Personalized medicine in epilepsy patients

The large number of different syndromes and seizure types together with an interindividual variable response to antiepileptic drugs (AEDs) make the treatment of epilepsy challenging. Fortunately, the last few years have been characterized by a huge interest in epilepsy genetics and two methods, genome-wide analyses and next-generation sequencing, have definitely given the possibility to write a new chapter in the book of treatment of epilepsy, the chapter on precision medicine. Epilepsy offers a good opportunity for the personalization of therapy if we consider that at least one third of epileptic patients do not achieve complete seizure control with the currently available pharmacological treatments, treatment is still often empirical and precise therapy, based on the pathogenesis and the mechanism of each AED is not generally possible because this mechanism often remains incompletely known. In addition, new drugs are often not targeted but developed using in vivo seizure models, to be potentially used by the largest number of patients. This method leads to a therapy aimed at treating the symptoms and the seizures rather than the single pathogenic mechanism of each seizure type or syndrome. In this narrative review, we summarize the established evidence regarding pharmacogenomics in epilepsy and discuss the basis of precision medicine.


INTRODUCTION
Epilepsy is a medical condition defined by recurrent, or likely recurrent, seizures due to excessive electrical discharges in a group of brain cells [1] .Nowadays, treatment is limited to a wide range of antiepileptic drugs (AEDs) with different mechanisms of action [2] , which can only provide control of symptoms (seizures).It is ineffective in a large percentage of patients and can sometimes also worsen seizures or cause adverse reactions [3] .The heterogeneous etiology of epilepsy, the large number of different syndromes and seizure types, together with an individually variable response to AEDs, make the treatment of this condition still challenging [4,5] .Moreover, adverse drug effects can be severe and life-threatening and some AEDs can even worsen seizure control and induce new seizure types [6] .It is now well established that genetic factors are the explication of the interindividual variability in the response to AEDs [5] ; different genes can be mutated thus affecting drug pharmacokinetics, drug pharmacodynamics or causing epilepsy itself.In addition, studies have shown that epigenetic mechanisms are involved in brain modifications due to epilepsy [7] .The term precision medicine aims to describe a personalization of treatments that ideally have to be targeted towards the precise molecular pathogenesis of disease [8] .Perhaps, the best realization of precision medicine is, to date, achieved in oncology specialties, which is called cancer precision medicine, a Barack Obama initiative in terms of funding.Epilepsy offers a good and challenging opportunity for the personalization of treatments for different reasons: it affects ~1% of worldwide populations at the age of 20 years and 3% at the age of 75 years [9] , many patients are still not seizure-free or have adverse drug reactions, and the genetic bases of many epileptic syndromes are well studied nowadays while new genes are discovered every day.
To write this manuscript, a literature search was conducted through the PubMed database using the terms "epilepsy", "pharmacogenomics", "antiepileptic drugs", "pharmacogenetics" and "diagnostic sequencing" from 1997 to 2018.Additional information was found in the reference lists of selected articles.

PHARMACOGENOMICS IN EPILEPSY
Genetic mutation can alter response to AEDs at both pharmacokinetic (e.g., polymorphism in gene involved in drug metabolism) and pharmacodynamic level (e.g., polymorphism in brain AED targets, such as ion channels).Other mechanisms involved are mutations in genes causing epilepsy or the modification of the expression of enzymes and other molecules involved in the pathogenesis of pharmacoresistance or adverse drug reactions [10,11] .Pharmacogenomics is the science that studies how these genetic differences affect drug response both in terms of efficacy and susceptibility to adverse drug reactions [11] .It is in the last two decades that advances in genetic testing have led to a systematic search for gene variations that could predict drug response and ultimately improve the efficacy and safety of epilepsy therapies.As we already know, adverse drug effects can be severe and life-threatening and some AEDs can even worsen seizure control and induce new seizure types [6] .

Genetic influences on AED metabolism
It is now well established that the clearance of most AEDs is linked to cytochrome P450 (CYP) enzymes activity.Polymorphisms of the gene encoding CYP enzymes can alter their activity, thus affecting serum AED concentrations and lead to drug toxicity [12] [ Table 1].A good example is phenytoin (first generation AED) which is metabolized primarily by CYP2C9 and also by CYP2C1.Some individuals have CYP2C9 polymorphisms, which cause a reduced activity of the enzyme thus leading to low phenytoin clearance, higher serum phenytoin concentrations and a greater risk of central nervous system adverse effects.CYP2C9*2 (rs1799853) and CYP2C9*3 [rs1057910(C)] are the best documented of these polymorphisms [13,14] .To the best of our knowledge, pre-treatment testing for CYP2C9 variants is not considered as routine practice in any center.In our opinion, the best practice remains the clinical monitoring of signs of toxicity and of serum drug levels and it is also important to consider drug interactions if patients are taking other AEDs [15] .
Another AED that has shown reduction of clearance due to genetic polymorphisms is phenobarbital (first generation AED).The enzyme involved is CYP2C19, with a difference of up to about 20%-50% in the reduction of clearance [16] .To the best of our knowledge, there is no evidence that genetic testing improves the outcome of phenobarbital therapy compared with clinical observation and monitoring of serum drug concentration [17] .A Japanese study showed that CYP2C19 and CYP3A5 polymorphisms are involved in the metabolism of zonisamide (second generation AED) leading to a reduction of clearance up to 16%-30% [18] .The clinical relevance of these changes is still to be clarified.In addition, drug interactions can increase zonisamide clearance, thus reducing the effects of the genetics variants.
Valproic acid (VPA) (first generation AED) is an AED that can cause severe adverse effects including severe hyperammonemia and non-alcoholic fatty liver disease.Liver damage due to VPA has been associated with the formation of the toxic 4-ene metabolite mediated by CYP2C9 [19] .It has been shown that genetic testing for CYP2C9 variant and subsequent different treatments significantly reduced VPA misdosing and hyperammonemia in a controlled trial [20] .A recent study showed that the content of 4-ene-VPA had no direct correlation with the incidence of liver dysfunction.In addition, VPA metabolism is also influenced by the polymorphism of ACSM2A, which can lead to higher levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) compared with wild-type subjects, however, the mutations had no effect on the VPA-related liver damage [21] .
Finally, VPA metabolism is also influenced by genetically determined variation of uridine diphosphate glucuronosyltransferase (UGT) enzymes.A study conducted on a pediatric cohort showed that -161C > T single nucleotide polymorphism in UGT2B7 gene led to significant differences in plasma VPA concentrations.Patients with the CC genotype had lower adjusted plasma VPA concentrations than those with CT or TT genotype (P = 0.028) [22] .Lamotrigine (second generation AED) is eliminated almost entirely by glucuronidation.An old study conducted in a small number of patients with Gilbert syndrome found that lamotrigine clearance was lower in these patients than in healthy control [23] .The clinical impact of this difference is still unknown.Other studies showed that UGT1A4 genetic polymorphisms also influence lamotrigine clearance, such as -219C > T/-163G > A mutations in the 5'-upstream regions of the UGT1A4 gene, which significantly increases lamotrigine (LTG) serum concentrations.However, other factors may play an important role in lamotrigine metabolism such as age, body weight and interaction with VPA [24,25] .
Another second-generation AED, retigabine, is also metabolized by N-glucuronidation and N-acetylation, but its clearance has been found to be unaffected in Gilbert's syndrome although arylamine N-acetyltransferase-2 acetylator status did influence the disposition of the weakly active metabolite N-acetyl-retigabine [26] .There is no evidence of any benefit of dose adjustments for genetic polymorphisms [27] .

Pharmacoresistance
It is well established that at least one third of epileptic patients do not achieve complete seizure control with currently available pharmacological treatments (AED) [28] .The cause of pharmacoresistance, is still not totally understood [29] , but it has been shown that several ATP-dependent transport proteins are involved in drug resistance.Drug transporters actively eliminate toxins from the cells, including many drugs.One of the most studied transporters is P-glycoprotein (P-gp), encoded by the ABCB1 gene [30] .In the brain, P-gp is expressed in astrocytes, endothelial cells and neurons, and there is evidence that its overexpression in epileptogenic tissue can be involved in pharmacoresistance to AEDs [31] .Several studies have shown that ABCB1 gene variants are involved in the response to treatment in epilepsy patients.A retrospective casecontrol study of C3435T variants reported that patients with pharmacoresistant epilepsy were more likely to have the CC genotype than the TT genotype ( (15.7% CC and 29.6% TT) [32] .Subsequently, other studies were made but were inconclusive [33,34] .In fact, two meta-analyses of these studies revealed no significant association between resistance to AEDs and the C3435T genotype [33,35] .Most recent studies continue to provide evidence of significant associations between drug resistance and ABCB1 3435 genotypes [36] .In addition, the results of a recent meta-analysis indicates that ABCB1 C3435T polymorphism, especially TT genotype, plays an important role in refractory epilepsy; the authors suggest that genetic screening of this genotype, before starting the treatment, may be useful to predict AED response [37] .Studies of gene variants for other transporter proteins such as multidrugresistanceassociated protein 2 (MRP2)/ABCC2 failed to provide evidence for a clinical impact of these tests in epilepsy therapy [38,39] .A recent meta-analysis of studies on the expression and cellular distribution of MRP1 suggest that MRP1 is overexpressed in both neurons and astrocytes of patients with drug resistant epilepsy and that its inhibition may lead to treatment response due to increased local drug availability [40] .Many AEDs exert their pharmacological effects, at least in part, by blocking voltage-dependent sodium channels.Polymorphisms in the genes coding for these channels have been studied to investigate their relationship with drug resistance.Many studies [41,42] have suggested that SCN1A polymorphisms influence the response to sodium channel blocking AEDs.However, other studies did not confirm this association [43,44] .In fact, some patients with Dravet syndrome show seizure aggravation after sodium channel blockers intake [45] .However, it has been shown that lamotrigine can improve seizure control in some of these patients [46] .This conflicting results show how useful improvements in the field of precision medicine in epilepsy will be.Studies on other drug targets such as the GABA-A receptor [47] , the KCNT1 potassium channel [48] and the synaptic vesicle proteins SV2A, SV2B and SV2C [49] , did not show any significant association.However, a recent Chinese study showed an association between some single-nucleotide polymorphisms of KCNJ10 gene and antiepileptic drug resistance [50] .In conclusion, to date, we still do not totally understand the real contribution of polymorphisms of AED target genes due to a lack of studies to evaluate the contribution of other factors to the individual variability of responses [51] .

Adverse drug reaction
A link between genetic polymorphisms and the risk of side effects is well established.In particular, certain human leukocyte antigen (HLA) alleles are associated with an increased risk of idiosyncratic adverse drug reactions [52] [Table 2].There is a special link between different types of drug and different alleles, like HLA-B*15:02 and HLA-A*31:01 for carbamazepine (first generation AED), HLA-B*57:01 for abacavir, HLA-B*13:01 for dapsone and HLA-B*58:01 for allopurinol.
There is evidence that in Han Chinese, Thai, and Malaysian populations the presence of HLA-B*15:02 is a genetic marker of risk for Stevens-Johnson syndrome (SJS) induced by carbamazepine, probably due to the activation of cytotoxic T-lymphocytes which is mediated by this allele [53,54] .A study showed that there is a significant decrease of carbamazepine-induced SJS-toxic epidermal necrolysis (TEN) if the subjects carrying the HLA-B*1502 allele were previously identified and carbamazepine was thus avoided as a therapy [55] .The frequency of this allele in the specific population is in the order of 1%-8% in residents of China and most South Asian countries [52] , with peaks as high as 15%-21% among Indonesians and 34% among Filipinos [56] , while instead the frequency of the HLA-B*15:02 allele is very low (< 0.5%) in people of European, or North East Asian (Korean and Japanese) ancestry [57] .For these reasons regulatory agencies and guidelines (like the Clinical Pharmacogenetics Implementation Consortium Guidelines for HLA-B genotype and carbamazepine dosing) recommend that patients from Han Chinese and other South Asian ethnic groups be routinely genotyped for HLA-B*15:02 before starting treatment with carbamazepine, and that carbamazepine be avoided if possible in carriers of the allele [56,58] .The HLA-B*15:02 allele has also been associated with an increased risk of SJS and TEN after therapy with other AEDs, including phenytoin (first generation AED) and, to a lesser extent, lamotrigine [59] , and oxcarbazepine (second generation AED) [60] .In fact, these AEDs have an aromatic ring just as carbamazepine (CBZ) does.
In addition, another allele, the HLA-A*31:01 has been linked with increased risk of carbamazepine-induced hypersensitivity reactions, such as maculopapular exanthema, SJS/TEN and also drug reaction with eosinophilia and systemic symptoms (DRESS) [61,62] .HLA-A*31:01 is frequent in many ethnic groups, both in Europeans and Orientals [63] .A recent meta-analysis confirmed a significant association of HLA-A*31:01 with carbamazepine-induced DRESS but a weaker association with CBZ-SJS/TEN, thus suggesting that HLA-A*31:01 is a genetic predictor for CBZ-DRESS but not for CBZ-SJS/TEN [64] .Studies have been made to find if, apart from a clinical benefit, genotyping for HLA-A*31:01, which reduces the incidence of cutaneous adverse drug reactions, could be economically convenient and the results show that this routine practice would be cost-effective [65] .
In conclusion, to date, there is limited evidence regarding the value of genotyping in predicting AED response.The best example of a useful pharmacogenetic variant in epilepsy is testing for HLAB*15:02 to prevent serious adverse cutaneous reactions in individuals from South Asian ethnic groups in whom initiation of carbamazepine therapy is considered.Unfortunately, there is no widely applicable genetic test to predict response to AED treatment in patients with the most common forms of epilepsy.However, we believe that genetic testing will help in preventing adverse drug reactions or to prescribe the correct dose of AED.Furthermore, it will help researchers better understand epilepsy genetics and approach new precision medicine.

PRECISION THERAPY
The treatment of epilepsy is still largely based on empirical science and the prescription of drugs for epileptic patients cannot be based on the mechanisms of action of these.The performance of a personalized therapy is limited by the broad clinical phenotypic spectrum and the underlying heterogeneous aetiology.However, recent scientific acquisitions about genetics mechanisms, studies of neuroimaging and epilepsy neurobiology are providing many indications about the choice between the drugs of the past or the newest ones, thus laying the foundations for a new era in the treatment of epilepsy, in which patients will benefit from therapies based on the etiological cause of diseases [68] .
The newest AEDs offer many therapeutic advantages compared to traditional and older generation therapies, in fact they have a lower risk profile of side effects and have by far fewer drug interactions.Despite this, treatment is still largely empirical and rational prescribing based on the mechanism of action in an individual patient is not generally possible.A fundamental problem is that the main mechanisms of action and biochemical effects of drugs are not yet known in depth or not completely clear.This is in part due to the fact that the exact molecular targets of current AEDs are largely unidentified.Therefore, drug discovery is not targeted and instead relies on developments using in vivo seizure models.Further, as a consequence of these models, the common antiepileptic therapies are limited to controlling the epileptic symptoms but not to preventing the epileptogenic events.It seems likely that we will need to understand epileptogenesis in order to devise novel therapeutic interventions.Accordingly, research is rather oriented towards the mechanisms of action and the complex molecular processes to which diseases are subjected [69] .
Genetic confirmation through the use of specific molecular diagnostic techniques in epileptic syndrome can provide an important contribution in establishing a more precise prognosis and in the evaluation of recurrence of epileptic seizures [70,71] .Furthermore, it offers an excellent opportunity to obtain better information on targeted treatments and the development of targeted drugs.The impact of increased knowledge is of paramount importance in particular in patients with epileptic encephalopathies, a group of neurodevelopmental disorders characterized by marked epileptic activity associated with regression of neurological development [72,73] .
Current drugs directly reduce neuronal excitability mainly by modulating ion channels and neurotransmitter receptors.Recent acquisitions have revealed further pathways that show different mechanisms such as synaptic vesicle traffic, mammalian target of rapamycin (mTOR) signaling, chromatin remodeling and transcription, thus offering new therapies [74,75] .
The genetics of epilepsy is still very complex, mutations of different genes can cause the same syndrome or even mutations in a single gene (for example, SCN1A) can be associated with a wide range of phenotypes, ranging from feverish convulsions to severe epileptic encephalopathies [76] .
It is also important to underline that there is a wide individual variability of response to antiepileptic treatment and the genetic differences between patients are most likely implicated in this variation [Table 3].

Drug use
On the basis of a particular form of epilepsy we can explain, in whole or in part, the answer, both positive and negative (paradoxical) to certain AEDs.For instance, the clinical picture of Dravet syndrome can be worsened by the use of carbamazepine and phenytoin, since the disease is caused by mutations in the sodium channel gene (SCN1A) and these drugs interfere on the mechanism of action mentioned blocking the channel [4,69,77] .In contrast, sodium channel blocking is considered the first choice therapy for the epileptic syndrome associated with mutations in SCN8A (another sodium channel gene) and KCNQ2 genes [4,49] .
LTG is a known blocker of the sodium channel and N-type calcium channels and its use has sparked controversy.Some works claimed it as a factor in exacerbation of seizures, therefore research has led to the avoidance of its use in patients with Dravet syndrome.On the other hand, other studies assert a positive effect in some patients with Dravet syndrome.This beneficial effect could be explained by the mechanism involving the cyclic-nucleotide channels activated by hyperpolarization processes [47,51] .At present, approved therapy for Dravet syndrome includes the use of three drugs in a polytherapy that are valproate, clobazam and often stiripentol.As regards stiripentol, this is the only drug used in Dravet syndrome for which an important randomized controlled trial has been performed (when combined with valproate and clobazam); but it is widely known that the use of stiripentol, valproate and clobazam can cause serious side effects and, as is well known, it is capable of reducing the number of critical episodes but not completely eliminating them [77] .
New and effective treatment strategies with possibly novel mechanisms are therefore needed.
Many studies confirm the efficacy of fenfluramine in Dravet syndrome [78,79] .This drug was initially developed as an appetite suppressant, but withdrawn from the market due to serious adverse effects, including cardiac and pulmonary problems such as valvular heart disease and pulmonary hypertension [80,81] .Fenfluramine has the ability to act on the serotoninergic cascade [82] , but unfortunately the specific mechanisms by which it carries out its anti-epileptic actions still need to be discovered.More recent works affirm that fenfluramine significantly reduced epileptiform discharges in SCN1A knock-out morphants [83,84] .
Retigabine (or ezogabine) (third generation AED), most often used in adult patients, is a drug that primarily acts as a positive allosteric modulator of KCNQ2-5 ion channels (Kv7.2-7.5) and is the first drug used to treat epilepsy, which exploits its action on the potassium channels of neuronal cells [47] .In vitro studies show that the most potent action of this drug is on the KCNQ2/3 heteromeric channels, which has been closely related to numerous forms of epileptic disorders from benign familial seizures to a severe form of epileptic encepha- m-TOR: mammalian target of rapamycin; USP9X: ubiquitin-specific peptidase 9 X-linked; ACTH: Adrenocorticotropic Hormone; nAChRs: nicotinic acetylcholine receptors Table 3. Precision therapy lopathy early onset [47] .Recent in vitro experiments have shown that retigabine opens Kv7 potassium channels and restores normal channel function in KCNQ2-related encephalopathy mutations [51] .
Regarding the treatment of GRIN2A-related epileptic disorder, new strategies are based on the use of N-methyl-D-aspartate (NMDA) receptor antagonists [57] : memantine, an NMDA receptor antagonist approved by the Food and Drug Administration, has shown the ability to reduce the frequency and onset of seizures in some types of encephalopathies affecting children defined early-onset epileptic encephalopathy, associated with a de novo missense mutation in GRIN2A (p.Leu812Met).However, the same drug was not effective in another case-report in which the authors demonstrated a different mutation (p.Asn615Lys) in the same gene, but with a completely different effect on the function of the protein [85] .
Quinidine, a well-known anti-arrhythmic drug, has been used to restore in vitro the hyperactivity of the KCNT1 mutant potassium channel in Xenopus oocytes [86] .A recent clinical case presented a child in whom oral administration of this drug led to an improvement in epilepsy and psychomotor skills which he suffered from because of focal epileptic seizures due to a lack of the protein product of KCNT1 [87] .However, the same drug had no efficacy in another patient with KCNT1 mutation and with severe secondarily generalized focal seizures.Therefore, we must use a great deal of attention and experience, in using quinidine, in patients with such genetic dysfunction [88] .The serious side effects on the liver of patients using valproate and presenting a POLG gene mutation are well known [89] .
Some papers have shown a strong correlation between the genetic polymorphisms that affect microsomal epoxide hydrolase (EPHX1) gene and pharmacokinetics of carbamazepine in Chinese patients suffering from some forms of epilepsy and in Kosovan people of Albanian ethnicity with epilepsy [90] .
A study demonstrated that in patients with SCN1A, ABCC2 and UGT2B7 genetic polymorphisms there is the possibility of making important changes to oxcarbazepine maintenance doses [91] .
In an Indian population the genetic contribution of CYP1A1 alleles on treatment outcome in people with epilepsy was studied.In particular, it has been demonstrated, through a study carried out on a population of Indian women with epilepsy, that the mutation in the variant rs2606345, which consists in a reduction of the CYP1A1 expression, determines a lack of response to the first line treatment with most used AEDs [92,93] .
Rapamycin (sirolimus) has been shown to decrease cortical dysplasia and tuber growth in patients with tuberosis sclerosis, also decreasing epileptic symptoms; moreover, the same drug is useful in patients with hemimegalencephaly.One possible explanation is that sirolimus blocks the mTOR complex, whose overexpression causes dysplasia in various organs and the formation of glial bands, subependymal nodules, tumors during both the fetal phase and subsequent central nervouse system development [2] .Another clinical trial was performed to understand the effects of everolimus (a derivative of sirolimus) on subependymal giant cell astrocytoma growth and showed a sustained effect on tumor reduction over ≥ 5 years of treatment, with no safety concerns [94] .
Very recently we learned that some forms of epilepsy listed below are associated with Dishevelled, Egl-10 and Pleckstrin domain containing protein 5 (DEPDC5) loss-of-function mutations, including autosomal dominant nocturnal frontal lobe epilepsy, familial focal epilepsy with variable foci, familial temporal lobe epilepsy, rolandic epilepsy and other non-lesional focal childhood epilepsies, and focal epilepsy associated with focal cortical dysplasia, both familial and sporadic.It has recently been demonstrated that mutations of DEPDC5 are associated with increased production activity of the mTOR signal cascade factors, because of the capability of DEPDC5 to reduce the mTOR activity.In global DEPDC5 knockout rats, in which the therapy was performed during the prenatal period, with rescued growth delay and embryonic lethality, prevented enhanced cell size and dysmorphism of neurons [95] .Numerous studies show that the reduction of the mTOR signaling pathway is certainly a basic mechanism underlying the pathophysiology of epilepsy in rodents and humans [96] .
GATOR1 complex gene mutations leading to mTORC1 pathway upregulation are closely related to the onset of focal epilepsy with cortical malformations.However, unfortunately, there is not enough scientific evidence to be able to state that treatment with mTOR inhibitors in patients with gene mutations affecting the GATOR1 complex subunit is efficacious [97] .
Mutations in the PRICKLE genes are associated with seizures in humans, zebrafish, mice, and flies, offering a seizure-suppression pathway that may be evolutionarily conserved.This path has never been studied in the past by the researchers who deal with antiepileptic therapy.The inhibition of ubiquitin-specific peptidase 9 X-linked (USP9X) can arrest PRICKLE-mediated seizures, so USP9X molecules may be evaluated as a new class of anti-seizure therapy [98] .
There are various associations between specific genetic mutations and non-pharmacological and rational therapeutic decisions.The benefits of using a ketogenic diet in patients with Glut1 deficiency syndrome due to mutations in SLC2A1 are widely known.The clinical spectrum of this condition is heterogeneous and includes a group of epileptic syndromes ranging from mild cases with no epilepsy to severe cases characterized by intractable epilepsy, infantile spasms and developmental delay.The treatment of first choice to resolve the symptoms due to neuroglycopenia consists in providing an alternative fuel to the brain through the ketones; this may be achieved through the use of a ketogenic diet [99] .It is of fundamental importance to start a ketogenic diet as early as possible, and therefore to make an early diagnosis to provide brain nourishment and control seizures [100] .However, the benefits on neurodevelopment seem controversial [101] .
Biallelic mutations of the ALDH7A1 gene cause a deficit of antiquitin and these mutations underlie pyridoxine (vitamin B6)-dependent epilepsy.The seizures are due to the fact that the lack of antiquininit is determined to accumulate 1-piperideine-6-carboxylate condenses with pyridoxal 5'-phosphate and inactivates this enzyme cofactor, essential component for the metabolism of the central nervous system and in specific neurotransmitters.The simple therapy using pyridoxine in most cases can lead to full control of convulsive episodes.ALDH7A1 analysis of gene could also be used for prenatal diagnosis of pyridoxinedependent epilepsy [102] .
Cannabis sativa was the first plant cultivated by humans for purposes other than food or other useful purposes.For thousands of years, extracts of the plant have been used for a variety of therapeutic conditions, including the treatment of epilepsy.In recent times, with the power of the new social media, the general population have been more and more interested in and approached the topics concerning the use of cannabisbased therapies and in particular their effectiveness as a therapy in drug-resistant epilepsy, demonstrating significant benefits for the therapeutic indication.The research has also been very much addressed in its use for a severe epileptic encephalopathy of childhood called Dravet syndrome.Given the low number of studies and research on the use of cannabidiol as a drug for the therapy of variable symptoms including epileptic ones, recently the research has been oriented with more attention in this regard.For example, some children were recruited to whom cannabidiol was administered in particular formulations, a botanically-derived pharmaceutical, with compassionate use programs and were placed in an open-label trial [103,104] .
In 2017 the results of a double-blind, placebo-controlled trial of cannabidiol in children with Dravet syndrome were published.The percentage of patients who had a greater than 50% reduction in convulsive seizure frequency was 43% with cannabidiol and 27% with placebo (OR 2.00, 95% CI 0.93-4.30,P = 0.08).These results have been much needed to meet the high expectations regarding cannabis therapies for epilepsies, and this trial will serve as a model for future studies, much more specific in particular in rare diseases [105] .
One therapy that must be mentioned is that of infantile spasms.This condition requires a specific therapy which should be carried out as early as possible to allow an adequate response to the therapy and avoid, as often happens, some cognitive sequelaes that can be determined if treatment is started too late.The two bestestablished therapies are hormonal treatments (steroids or adrenocorticotropic hormone) and vigabatrin (second generation AED); however, little international consensus exists on which treatment to use first.
In 2017 O'Callaghan et al. [106] published a work that reported the results of an International Collaborative Infantile Spasms Study, the most important open-label randomized trial for the treatment of infantile spasms carried out so far.For this multinational trial, 102 hospitals in five countries screened 766 infants over a 7-year period, 377 of whom were recruited to the study.The children were randomly divided into either the group in which a hormonal monotherapy was administered, or in the one in which a therapeutic association was combined with the AEDs.The patients were free of spasms between days 14 and 42.This outcome was achieved in 72% of infants in the combined treatment arm compared with 57% of those on hormonal therapy alone.The response to therapy was better in patients who did not have a well-established cause for their disease, without an identifiable etiology [106] .
The different genetic polymorphisms can determine variations both in the pharmacokinetic field (absorption, distribution, transport, metabolism, elimination) and in those of the pharmacodynamics (action sites, etc.).
The genetic test is the background of an individual variation in the response to antiepileptic treatment, in terms of efficacy and adverse reactions.
The KCNA1 mutations are associated with episodic ataxia type 1 [107,108] and homozygous Kv1.1 knockout mice have disrupted sleep, and seizures peak during times of light [109] .Almorexant, a dual orexin receptor antagonist used in sleeping disorders, improves sleep and reduces seizure severity, suggesting that it may be useful for patients with mutations in KCNA1 [110] .Treatment of homozygous Kv1.1 knockout mice with a ketogenic diet also reduces seizure frequency and has been shown to successfully extend life span [100] .Finally, homozygous Kv1.1 knockout mice with partial genetic ablation of NaV1.2 exhibit reduced duration of spontaneous seizures, and a significant improvement of survival rates argues that blockers of this Na + channel may have some treatment value [111] .
Heterozygous mutations in SCN8A are associated with an epileptic encephalopathy characterized by developmental delay, seizure onset within the first 18 months of life, and intractable epilepsy.These patients have multiple seizure types, including infantile spasms, generalized tonic-clonic seizures, absences, and focal seizures [112] .Mirroring the situation with SCN1A and SCN2A, where there are both mild and severe phenotypes, very recently there were two reports of familial benign infantile seizures, without cognitive impairment, and some with paroxysmal dyskinesias with missense variants in SCN8A [113,114] .The research provides a number of potential targeted therapeutic options.In particular, targeting the increased persistent Na + current makes sense.Significantly, GS967, a specific blocker of persistent Na + current, extends the survival of the NaV1.6 (N1768D/+) mouse [115] .Riluzole, a drug known for its mild efficacy in the treatment of amyotrophic lateral sclerosis, has proved to be an important therapeutic possibility, thanks to its capability as Na + current blocker.Riluzole is effective in blocking the early depolarization events seen in CA1 pyramidal neurons [116] , but as yet there is no published evidence of its efficacy in the NaV1.6 (N1768D/+) mouse.Patel et al. [117] have also demonstrated that increased persistent current of Nav1.6 mutant channels can be preferentially reversed with cannabidiol.
A further mechanism involving ionic channels has been extensively studied, is the one concerned with blocking the Na + /Ca 2+ exchanger which is implicated in the disease mechanism [116] .
De novo mutations in KCNA2 have been identified in cases of early infantile epileptic encephalopathy.In this pathology the onset of seizures is between 5 and 17 months, with a phenotypic spectrum including febrile and afebrile, hemiclonic, myoclonic, myoclonic-atonic, absence, focal dyscognitive, focal, and generalized seizures; mild to moderate intellectual disability; delayed speech development; severe ataxia [122] .The KCNA2 spectrum also encompasses milder familial epilepsy [123] .Some treatment options have been taken into consideration.The 4-aminopyridine is an approved K + blocker already being trialed in patients carrying gain-of function mutations.Xie et al. [124] have tested a non-targeted approach and reported that the carbonic anhydrase inhibitor, acetazolamide, is capable of rescuing the motor incoordination in Pingu mice.
Early evidence supported the idea that CACNA1A mutations were associated with genetic generalized epilepsies [125,126] .Micro-deletions that encompass CACNA1A and a single truncating mutation have been associated with severe epileptic encephalopathies that include infantile spasms and West syndrome.De novo missense mutations have been convincingly shown to cause severe epileptic encephalopathies with seizure types that typically include focal, tonic, and tonic-clonic seizures, severe intellectual disability and motor impairment [127,128] .It is known that acetazolamide and 4-aminopyridine are able to decrease, in tottering mice, the high-power low-frequency oscillations, which are thought to be a marker of cortical excitability, so these drugs are evaluated as treatment options for episodic ataxia 2 [129] .Spontaneous seizures in the CACNA1A knockout mouse can be abolished by knocking out CACNA1G [130] .This suggests that T-type Ca 2+ channel blockers, including ethosuximide, may be good therapeutic options [131] .
Numerous studies of literature on animal models have shown concrete evidence regarding the transcriptional changes of hyperpolarization-activated cyclic nucleotide-gate (HCN) channels in correlation with excitability, but there remains little evidence on the association between epilepsy and genetic change correlated to the mentioned channels.A study by Nava et al. [132] showed that HCN1 missense mutations were associated with the development of early infantile epileptic encephalopathy.In addition, these patients had symptoms associated with Dravet syndrome with intellectual impairment and autism.In gain-of-function disease, HCN1 blockers may be useful.Ivabradine is a use-dependent broad-spectrum blocker of HCN channels approved for use in angina pectoris.It is an important drug because it is well tolerated by most patients, however, its capacity is not known, nor is the timing to overcome the encephalic barrier [133] .The hypnotics propofol and ketamine, as well as the anesthetic isoflurane, are reported to inhibit HCN1 channels [134][135][136] .Finally, the AEDs, lamotrigine and gabapentin (second generation AED), have both been reported to enhance HCN currents [137,138] , potentially benefiting in particular patients with loss-of-function mutations.
In patients with mutation of the receptor for nACh, even if the type of mutation has been known for a long time, numerous precision therapy options have not yet been found.Only carbamazepine proved to be useful in patients with nicotinic acetylcholine receptors mutations with approximately 70% showing remission on low doses.Molecular and cellular studies argue that drugs that block nAChR should be effective in disease caused by mutations in CHRNA4, CHRNB2, and CHRNA2c [139,140] [Table 4].
In conclusion, cost-effectiveness of precision medicine was considered in this review.An interesting and recent study investigating the cost-effectiveness of a whole exome sequencing (WES)-based gene panel (targeted WES) in patients with severe epilepsies of infancy found that early targeted WES had lower total cost than a late WES.A diagnostic approach with early targeted WES and limited metabolic testing leaded to 7 additional diagnoses compared to investigation without targeted WES (46/86 vs. 39/86), with lower total cost ($455,597 USD vs. $661,103 USD) and lower cost per diagnosis ($9904 USD vs. $16,951 USD) [141] .
Another study was conducted in a neurogenetic clinic of a tertiary hospital in Argentina and confirmed the effective WES-based approach [142] .Additional studies on cost-effectiveness in precision medicine in epilepsy are needed, as healthcare system demands better allocation of its limited resources, still pursuing the best possible outcome for patients.

CONCLUSION
Mechanisms underlying epilepsy are multiple and it is very difficult to realize "the gold standard" of AED.Precision medicine is the future for antiepileptic treatment and can bring a better outcome also for some kind of epilepsy syndrome that in the past had been considered quite intractable.For example, in a particular

Table 4. Precision therapy: genetic influences and applied methods
USP9X: ubiquitin-specific peptidase 9 X-linked; ACTH: Adrenocorticotropic Hormone; nAChRs: nicotinic acetylcholine receptors type of epilepsy syndrome like Glut1 deficiency (chetogenic diet) or in GRIN2A mutations (memantine) or in tuberous sclerosis complex (rapamycin), precision medicine is already possible with good results.The future effort of the research must be to identify new drugs against specific pathogenic mechanisms, or a specific action of mutated proteins, up to a gene replacement therapy, and also the individual genetic polymorphism that could result in impaired effect of the conventional AEDs [143] .