Precision medicine and therapies of the future

Abstract

Precision medicine in the epilepsies has gathered much attention, especially with gene discovery pushing forward new understanding of disease biology. Several targeted treatments are emerging, some with considerable sophistication and individual-level tailoring. There have been rare achievements in improving short-term outcomes in a few very select patients with epilepsy. The prospects for further targeted, repurposed, or novel treatments seem promising. Along with much-needed success, difficulties are also arising. Precision treatments do not always work, and sometimes are inaccessible or do not yet exist. Failures of precision medicine may not find their way to broader scrutiny. Precision medicine is not a new concept: It has been boosted by genetics and is often focused on genetically determined epilepsies, typically considered to be driven in an individual by a single genetic variant. Often the mechanisms generating the full clinical phenotype from such a perceived single cause are incompletely understood. The impact of additional genetic variation and other factors that might influence the clinical presentation represent complexities that are not usually considered. Precision success and precision failure are usually equally incompletely explained. There is a need for more comprehensive evaluation and a more rigorous framework, bringing together information that is both necessary and sufficient to explain clinical presentation and clinical responses to precision treatment in a precision approach that considers the full picture not only of the effects of a single variant, but also of its genomic and other measurable environment, within the context of the whole person. As we may be on the brink of a treatment revolution, progress must be considered and reasoned: One possible framework is proposed for the evaluation of precision treatments.

Keywords: anti-seizure drugs; failure; genetics; personalized; pharmacogenetics; surgery.

Figure 1

Typical current PM scenarios. The same color scheme is used throughout for each of the steps crystalized in this figure. A, The idealized paradigm, with a linear progression from clinical description of an epilepsy, to determination of its genetic cause, a definition of necessary and sufficient disease mechanisms, establishing the basis of a rational treatment, subsequent clinical trials, licensing, and seizure‐free outcomes with improvements of comorbidities. B, Realization of the paradigm in tuberous sclerosis. It is worth noting the actual timeline, which is not depicted linearly. Also noteworthy is that although everolimus is licensed for use for seizures in tuberous sclerosis, the necessary funding mechanisms are not always in place. C, Application of the linear paradigm in Dravet syndrome due to mutations in SCN1A. The pathophysiology of Dravet syndrome has been shown to be more complex than originally reported, but the reasoned strategy of avoidance of sodium channel–blocking antiseizure drugs is still typical practice, with published, although not from formal trials, evidence of benefit., Agents for which randomized controlled trials have been undertaken in Dravet syndrome, such as stiripentol, cannabidiol, and fenfluramine, are not PM within the draft framework outlined at the end of this document. D, The paradigm in GLUT1 deficiency syndrome. Ketogenic dietary therapy (KD) was instituted in people with the clinical syndrome on biochemical grounds before discovery of the genetic cause. KD is considered standard treatment for GLUT1 deficiency disorder, although there have been no randomized controlled trials. Its frequently quoted position as a PM has been reconsidered. Abbreviations: EU, European Union; KD, ketogenic diet; PM precision medicine; RCT, randomized‐controlled trial; SEGA, subependymal giant cell astrocytoma

Figure 2

An example of complexity and potentially misleading genetic information. The proband had severe treatment‐resistant right occipital lobe epilepsy, referred on a combination of five antiseizure drugs, including three sodium channel–blocking drugs. His epilepsy was life‐threatening, with recurrent episodes of generalized tonic‐clonic status epilepticus, including an episode with acute renal failure. High‐resolution brain magnetic resonance imaging (MRI) was unremarkable. He had a personal and family history of febrile seizures in his sister and mother. At the time of referral, only single gene testing was available; gene panel, exome, and genome sequence were available, and neither ExAC nor gnomAD were in existence. In view of the history in the proband and nuclear family, SCN1A was Sanger sequenced in the nuclear family. The heterozygous missense mutation ( in A, dotted oval) c.1804G>A, p.Glu602Lys was identified in the proband, his mother, and sister, and not in the asymptomatic father. Current evaluation shows that the variant absent from gnomAD, has a CADD score of 13, and a REVEL (https://sites.google.com/site/revelgenomics/) score of 0.399: 75.4% of disease‐causing variants, and only 10.9% of neutral variants have a REVEL score of >0.5; variants proven to have a functional consequence, as the gold standard measure of pathogenicity had CADD scores >20 and REVEL scores >0.7 in a recent study of clinical prediction of the effect of variants in SCN1A. On bioinformatic grounds, this variant would be considered unlikely to be deleterious. These measures were unavailable at the time the proband was seen. Functional testing of the variant in vitro showed loss‐of‐function (B; Ref.69). At this point, the full pedigree (C, blue rectangle) became available through the proband's maternal grandmother. The SCN1A variant found in the proband was absent in the maternal grandmother, from whom a family history of right occipital lobe epilepsy (filled symbols) or migraine, inconsistently lateralized, without features of seizures (dotted symbols) was obtained. It emerged that a distant relative (dotted arrow) had had a right occipital lobectomy elsewhere with full control of generalized tonic‐clonic seizures, and the identification of a dysembryoplastic neuroepithelial tumor on histological examination (D). Pre‐surgical evaluation was then undertaken in the proband, with confirmation from intracranial electroencephalography (EEG) (E) of mesial right occipital lobe seizure onset: The electrode track is visible—seizure onset was from contacts colored red. Having regularly experienced ictal and postictal hemianopia, and with the knowledge of benefit from surgery in the distant relative, the patient selected an occipital lobectomy, and has been free of all seizures for over 10 years, except during one episode of diarrhea. He has a fixed right homonymous hemianopia and remains on the five antiseizure drugs on which he was referred, not wishing to come off any. Histology of the resection specimen was unremarkable: There was no evidence of a dysembryoplastic neuroepithelial tumour (F). Although the SCN1A variant identified causes loss‐of‐function in vitro, it is not the cause of the familial epilepsy shared by the proband; it may be responsible for the nuclear familial febrile seizures, but had only that information been acted upon (eg, with withdrawal of the sodium channel–blocking antiseizure drugs), it seems unlikely that he would have become seizure‐free, as he had a cluster of generalized tonic‐clonic seizures during the episode of diarrhea some years after surgery. The genetic cause of the epilepsy in the family has not yet been identified, despite exome sequencing of several family members

Figure 3

Real‐life examples of more complex PM scenarios. A, The linear PM development paradigm as detailed in Figure 1A. B, In a patient with frequent generalized tonic‐clonic seizures and a profound developmental and epileptic encephalopathy, with spastic quadriparesis, unable to walk and intolerant of most testing, with no syndromic diagnosis made, trio exome sequencing revealed a de novo KCNA2 mutation (c.894G>T, p.Leu298Phe; Patient 6 in Ref.70) demonstrated to be gain‐of‐function in vitro, reversible by 4‐aminopyridine. Because of regulatory and other issues, time to provision of this putative PM for the patient was >18 months and was eventually funded by the family. However, the frequency of generalized tonic‐clonic seizures increased significantly on 4‐aminopyridine, which had to be discontinued. A child with the same mutation treated elsewhere with 4‐aminopyridine became seizure‐free (H. Lerche, personal communication). Review of the exome sequence data did not reveal any other de novo mutations, but identified three unique inherited variants, one of which (CACNA1C p.Gly490Arg inherited from an asymptomatic parent) has been shown previously to cause loss‐of‐function with a Brugada syndrome phenotype and shortened QT interval. The proband's electrocardiography (ECG) showed lateral early repolarization with prominent U waves, but no Brugada‐like changes, with a short QTc interval. However, further testing of the electroclinical significance of this variant in the patient has not actually proved possible. The cause of seizure aggravation by a putative PM effective in another patient with the same mutation has remained unknown. C, A young man with longstanding epilepsy with focal‐onset bilateral tonic‐clonic seizures had tried several antiseizure drugs. His mother was seizure‐free on monotherapy. As a PM strategy, he underwent surgical evaluation, including high‐resolution magnetic resonance imaging (MRI; unremarkable) and intracranial electroencephalography (EEG) recording that demonstrated focal frontal onset. Surgical resection led to a brief period of seizure freedom. Histopathology demonstrated an unusual pathology with early lipofuscin accumulation in dysplastic neurons. Postoperatively, whole genome sequencing revealed a DEPDC5 stopgain mutation inherited from his mother. The different degrees of seizure control between mother and son are unexplained. Activation of the mammalian target of rapamycin (mTOR) pathway has already been demonstrated on tissue pathology in the proband. Given the life‐threatening nature of the proband's epilepsy, everolimus was considered. There is no trial or anecdotal basis for its use currently, no funding for its use, and no guidance for the duration of treatment that may be required, which were among the factors leading the patient to decline the offer to attempt to seek its individual use through clinical or research pathways. D, Two brothers were previously reported who were found to have homozygous mutation in GAMT. Seizure control was achieved with creatine supplementation. Antiseizure drugs were withdrawn without recurrence of seizures. Unmanageable behavioral deterioration necessitated reintroduction of some antiseizure drugs. In complex genetic epilepsies, it may be necessary to plan a PM approach in a multidisciplinary context, considering not only what might be done if the PM fails, but how life might change if PM succeeds and seizures are brought under control

Figure 4

Precision medicine (PM) beyond treatment of seizures alone. The discovery of mutations in ATP1A3 as the main cause of the ultra‐rare condition alternating hemiplegia of childhood (AHC), catalyzed interactions between existing communities of scientists working on the protein and gene, family organizations, and clinicians and scientists interested in AHC, leading to annual joint workshops since 2012. Delegates attending the 5th Annual Symposium on AHC are featured in the photograph (A), with the meeting brochure (B). Joint working has facilitated research into important aspects of ATP1A3‐related conditions, such as cardiac involvement (C & D, respectively showing ATP1A3 immunolabeling in explanted human heart tissue and episodes of asystole in an individual with AHC)—undertaking studies in such rare conditions can be very difficult without close interaction with family organizations. Such groups can also identify and promote research questions, such as the occurrence of apnea in AHC (E), a phenomenon of deep concern to parents of children with AHC

Figure 5

An example of difficulties in the PM approach (Top Panel) in an individual case. As documented in Table 1, quinidine was identified as a possible PM for seizures caused by gain‐of‐function mutations in KCNT1. Middle Panel: A child carrying a mutation studied in that first report was treated with quinidine to good effect, although the report carried information only on a short follow‐up period once seizure freedom had been obtained. Bottom Panel: A previously reported gain‐of‐function mutation in KCNT1, p.Gly288Ser (Ref.78), was detected in a man in his 40s. Parents declined testing for themselves, but based on the previous reports, the variant was considered pathogenic. Generalized tonic‐clonic seizures were considered by his parents and carers as the major burden in his life currently, repeatedly precipitating hospital admissions due to status epilepticus. However, with accumulating evidence of uncertain outcomes following quinidine treatment, including for this variant (see Table 1; Ref.93), concerns about cardiac toxicity, difficulties in obtaining serial electrocardiography (ECG) studies , and serum level data from the proband, who would not be able to voice symptoms himself, regulatory hurdles in the UK (which may differ from other jurisdictions), no decision has yet been made about use of quinidine in the patient