by Shobha N. Bhattachar, Yan He, Ph.D., John Morrison, Ph.D.
The AAPS Pharmaceutical Discovery and Preclinical Development Community (PDPD) hosted an online Ask The Expert (ATE) event in February 2020. It was the first of its kind on the topic of preclinical oral exposure variability. The experts responded to wide-ranging questions from partici- pants. The panel was comprised of four world-renowned experts:
Leslie Z. Benet, Ph.D., is a well-respected founding father of the modern pharmacokinetics with near 60 years of scientific excellence.
Natalia A. Hosea, Ph.D., has over 20 years industry experience. She had led a pharmacokinetic, dynamic, and metabolism group in Pfizer and currently leads a Drug Metabolism and Pharmacokinetics group for Takeda.
Raj Nagaraja, Ph.D., leads a DMPK group at Agios Pharmaceuticals. He has built a broad experience in both small and large molecules with his rich working experience at several top pharmaceutical companies.
Andrew M. Vick, Ph.D., the 2021 AAPS President, brought in a high-level perspective from CMO as he currently oversees a biomedical staff of 3000 with activities in both nonclinical and clinical development at Charles River Laboratories.
John Morrison, Ph.D., Punit Marathe, Ph.D., and Prathap Nagaraja Shastri, Ph.D., from the PDPD leadership team also joined in the live discussion. The panelists’ broad background and experience from academia, large pharma, biotech, and CMO made the event scientifically interesting and practically useful. The discussions were comprehensive with a wide breadth of perspectives. The questions and answers from the session are captured below.
What are the sources of oral exposure variability in pre-clinical species - specifically in rat and dog?
Nagaraja: Source of variability could be intrinsic (drug physico-chemical properties related) or extrinsic (like particle size, vehicle, formulation, etc.), and a function of pH conditions in GIT as well as the feeding status.
Morrison: Are you seeing exposure variability in rat/dog but not in other species? Do you typically dose suspensions, solutions, ASDs?
Attendee: My question is more general. What can possibly affect the exposure in pre-clinical species? Raj responded, those factors can be intrinsic and extrinsic. If we give the same drug to rat and dog what physiological factors in these two species would vary across labs? For example, blood flows, pH, enzyme levels, etc.?
Nagaraja: I am not sure about blood flow differences and variability. But variability could arise from differences in enzyme levels, specific metabolism issues, food status, etc.
What are the implications of rats lacking gall bladders? Does the higher level of intestinal bile salts and greater solubilization tend to lead to better oral absorption for suspensions?
Hosea: It may depend on the compound and differences in solubility with and without bile salts. I don't know of any evidence that suggests solubilization is greater in rats due to lack of a gall bladder.)
What would cause two peaks to occur in the plasma level vs. time profile upon oral dosing to rats, but only one peak to be seen in dogs and monkeys?
Benet: Is your compound BCS/BDDCS class 1 or 2, 3 or 4?
Attendee: The compound is BCS 4 and is known to be a P-gp substrate. It is also an acid with pKa = 5.
Benet: The double peaks could be due to potential differences in metabolism or transporters.
What is the best way to change the PK profile in the tox studies, decreasing Cmax and increasing T1/2?
Nagaraja: Is there a compelling reason to do that change? Is there a concern on acute safety profile due to high Cmax? And does increasing half-life refer to transiently reducing the clearance or only the 'apparent half-life' by extending the absorption phase? You could try material with larger particle size that could result in slower dissolution and/or less enabling formulations like suspension to apparently reduce Cmax. Care should be taken to establish that overall exposure is not compromised.
Benet: As Raj said, you really can't change half-life, except by making the formulation a flip-flop model and having absorption being the rate limiting step. But I don't think you want to do that in a tox study.
Shastri: Agree with Les. Formulation changes leading to change in PK can be misleading in tox studies to characterize PK. Unless you have special scenarios, where you may want to evaluate exposure and safety with and without formulation changes to understand safety at different exposure profiles.
What a timely discussion for me! I have just run my first capsule study in rats looking at different salts. Normally, I would have done this in dogs but compound supply was limited. My questions are around the variability of oral capsule results in rats compared to dogs. What potential issues which might lead to this variability (technical issues versus physiological)?
Marathe: In my experience, dosing capsules to rats is challenging. The amount to be dosed is small and a suitable formulation cannot be made. Why not just dose a solution or a suspension to rats?
Benet: I agree with Punit Marathe in that capsule dosing in rats is very difficult. But there is another issue. In rats you see the marked difference in CYP3A metabolism in males versus females (with females it being constitutive).
Decker: We wanted to get an idea if different salts made a difference in absorption. In solution, all salts should be the same.
Hosea: I would also add that capsules can sometimes get stuck in the rats stomach leading to limited exposure. This may be something to check at the end of the study.
Nagaraja: To add to the response, it also depends on whether it is just API in capsule or has other additives. Even using capsules, the API could crash-out in the stomach as soon as the disintegration of capsule has started.
For a zwitterion molecule with low to moderate effective permeability and good absolute oral bioavailability, but with considerable variability in oral exposure preclinically and clinically: 1) what are the causes of the variability in oral exposure, and 2) what are the strategies to reduce the variability -a prodrug?
Hosea: Possible reasons for variability could be gut and liver metabolism, and/or physchem dissolution related variability. The ECCS classification for zwitterions would suggest it will be a CYP3A substrate and therefore may have significant variability in gut metabolism. Can you clarify if some subjects have low oral F? Also, what is the pKa and pH dependent solubility?
Attendee: The oral F seems high but with variable Cmax. The molecule is water soluble (intrinsic ~ 1 mg/mL) with acidic pKa 4, and basic pKas of 2, 7, and 8. Attributing the differences to variability in gut and liver metabolism seems to make sense.
Anyone have words of wisdom on evaluation of enteric coated tablet/capsule in dogs, viz: Use same pH trigger as intended for human (say 5.5)? Is it necessary to use pentagastrin-stimulated dogs? Gastric emptying time issues? Overall: how to minimize variability.
Shastri: You have brought up all caveats in your question. There may be no general answer to this. Typically, there is no need to use in vivo model to study the release from enteric coated/capsule. You can do that using bio relevant dissolution systems. If the question is to study absorption and minimize variability, you can consider pentagastrin or intra-intestinal injection approaches. However, you will need to build in necessary controls. If you want to minimize concerns of delayed gastric emptying due to pentagastrin, you can use pH telemetry capsule in parallel to track pH and emptying time.
Hosea: When conducting studies to mimic human stomach pH, pentagastrin pre-treated dogs can be helpful to reducing variability. Variability could be due to other factors such as dissolution, solubility, active metabolism/transporter mediated effects, etc. Overall, the translation of dog to human for enteric coated tablets/caps should be taken with caution but can be useful for proof of principle or used in PD studies if necessary.
Attendee: Suggest pentagastrin pretreatment and use 1/4 of the human dose. Recommend coated tablets and not coated granules in a capsule (And do not use HPMC capsule as may give a false positive due to possible supersaturation with HPMC.)
Have you ever encountered modified release formulations in the Tox studies?
Marathe: Is there a reason to use modified release dosage form in tox studies? We are interested in achieving sufficiently high exposures and thereby a window relative to both Cmax and AUC. A MR formulation will decrease the Cmax.
Nagaraja: I have not. Salt form changes, and spray-dried dispersions have been used in tox studies to enable good exposure achievement, although another form was later developed for clinical use.
Recent in vitro studies have demonstrated significant differences between assays in the presence and absence of protein. These differences may affect permeability rates on cell monolayers or transporter uptake or efflux ratios. Do we have any data yet to determine whether the in vitro permeability and/or uptake rates in the presence and absence of protein may correlate to the preclinical oral absorption? Are there any species differences that have been identified/observed/correlated to in vitro and in vivo oral absorption?
Nagaraja: First, assay conditions including the presence of protein or its concentration are subject to the individual Company/Site/Lab/CRO. Second, in Discovery phase, the focus for many assays need not be to interpret the results as absolute numbers, but rather to measure the relative improvements made with structural modifications (lead optimization). One example is the permeability parameters A>B, B>A, and the calculated efflux ratio. My experience is rather mixed when trying to correlate the in vitro permeability numbers with the observed absorption or bioavailability extent.
Benet: Albumin mediated uptake is just in its development stage. Almost all the work has been done with substrates of OATPs and we really don't know if this occurs with highly protein bound drugs that are not transporter substrates. So, I don't think we have enough data to answer your question concerning preclinical differences in species.
Since oral exposure variability has long been noted in the clinical phase, would it be reasonable to correct for this by AUC estimates in individuals prior to then adjusting their dose, both in trials and in ultimate use for those drugs where it can matter. Should drug monitoring be taken more seriously? Challenge the making of TDM measurements better, faster, cheaper? Generic chemo drugs and antibiotics (sepsis) are possible examples where damage is done in hospitals.
Nagaraja: Variability is an inherent "property" for any compound in any study. It is the magnitude of it and the challenge is in interpreting the results given the sample size. It is not practical to pick the subjects (patients) to get smaller variability and it is not guaranteed that what is seen on one occasion is repeatable for every dose or day. TDM is more focused on the tolerability/safety than for addressing variability.
If rats eat their bedding material, could that lead to a pseudo-fed state and potentially oral exposure variability?
Nagaraja: It could. How long after the dosing they eat that bedding material is also relevant.
Vick: Yes, in theory it could. That being said, bedding (aspen, paper-based) is pretty inert and consumption of substrate is rare / unlikely. More likely is coprophagic, where animals may eat their own feces. In those instances, double peak phenomena could be observed. (by Andrew Vick)
Is it necessary to have a similar PK profile for the tox studies, compared to the clinical studies?
Nagaraja: Not necessarily. We should stick to the intended route of administration while performing tox studies. Almost always, formulation is very different (suspension in tox vs solid dosage form in the clinic) and it is bound to be a factor for seeing differences in PK profiles. And rodents could have a more rapid disposition than humans. So, it is more practical to compare key PK parameters (Cmax and AUC) and monitor the acute safety signals with Tmax.
Vick: ICH guidance requires nonclinical studies to support the intended clinical indication. This includes dosing regimen (route, frequency, etc). If the clinical dosing is BID, ideally the nonclinical tox study should be BID. Due to allometric differences in PK; however, we often need to increase frequency of administration in nonclinical studies to cover the targeted clinical exposure.
Shastri: Ideally yes. It will depend on mechanism of action of the drug, and if the safety is related to Cmax versus exposure versus accumulation etc.
In preclinical studies, what is the basis of animal model selection for each aspect of ADME and Toxicology studies? What is the exposure variability so that one can correlate or project human studies like Phase 1 and 2 clinical studies? For most cases BCS 4 > BCS 3 > BCS 2 drug molecules, the in-vitro data and pre-clinical or Ex-vivo data does not correlate with the human physiological data. That means IVIVC is difficult to achieve. In such cases pre-clinical studies are very subjective and not conclusive. Can the experts provide their comments and guidance about this?
Nagaraja: Smaller species (most often rat) is used for lead optimization related in vivo PK studies. Mouse studies are most often done to support pharmacology studies. Tox species (one rodent and one non-rodent) are selected based on the precedence, metabolite coverage, relevance, and other factors.
Variability can be higher in tox studies as they are made as API in a vehicle (most often suspension at higher doses) and this may have a different dissolution profile than the solid oral dosage form used in the clinic. This variability is exacerbated when dealing with compounds with very low solubility, pH dependent solubility, and/or low permeability.
The scope of IVIVC needs to be defined for each chemical series. The lack of correlation may depend on multiple factors (e.g. species unique clearance mechanisms, pH differences, formulation changes, dose, etc). Caco-2 assay results (like low A>B due to high efflux), for example, may not correlate with % bioavailability (or even absorption).
Pre-clinical studies are usually subjective and form the back-bone of lead optimization as we are looking for relative shift in the key parameters due to structural modifications. After identifying the compound with say, lower clearance and better bioavailability, additional studies are done in other species to gain confidence on the mechanism of clearance and the confidence in human PK projection.
Vick: First part of your question is well scoped in ICH M3 (for small molecules) and S6 for biologics. For small molecules, species selection is largely based on metabolic similarities between preclinical test species and the predicted clinical biotransformation.
Will the variability be translatable from species to species to human?
Nagaraja: In general, if variability is coming from intrinsic factors of the drug molecule such as physicochemical properties, solubility, dose, etc., then it may translate across species. However, if it is physiology related (species unique), then it may or may not, depending on the species used.
Hosea: Are there specific examples that you are interested in understanding variability? If so, please provide additional details. Generally speaking, variability can also be dose size related and be complicated when there is concentration dependent metabolism or transport. Thus, understanding the cause of variability can help to assess translation as Raj indicates.
Recently, a number of lead molecules are being developed using molecular simulations of ADME parameters followed by actual preclinical testing. However, there are considerable differences from the predicted to the actual values. How can we determine the oral exposure of the molecules effectively in vitro?
Nagaraja: I guess you are referring to modules such as ADME predictors that work based on structural features. As oral exposure is a culmination (net effect) of various factors that are both intrinsic (physicochemical properties) and extrinsic (particle size, vehicle, feeding status, etc.), trying to predict the oral exposure using structure based simulations is challenging. IV PK is a better bet for such comparisons of in silico or in vitro approaches.
Benet: IVIVE almost always significantly underpredicts both in animals and humans. This is a major effort for our field, and we are not there yet.
What is considered to be high variability, 30% or 50%, that is not acceptable?
Benet: It's all related to therapeutic index. There is no problem with high variability of a very safe drug. This is why the Agency had introduced weighting based on the variability of the innovator product.
When developing molecules for cancer and brain diseases, the variability of the induction of disease in experimental animals is an important consideration. How can this be accounted for during oral therapy and reduce the influence of false positives?
Nagaraja: For large molecules where target mediated disposition is a reality, the level/severity of disease could impact PK profile. For small molecules, it could be less critical. I am not sure what you mean by false positive. Can you please clarify that?
How can you tell if the source of variability is from drug or from the formulation?
Hosea: My recommendation is to characterize the pH solubility, particle size in the formulations and couple this information with pKa of the molecule and information about metabolism/transport etc. to evaluate using a PBPK model in either GastroPlus or SimCYP. This would allow you to establish some hypotheses and then test to confirm. Variability can be dependent on both and on dose size thus a PBPK approach can allow you to conduct sensitivity analysis on different parameters.
How do you deal with toxicity or PK difference in different preclinical species?
Nagaraja: Some species are more sensitive to certain tox signals than others. A comprehensive view is taken based on the tox studies in two species (most often) about the tox/safety profile and the margins.PK differences could come from a few different factors such as pH profile in GI tract, food status, species-unique metabolite profile, dose, drug batch/form, etc. Careful selection of tox species is of utmost importance.
Vick: Some level of species differences in PK / Tox between species is to be potentially expected. It's part of the rationale for a 2-species (rodent and large animal) testing strategy that is aimed at reducing clinical risk. Some of these differences represent normal allometric-based differences, while others may represent target / biological based differences that are key to understanding the most "sensitive" species to gauge clinical risk.
Can you decrease the variability if the source is from the drug properties?
Nagaraja: We could. For example, if poor solubility of API is the reason for higher variability (and lower bioavailability), then an amorphous suspension or nano-particles or spray dried dispersions, etc. could be used to create a super-saturated solution in the GI and hence a better absorption/bioavailability.
Conclusion and Next Steps
The panel has identified areas that merit follow-up discussion. These areas include: the impact of bile salt solubilization on absorption in rats (and implications for informing absorption in humans), PK variability with zwitterionic compounds, and the importance of PK (TK) profiles in tox studies. If you are interested in these topics, please reach out to us at AAPS PDPD community and please stay tuned for further communication on the follow up dialog.