Sail for Epilepsy, Part 4: Key Elements for Selecting a Proper Clinical Dose

In order to delve into this question of clinical dose selection, we are going to imagine the elements of a “perfect” scenario first. Then, in the next blog, we will unpack the obstacles experienced by many teams developing these therapies and some of the strategies to address these challenges. In practice, it is an unattainable goal to achieve all these elements, but considering the “perfect” case below provides us with a useful benchmark to compare to the typical scenarios.

Tropism or homing of the virus is conserved across species.

Ideally, whichever AAV serotype is being used therapeutically, it has been previously used in humans with the same route of administration. And that the data from those trials establishes that the distribution in the body closely mirrors down to the level of cell type, the results obtained after similar administration in rodent animal models and large animal models (like dogs or non-human primates). 

Transduction efficiency of a given dose of virus is conserved and predictable across species.

Thus, the distribution data noted above would establish the predictable “dose scaling” between rodents, large animals, and humans. This means that the same dose of virus (adjusted by body weight or tissue weight/volume) gives the same effect across all species. So, a dose in a rodent that provides efficacy is expected to transduce the same percent of cells in a larger animal and provide the same level of benefit.  And on the flip side, the dose that results in toxicity in a rodent would do the same in larger animals. This predictability provides confidence that the doses under study in the clinic are likely to provide benefit and avoid negative consequences.

The therapeutic window (e.g. the difference between the efficacious dose and maximum tolerated dose) is wide and conserved across species.

Also, in an ideal world, the dose required to provide optimal benefit would be much lower than the dose that results in immunotoxicity or other tolerability concerns. This provides confidence that even if a higher dosing turns out to be required in the clinic for optimal efficacy than was predicted by rodent data, there would still be room to safely increase the dose. 

Gene activity and the target cells for correction are conserved across species.

It would also be ideal if the role of the gene’s product and the cells in which the gene is active (e.g. neurons, glia, both?) are the same across species. In a perfect world, this would be established from data in multiple animal species and confirmed by data that shows that the loss of the gene product results in the same disease progression. Also, it would add confidence to dose selection if the percent of cells needed to be corrected is the same percentage across multiple species as this would rule out species-specific differences in cross correction.

Overexpression of the gene in the right cells or misexpression in the wrong cells is tolerated.

Some gene products result in toxicity if they are expressed too highly in the right cells or expressed in the wrong cell type. However, in the case of genes where overexpression and/or misexpression is tolerated, this provides greater confidence that, if there is a difference in tropism or levels of transduction in humans, it may be still tolerated. 

The mutations are truly “loss of function” and likely to benefit from simple gene replacement.

Pursuing a gene target that causes disease through an unambiguously “loss of function” mechanism simplifies the therapeutic approach. However, often when diseases are caused by missense mutations and sometimes with nonsense mutations, it turns out that the mutated gene product causes problems for the cell such as dominant negative effects. This means that simply adding more of the normal gene product might not necessarily correct things in the cell unless you knock down the aberrant version’s activity at the same time. There are emerging technologies that allow one to knock down faulty genes at the same time as providing a functional version, however these are still being evaluated to understand the nuances of their utility across gene targets and indications. 

Therapeutic benefit can be achieved even after symptoms are developed.

In a perfect world, every monogenic disease would already have an existing newborn genetic screen to allow identification of patients at birth (when they are likely not yet symptomatic). Unfortunately, newborn screening is often not implemented as a matter of course until after there is a therapy available to treat patients. This creates a catch 22 for indications that need pre-symptomatic treatment. So, in these cases, it is preferable to establish that there is still a therapeutic benefit even after symptomatic presentation based on evidence in rodent and large animal models. In this way, even if patients are diagnosed later, there is stronger confidence that they may still benefit from gene correction.

Manufacturing for nonclinical studies is fully representative of the planned clinical manufacturing.

Lastly, throughout all these studies noted above, the hope is that one would be using AAV vector produced by the same process and evaluated by the same analytical tests as what will be used in the clinic. Importantly, because of the nature of viral therapies, analytical assays for quantifying and characterizing vector often have a great deal of variability (e.g. infectivity, qPCR/ddPCR, empty/full capsid, etc). Reducing assay variability helps to ensure that the nonclinical doses administered can be taken at face value when calculating doses for the clinic. Of course, this often means significant investment in manufacturing and assay development prior to having all the key data from nonclinical studies to support clinical advancement.

If all the above elements are in place for a development program, there would be significant data in your favor to decide on the clinical dose. From prior clinical experience with that serotype, you would know the maximum tolerated dose in the clinic, thus avoiding any doses that are too high. And you could infer from the animal model data the doses that provided benefit; even in the first patient, the starting dose might provide them benefit (minimal effective dose). Ideally that minimal effective dose would be far below the maximum tolerated dose, as that would give you a wide therapeutic window in case the nonclinical data weren’t perfectly predicting the clinical efficacy. However, as noted at the beginning of this blog, development of gene therapies is never this simple. 

In the next blog we will discuss a number of typical scenarios that development teams are faced with, and the unknowns that often remain as novel gene therapies enter the clinic for the first time.

Emily Walsh Martin is a volunteer crew member for Sail For Epilepsy’s Atlantic crossing on the vessel Ingwe. When she’s not sailing, she is a consultant for gene and cell therapy companies and investors who are seeking to advance novel therapies in the clinic.

Read the Series

Part 3: How Would Gene Therapy Work to Address Monogenic Epilepsies?

Part 2: An Overview of Syndromic and Monogenic Non-syndromic Epilepsies

Part 1: A Worthy Excuse for Missing the ASGCT Annual Meeting