Abstract
The use of lipid formulations has greatly improved the ability to effectively deliver oligonucleotides and has been instrumental in the rapid expansion of therapeutic development programs using oligonucleotide drugs. However, the development of such complex multicomponent therapeutics requires the implementation of unique, scientifically sound approaches to the nonclinical development of these drugs, based upon a hybrid of knowledge and experiences drawn from small molecule, protein, and oligonucleotide therapeutic drug development. The relative paucity of directly applicable regulatory guidance documents for oligonucleotide therapeutics in general has resulted in the generation of multiple white papers from oligonucleotide drug development experts and members of the Oligonucleotide Safety Working Group (OSWG). The members of the Formulated Oligonucleotide Subcommittee of the OSWG have utilized their collective experience working with a variety of formulations and their associated oligonucleotide payloads, as well as their insights into regulatory considerations and expectations, to generate a series of consensus recommendations for the pharmacokinetic characterization and nonclinical safety assessment of this unique class of therapeutics. It should be noted that the focus of Subcommittee discussions was on lipid nanoparticle and other types of particulate formulations of therapeutic oligonucleotides and not on conjugates or other types of modifications of oligonucleotide structure intended to facilitate delivery.
Introduction
S
Another strategy involves delivery of the oligonucleotide within a particulate formulation consisting of various lipid and/or other components designed to insulate the oligonucleotide against nucleases in the blood compartment. In addition, such formulations serve to facilitate uptake into tissues and subsequent release of the oligonucleotide payload within the cytoplasmic compartment of individual cells. In some instances, targeting agents have been incorporated into the formulation to impart organ-specific uptake. For many of these “formulated oligonucleotide” drug products, the nucleic acid payload can be unmodified or minimally chemically modified (ie, largely retaining the natural phosphodiester backbone), thereby imparting a generally low order of intrinsic toxicity. However, many oligonucleotide formulations contain charged lipids or other molecular entities that possess substantial toxicity relative to the payload. While there may occasionally be adverse effects stemming from the pharmacologic activity of the nucleic acid component or from the nucleic acid chemistry itself, the toxicity profile of formulated oligonucleotide drug products is often dominated by the toxicity of the formulation excipients. In addition, the pharmacokinetics, tissue distribution, extent of functional delivery, and other properties of formulated oligonucleotides may differ dramatically from what has generally been observed with unformulated oligonucleotides.
The Oligonucleotide Safety Working Group was established as an outgrowth of the inaugural FDA-DIA Oligonucleotide-Based Therapeutics Conference in 2007, to foster regular discussions of emerging scientific data among academia, industry, and regulatory authorities [1]. A number of subcommittees were established for the focused discussion of specific areas of safety assessment within the oligonucleotide development field, and several white papers have been published in recent years as the direct result of the work of these subcommittees [2–5]. A recently established subcommittee of the OSWG, referred to herein as the Formulated Oligonucleotide Subcommittee (or henceforth the “Subcommittee”), was formed to discuss best practices for the characterization of drug disposition and the safety evaluation of formulated oligonucleotide therapeutics. This white paper represents the consensus opinions of the members of the Formulated Oligonucleotide Subcommittee. It should be noted that the focus of Subcommittee discussions was on lipid nanoparticle (LNP) and other types of particulate formulations of therapeutic oligonucleotides and not on conjugates or other types of modifications of oligonucleotide structure intended to facilitate delivery.
Characterization of PKs and Metabolism
One of the major goals for using a formulated oligonucleotide is to improve the pharmacokinetic profile and tissue distribution of a nucleic acid payload. LNP formulations have been a commonly used approach, in particular, for the development of small interfering RNA (siRNA)-based therapeutics. However, with the additional complexity of a lipid-formulated drug product, there are as yet no standards of rational, science-driven approaches and directly applicable regulatory guidance documents for evaluating the pharmacokinetic parameters of these therapeutic candidates.
Regulatory guidance documents for PK evaluation
There are several guidance documents or reflection articles that have been considered in defining the development path for formulated oligonucleotide programs, although none are specific for formulated oligonucleotides. Members of the Subcommittee have turned to both the US FDA guidance and the EMA/CHMP reflection article on liposomes [6,7]. These documents provide recommendations for the pharmacokinetic evaluation of liposomal drugs. However, the Subcommittee agrees that most elements of these documents are minimally applicable, as they focus primarily on comparability and bioequivalence with previously developed conventional small molecule drugs or other liposomal products. An additional relevant regulatory document is the 2011 AFSSAPS Recommendations for Toxicological Evaluation of Nanoparticle Medicinal Products [8]. This document details specific pharmacokinetic study recommendations, including evaluation of particle stability, the “active principle”, and the material comprising the nanoparticle. However, in this document, and in agreement with the approaches and recommendations of the Subcommittee members, a case-by-case approach is suggested for determining essential pharmacokinetic studies and endpoints for formulated oligonucleotide therapeutics.
In vivo pharmacokinetics
One major decision when evaluating the pharmacokinetic properties of a formulated oligonucleotide is to define what components of the drug product to quantify. It is the consensus opinion of the Subcommittee that the primary analyte, particularly in the early stages of drug development, is the oligonucleotide payload itself. The pharmacokinetics and localization of the oligonucleotide have been reported in several nonclinical programs using a variety of techniques, including hybridization ELISA, qRT-PCR, branched-chain DNA, and radio- or fluorescent labeling [9–13]. In clinical studies for ALN-VSP, ALN-TTR01, and ALN-TTR02, quantification of the siRNA payloads within the LNP formulation was determined using a hybridization ELISA or a probe-based HPLC method [14–17].
Although it is the Subcommittee's expectation that, at some point in development, additional characterization of any xenobiotic lipid excipients may be required by regulators, there is not a consensus among the Subcommittee members on the exact timing when such evaluations should be completed. There are examples in the literature of the quantification of formulation excipients both nonclinically and clinically. Evaluation of lipid pharmacokinetics and metabolism in nonclinical models has relied primarily on the use of radiolabels [9,18,19] or LC-MS-based methods [11,20]. Several lipid excipients in Atu027 were quantified in human plasma using an LC-MS/MS method, which was used to demonstrate the much slower elimination profile of the lipids compared to the siRNA payload in a Phase 1 study in patients with advanced solid tumors [21]. Despite these examples, there is lack of regulatory clarity on the necessity of performing such analyses at specific time points within a development program, and as they do add cost and complexity, sponsors may elect to defer them until later in development.
The Subcommittee agrees that rigorous characterization of the tissue distribution characteristics of formulated oligonucleotide products is generally not expected by most regulatory agencies in the early stages of a program, although minimal characterization may be required. More detailed exploration may be expected as a program progresses into later clinical development or if unexpected toxicities are observed in nonclinical studies. The highly predictable nature of particulate tissue distribution means that oligonucleotide concentrations are expected to be greatest in liver and spleen in all species. Therefore, the consensus approach taken by Subcommittee members in characterizing tissue distribution is to perform such analyses in one species, typically as part of the IND-enabling rodent toxicity study, and on limited tissue sets (eg, liver, spleen, kidney, lung, and heart) in the absence of other unexpected findings or additional information regarding a specific formulated oligonucleotide product. The primary reason to select the rodent toxicity study is that dedicated cohorts of animals are very commonly included for toxicokinetic blood sampling at specific postdosing time points, and tissue samples can easily be obtained at selected time points that are optimal for characterizing maximal tissue uptake. It may also be prudent to collect tissue samples from the nonrodent toxicity study for possible analysis (eg, if there is unexpected toxicity in one or more of the collected tissues). In the rodent study, it may be prudent to collect and analyze samples from multiple tissues at a single time point (eg, 6–8 h following the first dose), followed by sampling at other time points (eg, at later postdose time points or after repeated dosing) only from liver and spleen, to track clearance, persistence, or accumulation in the primary organs of uptake. It may also be useful to collect liver and spleen samples from a small subset of the toxicity study animals at the primary necropsy (eg, one to several days after the last dose) and the recovery necropsy, to supplement the characterization of tissue clearance and/or persistence.
As the lipid formulation and the oligonucleotide payload are inherently linked with regards to cellular uptake, analysis of the oligonucleotide component is considered sufficient to characterize distribution of the formulated product. An alternative approach used by some of the Subcommittee members involves incorporation of a radiolabel into the lipid formulation, akin to a classical tissue distribution study for small molecule drugs.
Formulation stability
The regulatory guidance documents or reflection articles mentioned previously [6–8] suggest that both unencapsulated (ie, released or free) and encapsulated drug substance (payload) should be measured in pharmacokinetic studies as an indicator of the in vivo stability of the formulation. The 2015 FDA liposome guidance document [6] recommends use of bioanalytical methods that can quantify both free and encapsulated drug substance. Similarly, the 2013 EMA reflection article [7] also recommends evaluating tissue distribution and excretion of both the encapsulated and unencapsulated drug substance, when feasible.
The aforementioned guidance document positions were generally developed based on experience with small molecule drugs, many of which are cytotoxic anticancer agents with significant toxicities, formulated in liposomal carriers. A primary objective of many of these liposomal formulations is to alter the distribution and toxicity profile of the drug, thereby enabling higher dose levels to be given. In these instances, developing bioanalytical methods to measure encapsulated and released drug are warranted, given the concern regarding possible “dose dumping” (ie, rapid release of the encapsulated drug from the liposome within the blood compartment). However, as noted in the introduction, many of the nucleic acid-based drugs used in formulated products are not extensively modified and have low intrinsic toxicity. In an unformulated state, these nucleic acid payloads are very rapidly cleared from the blood (usually within minutes) [9,22], in part, through nuclease-mediated metabolism, and to a greater extent, by rapid glomerular filtration and urinary excretion.
Thus, it is the Subcommittee's view that it is unnecessary to develop bioanalytical methods to distinguish between unencapsulated and encapsulated oligonucleotide, unless there is a very specific safety concern regarding a particular oligonucleotide modification. This position is reinforced if the stability of the product (ie, a lack of release of the oligonucleotide from the formulation) has been demonstrated in vitro and/or in vivo. Suitable demonstration of formulation stability could include the following multiple options: (1) in vitro stability assessments in blood or plasma/serum that indicate maintenance of parent (full-length) oligonucleotide concentration and/or a lack of metabolites (“shortmers”); (2) in vivo measurements of oligonucleotide-to-lipid ratios in blood or plasma/serum; or (3) demonstration that the half-life of the formulated nucleic acid is far greater than the half-life of the free unformulated oligonucleotide (eg, a half-life of hours for the complete formulation vs. several minutes for the unformulated oligonucleotide [9,22]). Option 2 could be accomplished by measuring the nucleic acid payload and a key formulation excipient directly using nonradioactive bioanalytical methods, or using radiolabeled markers, or a combination of these approaches. Parallel pharmacokinetic assessment of the oligonucleotide and a key formulation excipient in blood or plasma/serum over time could provide sufficient evidence of a stable formulation. In some cases (consistent with Option 3), sponsors have simply documented that the free oligonucleotide has a very short half-life (typically several minutes, if there is no backbone modification [9,22]) and that the formulated oligonucleotide has a much longer half-life (hours), thus providing indirect evidence for the stability of the formulation in vivo.
As noted above, when the unformulated nucleic acid payload is unmodified or minimally modified, its intrinsic toxicity is usually quite low (except in cases of toxicity induced by exaggerated pharmacological responses [5]), owing in large part to extremely rapid clearance of the free nucleic acid from the blood. In fact, the initial plasma half-life of an unformulated oligo with no or minimal chemical modification (eg, 2′-O-methyl) can be as short as several minutes [9,22], and toxicity from such unformulated oligonucleotides is only observed at very high dose levels [22] far above the typical dose range of formulated oligonucleotides. Therefore, it is the Subcommittee's view that the safety concerns regarding instability of the formulation in blood should be low for formulated oligonucleotide products containing minimally modified payloads, relative to more conventional liposomal products that typically contain payloads with much higher intrinsic toxicity (eg, cytotoxic anticancer agents).
Metabolism
In developing formulated oligonucleotide products, regulatory questions may arise regarding the metabolism of the oligonucleotide payload, as well as metabolism of the formulation excipients (eg, lipid components). Regarding oligonucleotide metabolism, the Subcommittee agreed that there may often be a low impetus for generating such data at an early stage of development. The rationale for this view is partly based on the fact that the metabolic pathways for nucleic acids with the most widely used chemistry (eg, those with 2′ ribose modifications and/or phosphorothioate backbone modifications) are well characterized, involving exonuclease-mediated chain-shortening and/or endonuclease-mediated cleavage to progressively smaller oligomers. Although this information is unpublished, the Subcommittee is aware of data from a few sources indicating that, at least for the more “familiar” chemistries mentioned above, the severity of toxicity is expected to be related to chain length, such that shorter metabolites would be less toxic than the longer parent molecules. For many formulated oligonucleotide products, the payload may contain either a natural nucleic acid structure or chemical modifications that are believed to be non-xenobiotic, such as 2′-O-methyl substitutions on the ribose units. Formation of shorter oligomers and/or monomers from such payloads should theoretically not yield toxic metabolites. Even with highly modified oligonucleotide products (such as the numerous nonformulated oligonucleotides with full-length phosphorothioate backbones or other unnatural modifications that have been advanced into clinical testing), the Subcommittee is unaware of any published reports documenting the formation of metabolites that are more toxic than the parent structure. However, it cannot be ruled out that metabolic transformation of novel chemistries or certain types of substitutions (such as the addition of fluorine in particular locations) may pose a risk of greater or unique toxicities. It should be noted that the resources and time commitment required to tease out the myriad of possible metabolite structures of an oligonucleotide, to distinguish these from endogenous pools of nucleic acids and to investigate their individual toxicities, are a highly daunting undertaking and may not actually be technically feasible. Therefore, the Subcommittee adopted the position that, for most formulated oligonucleotide programs, investigation of metabolism at an early stage of development may not be warranted.
Investigations of formulation excipient metabolism are also challenging in terms of bioanalytical methodologies and resources required. Although based on a lesser scope of experience (relative to oligonucleotide metabolism), the Subcommittee considered it likely that the metabolites of lipid excipients would not be dramatically more toxic than the parent molecules. Of course, sponsors should make reasonable judgments as to whether investigations of excipient metabolism should be undertaken depending on the presence of known structural alerts and/or observations of novel excipient-related toxicities. It is also noteworthy that regulatory authorities have thus far not expressed significant concern regarding the metabolism of formulation excipients at an early stage in development. A few exceptions are known to the Subcommittee, although these regulatory inquiries seemed to be based less on a particular concern than on the conventional expectations of small molecule drug development. Thus far, to the best of the Subcommittee's knowledge, the relative absence of information on metabolism of formulation excipients has not resulted in safety issues during the conduct of clinical trials.
Protein binding
The 2002 FDA liposome guidance indicated that protein binding (including endogenous lipoprotein complexes) should be evaluated for both the drug substance and drug product, including identification of major binding proteins. Although this is no longer mentioned in the updated 2015 FDA draft guidance on liposome products [6], requests from regulatory agencies for sponsors to evaluate protein binding are still frequent, particularly for LNP products. Although understanding protein binding has important pharmacokinetic and pharmacodynamic implications for small molecule drugs, there is little scientific rationale for measuring protein binding for most formulated oligonucleotide products. In many formulated oligonucleotide products, the oligonucleotide payload is sequestered or encapsulated within the formulation excipients and is thus inaccessible to plasma proteins. Moreover, cellular uptake and delivery of formulated oligonucleotides are typically driven by the binding of the formulation to endogenous (eg, apolipoprotein E) or exogenous (eg, N-acetylgalactosamine) ligands [23] and not through direct interactions of the oligonucleotide. This is in contrast to most unformulated oligonucleotides containing stabilizing chemistries (eg, phosphorothioate backbone modifications), which are extensively protein bound, although reversibly [24–26]. Consequently, measuring protein binding to the oligonucleotide payload is not relevant for most formulated oligonucleotides, particularly those for which the oligonucleotide is encapsulated within the formulation.
Determining protein binding of the oligonucleotide formulation may be of academic or mechanistic interest to understand cellular uptake processes of particle-based formulations [23] and may provide some information on the potential for drug–drug interactions. However, the standard methods for assessing protein binding (eg, equilibrium dialysis and ultrafiltration) are not appropriate for many formulated oligonucleotide products. Consequently, assessment of protein binding for particulate formulations has been challenging and, when undertaken, has typically utilized a modified product or formulation (eg, incorporation of a biotinylated lipid). For these reasons, these studies are not recommended in early development, if at all, but may be warranted on a case-by-case basis later in development (eg, if the formulated product is to be used in combination with small molecule drugs with high protein binding and narrow safety margins).
Assessment of interactions with the cytochrome P450 system
In general, nucleic acids are not known to interact with the hepatic cytochrome P450 system. Intracellularly, they are metabolized by nucleases (mainly endonucleases) and do not yield products that are substrates for or inducers of cytochrome P450s. However, it has typically been requested by regulators that sponsors conduct conventional studies to assess potential inhibition or induction of P450 isoenzymes, most often in human hepatocyte cultures. For formulated nucleic acid products that contain xenobiotic excipients, it is reasonable to expect that one or more of these constituents could interact with the P450 system, particularly if the formulation is designed to achieve high uptake into hepatocytes. No examples of P450 interactions with formulated nucleic acids are known to the Subcommittee, and enzyme induction is considered an unlikely effect. This is, in part, based on the lack of observed anatomic changes that would reflect enzyme induction in the livers of animals treated with formulated oligonucleotide products, including hepatocellular hypertrophy and increased liver weight. (Notably, delivery of significant amounts of formulated drug to the liver may result in hypertrophy and increased liver weight, although this histological picture is quite distinct from that described with regards to traditional enzyme induction). Therefore, the Subcommittee concludes that assessment of P450 interactions for formulated oligonucleotide products need not be addressed at an early stage of development (eg, pre-IND), but should be considered for the final regulatory submission, or perhaps sooner if any of the excipients contain structural alerts (ie, moieties known to have P450 inhibitory or inducing potential).
Nonclinical Safety Assessment
Species selection
In general, it is the recommendation of the Subcommittee and the expectation of regulators that a two-species toxicology program be used for the safety assessment of formulated oligonucleotide therapeutics. It is also the consensus of the Subcommittee that at least one pharmacologically relevant species (ie, one species that is responsive to the pharmacologic activity of the oligonucleotide payload) should be used for safety assessment of formulated oligonucleotide products that are directed against ‘host targets’ (ie, human gene products or pathways). This opinion is consistent with the recommendation of the Exaggerated Pharmacology Subcommittee of the OSWG [5] regarding appropriate strategies to assess the potential for adverse effects induced by the intended pharmacologic activity of an oligonucleotide.
It is the experience of the Subcommittee that the pharmacological activity of the oligonucleotide drug is often achievable only in one species, owing to the specificity of the oligonucleotide sequence and the requirement for precise sequence homology to achieve downregulation of an expressed gene [27]. More often than not, the only available pharmacologically relevant species for toxicological assessment of formulated oligonucleotides is the nonhuman primate. If a second pharmacologically relevant species is not available (eg, mouse or rat), then the second species should generally be used to test solely for chemistry-related effects or other types of non-mechanism-based effects, both of the formulation excipients and of the oligonucleotide drug itself. As is the case with antibody therapeutics, there may be circumstances in which a scientific argument can be advanced for avoiding the toxicological testing of a particular candidate in a nonpharmacologically relevant species (eg, previously well-studied molecule classes such as antisense oligonucleotides [27]).
The experience of the Subcommittee, as well as that of the field in general, is mixed with regard to the choice of rodent species for the safety assessment of formulated oligonucleotide drugs. In some cases, rats have been observed to be more sensitive to specific types of formulated oligonucleotide products than mice, based upon maximum tolerated dose (MTD) determinations [28]. More specifically, rats may be more sensitive to the innate immune response that is a primary trigger of systemic toxicities elicited by specific types of formulated oligonucleotide products, such as those using cationic lipids with a positive surface charge at a physiologic pH [29,30]. In contrast, the collective experience of the Subcommittee suggests that for many types of formulations, including other types of LNP formulations, there is no appreciable difference between rats and mice in the sensitivity to formulated oligonucleotide product toxicity. Thus, despite specific reports of differences in safety outcomes across species, the consensus of the Subcommittee is that there is no basis to assume a priori that any particular species (rodent or nonrodent) may be more or less sensitive to a novel formulated oligonucleotide product, unless there is prior experience with a particular type of formulation that has clearly defined a species difference. For some programs that have utilized murine models for pharmacology evaluation of a formulated oligonucleotide, the mouse may be a logical choice of rodent species to enable direct comparisons to pharmacology study outcomes [27], but there is otherwise no obvious scientific rationale for selecting the mouse over the rat for preclinical safety assessment of novel formulated oligonucleotide products.
The nonhuman primate has been the default choice of nonrodent species for the assessment of oligonucleotide therapeutics, including for formulated oligonucleotides [28,31–33]. This choice is based upon the ability to obtain complete or nearly complete nucleotide sequence homology of human and nonhuman primate mRNA target regions, as well as the accumulated body of knowledge regarding the expected toxicological profile of formulated oligonucleotide drugs in the nonhuman primate versus other nonrodent species. There is a relative lack of oligonucleotide safety assessment experience in the dog, either published or among the Subcommittee members, and thus, the toxicological profile of formulated oligonucleotides in the dog remains poorly characterized. The pig is not considered relevant for the toxicological assessment of particulate drugs in general, as a major component of the mononuclear phagocyte system in the pig resides in the lung as pulmonary intravascular macrophages. As particulate formulations are substantially cleared by mononuclear phagocytes, such products may exhibit extensive pulmonary localization when administered intravenously to pigs [34], a phenomenon that does not occur in humans or nonhuman primates. Finally, charged lipid formulations commonly used to encapsulate oligonucleotide payloads may cause different effects across different species (primates versus rodents), but these differences are currently not well characterized and, therefore, should not play a major role in the decision regarding nonrodent species selection.
Finally, in the case that an oligonucleotide drug, whether formulated or not, is designed to modulate a nonhuman target (eg, an antimicrobial application), the selection of species for safety assessment cannot be based on pharmacological relevance and should therefore be made based on historical precedent and relative potential for various species to reveal nonpharmacologic effects of the product. The Subcommittee also refers to the opinions of the OSWG Exaggerated Pharmacology Subcommittee regarding this topic, as this scenario was briefly addressed in the Exaggerated Pharmacology Subcommittee white paper [5].
Appropriate use of formulation or excipient controls in nonclinical studies
A major focus of the Subcommittee's discussions concerned the selection of appropriate controls for distinguishing toxicity of formulation excipients from that of the nucleic acid payload, particularly LNP formulations commonly used as delivery systems for oligonucleotides. The inclusion of appropriate controls for evaluation of toxicity and pharmacology is often scientifically valuable and may be expected by regulatory authorities. The value of including such excipient or formulation control groups in pivotal toxicity studies is not entirely clear to the Subcommittee. At best, these control groups enable the differentiation of toxicities attributable to excipients or the particulate nature of the formulation versus those stemming from the oligonucleotide payload. However, as discussed previously in this article, nearly all manifestations of toxicity for formulated oligonucleotides have been shown to be associated with the lipids or the particulate structure of the formulation and are rarely attributable to exaggerated pharmacology of the payload. Hence, it is reasonable to consider that the inclusion of excipient or formulation control groups does not provide critical information to enable selection of a safe starting clinical dose level or implementation of proper safety monitoring.
With the array of oligonucleotide molecules used for pharmaceutical applications (siRNA, microRNA, antisense, mRNA, etc.) and the variety of lipid formulations used as delivery systems, it is challenging to recommend an approach that is appropriate for all programs. In general, two basic types of formulation excipient-containing control articles have been used across most programs: a formulation containing no oligonucleotide (“empty” particles, sometimes referred to as an “excipient control article”) or formulations containing a nontargeting oligonucleotide, such as an inactive or “scrambled” sequence, or an active oligonucleotide that targets a nonmammalian gene such as luciferase [28], which is collectively referred to as the “formulation control article”. There is no experience among the Subcommittee with the testing of novel formulation excipients in isolation (eg, independent of LNPs) as controls for toxicity evaluation, nor does the Subcommittee believe this to be a scientifically rational undertaking.
Finally, species-specific analogues may be considered for cases in which the test article itself is not active in the species utilized for safety studies and for which one aims to distinguish exaggerated pharmacology-related toxicity from excipient-related toxicity. A detailed discussion of this approach was outlined by the OSWG's Exaggerated Pharmacology Subcommittee in their published white paper [5].
Empty nanoparticle formulations may serve as relevant controls to delineate the toxicological effects of the formulation excipients from those exerted by the oligonucleotide. The relevance and utility of an empty nanoparticle control increase when it maintains characteristics similar to its oligonucleotide-formulated counterpart (ie, lipid composition, mean particle size and distribution). If there are no toxicities related to exaggerated pharmacologic action or chemistry of the oligonucleotide payload, the empty particle control should theoretically exhibit a toxicity profile similar to the profile for the formulated oligonucleotide, thus demonstrating that all manifestations of toxicity are caused by the formulation excipients. However, there are often differences in the severity of such excipient-related effects between empty particles and oligonucleotide-containing particles, which may be attributable to differences in charge, particle size, and/or in vivo stability of the particles.
In some cases, excipient-related toxicity may be absent in the empty nanoparticle control group due to the aforementioned factors, which that may result in empty particles behaving differently than oligonucleotide-containing particles. Therefore, one should not readily conclude that the absence of a particular toxicity with the empty particle control group that is observed with the oligonucleotide-containing particles is absolutely attributable to the payload. If the oligonucleotide is not chemically modified (or is minimally modified) and if the observed effect appears to be completely unrelated to pharmacologic activity, it may be a manifestation of the particulate nature of the product and, therefore, absent with less stable empty particle control material. Conversely, there is experience among the Subcommittee members that, in some cases, empty particles bearing positively charged cationic lipids that are not compensated for by the negative charge profile of an oligonucleotide payload exhibit higher toxicity than the oligonucleotide-formulated particles. Such differences may be less pronounced when the nanoparticle formulations are composed of ionizable lipids, and the overall net charge of the lipid formulation is close to zero at physiologic pH.
Similar considerations apply to control formulations that contain nucleic acids. The inclusion of an inactive or pharmacologically irrelevant oligonucleotide in control formulations may overcome issues with particle instability and charge differences. Ideally, both the mean particle size and the particle size distribution should be as well matched as possible between the control and active formulations. However, despite efforts to achieve such similar formulations, there are cases known to the Subcommittee in which the toxicity profile of the formulation containing the target oligonucleotide and the control formulation containing a nontargeting or scrambled oligonucleotide did not fully align in terms of severity.
To aid in the decision regarding whether to use an empty particle versus oligonucleotide-containing control formulation, the pharmacokinetics and/or tissue distribution can be compared for these two types of control formulations. This strategy is dependent on the availability of a bioanalytical method for one of the excipients at an early stage in the program, before initiating pivotal toxicity studies, and invokes the need for exploratory PK studies, all of which require substantial resource allocation and thus may not be warranted at an early stage.
In a small number of examples known to the Subcommittee, novel formulation excipients were tested for toxicity as isolated entities, apart from the formulated oligonucleotide product. Such investigations have occasionally been requested during regulatory review, in some cases reflecting guidance documents focused on the safety assessment of combination drug products. While there may be specific circumstances where such investigations could provide useful safety information (eg, in genetic toxicity studies, to investigate the causative agent for a positive response), in general, the Subcommittee believes that toxicity studies conducted with isolated excipients are not scientifically justified, as unformulated excipients are likely to exhibit markedly different properties than when present in the context of a formulated oligonucleotide product and/or would be a challenge to properly administer in an in vivo toxicity study (due to CMC issues regarding solubility, particle size, stability, etc.).
In summary, although inclusion of excipient or formulation controls in nonclinical toxicity studies can be of informational value, the choice of which control is best or required will depend on the formulation under investigation and the specific questions to be addressed concerning the formulation components or nucleic acid payload.
Dose scaling and clinical translation of nonclinical toxicology assessments
The best approach for extrapolating a no-observed-adverse-effect-level (NOAEL) or no-observed-effect-level (NOEL) from animal toxicity studies to a safe starting dose level for human subjects or patients is currently not well defined for formulated oligonucleotides. To the Subcommittee's knowledge, the only regulatory guidance document issued on this general topic is the FDA's publication titled ‘Estimating the Maximum Safe Starting Dose in Initial Clinical Trials for Therapeutics in Adult Healthy Volunteers’ [35]. This guidance describes two major paradigms for estimating a human equivalent dose (HED) from the animal NOAEL/NOEL, either a direct body weight-based extrapolation, or a conversion based upon body surface area (BSA). The latter means of extrapolation (BSA based) represents a more conservative approach, as the conversion of milligram per kilogram (mg/kg) dose levels in animals to dose per BSA (mg/m2) translates into a smaller HED, relative to a body-weight based extrapolation. The magnitude of this difference depends on the species from which the HED is extrapolated. For example, the BSA conversion factor is 0.33 for nonhuman primate to human and 0.08 for mouse to human. Thus, if the NOAEL in mouse is defined as 1 mg/kg, the extrapolated HED would be 0.08 mg/kg (12.5-fold lower than the dose extrapolated directly from body weight).
This conservative BSA-based scaling paradigm has been the default means of calculating the HED for most classes of drug products, but the rationale for broad adoption is not clear. The original impetus to adopt BSA-based scaling derived from a body of work conducted with cytotoxic anticancer agents. Resulting publications demonstrated that direct extrapolation from the body weight-relative MTD underpredicted human sensitivity and that better correlation of MTDs across species was obtained by expressing doses per BSA [36,37]. The primary reason for the lesser sensitivity of lower species (eg, rodents) to small-molecule anticancer agents is that rodents tend to metabolize such molecules through the hepatic cytochrome P450 system faster than higher species and/or exhibit faster clearance from the blood compartment. This collectively contributes to more rapid or extensive detoxification than in higher species. This phenomenon does not necessarily translate to other drug classes, and some investigators have actually shown that cross-species MTD correlations for various small molecules are not precisely related to BSA [38,39]. Nevertheless, BSA-based scaling has been adopted for most classes of drugs [40] and was supported by the FDA's 2005 guidance document [35], although the guidance does state that a body weight-based extrapolation or other approach for determining an appropriate HED from animal toxicity data can be utilized, if justified.
It has been the experience of the Subcommittee members that most divisions of the FDA have maintained adherence to BSA-based scaling for defining the HED with formulated oligonucleotide products, sometimes citing the 2005 guidance document [35], but often without any clear scientific rationale. In some cases, despite evidence presented by a sponsor in support of a direct body weight-based extrapolation, the Subcommittee's experience is that conventional BSA-based scaling is often imposed on the selection of a starting human dose level at the IND stage. Notably, for many of the formulated oligonucleotides having advanced through preclinical development, the NOAEL in rodents tends to be similar to or lower than those in nonhuman primates, which is not consistent with the rationale for BSA-based scaling. In addition, formulated oligonucleotide products would be expected to behave differently than small-molecule anticancer drugs, with regards to P450 metabolism. The formulations themselves traverse the bloodstream in particulate form, thereby exhibiting unique pharmacokinetic and clearance pathways that are unlike small molecule drugs.
Therefore, it is the Subcommittee's opinion that determination of the HED should not routinely be based on conventional BSA-based scaling and that attention should be focused on the actual NOAELs defined in the most relevant toxicity studies. If the NOAEL in rodents is similar to or lower than the NOAEL in the nonrodent species, consideration should be given to defining the HED based on the mg/kg NOAEL in the more sensitive species (assuming that the toxicity profile in both species is considered to be clinically relevant and that the findings in one or the other species have not been previously shown to over- or underpredict human safety). Particularly for applications of formulated oligonucleotides involving initial clinical testing in patients with life-threatening conditions, such as cancer or other diseases with high unmet medical need, unfounded adherence to BSA-based scaling for determining the starting human dose level may result in administration of doses to the first patient cohorts that are well below the anticipated therapeutic range, thus offering no clinical benefit.
The position of the Subcommittee applies specifically to the estimation of the appropriate HED and not to the safety margin that should be applied to selection of a safe starting dose level. For example, the Subcommittee is aware of a growing body of unpublished information on the potential for LNP-formulated siRNA products containing cationic lipids to elicit infusion reactions and/or delayed immunologic responses in patients at dose levels that did not produce such effects in animal toxicity studies (either because of a lesser sensitivity of the animal species and/or because such responses are not easy to characterize in nonclinical toxicity studies). This experience may justify imposing a larger safety margin in initial trials with such products, but the application of BSA scaling is not a scientific means to achieve a more conservative starting dose level. It would be of greater scientific soundness to cite direct concerns regarding safety margins as the rationale for setting a lower starting clinical dose level, rather than achieving that end by dogmatically applying a BSA-based scaling of the animal NOAELs to arrive at a lower HED.
Immunological evaluations in nonclinical toxicity assessments
Cytokines
It is well-established that nucleic acids can activate the mammalian innate immune system through a variety of nucleic acid receptor-mediated pathways, notably the Toll-like receptors (TLRs) and RIG-I-like receptors [41,42]. Activation of these pathways can trigger interactions between immune cell types that may ultimately translate into various manifestations of immunopathology. However, a range of responses of widely differing severity may be elicited, depending on the nucleotide sequence, its tertiary structure, in vivo stability, chemical modifications, species differences, and other as yet poorly characterized properties. Hence, it is the consensus of the Subcommittee that, at least for those formulated oligonucleotide products that are expected to be immunostimulatory (eg, if they have been intentionally synthesized with a known immunostimulatory structural motif or are structurally related to other oligonucleotides that have exhibited strong immunostimulatory character), it may be of value to conduct cytokine screening early in development. In fact, lead optimization efforts may include an iterative process of incorporating chemical modifications into candidate oligonucleotides to augment (if desired) or, more commonly, minimize the potential to induce immune responses (eg, 2′-O-methyl substitution at cytosine residues [43]), followed by in vitro and/or in vivo cytokine screening to identify and eliminate those candidates with undesirable properties.
Despite the implementation of chemistry-based alterations to dampen immune responses, elevations in cytokines and inflammatory biomarkers are still commonly observed to varying degrees with formulated oligonucleotide products in nonclinical and clinical programs. This is likely related to a constellation of factors, including uptake of the formulation by immune cells such as monocytes and macrophages, concentration of oligonucleotides into intracellular compartments expressing TLRs, as well as the potential for immune stimulation from the excipients themselves. As such, there is an increasing regulatory expectation to evaluate cytokines and inflammatory biomarkers during the nonclinical (and clinical) safety assessment of formulated oligonucleotides.
Much of the interest in conducting cytokine and inflammatory biomarker evaluations in nonclinical studies derives from increasing experience with the use of formulated oligonucleotide products in humans. LNP-formulated siRNAs are arguably the most extensively studied formulated oligonucleotides in the clinic to date. Many of these clinical studies have observed dose-dependent inflammatory cytokine elevations, both in the presence and absence of steroid premedication. Adverse events related to cytokine release have typically been mild-to-moderate and included symptoms such as fever, nausea, chills/rigors, tachycardia, headache, rash, dyspnea, and occasionally, transient hypotension [15,44]. Typically, these effects occurred between 4 and 8 h following drug infusion and have correlated with peak cytokine levels (eg, IL-6 levels >1,000 pg/mL, or approximately 20- to 100-fold increases from baseline). In nonclinical studies, nonhuman primates exhibit dose-dependent cytokine release similar to humans following LNP-formulated siRNA administration, but monkeys appear generally less sensitive to the physiological consequences of cytokine release compared to humans. It is the Subcommittee's experience that monkeys typically show no clinical signs in response to cytokine elevations, although body temperature increases are often observed at 4 to 6 h postdose at higher dose levels. In addition, the dose levels at which cytokine increases are observed in monkeys are typically much higher than in humans (more than 5- to 10-fold, relative to body weight). Mice can also exhibit dose-related cytokine responses, but are the least sensitive relative to humans and monkeys and typically require dose levels approaching those where overt toxicities, in particular hepatoxicity, are already evident. Rats and monkeys tend to be similar with regard to doses where cytokine responses are observed.
It is the Subcommittee's opinion that careful consideration should be given to which and how many cytokines to monitor in nonclinical species, the optimal time points for monitoring, and whether to monitor in one or two species. Extensive panels of cytokines and inflammatory biomarkers are available in multiplex platforms, particularly for human and monkey. However, it is the Subcommittee's experience that many of these do not change after dosing with formulated oligonucleotides, and there are some cases in which changes were observed, even in low dose groups, in the absence of clinical or anatomic pathology findings. These situations raise new questions and result in unnecessary resources being directed toward evaluating and addressing large amounts of data, many of which do not offer additional insight toward the safety assessment of an oligonucleotide therapeutic. It is therefore advisable to restrict these assessments to a small number of key cytokines that are typically identifiable in early in vitro screens (discussed below). For example, experience with LNP-formulated siRNA products in monkeys and humans indicates that IL-6, MCP-1, IL-1RA, and IL-8 are reliable and generally consistent across species. Depending on the oligonucleotide used, other markers such as IP-10 and TNFα may also be appropriate. Experience has indicated that cytokine evaluations at a few time points are generally appropriate to capture changes, if any, for a variety of formulations and across multiple species (rodent, monkey, and humans). With LNP-formulated siRNA, peak cytokine levels typically occur within 6 h in monkeys and humans, but can be earlier in rodents, and most cytokines return to baseline levels by 24 h postdose. It is the Subcommittee's consensus that evaluating cytokine responses in one species is usually sufficient, as the results across species have typically been consistent. The Subcommittee's recommendation is to evaluate cytokine changes in monkeys, given the phylogenetic closeness between monkeys and humans, as well as the similar cytokine profile and timing of elevations compared to humans.
It is the Subcommittee's general recommendation that cytokine elevations alone, in the absence of additional clinical and/or histopathology effects, should not be used to define a NOAEL. For example, IL-6 is sometimes elevated at low formulated oligonucleotide dose levels in monkeys [28], typically in the absence of other findings (eg, no body temperature elevations or clinical or anatomic pathology changes). In humans, significant elevations in IL-6 can be achieved simply through exercise [45], and it is relatively common to observe > 20-100-fold increases in IL-6 after LNP-siRNA administration in humans without any adverse consequences [28]. This is consistent with studies involving infusion of recombinant IL-6 in humans, in which up to 35-fold increases in plasma IL-6 levels were observed without any changes in body temperature, mean arterial pressure, or heart rate, providing further support to the above position [46].
In vitro assays to evaluate cytokine responses to oligonucleotides and other drugs have been in use for many years, most commonly peripheral blood mononuclear cell (PBMC) assays and whole blood assays (WBA), and have mainly served to facilitate the selection of a lead candidate with minimal pro-inflammatory properties. In recent years, the human WBA has been developed as a more predictive alternative to PBMC assays for screening against immunostimulatory properties [47–50].
During the earliest clinical trials with LNP-formulated siRNA products, it became apparent that testing for cytokine induction in human PBMCs was not adequately identifying siRNAs with minimal immunostimulatory potential (Arbutus Biopharm, unpublished observations). The WBA involves minimal manipulation and processing steps to maximize cell viability and more fully recapitulate the in vivo response. The assay typically involves the incubation of fresh blood samples (from multiple human donors) with test compounds (eg, chemically modified siRNAs) at clinically relevant concentrations for 24 h at 37°C, followed by recovery of the supernatant and analysis for a panel of cytokines/chemokines. When screening new siRNAs or oligonucleotides, these compounds are formulated in an LNP or commercially available transfection reagent with proven capacity to deliver the nucleic acids into blood monocytes and plasmacytoid dendritic cells. Interestingly, screening of the same compounds in analogous in vitro studies in mouse and rat whole blood resulted in significantly lower levels of released cytokines compared with human whole blood, suggesting that the human system is particularly responsive to LNP-mediated immune stimulation (Arbutus, unpublished observations). Notably, this may also account for the absence of significant cytokine release or the requirement for much higher dose levels in mice treated with LNP-formulated siRNA. Overall, the human WBA is a sensitive method for screening new siRNAs and formulations to minimize the potential for innate immune responses.
The consensus of the Subcommittee is that the precise mechanisms responsible for cytokine release after administration of formulated oligonucleotides are not precisely understood and may vary by formulation and oligonucleotide. Moreover, cytokine elevations have not played a significant role in determining NOAELs, but limited evaluations of cytokines and inflammatory biomarkers, particularly in monkeys, may be useful in determining the likelihood of cytokine release in humans, as well as which cytokines to monitor in humans. Notably, in the Subcommittee's experience, there is limited correlation between cytokine release observed in nonclinical toxicology studies and the “infusion” or delayed immunological reactions observed clinically. The Subcommittee is not aware of any systematic investigations of either infusion-related reactions or the exploration of nonclinical models for these reactions.
Complement activation
The potential for oligonucleotides with charged backbones (such as those with phosphorothioate modifications) to elicit alternative complement pathway activation in monkeys is well documented [51–55]. The observed alternative pathway activation for this class is correlated with blood concentrations, and the molecular mechanism has been elucidated and involves direct interaction of the polyanionic oligonucleotide with a cationic site in Factor H, a regulatory protein in the alternative complement pathway [51]. Because of this legacy with phosphorothioate oligonucleotides, most sponsors of formulated oligonucleotide products either assumed that testing for complement activation would be required by regulatory agencies or were specifically instructed to do so by the FDA. In addition, there was concern that one or more of the charged excipients (eg, cationic lipids) might possess a potential for complement activation. In fact, there are a number of literature reports of complement activation by anionic and cationic liposomes (and cationic lipoplexes) in various species, including human [56–58]. In many of these reports, cationic liposomes activate the classical pathway, although there are also some reports of alternative pathway activation [56,59–62]. Conversely, most studies agree that charge-neutral liposomes or nanoparticles do not activate the complement system [56,58].
Early experience with several formulated oligonucleotide programs (ie, mainly those using LNPs and one or more charged lipids) confirmed complement activation in nonclinical assessments. However, the disconnect between the type and timing of peak split product formation and blood levels of the formulated oligonucleotide indicated that the mechanism of complement activation induced by formulated oligonucleotide products is distinct from the alternative pathway activation triggered by unformulated phosphorothioate oligonucleotides. In these cases, complement activation was associated with the lipid excipients and not with the oligonucleotide payload, as the activation was also observed with empty nanoparticle control formulations.
Based largely on experience accrued to date with formulated oligonucleotides, it is generally agreed by the Subcommittee that assessment of the potential for complement activation by formulated oligonucleotides should be conducted as part of the nonclinical safety assessment program and that this evaluation is best performed in monkeys. The Subcommittee further agrees that appropriate markers of complement activation should include: the Bb split product, a marker of alternative pathway activation derived from the cleavage of factor B (Bb split product is relatively stable once produced and can accumulate), and C4a split product, a marker of classical pathway activation derived from the cleavage of C4. If neither marker is elevated (ie, ≤3-fold increase), then there is little or no value in further assessments. If one or both of these markers are elevated, then it is recommended to evaluate the C5a split product. C5a is a terminal pathway (also known as the common pathway) marker produced from the cleavage of C5 to C5a (an anaphylatoxin) and C5b (which forms part of the terminal membrane attack complex). An increase in C5a is noteworthy, as it is a potent anaphylatoxin (roughly 20-fold more potent than C3a), is directly cardiotoxic, and can activate neutrophils, which collectively can lead to a sequelae of adverse effects, including hemodynamic changes (eg, hypotension).
This tiered approach to complement testing, including C5a measurement if other split products are elevated, is intended to address the toxicologic significance of observed complement activation, as C5a is believed to be the primary mediator of adverse effects of complement activation in monkeys [63]. However, C5a is an extremely labile split product and will not accumulate in blood (plasma) unless there is a burst of complement activation. Hence, it has been repeatedly shown that formulated oligonucleotides may cause activation of the alternative or classical pathways (or both), but that there are no adverse hemodynamic consequences owing to the absence of accumulation of C5a. Generally speaking, changes in C5a of <2-fold from baseline should not be considered biologically relevant.
The Subcommittee's experience with complement activation assessments of formulated oligonucleotides has largely been restricted to nonhuman primates. The results of these assessments have been variable, depending on the formulation and route of administration. Effects on complement are frequently observed with formulated oligonucleotides, although these typically occur at higher dose levels, with effects occasionally observed at lower dose levels. In some cases, increases in split products are more apparent with repeated dosing. For unformulated oligonucleotides, complement activation is typically observed acutely, with peak levels of split products generally well correlated with plasma Cmax (unpublished observation). This is frequently not the case for formulated oligonucleotides. Some split products may have small increases at the end of infusion (eg, C4a), and others may peak at later time points (eg, Bb at 24 h postdose in monkeys [28]). Regarding the timing of complement monitoring, the Subcommittee agrees that complement activation is ideally evaluated at predose or preinfusion, at the end of infusion (corresponding to Cmax), between 2 to 6 h, and at 24 h postinfusion (as a minimum). There can also be a certain level of variability and procedure-related changes in these markers, particularly for chair-restrained monkeys, so it is essential for proper interpretation to evaluate complement activation in a concurrent control group.
As discussed above, it is the Subcommittee's experience that complement effects for formulated oligonucleotides have not typically involved the terminal pathway and C5a generation. In humans, evidence of complement activation has been observed in some subjects following infusions of LNP-formulated siRNA products [14,16,28]. These increases, which do not show a strong dose relationship, have mainly consisted of elevations (typically 3- to 10-fold above baseline) in the Bb split product during or at the end of infusion. The relationship between elevated Bb levels and acute infusion reactions remains unclear. Although some of the subjects that experienced an acute infusion reaction also tended to have higher Bb levels, many subjects with elevated Bb levels did not experience an acute infusion reaction.
The Subcommittee acknowledges that some of the complement observations (eg, the late and significant accumulation of Bb for some LNP-formulated siRNA products) observed for formulated oligonucleotides could be secondary to other physiologic or toxicologic processes. However, given the evidence of complement activation for formulated oligonucleotides in monkeys and occasionally in humans, it seems appropriate to evaluate complement activation during the nonclinical safety studies.
In vitro complement assays have sometimes been used to assess the potential for complement activation by various compounds, including liposomes and formulated oligonucleotides. In general, these assays can be run in human serum, and some are applicable or can be adapted to other species as well. However, careful consideration must be given to the implementation and interpretation of these assays when using formulated oligonucleotides as the in vitro test system may not adequately reflect the in vivo system. One notable example of this is products that incorporate an exchangeable PEGylated lipid, which would sterically shield the particle surface against complement effects in vitro [64], possibly resulting in false negative results, but would rapidly exchange out of the particles in vivo and expose the particle surface. As such, the Subcommittee position is that an in vivo assessment of complement activation is more appropriate and relevant in the initial nonclinical safety studies.
Immunogenicity investigations
Immunogenicity testing for formulated oligonucleotides may be more complex than for other drug product classes, mainly owing to the multiple chemical entities in these products, including the nucleic acid payload and delivery formulation excipients. In the present discussion, immunogenicity implies the assessment of a primary antibody response to any of the constituents, as distinguished from the more general potential for particulate formulations to interact with immune cells, either through phagocytosis of particles by mononuclear phagocytes or by direct stimulation of the innate immune system.
Notably, the question of whether oligonucleotides are immunogenic and can elicit a primary antibody response has been controversial for many years. Early experience suggested that conventional phosphorothioate-modified antisense oligonucleotides are not immunogenic [65,66]. However, more recent investigations with an approved antisense oligonucleotide from Ionis Pharmaceuticals, mipomersen, and related molecules have demonstrated a potential for second-generation phosphorothioate oligonucleotides [67] to induce immunoglobulins with antibody-like affinity for the oligonucleotide. It has been noted that this immunoreactivity was not nucleotide sequence-specific, which is uncharacteristic of a true primary antibody response. More in-depth discussion of this topic will be presented in a white paper from the OSWG's Immunomodulatory Subcommittee (article in preparation). Importantly, if antidrug antibodies are formed from repeated administration of phosphorothioate oligonucleotides, they don't appear to have any significant effect on pharmacokinetics or pharmacologic activity (ie, there is no substantial loss of exposure or activity with chronic administration).
It is the Subcommittee's experience that the FDA has on several occasions requested investigation of the immunogenicity/antigenicity of formulated oligonucleotides during pre-IND interactions, often without clarity as to whether studies should be focused on potential antibody formation against the nucleic acid payload and/or to excipient constituents of the formulation. Regarding the potential for immunogenicity of the oligonucleotide, many payloads are chemically modified to minimize immunostimulatory activity (ie, interaction with TLRs and other triggers for innate immunity), but such modifications would not necessarily preclude an antibody response. In addition, given the particulate nature of the formulations, there is potential for uptake by mononuclear phagocytes. While macrophages are immunocompetent cells, it is uncertain whether such uptake would achieve proper antigen presentation or even whether the relatively small size of the nucleic acid payload would prevent antigenicity. Unformulated oligonucleotides with no chemical alteration of the backbone have very short half-lives, on the order of minutes, when injected intravenously [9,22], so there is little concern for immunogenicity of any oligonucleotide that is released into circulation due to particle instability.
Similar considerations as outlined above apply to formulation excipients. In general, these are not large molecules, with the exception of the PEGylated lipids that are often used. Some PEGylated molecules have been shown to induce anti-PEG antibody formation, and some human subjects with preexisting anti-PEG antibodies have been reported to exhibit hypersensitivity reactions to certain PEGylated proteins [68–71]. Anti-PEG antibodies have also been reported in response to repeated systemic administration of PEGylated lipid formulations, containing plasmid DNA or antisense molecules, in mice [19,72]. However, not all PEGylated molecules present this risk, and the collective experience of the Subcommittee suggests that PEGylated lipid excipients utilized in oligonucleotide formulations have thus far not posed a significant clinical safety issue with respect to interaction with preexisting or de novo anti-PEG antibodies in patients.
Based on the experience of the Subcommittee, no dramatic changes in pharmacokinetics or pharmacologic activity of nucleic acid payloads have been observed with repeated administration of formulated oligonucleotides. This suggests that an antibody response to the nucleic acid payload, excipients, or particles is not occurring or that any antibodies formed are not affecting exposure or activity [73]. Therefore, the Subcommittee does not perceive a significant issue regarding immunogenicity of formulated oligonucleotide drugs, with the largest theoretical concern being the potential for patients with preexisting anti-PEG antibodies to develop an immune response to a PEGylated excipient. Given the very low clinical incidence of such reactions documented for other PEGylated molecules, this phenomenon likely cannot be addressed nonclinically.
Genetic toxicity testing
The Subcommittee recognizes the recent publication from the OSWG's subcommittee on genotoxicity regarding their general perspective and recommendations for genotoxicity testing of oligonucleotides as a broad therapeutic class [4]. However, the genotoxicity subcommittee did not specifically address the applicability of this publication to formulated oligonucleotide products, and the following discussion summarizes the Subcommittee's recommendations for the assessment of genotoxicity for this specific subclass of oligonucleotide therapeutic drugs.
The experience of the Subcommittee members indicates that, at the time of IND filing, the standard in vitro genetic toxicity studies (ie, Ames/bacterial mutagenicity assay and in vitro chromosomal aberrations or in vitro micronucleus assays) should be conducted using the unformulated nucleic acid component of the drug product. In some cases, and with more recent programs, sponsors have been asked to conduct such testing both with the unformulated nucleic acid, as well as with the formulated nucleic acid. It is the Subcommittee's view that a decision about whether to test the formulated oligonucleotide in vitro (in addition to or apart from testing the nucleic acid payload) should depend on the novelty of the formulation excipients. For example, if in vitro genetic toxicity studies with a particular drug product have been conducted with the formulation containing a nucleic acid payload, and if the study results were “negative”, it may not be necessary to test the full formulation for another program with a different payload that utilizes the same excipient composition. In this case, it should suffice to perform testing only with the unformulated nucleic acid(s).
Regarding appropriate dose levels for in vitro genotoxicity assays, toxicity to the test systems resulting from the unformulated nucleic acid is usually very low, and it would be appropriate to test concentrations up to the “limit doses” recommended by regulatory guidelines [74]. For a formulated oligonucleotide, it is appropriate to consider the entire weight of the test material, in terms of lipids, nucleic acid(s), and other excipients, in defining the appropriate limit dose. Most likely, attempts to reach limit doses based on the nucleic acid content for a formulated oligonucleotide will be confounded by the density of the suspension and possibly interfere with evaluation of the assay results.
In vivo genetic toxicity studies (eg, the in vivo micronucleus assay) should be conducted with the complete formulation using the relevant clinical route. Dose levels should be selected in the manner described in the applicable ICH guidance [74] and may require a preceding range-finding exercise to identify appropriate dose levels (if not previously defined in general toxicity studies).
The Subcommittee agrees that genetic toxicity testing of individual excipients is considered unnecessary, as these excipients will likely behave entirely differently in isolation, relative to their properties within the complex milieu of the formulation, particularly in vivo. Although highly unlikely, it is possible that incubation of the formulated oligonucleotide product over an extended period (eg, 24 h), as is typically done in in vitro genotoxicity assays, would result in some degradation of the formulation and some exposure of the test system to individual excipients. However, given the uncertainties about exposure of the test systems in vitro, sponsors might consider that the in vitro assays are better suited for testing the unformulated nucleic acid and that in vivo genetic toxicity studies are more relevant for the final formulated drug product.
Conclusions
While there are general considerations regarding the nonclinical characterization of the broad therapeutic class of oligonucleotide drugs, oligonucleotides encapsulated within a particulate formulation require unique considerations across a number of topics regarding pharmacokinetic and toxicological characterization. Specific international guidances for the development of this class of drugs have not been generated, and while guidances for both small molecule and large molecule drug development may often be applicable, these are neither sufficient nor complete enough to guide developers through the many unique aspects to be considered with oligonucleotide drug development. Several white papers from various OSWG subcommittees have been published in recent years to address these gaps using the breadth and depth of experience of OSWG members [1–4,63,75]. Many aspects of these white papers are as well applicable to the characterization of formulated oligonucleotide drugs in many respects, but the large number of unique and divergent aspects of formulated oligonucleotide drug development warranted a dedicated consideration and set of specific recommendations by a group of OSWG experts uniquely positioned to provide these perspectives. We aspire that the recommendations provided within may someday form the basis of guidances specifically dedicated to considerations for the early nonclinical development of formulated oligonucleotide therapeutic drugs.
Footnotes
Acknowledgments
The authors thank O. Almarsson, R. Brown, J. Heidel, R. Oleson, J. Petrick, J. Senn, S. Shrewsbury, P Smith J. Stoudemire, H-P. Vornlocher, L. Whiteley, and H. Younis for their contributions to the various discussions that formed the basis of this white paper and S. Voytek for her thoughtful review of the draft manuscript.
Author Disclosure Statement
The submitted article is a committee position document that does not promote any particular product(s), and no competing financial interests exist for any of the authors. This article represents the views of the authors and the contents do not necessarily reflect the views or policies of the authors' affiliated institutions.