Regulatory Forum Opinion Piece*

Abstract

Adding a few animals to a toxicology study in order to document whether any lesions observed are reversible and at what level is a common practice but it is often unjustified, as most of the findings observed, particularly in the early stages of drug development, are easily categorized in regard to reversibility potential. The scientific application of all knowledge regarding the compound and the lesion in question is necessary to develop specific studies to address reversibility when this is a necessary step.

There is an increasing expectation in the drug development and regulatory communities to understand whether changes observed in toxicology studies will regress, which has led to the widespread use of “reversibility groups” in the design of toxicology studies. Currently, there is no consensus on when, where, and how to perform these studies. Also, there are a multitude of practices regarding reversibility groups that have led to an increase in the cost and duration of toxicology studies, increases in the number of animals used, and/or which generates imperfect data to address the issue. In the following paragraphs, I will address the “hows” and “how-nots” of the use of reversibility groups within toxicology studies.

The groups that are added to the studies are labeled “reversibility” or “recovery” groups, according to the preference of the individual laboratory or study director. There has also been a “mission” creep that these groups are necessary to document “late appearing” adverse effects. The argument that the addition of “reversibility” animals is needed to address delayed effects is not supported by current evidence or published literature. There are no significant numbers of studies that show unexpected delayed effects in the recovery period and no data to suggest that this additional information will increase clinical safety. Delayed effects are also detected in long-term studies as the compound progresses. Therefore, this argument will not be considered further in this discussion.

In a majority of the cases, the reversibility of changes observed in toxicologic studies can be adequately addressed by the study pathologist (clinical and morphologic) or the study director based on collective experience. Some of the more common changes in toxicology studies that have been proven to be reversible include, but are not limited to, the following: mild liver vacuolation, single cell necrosis of the liver (obviously, the individual cell that disappears does not come back, but the organ can repair this loss fairly quickly when the rest of the organ is in good shape), subabsolute decreases in specific cell population of the bone marrow, or even some peripheral nerve damage. I would propose that in these conditions there is no need for experimental documentation of their reversibility. Certainly, we also know that some changes that are not reversible (i.e., damage that produces fibrosis, such as liver cirrhosis, or loss of large portions of an organ, such as an infarct) can also be adequately addressed by the pathologist based on experience and scientific judgment without the need for reversibility data.

There are other conditions where the lack of reversibility does not constitute an increase in the risk. An example of this is the stress-related early involution of the thymus that is so common in animals under stress (yes, there is a possibility of repopulation of the thymus, but in general, particularly in long-term studies, the thymus will not regain its original size). However, T-lymphocyte function can recover once the adverse effect of endogenous glucocorticoids is removed. Also, if a compound increases the incidence of rodent chronic progressive nephropathy or cardiomyopathy, it is quite common that these changes will not reverse. The conclusion is that conditions that increase with age will not regress once started, and the lack of their reversibility does not significantly increase risk. Again, for these conditions, there is no need to have experimental evidence of their lack of reversibility.

This leaves a small subset of conditions where we really do not know whether they are reversible or not, and where reversibility information can be of importance to assess the impact on risk management. Obviously, serious conditions that are not reversible have a more serious effect on drug development than conditions that are reversible after the compound administration is stopped.

The decision of documenting reversibility also brings up the question of “when” a change is considered reversed. Many times in the present practice of “blind” reversibility group assignation, schedule constraints dictate that animals have to be sacrificed before one knows whether the change has reversed or not. Sometimes, there is a biomarker that allows for monitoring of the adverse change through the recovery period; however, more often there is no such biomarker and one is forced to bring the reversibility group to necropsy without a clear indication of the status of the change. This leads to a designation of “non-reversible” or “partially reversible” for changes that we know very well would resolve if allowed enough time for completion of the repair mechanisms. Presently, it seems that the toxicology and pathology communities “estimate” the time that should be allowed for reversibility based on the length of administration of the compound. It is very common to plan in the protocol for a reversibility length of 2 weeks when the compound administration is for 2 weeks or 1 month when the administration is 3 or 6 months. There is no scientific basis for these decisions where there is often either insufficient or too much time allowed for resolution. Sometimes, I have encountered reluctance to change sacrifice dates in the reversibility group based on a known biomarker returning to control levels. This resistance is often due to schedule conflicts or a perception that a change in reversibility duration is viewed negatively by regulatory agencies.

In order to address reversibility of the damage, the most common practice is to add either two (dogs or nonhuman primates) or five (rats or mice) animals to the high dose and control groups. However, this small number of animals is frequently a problem when the change is of low incidence. Without a biomarker to follow through the recovery portion of the study, it is often impossible at the end of dosing to assume that the two (or five) recovery animals actually presented with the same change as the high-dose group that is the object of the reversibility evaluation. This leads to a designation of “fully reversed” when we do not really know that these animals had anything to reverse. The numbers issue becomes more interesting when people insist on having recovery animals in the lower dose groups in order to find a minimum-effect level for the reversibility of the change. This sometimes creates complications when the finding (sometimes something very mild) is found to be reversible in the high dose but not in the mid-or low-dose animals.

Many of these complications seem to stem from the practice of “blind reversibility” where reversibility groups are arbitrarily added without knowing whether this information is really needed. Of course, the major impetus behind this practice is to gain time for compound development plans. On rare occasions, when we do not know the likelihood of reversibility of the change, addressing the reversibility may lead to producing additional experimental data, which may delay the compound development. However, this delay can be ameliorated by continuing the project with some restrictions while this information is produced.

If we need to develop reversibility data, then “when” in the development process should we plan to obtain this information? This of course will depend on the kind and severity of the damage as well as the therapeutic indication of the compound. The International Conference on Harmonization recently published a Questions and Answers document (February 2012) that addresses some of these points. They indicate that there is no need to continue to demonstrate reversibility if a change is considered likely reversible or if the reversibility has been documented in earlier studies. The time and resources costs of developing additional information also should enter into consideration. Reversibility groups are generally not needed when our collective understanding of the risks of the most common lesions seen in toxicity studies (i.e., those that can be readily monitored in humans at early stage, irrelevant to humans, or similar to that induced by related agents) considers them low and manageable. If the change is severe and strange enough to the scientific community that we cannot figure out its reversibility, two or even five extra animals often will not provide enough information to solve the problem and additional work will be needed. Whether this additional work should be performed blindly with the addition of reversibility groups within longer term toxicity studies or as a specialized toxicity study is debatable. Although conditions that appear after increased length of administration are generally few and of a known nature, the evaluation of reversibility of a totally new condition would be costly to duplicate. In such cases, it may be useful to have a few recovery animals in chronic studies to address the effects after dose stoppage, especially for new classes of therapeutics. It is frequently difficult to determine the length of the recovery period as well as the number of animals to include in recovery groups.

In conclusion, the resolution of the issue of reversibility groups lies in what is the most scientifically reasonable and realistic approach for the program under evaluation.

Clearly, one solution is not going to fit all development programs, and the production of reversibility data should fulfill the needs of the program. A careful review of all the information available will help in designing a successful reversibility experiment or an arm to a toxicology study. This information includes, among others, the nature of the event to be reversed, the incidence and severity of it, whether the effect is directly or indirectly compound related, the pharmacokinetic parameters that may influence the presentation of the effect, including the length of exposure after suspension of dosing, the length of treatment necessary to induce the change, and the chronicity of the change. Therefore, proper scientific assessment for the need of reversibility of a change should take precedence over routine practice of addition of recovery groups to toxicity studies.