Biology, (Greene Publishing and Wiley-Interscience, New York).
The exemplary computerized methods and/or computer-assisted methods (including software algorithms) of the invention may employ the following rationale. The pharmacokinetic characteristics of a compound, particularly a neuropsychiatric drug, affect the initial dose of a compound more than the compound's pharmacodynamic properties. A compound's pharmacokinetic profile is a dynamic summation of its absorption, distribution, metabolism, and excretion. Genetic differences in drug metabolizing enzymes (DME) that affect enzyme activity and thus drug metabolism constitute a major component of most compounds' pharmacokinetic variability. DMEs include, but are not limited to, a) medication specific metabolizing enzymes, b) medication specific transporters, c) medication specific receptors, d) enzymes, transporters or receptors affecting other drugs that interact with the medication in question or e) body functions that affect that activities of the medication in question. Most compounds' absorption, distribution, and excretion characteristics are independent of the genetic variability in DMB activity. Specific DMB polymorphisms affect the metabolism of most compounds in a reproducible, predictable, uniform manner. Typically a detectable polymorphism in a specific DME will either have no effect or will reduce enzyme activity. Thus, the subject will have either.
1. two functional alleles (a wild-type, normal, or extensive metabolizer);
2. one functional allele (an intermediate metabolizer); or
3. no functional alleles (a poor metabolizer).
Additionally for certain genes, such as CYP2D6, multiple copies of the gene may be present. In such instances, the presence of more than two functional alleles for a particular gene correlates with an ultrarapid metabolizer state.
Frequently more than one DMEs working either in series or in parallel metabolize a particular compound. The effect of genetic variability for each DME can be determined independently and combined. The invention provides methods of combining or integrating the genetic variability effect for each DMB or DMBs that function sequentially or concurrently. The methods of the invention utilize Bayesian population pharmacokinetic modeling and analysis to integrate and predict the effects of multiple DMEs on metabolism of a particular compound.
Also, the concurrent use of more than one compound can affect the activity of a subject's DMEs. Again, the effect of genetic variability for each DME can be determined independently for each compound. The computerized methods and/or computer-assisted methods (including software algorithms) of the invention utilize Bayesian population pharmacokinetic modeling and analysis to integrate and predict the effects of multiple compounds on one or more DMEs. The methods of the invention allow the integration of information about the genetic variability of one or more DMEs and one or more compounds to generate an area under the time concentration curve (AUC) value. The AUC value reflects the amount of a particular compound accessible to a patient and is the clinically important variable.
The AUC value is determined by drug dose and patient specific pharmacokinetics. Prior to this invention, medical practice utilized a “one size fits all” approach that kept the drug dose constant. In the “one size fits all” approach, variability in pharmacokinetics among patients leads to variability in AUC that results in interpatient clinical variability such as side effects or variable efficacy levels. Thus the methods of the invention provide a means of selecting compound dosing regimens that provide patients with similar AUC values. The methods of the invention integrate the number of genetic variations to be included, the population frequency for each genetic variation, and AUC data for each genetic variation. The methods of the invention transforms a heterogenous population into multiple homogenous subpopulations. Such homogenous subpopulations, suitable dosing regimens, and suitable compounds can be described in a population profile of the invention.
By “dosing regimen” is intended a combination of factors including “dosage level” and “frequency of administration”. An optimized dosing regimen provides a therapeutically reasonable balance between pharmacological effectiveness and deleterious effects. A “frequency of administration” refers to how often in a specified time period a treatment is administered, e.g., once, twice, or three times per day, every other day, every other week, etc. For a compound or compounds of interest, a frequency of administration is chosen to achieve a pharmacologically effective average or peak serum level without excessive deleterious effects. Thus, it is desirable to maintain the serum level of the drug within a therapeutic window of concentrations for a high percentage of time.
The exemplary software program of the invention employs Bayesian methods. The Bayesian methods allow fewer drug measurements for individual PK parameter estimation, sample sizes (e.g. one sample), and random samples. Therapeutic drug monitoring data, when applied appropriately, can also be used to detect and quantify clinically relevant drug-drug interactions. These methods are more informative, cost-saving, and reliable than methods relying on simply reporting results as below, within or above a published range.
Determining a Predictive Index Called the “Simplicity Index” DEFINITIONSThe following abbreviations and definitions will be used in the construction of the simplicity index—the variables are grouped by common themes:
Preclinical Taxicity variables
-
- 1. TD50=called “50% therapeutic dose”=the dose of the medication that results in 50% of the animals tested achieving the desired therapeutic outcome
- 2. LD50=called “50% lethal dose”=the dose of the medication that results in 50% of the animals tested dying
- 3. TI—called therapeutic index=the ratio of LD50/TD50=a measure of the drug's inherent toxicity
-
- 4. F=Bioavailability=fraction of the dose which reaches the systemic circulation as intact drug
- 5. fu=The extent to which a drug is bound in plasma or blood is called the fraction unbound=[unbound drug concentration]/[total drug concentration]
- 6. f-BIND-T=fraction of drug that is a substrate for a drug-specific efflux transporter “T”
- 7. PTX=percentage of transporter “T” with functional polymorphism “X”
- 8. ATA=number of functional non-wild type transporter polymorphisms for the specific patient
- 9. MET-NonL=drug with non-linear metabolism
- 10. MET-L=drug with linear metabolism
- 11. f-MET-=fraction of drug that is metabolized by drug metabolizing enzyme “E”
- 12. PBX=percentage of drug metabolizing enzyme “E” with functional polymorphism “X”
- 13. ABA=number of functional non-wild type drug metabolizing enzyme polymorphisms for the specific patient
- 14. AUC=Total area under the plasma drug concentration-time curve=mg*hour/L
- 15. CL=clearance=the volume of blood cleared of drug per unit time=(liters/hour), CL=dose/AUC
- 16. CLcr=creatinine clearance=the volume of blood cleared of creatinine per unit time=(liters/hour)
- 17. MED-IND concurrent use of medications that induce metabolizing enzymes
- 18. MED-INH=concurrent use of medications that inhibit metabolizing enzymes
- 19. DIET-IND=concurrent use of dietary supplements that induce metabolizing enzymes
- 20. DIET-INH=concurrent use of dietary supplements that inhibit metabolizing enzymes
-
- 21. NNT-EFF=number need to treat=the number of patients who need to be treated to reach 1 desired outcome
- 22. OR=odds ratio=a measure of the degree of association; for example, the odds of reaching the desired outcome among the treated cases compared with the odds of not reaching the desired outcome among the controls
- 23. META-EF=results from an efficacy meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder
-
- 24. NNT-TOX=number need to treat—the number of patients who need to be treated to have 1 toxicity outcome
- 25. OR=odds ratio=a measure of the degree of association; for example, the odds of reaching the drug toxicity among the treated cases compared with the odds of not reaching drug toxicity among the controls
- 26. META-TOX=results from a toxicity meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder
-
- 27. IDR=rate of idiosyncratic reactions
-
- 28. FORM=formulation
- 29. FREQ=frequency of daily drug administration
- 30. MAT ED=maternal education level
- 31. SES=socio-economic class
- 32. TRANS=method of transportation to/from clinic
An algorithm can be used to rank the most appropriate medications for an individual patient. The design of the algorithm requires the initial identification of the phenotype, which provides a preliminary identification of the universe of possible medications. At the next step of the algorithm, the results of the target gene analyses can be sequentially entered. The algorithm that produces the predictive index (called the “simplicity index”) combines the above factors using the following principles:
-
- 1. Each factor contributes differentially based on weighting and scaling variables determined during the validation process.
- 2. The following variables contribute linearly to the final ranking score: TI, F, fu, f-BIND-T, MET=L, f-MET-E, PEX, CLcr IDR, FORM, FREQ, MAT ED, SES, TRANS
- 3. The following variables contribute exponentially to the final ranking score: ATA, MET=NonL, AEA, MED-IND, MED-INH, DIET-IND, DIET-INH, NNT-EFF, META-EEF, NNT-TOX, META-TOX
The algorithm produces a rank list of medications based on the above patient specific genetic factors, non-heritable patient factors and drug specific factors. An exemplary software tool for determining such a predictive index, called the “simplicity index,” is described in detail below.
The following abbreviations and definitions will be used in the determination of the initial starting dose:
Abbreviations:Dpop=the perceived usual drug dosage for the general population
Extensive MetabolizersEM=extensive metabolizer
fEM=frequency of extensive metabolizers in the general population
DEM=Drug dosage for extensive metabolizer subpopulation
AUCEM=Area Under the Time Concentration Curve for extensive metabolizer subpopulation
IM=intermediate metabolizer
fIH=frequency of intermediate metabolizers in the general population
DIM=Drug dosage for intermediate metabolizer subpopulation
AUCIM=Area Under the Time Concentration Curve for intermediate metabolizer subpopulation
PM=poor metabolizer
fPM=frequency of poor metabolizers in the general population
DPM=Drug dosage for poor metabolizers subpopulation
AUCPM=Area Under the Time Concentration Curve for poor metabolizers subpopulation
The following section describes how the dosing for the more homogeneous subgroups is determined; the dosing results are expressed as a fraction of the clinician's usual heterogeneous whole group dosages.
For any one specific polymorphic DME (assuming all other relevant polymorphic DME have normal activity), the usual drug dose seen in a population is the weighted summation of the drug dosages in each genetic different subpopulation expressed in equation 1: (See Kirchheiner Acta Psychiatr Scand 2001:104: 173-192 BUT note authors made mistake in non-numbered equation between Equations 1 and 2, page 178):
Dpop=fPM*DEM+fIM*DIM=fPM·DPM (Equation 1)
Assuming the goal is to maintain the same AUC for all three subpopulations of patients, the following subpopulation dosing relationships hold:
DPM=DEM*(AUCEM/AUCPM) OR DEM*R if R=(AUCEM/AUCPM) (Equation 2)
DIM=DEM*(AUCEM/AUCIM) OR DIM=DEM*S if S=(AUCEM/AUCIM) (Equation 3)
By substituting equations 2 and 3 into equation 1, and then rearranging the equation to solve for the percent dose adjustment needed for each subgroup relative to the population dose:
DEM(%)=100/(fEM+fIM*S+fPM*R) (Equation 4)
DPM(%)=R*DEM (Equation 5)
DIM(%)=S*DEM (Equation 6)
Equations 4, 5, and 6 show bow the dosing for the more homogeneous subgroups is determined and how the dosing results are expressed as a fraction of the clinician's usual heterogeneous whole group dosages.
Determining “Minimal Dose Adjustment Units”The cumulative effect of various genetic or environmentally based alterations in DME activity will result in interpatient variability in subsequent drug dosing requirements. If the variability is large enough, then “one size fits all” dosing approach can cause noticeable toxicity in some patients and lack of efficacy in others. In this situation, clinicians alter their drug prescribing or drug dosing behavior. We define the smallest clinically relevant dosing change used by clinicians to compensate for this interpatient variability as the “minimal dose adjustment unit” (MDA unit). The MDA unit for neuropsychiatric drugs is 20%. This means that a clinician will alter their dosing of neuropsychiatric medications in response to specific information if the dosing change is 20% or greater. Perturbations that either singly or in combination suggest a <20% change in dosing of neuropsychiatric medications are usually ignored.
MDA units are additive—so that a patient with one MDA unit from a genetic polymorphism and one MDA unit from a drug interaction needs a 40% reduction in dose.
Example: The approach in the previous section leads to individualized initial drug dose recommendations for each of the 3 subgroups (extensive, poor and intermediate metabolizers). Each subgroup represents a specific number of functional alleles for the specific DME (extensive metabolizers have 2 functional, intermediate metabolizers have 1 functional and poor metabolizers have 0 functional). The resultant dosing recommendations are expressed as percentages of the clinician's usual starting dose. It is possible to investigate the effect of increasing numbers of non-functional alleles using these new dosing recommendations. For example, if DRx % is the dosing recommendation for subgroup X expressed as a percentage of the clinician's usual starting dose then the following are true:
Effect of claim 1 non-functional allele=(DREM %−DRIM %)/DRFM %
Effect of 2 non-functional allele=(DREM %−DRPM %)/DREM %
Below is a spreadsheet (Table 3) that examines this for CYP2D6, CYP2C19 and CYP2C9. The summary table below demonstrates:
-
- a. it is apparent that each additional nonfunctional allele alters dosing recommendation by at least 20%
- b. there is a “genetic dose”—“dosing reduction” relationship that appears constant across these 3 CYP450 genes. This approach can be used to solidify the importance of subsequent DM genes and to quantify their effect in MDA units.
- c. 2D6 and 2C19 have 1 MDA unit per non-functional allele
- d. 2C9 has 2 MDA units per non-functional allele. This implies that drug metabolized through 2C9 have very large variability in dosage requirements. This confirms the clinical impression about these drugs (warfarin, phenytoin).
For some drugs, there is very little pharmacokinetic genetic variability but rather clinically relevant pharmacodynamic genetic variability most likely at the drug's receptor. For these medications, the impact of genetic testing will be reflected in the final dosage requirements instead of the initial dosage requirements.
Studies that demonstrate this genetic-pharmacodynamic effect will be captured in the software that encodes the calculations used to derive the simplicity index described earlier. This invention will incorporate this information and report not only the rank simplicity index of the potential drug candidates but also those candidates that would require a higher than expected dosing requirement to achieve the desire effect.
Population Models
The purpose of population pharmacokinetic modeling is to describe the statistical distribution of pharmacokinetic parameters in the population under study and to identify potential sources of intra- and inter-individual variability among patients. Population modeling is a powerful tool to study if, and to what extent, demographic parameters (e.g. age, weight, and gender), pathophysiologic conditions (e.g. as reflected by creatinine clearance) and pharmacogenetic variability can influence the dose-concentration relationship. A population pharmacokinetic analysis is robust, can handle sparse data (such as therapeutic drug monitoring data) and is designed to generate a full description of the drug's PK behavior in the population. A “population model” of the invention provides a description of the statistical distribution of at least one pharmacokinetic parameter in a given population and identifies at least on potential source of variability among patients with regards to a particular compound or agent. A population model of the invention may further provide mean parameter estimates with their dispersion, between subject variability and residual variability, within subject variability, model misspecification and measurement error for a particular compound.
An embodiment of the invention provides several novel population models for predicting a medication concentration-time profile and for selecting a dosing regimen based on a user-entered target range (see examples). The computerized methods and/or computer-assisted methods (including software algorithms) of the invention employ population models such as, but not limited to, the novel population models of the invention and externally developed population models. In an embodiment, such externally developed population models are adjusted or rearranged in such a manner that they can be programmed into the software of the invention.
In various embodiments, the computerized methods and/or computer-assisted methods (including software algorithms) of the invention comprise the step of monitoring a biomarker. By “biomarker” is intended any molecule or species present in a patient that is indicative of the concentration or specific activity of an exogenous compound in the subject. Biomarkers include, but are not limited to, a compound, a metabolite of the compound, an active metabolite of the compound, a molecule induced or altered by administration of the compound of interest, and a molecule that exhibits an altered cytological, cellular, or subcellular location concentration profile in after exposure to a compound of interest. Methods of monitoring biomarkers are known in the art and include, but are not limited to, therapeutic drug monitoring. Any method of monitoring a biomarker suitable for the indicated biomarker known in the art is useful in the practice of the invention.
Exemplary computerized methods and/or computer-assisted methods (including software algorithms) of the invention use data generated by therapeutic drug monitoring (TDM). TDM is the process of measuring one or more concentrations of a given drug or its active metabolite(s) in biological sample such as, but not limited to, blood (or in plasma or serum) with the purpose to optimize the patient's dosing regimen. The invention encompasses any means of measuring one or more concentrations of a given drug or its active metabolite(s) in a biological sample known in the art. By “biological sample” is intended a sample collected from a subject including, but not limited to, tissues, cells, mucosa, fluid, scrapings, hairs, cell lysates, blood, plasma, serum, and secretions. Biological samples such as blood samples can be obtained by any method known to one skilled in the art.
The following examples are offered by way of illustration and not limitation.
EXPERIMENTAL Example 1 Optimization of Compound Dosage in an Autistic PatientAn 11-year-old boy with autism was started on risperidone (Risperdal®) therapy, at 0.5 mg two times a day. The patient's pressured speech and labile mood did not improve with time. The lack of efficacy could be due to insufficient coverage or to non-compliance. The patient's dosing regimen was analyzed by the methods of this invention.
Step 1 Dose Appropriateness Analysis.
The patient demographic data (age, sex, weight) and the risperidone dose and times of administration were entered into the program. A population model was selected. The population model selected was a Risperidone model based on data of pediatric psychiatry patients. As risperidone is metabolized by CYP2D6, there are 3 models: one for extensive metabolizers (EM model), one for intermediate metabolizers (IM model) and one for poor metabolizers (PM model).
The genotype of the patient was determined and found to be CYP2D6 *1/*1. This genotype fit the extensive metabolizer (EM model). The patient's data and the genotype were analyzed by an algorithm of the invention and a drug concentration profile for the patient was generated. An exemplary pharmacokinetic model-based simulation of the risperidone concentration time profile based on this patient's data is shown in
Step 2. Integration of Biomarker Evaluation in Recommended Dosage Regimen
The risperidone dose was increased to 1 mg given twice a day (morning and evening). In addition, a biomarker evaluation was performed. Drug levels were ordered and therapeutic drug monitoring were performed. The pre-dose level and two post dose levels (1 h after dose) and (4 h after dose) were measured. These data were entered in the software program. The software program performed a Bayesian recalculation based on the a priori information from the model in combination with the new patient specific information (i.e. the drug levels). Exemplary results of this Bayesian update are shown in
The above-described methods according the present invention can be implemented on a computer system such as a personal computer, a client/server system, a local area network, or the like. The computer system may be portable including but not limited to a laptop computer or hand-held computer. Further the computer may be a general purpose system capable of executing a variety of commercially available software products, or may be designed specifically to run only the drug identification and selection algorithms that are the subject of this invention. The computer system may include a display unit, a main processing unit, and one or more input/output devices. The one or more input/output device may include a touchscreen, a keyboard, a mouse, and a printer. The device may include a variety of external communication interfaces such as universal serial bus (USB), wireless, including but not limited to infrared and RF protocols, serial ports and parallel ports. The display unit may be any typical display device, such as a cathode-ray tube, liquid crystal display, or the like.
The main processing unit may further include essential processing unit (CPU) In memory, and a persistent storage device that are interconnected together. The CPU may control the operation of the computer and may execute one or more software applications that implement the steps of an embodiment of the present invention. The software applications may be stored permanently in the persistent storage device that stores the software applications even when the power is off and then loaded into the memory when the CPU is ready to execute the particular software application. The persistent storage device may be a hard disk drive, an optimal drive, a tape drive or the like. The memory may include a random access memory (RAM), a read only memory (ROM), or the like.
Exemplary Simplicity Index Software ToolAs introduced above an algorithm used to construct the drug predictive index (“simplicity index”) utilizes an initial identification of the disease phenotype (e.g. epilepsy, depression, etc.), which provides a preliminary identification of the universe of possible medications for that condition. An exemplary software tool for producing the simplicity index uses linear algebra computational science to integrate disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology principles, and patient specific environmental and genetic factors to produce a ranking of potential medications for an individual patient based on these factors. There are separate algorithms for each disease phenotype but the algorithms can be run simultaneously. Further, in the exemplary embodiment, there are three components used to produce the final ranking score: a disease matrix, a patient vector and a weighting vector. Each of the five factors and three components will be defined below followed by an example with a sample output. The output contains both the drug predictive index and an adherence score.
Definitions:
Disease Specific Evidence Based Medicine Data
Disease specific evidence based medicine data consists of disease specific efficacy and tolerability data for potentially effective medications. This disease specific efficacy and tolerability data may exist for age or disease subgroups; each age or disease subgroup is considered separately. For example in epilepsy, evidence based data exists for five age groups (neonates, infants, children, adults, and elderly adults) along with four disease subgroups (partial onset seizures, generalized tonic clonic seizures, absence seizures, and myoclonic seizures). In this example, there would be a maximum of 20 separate evidence based data sets covering all age-seizure type combinations.
The first step in the evidence based approach is to identify all relevant scientific information about the efficacy and tolerability of any potential therapeutic modality (medical, surgical or dietary). Articles are identified through multiple methods including, but not limited to, electronic literature searches of the medical literature, hand searches of major medical journals, the Cochrane library of randomized controlled trials, and the reference lists of all studies identified from the electronic literature searches. These articles may include, but are not limited to, randomized control trials, nonrandomized controlled trials, case series, case reports, and expert opinions. Supplementary data is found in package inserts of individual drugs.
The data in each article is evaluated for drug specific efficacy and tolerability data. The analysis is performed using the grading system used by the national scientific organization associated with that specialty. If there is no national scientific organization associated with the specialty then the default grading system is the American Academy of Neurology evaluation system. After the evidence based analysis is complete, the efficacy and tolerability data for each potential drug (stratified by age and disease subgroup) is summarized according to the following Table 5 using a scale from 1+ to −1.
Drug Specific Basic Pharmacology Characteristics
Drug specific basic pharmacology characteristics are evaluated in three categories: Preclinical toxicity, fundamental clinical pharmacokinetic variables and drug safety. An example in the preclinical toxicity category is a drug's therapeutic index. This is defined as the ratio of LD50/TD50 where TD50 is the dose of the medication that results in 50% of the animals tested achieving the desired therapeutic outcome while LD50 is the dose of the medication that results in 50% of the animals tested dying. Fundamental clinical pharmacokinetic variables include, but are not limited to,
-
- i) a drug's bioavailability (fraction of the dose which reaches the systemic circulation as intact drug),
- ii) the fraction of the drug circulating unbound (defined by the extent to which a drug is bound in plasma or blood=[unbound drug concentration]/[total drug concentration]),
- iii) the type of metabolism the drug undergoes (whether linear or non-linear),
- iv) the type of elimination the drug undergoes (e.g. percentage of drug renally excreted or hepatically metabolized) and
- v) the drug's half-life.
Drug safety includes, but is not limited to, the risk of life threatening side effects (idiosyncratic reactions) and the risk of teratogenicity. For each drug under consideration, each variable in the three categories is scored on a scale from +1 (most favorable) to −1 (most unfavorable).
Patient Specific Advanced Pharmacology Factors
Patient specific advanced pharmacology factors include i) bidirectional pharmacokinetic or pharmacodynamic drug-drug interactions and ii) bidirectional pharmacodynamic drug-disease interactions. A pharmacokinetic drug-drug interaction is considered potentially clinically significant if there is a documented Interaction that shows one drug either induces or inhibits the activity of a specific enzyme associated with the metabolism of the other drug by ≧20%. Only concomitant medications actually being taken at the time of the analysis are considered in the analysis. For drug-disease interactions, the word “diseases” refers to all forms of altered health ranging from single organ dysfunction (e.g. renal failure) to whole body illness (e.g. systemic lupus erythematosus). The potential for drug-drug or drug-disease interactions is evaluated on a scale from +1 (most favorable) to −1 (most unfavorable).
To clarify using an example: In a specific patient, assume drug A is being evaluated for use in disease D. The patient is currently taking oral contraceptives, a statin for hypercholesterolemia and is overweight. To evaluate the “Patient specific advanced pharmacology factors” for drug A for this patient there are 8 potential drug-drug interactions and 4 potential drug-disease interactions to evaluate: i) pharmacokinetic effect of drug A on oral contraceptives, ii) pharmacokinetic effect of oral contraceptives on drug A, iii) pharmacokinetic effect of drug A on statin medications, iv) pharmacokinetic effect of statin medication on drug A, v)-viii) the same four combinations mentioned previously but examining the pharmacodynamic interactions between drugs, ix) pharmacodynamic effect of drug A on hypercholesterolemia, x) pharmacodynamic effect of hypercholesterolemia on drug A, xi) pharmacodynamic effect of drug A on weight, xii) pharmacodynamic effect of weight on drug A. If Drug A has i) a clinically significant negative effect on statin pharmacokinetics and ii) causes weight gain then Drug A would receive a score of −1 for these two assessments and a score of 0 for the remaining 10 evaluations. This approach is repeated for each drug under consideration (e.g. drugs B, C, . . . etc).
Patient Specific Environmental Factors
Patient specific environmental factors involve unidirectional, pharmacokinetic or pharmacodynamic, drug-environment interactions. Unidirectional refers to the effect of the environmental agent on the drug. A pharmacokinetic drug-environment interaction is considered potentially clinically significant if there is a documented interaction that shows the environmental agent either induces or inhibits the activity of a specific enzyme associated with the metabolism of the drug by ≧20%. A pharmacodynamic drug-environment interaction is considered potentially clinically significant if there is a documented interaction that shows the environmental factor alters (either positively or negatively) the action of the drug by ≧20%. Only environmental factors occurring at the time of the analysis are considered in the analysis. For drug-environment interactions, the word “environment” refers to all forms of exposure ranging from food (grapefruit juice) to herbal/vitamin supplements (e.g. St. Johns wort) to voluntary toxic exposures (e.g. smoking or alcohol) to involuntary toxic exposures (second hand smoke, pesticides). The potential for drug environment interactions is evaluated on a scale from +1 (most favorable) to −1 (most unfavorable).
Patient Specific Genetic Factors
Patient specific genetic factors involve unidirectional, pharmacokinetic or pharmacodynamic, drug-gene interactions. Unidirectional refers to the effect of the genetic variation on the pharmacokinetic or pharmacodynamic action of the drug. A pharmacokinetic drug-gene interaction is considered potentially clinically significant if there is a documented interaction that shows the genetic factor either increases or reduces the activity of a specific enzyme associated with the metabolism of the drug by ≧20%. A pharmacodynamic drug-gene interaction is considered potentially clinically significant if there is a documented interaction that shows the genetic factor alters (either positively or negatively) the action of the drug by ≧20%. For drug-gene interactions, the word “gene” refers to all forms of genetic variability including DNA variability, mRNA variability, protein alterations or metabolite alterations. The potential for drug-gene interactions is evaluated on a scale from +1 (most favorable) to −1 (most unfavorable).
Disease Matrix
An example (very small) segment of a disease matrix is provided in
Referring to the example disease matrix segment in
Patient Vector Column (Matrix)
A patient vector is constructed for each individual patient. In the exemplary embodiment, the patient vector is a column (not shown in
Weighting Vector
A weighting vector is constructed for each disease matrix. In the exemplary embodiment, the weighting vector is a column (not shown in
Algorithm Output
The main output of the algorithm is a ranking of all potential therapies (medications, surgeries or diet) for that specific disease ranging from most likely to be successful (highest score) to least likely to be successful (lowest score). Each drug's score is the product of the patient vector, the weighting vector and the particular drug's column value in the disease matrix. The dosing for the drug is determined by the algorithm described above. In the exemplary embodiment, the output display includes the top 5 factors contributing and the lowest 3 factor detracting from the score are included for evaluation. Above the ranking is an adherence score reflecting the likelihood the patient will adhere to the proposed treatment regimen. The determination and interpretation of this number is described in the Adherence score section.
Adherence Score
The adherence score is determined in a similar fashion to the simplicity index: the score is the product of an “adherence matrix”, a patient vector and a weighting vector. For each disease, potential adherence problems are assessed using a series of approximately 10 yes/no/unknown questions. If all questions are answered unknown then the adherence score will be 50% implying a 50% chance the patient will adhere to the treatment regimens. The more questions that are answered “no”, the higher the adherence score and the greater the chance the patient will adhere to the prescribed treatment regimen. The more questions answered “yes”, the lower the adherence score and the greater the chance the patient will not adhere to the prescribed treatment regimen.
Patient Example:
-
- History: The patient is a 7 year old male presenting with frequent staring episodes lasting 30-60 seconds associated with unresponsiveness, facial twitching and extreme tiredness afterwards. He develops a funny taste in his mouth in the few minutes before the events occur. He has had about 10 of these in the past year with 3 in the last month. The patient does not have depression, ADHD or anxiety but does have frequent migraines. The patient is currently taking erythromycin for an infection but takes no chronic medications, There is no family history of epilepsy. The patient loves to drink grapefruit juice. The family has insurance, no transportation problems and no identifiable stressors.
- Physical examination: Normal in detail except the patient is very overweight
- Lab tests: EEG shows normal background and focal discharges in the temporal lobe. MRI of the brain is normal. Pharmacogenetic testing shows a CYP2C9 polymorphism that makes him a poor metabolism for drugs metabolized by CYP2C9.
- Diagnosis: Newly diagnosed idiopathic partial epilepsy characterized by partial onset seizures.
- Need: Determine the best antiepileptic medications for this specific patient.
Step 1: As can be seen if
Step 2: As can be seen if
Step 3: As can be seen in
Step 4: As can be seen in
Step 5: As can be seen in
Step 6: As can be seen in
Step 7: As can be seen in
Step 8: As can be seen in
Step 9: As can be seen in
Step 10: As can be seen in
Step 11: As can be seen in
Having described the invention with reference to the exemplary embodiments, it is to be understood that it is not intended that any limitations or elements describing the exemplary embodiment set forth herein are to be incorporated into the meanings of the patent claims unless such limitations or elements are explicitly listed in the claims. Likewise, it is to be understood that it is not necessary to meet any or all of the identified advantages or objects of the invention disclose herein in order to fall within the scope of any claims, since the invention is defined by the claims and since inherent and/or unforeseen advantages of the present invention may exist even though they may not be explicitly discussed herein.
Finally, it is to be understood that it is also within the scope of the invention to provide any computer, computer-system and/or computerized tool as is known by one of ordinary skill in the art that is designed, programmed or otherwise configured to perform any of the above-discussed methods, algorithms or processes.
All publications, patents, and patent applications mentioned in the specification are indicative of the level of those skilled in the art to which this invention pertains. All publications, patents, and patent applications are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually incorporated by reference.
Claims
1-39. (canceled)
40. A method for producing a ranked list of drugs for administration to a patient in need thereof, the ranked list produced by calculating for each drug a score that is the product of a patient vector, a weighting vector, and the drug's column value in a disease matrix, the method comprising the steps of:
- (a) receiving at a processor the patient's disease phenotype;
- (b) providing a disease matrix comprising a first plurality of drugs based upon the patient's disease phenotype; the disease matrix comprising columns and rows, each column representing a drug in the plurality of drugs and consisting of a set of numbers, each number consisting of a computational value relating the drug to a factor represented by a row of the matrix such that each of the first plurality of drugs is related to a plurality of factors in the disease matrix, the plurality of factors in the disease matrix comprising factors from a plurality of the following categories: (1) disease specific evidence based medicine data, (2) drug specific basic pharmacology factors, (3) patient specific advanced pharmacology factors, (4) patient specific environmental factors, and (5) patient specific genetic factors, wherein (1) the plurality of factors for disease specific evidence based medicine data is presented for each combination of age (neonates, infants, children, adults, and elderly adults) and disease subgroup from a plurality of variables selected from the group consisting of
- Drug efficacy score (score ranging from −1.0 to 1.0 based on the level of evidence for that drug's efficacy in that age and disease subgroup)
- Drug toxicity score (score ranging from −1.0 to 1.0 based on the level of evidence for that drug's toxicity in that age and disease subgroup)
- OR (odds ratio, a measure of the degree of association),
- NNT-EFF (number need to treat, the number of patients who need to be treated to reach 1 desired outcome),
- NNT-TOX (number need to treat, the number of patients who need to be treated to have a 1 toxicity outcome),
- META-EFF (results from an efficacy meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder), and
- META-TOX (results from toxicity meta-analysis of clinical trials involving medications used to treat a neuropsychiatric disorder); (2) the plurality of drug specific basic pharmacology characteristics comprises a plurality of preclinical toxicity variables, clinical pharmacokinetic variables, and drug safety variables selected from the group consisting of
- TD50 (50% therapeutic dose=the dose of the medication that results in 50% of animals tested achieving the desired therapeutic outcome),
- LD50 (50% lethal dose=the dose of the medication that results in 50% of the animals tested dying),
- TI (therapeutic index, the ratio of LD500/TD50=a measure of the drug's inherent toxicity),
- F (Bioavailability, the fraction of the dose which reaches the systemic circulation as intact drug),
- fu (the extent to which a drug is bound in plasma or blood is called the fraction unbound=[unbound drug concentration/[total drug concentration]),
- f-BIND-T (fraction of drug that is a substrate for a drug-specific efflux transporter “T”),
- MET-nonL (drug with non-linear metabolism),
- MET-L (drug with linear metabolism),
- T1/2 (drug half life)
- f-MET-E (fraction of drug that is metabolized by drug metabolizing enzyme “E”),
- AUC (total area under the plasma drug concentration time curve),
- CL (clearance, the volume of blood cleared of drug per unit time),
- IDR (rate of idiosyncratic reactions),
- TERATO (rate of teratogenicity)
- FORM (formulation), and
- FREQ (frequency of daily drug administration; (3) the plurality of patient specific advanced pharmacology factors comprises a plurality of variables that include clinically significant i) bidirectional pharmacokinetic or pharmacodynamic drug-drug interactions and ii) bidirectional pharmacodynamic drug-disease interactions selected from the group consisting of
- CLCR, (creatinine clearance=the volume of blood cleared of creatinine per unit time=(liters/hour)),
- MED-IND (concurrent use of medications that induce metabolizing enzymes),
- MED-INH (concurrent use of medications that inhibit metabolizing enzymes), and
- DISEASES (all forms of altered health ranging from single organ dysfunction to whole body illness); (4) the plurality of patient specific environmental factors comprises a plurality of variables that can cause clinically significant unidirectional pharmacokinetic or pharmacodynamic drug environment interactions selected from the group consisting of
- DIET-IND (concurrent use of dietary supplements that induce metabolizing enzymes),
- DIET-INH (concurrent use of dietary supplements that inhibit metabolizing enzymes),
- Food
- Herbal/vitamin supplements,
- Voluntary toxic exposures, and
- Involuntary toxic exposures;
- and (5) the plurality of patient specific genetic factors comprises a plurality of variables that can cause clinically significant unidirectional pharmacokinetic or pharmacodynamic drug gene interactions selected from the group consisting of
- PTX (percentage of transporter “T” with functional polymorphism “X”
- ATA (number of functional non-wild type transporter polymorphisms for the specific patient),
- PEX (percentage of drug metabolizing enzyme “E” with functional polymorphism “X”), and
- AEA (number of functional non-wild type drug metabolizing enzyme polymorphisms for the specific patient),
- (c) creating a patient vector for the patient with the same row labels as the disease matrix including the categories of disease specific evidence based medicine data, drug specific basic pharmacology characteristics, patient specific advanced pharmacology factors, patient specific environmental factors, and patient specific genetic factors in (3), (4), and (5);
- (d) providing a weighting vector with the same row labels as the disease matrix;
- (e) calculating, via a processor, a score for each drug in the disease matrix, where the score is the product of the patient vector, the weighting vector, and the drug's column value in the disease matrix; and
- (f) outputting a ranked list of drugs from highest to lowest of the scores calculated in step (e).
41. The method of claim 40, wherein the following variables contribute linearly: TI, F, fu, f-BIND-T, MET-L, f-MET-E, PEX. CLCR, IDR, FORM, FREQ, MAT ED, SES, and TRANS.
42. The method of claim 40, wherein the following variables contribute exponentially: ATA, MET-NonL, AEA, MED-IND, MED-INH, DIET-IND, DIET-INH, NNT-EFF, META-EEF, NNT-TOX, and META-TOX.
43. The method of claim 40, wherein step (b)(3) comprises a step of assigning computational values for patient specific advanced pharmacology factors selected from clinically significant (i) bidirectional pharmacokinetic drug-drug interactions and (ii) bidirectional pharmacodynamic drug-disease interactions, wherein
- (i) a significant pharmacokinetic drug-drug interaction is defined as one for which there is a documented interaction that shows one drug either induces or inhibits the activity of a specific enzyme associated with the metabolism of the other drug by ≧20%, and
- (ii) a significant pharmacodynamic drug-disease interaction is defined as one for which there is a documented interaction that shows a ≧20% positive or negative effect of a drug on a disease.
44. The method of claim 40, wherein step (b)(4) comprises a step of assigning computational values for patient specific environmental factors selected from clinically significant (i) unidirectional pharmacokinetic or (ii) pharmacodynamics drug-environment interactions, wherein
- (i) a significant pharmacokinetic drug-environment interaction is defined as one for which there is a documented interaction that shows the environmental agent either induces or inhibits the activity of a specific enzyme associated with the metabolism of the drug by ≧20%, and
- (ii) a significant pharmacodynamic drug-environment interaction is defined as one for which there is a documented interaction that shows the environmental factor alters (either positively or negatively) the action of the drug by ≧20%.
45. The method of claim 40, wherein step (b)(5) comprises a step of assigning computational values for patient specific genetic factors selected from clinically significant (i) unidirectional pharmacokinetic or (ii) pharmacodynamic drug-gene interactions, wherein
- (i) a significant pharmacokinetic drug-gene interaction is defined as one for which there is a documented interaction that shows the genetic factor either increases or reduces the activity of a specific enzyme associated with the metabolism of the drug by ≧20%, and
- (ii) a significant pharmacodynamic drug-gene interaction is defined as one for which there is a documented interaction that shows the genetic factor alters (either positively or negatively) the action of the drug by ≧20%.
46. The method of claim 40, further comprising a step of generating an adherence score corresponding to a predicted likelihood that the patient will adhere to a scheduled therapy or prescription.
47. The method of claim 40, wherein the factors are obtained utilizing an interactive computerized tool.
48. The method of claim 47, wherein the interactive computerized tool collects factors utilizing a hierarchical question and answer session.
49. The method of claim 40, further comprising a step of selecting one or more drugs from the ranked list for administration to the patient and administering the one or more selected drugs to the patient.
50. The method of claim 40, wherein the plurality of factors in the disease matrix comprises all variables listed in (1) through (5) for which data is available.
51. The method of claim 40, wherein the patient vector is a column of the disease matrix.
52. The method of claim 40, wherein the patient vector is a 1 by N matrix.
53. The method of claim 40, wherein the weighting vector is a column of the disease matrix.
54. The method of claim 40, wherein the weighting vector is a 1 by N matrix.
55. The method of claim 46, wherein the factors included in the adherence score include one or more, or all, of the following factors:
- MAT ED (maternal education level),
- SES (socio-economic class), and
- TRANS (method of transportation to/from clinic).
Type: Application
Filed: Oct 14, 2013
Publication Date: Jan 1, 2015
Inventors: Tracy A. Glauser (Cincinnati, OH), Richard J. Wenstrup (Cincinnati, OH), Alexander A. Vinks (Cincinnati, OH), John Pestian (Cincinnati, OH)
Application Number: 14/053,220
International Classification: G06F 19/00 (20060101);
