The Road To The IND: The Fine Art of Focused Preclinical Development

The Investigational New Drug (IND) application is a critical stage of drug development, serving as the gateway to the clinic (and often, to further funding). In lean times, getting to the finish line with the IND can be decisive for program outcomes. In this article, part two of a three-part series, we deep dive into the nuts and bolts of the IND package and explore how modeling can be used to keep the preclinical work on track and under budget.

So, first things first, what are we talking about in terms of budget? It costs around $1.2B to bring a new drug to market, but around 90% of that price tag is the cost of clinical failures and the time value of money. Interestingly, a typical pharma or biotech?spends ?about $6 to $8 million to get a molecule to the clinic, a number that’s surprisingly small and?similar ?across a range of modalities and therapeutic areas. Now, that still sounds like a tidy chunk of change for a preclinical package, but around half of that ($3.8M) goes to figuring out the right formulation and manufacturing the drug product for the clinical trial. So, roughly speaking, this leaves around $1.8M for the toxicology studies, and anywhere between $1 and $2M for the pharmacology studies. Trouble is, there’s a long “laundry list” of studies that have to be completed to put together an IND package. There are exploratory studies, and there are GLP (Good Laboratory Practice) studies, often with the same goals. And GLP toxicology studies can come with a hefty price tag- each one will each cost hundreds of thousands of dollars. Often, the budgeting assumes that things (mostly) work right the first time too- there’s very little margin for error.

Bringing a preclinical campaign to fruition with a successful IND application under tight budgetary and time constraints is far from trivial! Here we will explore the fine art of focus, a critical prerequisite for a successful outcome at this stage of drug development.

(This article, and the others in the series, are complementary to the “Shoestring IND” webinar which we ran recently and plan to run again in the near future, so stay tuned!)

A strong IND package mirrors a realistic view of the drug’s capabilities

It might sound obvious, but a focused approach to putting an IND package together requires a focused view of the process. And the first part of that focused view is to understand why we do—and don’t – put IND packages together.

In a recent LinkedIn article ,? intended as a ‘Beginner’s Guide to the IND’, we outlined the design choices that go into the IND package. While there are multiple decisions that get made at this stage, a few crucial ones in the pharmacology/toxicology space are:

-?????????What is the proposed starting dose?

-?????????What is the proposed dose escalation scheme?

-?????????What is right dose route and schedule?

-?????????What is the right patient population?

-?????????Has the biomarker strategy been thought out thoroughly?

When putting together an IND data package, most project teams’ instinctive reaction is to seek out data that provides support for the design choices that have already been made.?So, for example, a project team may have decided at the inception of the project that the Target Product Profile for their me-too drug would be a once-daily, oral drug. The team will then seek out data that justifies this choice. Unfortunately, this leads to situations where the bass all too frequently ends up pointing ackwards. While it may be true that -all other factors being equal- a once-daily oral drug would have the largest commercial potential, all other factors are rarely equal. A drug that lacks therapeutic potential is a drug that lacks commercial potential. Ignoring the reality of what the drug is capable of delivering amounts to shoehorning a development candidate into a TPP that it will not achieve. This is an all-too-common cognitive trap, one which sets up project teams for particularly insidious failures that sink years of development time and billions of dollars before they play out. (We dive into this topic in great detail in an another LinkedIn article? of ours, which is well worth a read, by the way!)

The push-and-pull between what Sales wants to sell and what Engineering is able to build is an age-old one in tech R&D, and it is one that applies to pharma R&D as well. Instead of framing the IND as an exercise in justifying the TPP, the IND should seek to refine the TPP to position the molecule for its greatest likelihood of success in the clinic.

The therapeutic window is the quantitative basis for all rational decision-making

Before we dive into the nuts and bolts of how to focus the pharmacology and toxicology package for the IND, let’s quickly recap the basics. In brief, the pharmacologist’s view of a biological pathway is focused on the abstract set of events that the biology defines. For a pharmacologist, the most useful information is the set of dose-response relationships for drug binding to target, downstream biology and the ultimate outcome of interest (for both efficacy and toxicity).

The therapeutic window (the gap between the efficacious and the toxic dose) is a critical driver of outcomes in drug development. This therapeutic window is associated with the TPP. Thus, a drug may have a favorable therapeutic window via one dose route/patient population/dose schedule but not others.?For this reason, IND pharmacology and toxicology is best focused on estimating the therapeutic window of the choices made in the TPP, and being prepared to refine the TPP iteratively to take account of emerging information if a better therapeutic window exists outside the current TPP. In this way, the process of putting together an IND package helps increase the likelihood of a successful clinical trial. (For a deeper understanding of this background material, check out the previous article? in this series!)

?

Modeling can help provide rigorous answers to critical IND questions

Now, let’s take a closer look at how therapeutic window assessments can be leveraged in assembling an IND package which both refines the TPP and informs clinical trial design. Each of the five critical questions that we listed above can be framed and answered rigorously, if you have the dose-response curves for efficacy and toxicity in hand.

-?????????What is the proposed starting dose for the candidate drug? There are two ways to do a starting dose projection- based on toxicology (based on the NOAEL, or No Adverse Event Level) or on PD (MABEL or Minimal Anticipated Biological Effect Level). The NOAEL approach is traditionally performed without modeling support. However, as mentioned previously, GLP toxicology studies can be extremely expensive, and they come with a high risk of outright study failure if the right dose levels are not selected. Running a pilot study to build a model-based understanding of the dose-toxicity relationship(s) can be extremely helpful in designing an effective GLP tox study. The second approach is the MABEL approach, which relies on building a pharmacokinetic/ pharmacodynamic (PK/PD) relationship with a causal biomarker for drug effect (this can be something as simple as drug-target occupancy). The PK/PD relationship should ideally be established in a mechanistically relevant setting (e.g. an ex vivo human blood assay for a target that is in the blood, or functional imaging for a CNS target), and care should be taken to ensure that human variability is captured appropriately. With that said, the MABEL approach is often far cheaper to implement in the IND setting, can provide more precise estimates of the starting dose, and also provides a basis for a rational dose-escalation scheme. Implementing a MABEL approach requires two things- a translationally relevant PK/PD relationship, a projection of the dose-to-PK relationship in humans. In addition, having the biomarker assay that was used to set the MABEL be capable of being deployed during clinical dose escalation can be very useful!?

(In the next blog post in this series, we will deep dive into the use of modeling in setting the starting dose and supporting the dose escalation scheme.)

-?????????What is the proposed dose escalation scheme? There are again two ways to do this. Based on the steepness of the projected dose-toxicity relationship, you can decide whether a traditional 3+3 scheme is acceptable. If not (and it’s usually not), you can leverage the projected dose-toxicity relationship (either using the PK/PD relationship above, or separately modeling clinical adverse events as a function of dose) as a Bayesian prior and design a dynamic dose escalation scheme to escalate quickly and safely. (That’s a high-level overview, if you’d like to learn more, check out the next blog post as I mentioned).

-?????????What’s the right dose route and schedule? Interestingly, the dose-response relationships for efficacy and toxicity can also be leveraged to select the right dose. This approach has been used for generations now in the anti-infectives field. The way to do this is to first design careful pharmacology studies that allow you to get at the pharmacokinetic (PK) parameter that is best correlated with efficacy (the PK “driver” of efficacy). Typically, this step can be implemented after having a preliminary PK model in hand and using a little fancy footwork. (Simply running a set of ascending doses on one schedule won’t cut it, as peak drug concentration (Cmax) and total drug concentration over time (AUC, or area under the curve) are correlated on a single schedule). Using a carefully designed efficacy study, the PK driver of efficacy can usually be nailed down in a single experiment. The PK driver of toxicity has to be established in a similar manner. If a translationally relevant toxicology model exists for the most likely toxicities, then the approach described above maps over directly. If the mechanism and/or toxicity are clinically precedented, retrospective modeling can be performed to draw out the PK driver of toxicity (as we did here ).

In either case, one can now leverage the understanding of the PK drivers of efficacy and toxicity in identifying the dose route and schedule that provide the greatest therapeutic window. For example, if Cmax drives toxicity, while AUC drives efficacy, a once-weekly i.v. infusion may be a better idea than a single i.v. bolus delivered every three weeks. A difficult side effect profile will make it harder to achieve the full potential of the drug and will negate the added convenience of the single i.v. bolus dose.

-?????????What are the right time points and dose levels for biomarker collection? Designing the correct biomarker strategy for the clinic can be tricky, as the PK between preclinical and clinical species can be very different. When projecting biomarker behavior in the clinic,?the PK/PD relationship is linked to the projected human PK, in order to examine how the system is likely to behave. (We will deep dive into the use of modeling in biomarker strategy design in a subsequent blog post).

?-?????????What is the right patient population? The first question to answer is- has the patient population been identified already, or will it be identified during Phase I? If the patient population has already been identified, one can construct PK/ efficacy dose-response relationships in a range of translationally relevant animal models. (Usually, for Oncology, this can be done in patient-derived xenograft models, which have very high predictive value for this sort of work). If the patient population hasn’t been identified, you can use the projected dose-response relationship for efficacy as a Bayesian prior in a retrospective analysis at the end of a Phase I/II trial to zero in on sensitive subpopulations. The plans for this should ideally be in place before the start of the Phase I trial. In either case, a model-based approach will allow you to squeeze more information out of the preclinical and clinical datasets for patient selection.

?

Model-based INDs help you avoid pitfalls

Hopefully, the preceding section spelled out the ways in which modeling can be used to tighten up the IND package. The crucial point here is this: while it can be tempting to make design choices for clinical trials (or the TPP) first and then to do the pharm/tox studies that justify these choices, this is a mistake. Locking in on one’s choices in this way sets the project up for clinical failure that can often be extremely subtle in how it manifests, wasting years of time and millions (or more) dollars in the process. The pitfalls of drug development pretty much all relate to choices that were made in haste and repented at leisure in the years or decades that followed.

A model-based approach allows us to turn each design choice into a focused pharmacology experiment. Taken together, these experiments also help build a data package that can provide a clear quantitative understanding of how the system works. Such an understanding can be worth its weight in gold (or bitcoin, if you prefer), particularly when troubleshooting situations in the clinic. To provide just one simple example of this- say you’re dose escalating in the clinic, and the third cohort of patients has two patients showing weak signs of a toxicity that is often ascribed to the protein that your drug targets. Knowing whether or not you can expect target engagement by your drug in this particular dose cohort can be absolutely decisive in what happens next.?There are many other situations where a quantitative picture of what’s happening in the clinic can be crucial, making the model-based approach far more nimble and amenable to rational troubleshooting.

Thus, modeling can help you build a stronger, more focused package for the IND. In sum, modeling enables project teams to build a quantitative understanding of the dose-PK-PD-efficacy (or toxicity) relationship. These relationships are invaluable for doing any sort of projection of the consequences of design choices in clinical trials. (Ironic, given the framing of focused pharmacology as fine art, but)?modeling converts drug development into an engineering exercise, allowing rational choices to be made at each step along the way.

This was a high-level overview of a very complex topic, and hopefully the references to other articles of ours sprinkled in here (both completed and upcoming) provide you with more to dig into as you have questions.

If you found this article useful, you can also attend the other complementary webinars in our ongoing series on model-based drug development. We touch base on a different aspect of drug discovery and development, diving into concrete, practical ways in which modeling can be used to add value and increase cost-effectiveness.

For example, in next week's webinar (Tuesday 03/28, noon EDT/9am PDT), we will delve into how to use modeling to design cheaper in vivo studies that are less likely to fail.

You can also reach out directly to us on LinkedIn , or head on over to our webpage , where we have a large number of white papers covering various topics in model-based drug development.