THE DEBATE
Earlier this month, the epilepsy community gathered in Barcelona for the 32nd International Epilepsy Congress. There were three interconnected topics that dominated much of the program: genetics of epilepsy, rare epilepsy syndromes and personalized medicine. The other large topic that dominated the agenda was the new epilepsy and seizure classification by the ILAE, and it felt like two separate conferences.
While most scientists are looking inside the cell, looking for genetic changes, many practicing physicians are looking at what they can see, and creating new nomenclatures to define seizure types. And from the patient side this makes a large difference, because the same child with an SCN2A mutation might be diagnosed by a specialist as having SCN2A encephalopathy, and as having Dravet syndrome by another physician based on clinical symptoms.
There has been a trend in the recent years towards creating new syndromes that group patients by the gene that they have affected when their epilepsy is found to have a genetic cause. Some examples of this are CDKL5 deficiency disorder or STXBP1 encephalopathy, which until recently would have be classified as atypical Rett syndrome and Lennox-Gastaut or Ohtahara syndrome respectively, at least in most cases. And that is the catch, that the same gene mutation can lead to different phenotypes in different patients and actually manifest itself as a complete different syndrome.
We see this mismatch between gene and syndrome every day. For example, the nigh before I few to the congress a mother had contacted asking how it was possible that her daughter with Dravet syndrome had exactly the same gene mutation that another patient diagnosed with Ohtahara syndrome had. Then at the congress, Professor Jackie French from NYU presented the case of a family with an inherited mutation in SCN1A that included individuals with mild forms of epilepsies, for example just febrile seizures, but also one child with the much more severe Dravet syndrome. As a physician, she would adjust their medication to their seizure type and treat more aggressively the individual with Dravet syndrome, because their diseases are indeed different ones.
So what is the best classification approach? Should we talk about syndromes based on the gene that causes them or should we talk about them (and treat them) based on the clinical characteristics that they display?
This debate is not just about semantics, it affects a lot how well we develop new medicines and how many patients will benefit from them. So it is important that we get it right.
What follows is my personal view of which syndrome classification is best, when to use it, and what regulatory changes are needed in order to get more medicines to all patients with rare genetic epilepsy syndromes.
OVERLAP BETWEEN SYNDROMES AND GENES
The way I see the debate between genetic and symptom-base syndrome classification is that they are not exclusive; in fact, they form a matrix where many of the epilepsy genes can produce phenotypes that fit into many of the classical (phenotypic) syndrome boxes.
For illustration purposes, I’ve selected a handful of genes and syndromes and created the matrix below. While not all of these genes are found in all of these syndromes, what we have to remember is that many of the epilepsy genes are present across a large number of epilepsy syndromes.
With this in mind, what would be the correct diagnosis for the child of the previous story that had a diagnosis of Ohtahara syndrome before performing genetic testing but was later found to have an SCN1A mutation (typical of Dravet syndrome)? Should the genetic finding change the diagnosis to Dravet syndrome?
I would argue that both a genetic and a clinical diagnosis are needed, so the best diagnosis for this child, if his clinical symptoms are indeed a better match for Ohtahara than for Dravet syndrome, would be “Ohtahara syndrome caused by SCN1A deficiency”. And the other child in the same story would have “Dravet syndrome caused by SCN1A deficiency”. Except for a few truly monogenic syndromes, most epilepsy syndromes have many possible genetic (and sometimes non-genetic) etiologies, and most epilepsy genes produce multiple phenotypes, so I don’t need we need to prioritize the symptomatic classification over the genetic one or vice versa; both are useful and both are necessary. Instead, we should point to the right box in the matrix when diagnosing a patient, which provides both sets of genetic and clinical data.
My purpose when creating this visual matrix is not to break down already rare syndromes into even smaller diseases. On the contrary, the purpose of this exercise is to make the syndrome classification more flexible, so that we can cluster syndromes, and create more flexible indications for developing new drugs and for treating more patients.
REGULATORY IMPLICATIONS
For as long as we didn’t have a good understanding of the genetics behind epilepsy, we have been perfectly OK with symptomatic epilepsy classifications. But the progresses in genetics in the recent decade, and the appearance of many new “genetic epilepsy syndromes”, has opened the door to the development of new drugs that specifically target those genetic defects, creating a new regulatory landscape where our former classification of syndromes based on symptomatology falls short.
The patient groups are already confortable using the genetic syndrome classification, the conference rooms talk more and more about personalized medicine and genetic syndrome classification, and companies pipelines start getting populated with programs that specifically target those abnormal genes and proteins that cause these epilepsies. The next step is to start using the genetic syndrome classification as a suitable drug indication in the drug development process as well.
Let’s go back again to the genes versus (phenotypic) syndromes matrix and see the different therapeutic implications of using one classification versus another.
The first example (scenario A) corresponds to the current preferred way of selecting drug indications. There are 8 different products that have received the Orphan Drug Designation to treat Dravet syndrome. This makes sense because Dravet syndrome has been recognized as a separate clinical syndrome since 1978, way before the genetic mutations that give rise to this phenotype were uncovered. It also makes sense because all of the drugs approved or in clinical trial for Dravet syndrome are symptomatic, meaning that they are not treating any genetic problem.
In short: if the drug treats a particular cluster of symptoms that corresponds to a syndrome, then the best indication for that drug is indeed the classical (phenotypic) syndrome.
But things change when a drug in development specifically targets a disease gene, and this is what is happening right now with a program by OPKO Health designed to increase SCN1A gene expression levels. This product has received the Orphan Drug Designation by the FDA to treat “Dravet syndrome”, because about 80% of patients with Dravet syndrome have mutations or deletions in SCN1A, but would have been more appropriate to match the gene-targeting therapeutic with a genetic patient classification, and target “SCN1A deficiency” as the product indication, like in the scenario B.
If we stick to the classical (phenotypic) syndrome classification for the programs in development that target the gene or protein that are defective in patients, we will be unnecessarily limiting the number of patients that will benefit from that drug. It is clear that in the case of the OPKO program, the indication cannot be simply “Dravet syndrome”, but a subset of patients with Dravet syndrome caused by SCN1A deficiency. But how about the people that having the same SCN1A deficiency have received a different clinical diagnosis based on their symptoms? With the current indication, those people will not have access to such treatment because it will be off-label.
The specific program that OPKO is developing is oligonucleotide-based and therefore very invasive, requiring intrathecal administration, so it is not likely to be used in people with SCN1A mutations and milder phenotypes. But there are small molecule approaches also in development for SCN1A deficiency, and these should not have an indication restricted to a subset of patients with Dravet syndrome. The same will happen with drugs that modulate SCN2A, a related channel that produces multiple phenotypes when mutated and leads to multiple diagnosis. Asking the drug developer to seek approval for only a subset of patients in a rare disease would artificially restrict the number of patients that would benefit from that drug.
This is why we need to get confortable with drugs being developed for a genetic indication that overlaps but doesn’t match the classical (phenotypic) syndrome classification. At the other side of the approval line, physicians must also get confortable with patients having both a genetic and a symptomatic diagnosis so that they can receive both types of medications.
I do believe that indications based on a gene defect (gain-of-function or loss-of-function), when the drug treats the gene or protein that is altered in those patients, will be acceptable and soon become common. What I am more worried about is the need of a scenario C (see next) and the mismatch between the clinical practice and the regulatory pathway.
ADDING COMPLEXITY – SCENARIO C AND MULTI-SYNDROME INDICATIONS
There are drugs that don’t treat the disease (gene or protein) but the symptoms, in the case of epilepsy they treat the seizures, yet they get approved for treating a particular syndrome only. Staying on the example of Dravet syndrome, this would be the case of stiripentol (Diacomit), which is only approved for treating Dravet syndrome but it is simply a GABA-ergic modulator that has no biological reason to work only – or even preferentially – in this syndrome. It is basically a strategic corporate decision to focus on a given orphan indication. Another example is ganaxolone, also a GABA-ergic modulator, that has received Orphan Drug Designations for PCDH19 epilepsy and CDKL5 disorder and is in development for these indications, while there is no obvious biological reason to target those syndromes and no others.
I understand the appeal of orphan indications with limited competition, and how the incentives that the orphan designation brings make these drugs possible so that patients ultimately benefit from them. It is possible that without these incentives these drug developers would have never had the resources to bring forward their drugs to a broader (non-orphan) market. But things start changing when the same developer goes after many orphan epilepsies with a broad symptomatic drug, and we are seeing this right now with Epidiolex (cannabidiol, by GW Pharmaceuticals).
Let me first set one thing straight, I don’t think that GW Pharma is abusing the orphan regulatory route with Epidiolex, they simply have no choice. If they want to develop their product for a large number of patients with orphan refractory epilepsy syndromes, which happen to be separate indications under the traditional classification, they must develop a separate clinical development program for each syndrome. And because they have a broad-spectrum drug with the potential to treat many types of refractory epilepsy syndromes they are running clinical trials for Lennox-Gastaut syndrome, Dravet syndrome, Tuberous Sclerosis Complex and West syndrome. That’s four separate indications.
I wonder how many syndromes should a drug be approved for before it becomes clear that it should have a broader indication and label. In this case, the most appropriate indication would have been “epileptic encephalopathies” or “developmental encephalopathies” or similar, given that the drug appears to behave similarly in all of them. This is what the scenario C is about, drugs that treat a general symptom and should therefore have an indication that is broader than a single individual syndrome (or four).
Recently Takeda started a Phase 2 trial in collaboration with Ovid Therapeutics that is recruiting patients diagnosed with any type of epileptic encephalopathy for treatment with their drug TAK-935. This is what is known as a “basket trial”, where multiple indications are combined. Basket trials started in oncology but this is the first time that we see it run with a mix of epilepsy syndromes. The Takeda trial is not a pivotal trial so it won’t lead to an approval, but it gives me hope that we will one day see this mix of refractory epilepsy syndromes considered as a single indications for drugs that, like the Takeda drug, have mechanisms of action that don’t make them specific for one particular syndrome. As I see it, the basket trial approach should be the future trial design for such drugs, and the broad-label indication should also be the preferred indication unless the product has reasons to only work in a particular syndrome or seizure type.
An interesting consequence of broadening the indication towards this “umbrella” indication might be that it is not longer considered orphan, and therefore the product is no longer eligible for the orphan development pathway and incentives. The answer to this question has important regulatory and commercial implications.
Last, one of the most important aspects of this broad-label scenario C is not the benefit it might bring to drug developers, but the impact it will have towards making sure that all patients with rare epilepsies get access to new medications. Some of the syndromes have patients that can be counted in the low hundreds or even less. The best way to facilitate the development of new medicines for these ultra-rare syndromes is making it possible for them to go together with the other syndromes that share common phenotypes when it comes to evaluating and approving symptomatic drugs. This approach makes medical sense because it is also the same way that physicians work with these syndromes in their regular practice, and it will enable them to stop relying on off-label use of medications.
THE FUTURE
In conclusion, we have a few things to work on when it comes to the debate between genetic and symptom-base syndrome classification:
– First, we need to understand that both classifications are needed because they address different aspects of the disease: the cause and the clinical manifestation.
– Second, as drugs that treat the genes/proteins and not just the symptoms get into clinical trials, we need to be willing to switch from indications that follow the classical (phenotypic) syndrome classification to indications that reflect the drug mechanism so that it can be used across multiple classical syndromes.
– And last, we need to identify better regulatory pathways to make sure that the ultra-rare syndromes can get drugs approved for them as well. In the case of epilepsy this means considering the umbrella indication of epileptic encephalopathies or similar as opposed to artificially slicing indications unless the drug has a syndrome or seizure type-specific mechanism of action.
I hope all of these points will also create a new debate and have an important position in the agenda of the next the International Epilepsy Congress. Until then, I would appreciate your comments and thoughts.
Ana Mingorance PhD