Abstract
Drug-induced phospholipidosis (PL) is an excessive accumulation of phospholipids and drug in lysosomes. Phospholipidosis signals a change in cell membrane integrity and accumulation of intracellular drug or metabolite in tissues. The sensitivity and susceptibility of preclinical models to detect PL vary with therapeutic agents, and PL is expected to be reversible after discontinuation of drug treatment. The prevailing scientific opinion is that PL by itself is not adverse; however, some regulatory authorities consider PL to be adverse because a small number of chemicals are able to cause PL and concurrent organ toxicity. Until a greater understanding of PL emerges, a well-thought-out risk management strategy for PL will increase confidence in safety and improve selection and development of new drugs. This paper provides a tiered approach to risk management of drug-induced PL. It begins with use of in silico and in vitro tools to design and select compounds with reduced potential to produce PL. Early in vivo studies in two species are used to better characterize potential for toxicity and PL. Finally, routine risk management tools (i.e., translational biomarkers, assessment of reversibility) are used to support confidence in safety of compounds that induce PL in animals.
Introduction
Phospholipidosis (PL) is a lipid storage disorder in which phospholipid–drug complexes accumulate within lysosomes as lamellar inclusion bodies. A condition that resembled Niemann-Pick disease, first noted in Japanese patients, was characterized by hepatosplenomegaly, foamy cells, and vacuolated peripheral lymphocytes after treatment with the antianginal medication 4,4′-diethylaminoethoxyhexestrol (DH) (Yamamoto et al. 1971). Electron microscopy confirmed the presence of multilamellar bodies in multiple organs including the liver, spleen, and lymph nodes. Since this first report, more than fifty marketed drugs have been described with the potential to cause PL in animals. Several excellent reviews have been published (Halliwell 1997; Hostetler 1984; Kacew 1987; Kodavanti and Mehendale 1990; Lüllmann-Rauch 1979; Matsuzawa et al. 1972; Reasor et al. 1988; Reasor et al. 1996; Reasor, Hastings, and Ulrich 2006; Reasor and Kacew 2001). It is generally agreed that in most cases, the process involves drugs, commonly referred to as CADs, with a cationic amphiphilic structure. There is also general agreement that inhibition of lysosomal phospholipase A1, A2, and/or C contributes to the accumulation of CAD–phospholipid complexes.
Multiple theories have been proposed to explain how phospholipase inhibition leads to buildup of CAD and phospholipid complexes (Reasor, Hastings, and Ulrich 2006). In vitro studies performed by Scuntaro et al. (1996) showed that PL–CAD accumulation can be potentially inhibited by D-α-tocopherol (vitamin E). This research suggested that competition between vitamin E and drug for cell membranes prevented the formation of phospho-lipid–drug complexes. Recent toxicogenomics research by Sawada, Takami, and Asahi (2005) provided additional insight into potential mechanisms of PL at the genomic level. Gene expression changes following CAD administration suggested inhibition of phospholipase activity, as indicated by upregulation of phospholipid degradation–related genes; inhibition of lysosomal enzyme transport by downregulation of genes involved in lysosomal enzyme transport; enhanced phospholipid biosynthesis, as indicated by the upregulation of fatty acid biosynthesis–related genes; and enhanced cholesterol biosynthesis, as indicated by upregulation of cholesterol biosynthesis–related genes. Although PL has not been shown to be a direct toxic end point, it is predictive of drug and/or metabolite accumulation within cells of affected tissues.
Most often, PL is detected histologically during exploratory preclinical in vivo studies. Since phospholipids are a part of the building block of cell membranes for cells of all organs, drug-induced PL can potentially be observed in any organ in the body. Lung and liver are most often involved in PL seen in pre-clinical studies. Some compounds that cause PL also cause concurrent inflammatory and/or degenerative changes in the same, or different, tissues (Stebbins et al. 2002; Vonderfecht et al. 2004). Phospholipidosis alone, or appearing concurrently with other changes, could potentially delay the drug development process if not properly assessed with an effective risk management strategy. Additionally, for programs in which PL findings may have an adverse impact on the success of the project, it is desirable to be able to rapidly identify and screen out those compounds with the potential to induce PL early in the discovery process. Transmission electron microscopy (TEM) is considered to be the most sensitive method for confirming the presence of PL, although it is considered low throughput and resource intensive. Therefore, low-cost, low-bulk, high-throughput predictive assays, applicable in the discovery cycle, would be important tools in minimizing the impact of PL on a drug development program. When PL is found in animal toxicity studies, risk assessment should be based on other evidence of toxicity, not on the presence of PL, because many marketed drugs cause PL without concurrent organ toxicity (Reasor, Hastings, and Ulrich 2006). The overall goal of this paper is to provide recommendations for accurately assessing the likelihood of compounds’ potential to induce PL; and to provide a plausible, three-tiered risk management strategy to guide portfolio development of drugs that could potentially cause PL with or without concurrent target organ toxicity in humans.
A Plausible Risk Management Strategy for PL
Pharmaceutical companies seek to apply appropriate strategies to select, develop, and market safe and effective compounds that benefit the health of patients. The strategy in this article encourages potential for development success through the use of three major decision-making stages (“tiers”) for managing a project with potential or known PL issues (Figure 1).
Tier 1. In the chemistry stage, use in silico (computer-based) predictive tools supplemented with in vitro assays to select compounds with minimal PL liability. Tier 2. If the best compounds in the chemistry stage still have PL concerns, use in vivo studies (at least fourteen-day oral rodent and nonrodent studies) as early as possible in the development process to determine confidence in safety, namely, to identify target organs, therapeutic index, and compounds that will be deemed potentially toxic and unsafe for continued development. Tier 3. Using de-risking tools or “points to consider” to manage and provide further confidence in the safety of a new drug candidate with PL liability.
Once a risk management plan is established for a specific program, it should be reevaluated as new preclinical and clinical information emerges.
Tier 1: Chemistry Selection of Compounds with Minimal PL Liability–In Silico Approach
The goal for the “first tier” is to select a compound with no potential, or minimal potential, to induce PL while respecting other significant features (i.e., safety, pharmacology, efficacy, bioavailability, genetic toxicity, and pharmacokinetics) required of a drug in the earliest discovery phase. This step also requires the knowledge of structure activity relationships (SAR) and recognition that cationic amphiphilic characteristics frequently indicate the drug will have a propensity to induce PL. Generally, discovery chemists will make use of an iterative process to define compounds that have the necessary characteristics to be nominated for further drug development. The iterative process makes use of quantitative in silico (computer-based) predictive tools based on chemical structure, physiochemical properties, results from in vitro models, and knowledge of other compounds (Table 1). A thorough review of in silico methods and in vitro assays for prediction of PL can be found in Monteith, Morgan, and Halstead (2006). In some programs, it may be preferable to screen all compounds in an in silico system. In vitro assays may be used to back up the in silico method or to confirm the best selection of compounds with little if any potential to induce PL.
Once it is has been determined that a lead compound has potential liability of inducing PL, subsequent molecular designs can be focused on modifying the physiochemical properties known to be associated with PL. The physiochemical properties associated with PL are primarily lipophilicity (affinity for lipids or ability to dissolve in lipids, including cell membranes) and positive charge. The greater the lipophilicity (higher calculated log of the octanol/water partition coefficient, or clogP), and the more basic the molecule (higher pKa), the greater its PL risk. Lipophilic, basic amines are the type of molecules that are most commonly associated with PL. These molecules are also often called cationic amphiphilic drugs, or CADs. Molecules that contain more than one basic amine have a dramatically increased risk for PL because they have additional positive charge. Designs should be undertaken that reduce the number of basic amines to the minimum number required for pharmacologic activity.
The shape of a molecule and the distribution of its polar functionality should also be analyzed when designing molecules with reduced PL liability. Molecules that are rod shaped, have polarity concentrated on one end, and are lipophilic on the other end have the highest risk for PL, especially as the distance separating the polar and lipophilic ends increases. The degree of polarity and the degree of lipophilicity combined with the distance separating the two ends of the molecule are described by the physiochemical property of amphiphilicity, and this property can be calculated (Fischer, Kansy, and Bur 2000). Molecular designs that reduce amphiphilicity should be favored. Molecules that have a more even distribution of polarity and lipophilicity, rather than having polarity or lipophilicity concentrated at one end, will have lower amphiphilicity and are thereby considered more favorable. Making molecules less linear and more ball-like has been shown to reduce amphiphilicity.
In silico models can be used in conjunction with an iterative process to design PL liability from a lead series. Starting with a lead that is active for PL, modifications can be made to the chemical structure to alter properties linked to PL induction such as clogP, pKa, reducing the number of basic nitrogen atoms, and making the molecule less linear. Ploemen’s model (Ploemen et al. 2004) as well as Pelletier’s modification (Pelletier et al. 2007) of the model can be used to predict PL liability of a molecule.
Tier 2: Determine Confidence in Safety–In Vivo Confirmation
The first tier focuses on ensuring that the drug candidate to be advanced displays the appropriate physical-chemical properties representative of lead compounds for development. Occasionally the best compound (acceptable hERG [the
If PL is anticipated based on in vitro assay results, in vivo studies should be conducted early in the development process to confirm the presence or absence of PL and assess potential for other toxicities. Often, these evaluations can be conducted in conjunction with an in vivo exploratory study prior to GLP studies and will guide the design of subsequent regulatory toxicology studies. In vivo studies offer information regarding the metabolic and physiologic challenges of the drug that would not be found in routine in vitro screens. For instance, a drug might be negative for PL in vitro but positive in vivo, suggesting that a metabolite may be the cause of PL (Reasor, Hastings, and Ulrich 2006). Preferably, rodent and nonrodent systems should be considered concurrently or sequentially for initial in vivo evaluation. The selection of dose level, dosing duration, sex, and relevant tissues to be collected and evaluated are determined on an individual compound basis, but selected tissues should always include the major and /or most sensitive organs such as liver, lung, kidney, and heart for microscopic examination. It is recommended that additional tissues be collected and held for possible evaluation based on initial histopathology examinations. Further, if histological changes such as foamy macrophages (Figure 2) suggestive of PL were observed with light microscopy in a previous study with the current or similar compound, the same tissue should be collected in the next study (e.g., representative sampling of the previously identified target organs) from control and treated animals in a manner that would permit confirmation of PL by TEM (Figure 3A, B, and C).
The authors have found tissues fixed by immersion in formalin to be acceptable for TEM examination following appropriate processing and post-fixation in osmium tetroxide to preserve the phospholipid component. Confirmation by electron microscopy is recommended, and regulatory agencies may request TEM evaluation. Routine examination of blood smears is also recommended. Phospholipidosis has been observed in circulating leukocytes following a single dose and in tissues twenty-four to seventy-two hours following intravenous administration. The presence of PL in peripheral leukocytes is generally compound specific and is often predictive of PL in solid tissues. The presence of PL in solid tissues is not highly predictive of PL in peripheral leukocytes, that is, there often is not a one-to-one correlation. However, a better correlation may be expected when there is involvement of multiple organ systems.
Performing fourteen-day rodent and nonrodent exploratory studies will identify potential target organs for PL, the drug exposures at which PL is observed in various organs, a no-observed-effect level (NOEL) for PL, and concurrent target organ toxicities. The duration of the study could vary based on knowledge of the chemicals being tested—fourteen days may be longer than is needed to detect PL for some chemicals and insufficient for others. One would expect to see PL in shorter-term, maximal-tolerated-dose studies. However, in the authors’ experience, occasionally a compound may induce PL only after a long-term exposure, namely, a three-month study. This variability in time to detection may be considered another example of the unpredictable behavior of some PL-inducing agents. This variability is also the basis of the recommendation to perform studies at least fourteen days in duration to confidently characterize the potential for a drug to induce PL.
Additionally, Tier 2 will most likely be the stage at which toxicity of a compound is unveiled, and the team is confronted with the difficult decision of potentially ending further development. Toxicity and mortality should be evaluated separately from PL. Phospholipidosis alone, with no concurrent target organ toxicities, may not represent a risk to patients and may not interfere with drug development. Phospholipidosis without concurrent organ toxicity may raise concerns when it occurs in animals at efficacious exposures or when there is marked PL in critical organs (i.e., heart, brain, and retina) with low exposure margins. A decision to continue or discontinue a compound in development based on PL alone should take into consideration the duration of treatment (more favorable for short-term-use drugs), target population, organs affected, and risk:benefit ratio, and so on. The “Points to Consider” in Tier 3 may be used to help guide difficult development decisions.
Tier 3: “Points to Consider” to Manage Development of New Drug Candidates
“Points to Consider” (Table 2) are de-risking tools that are routinely used to manage development of a new drug candidate. These tools include testing for presence of metabolites, determining reversibility, using available biomarkers of PL, consideration of margin of safety, benchmarking against other related chemicals or classes, managing concurrent target toxicity, and so on. Confirmation of PL by TEM is recommended. These tools help in making GO/NO GO decisions in drug development.
Presence of Metabolite
The parent compound or a metabolite may induce PL. In vitro assays can provide rapid analysis to determine the molecule likely responsible for PL. Consideration of the relative concentrations and chemical structures of metabolites in each preclinical species and in humans will help guide risk assessment. In general, assessment of the potential for metabolites to induce PL does not require additional or special in vivo studies. In specific cases in which a metabolite is suspected of producing PL, the exposure margins for the parent compound between the animal model and humans are low, and clear differences between species exist in exposure to the metabolite of interest. The value of special studies would be to demonstrate why PL may or may not be a risk in humans by showing that the metabolite responsible for PL in animals is not formed in humans or is formed in much smaller quantities in humans. In such a case, special studies with a metabolite may contribute to risk assessment.
Reversibility
Reversibility of phospholipids–drug complex formation within a reasonable time after treatment cessation is an important safety consideration in risk assessment. It is expected that PL will eventually reverse following cessation of drug administration (Halliwell 1997; Reasor, Hastings, and Ulrich 2006).
The authors recommend that reversibility or trend toward reversibility of PL in each tissue and in each species should be demonstrated in at least one multidose animal toxicity study prior to New Drug Application (NDA) filing. A minimum of four weeks should be allowed to demonstrate reversibility or a trend toward reversal, since it is not uncommon for complete reversibility to take more than a month (Reasor, Hastings, and Ulrich 2006). Ideally, reversibility would be conducted once for each affected species, but if new information such as new targets or concurrent toxicities is identified in later studies, additional reversibility assessments may be necessary. Confidence in safety is further increased if the exposure margin is high (i.e., >50× anticipated efficacious exposure in humans, hypothetically) at the lowest dose that produces PL, there is no concurrent tissue toxicity present at this dose, and reversibility has been demonstrated. Since demonstration of reversibility of animal toxicity findings is generally required for international registration, a study of reversibility of PL should require little additional resources.
Therapeutic Index (TI)
Phospholipidosis often occurs in the absence of any evidence of concurrent general or target organ toxicity or organ dysfunction. In drug development, PL alone is generally not considered a manifestation of toxicity, but rather an adaptive response (Reasor, Hastings, and Ulrich 2006). Based on the findings of in vivo studies, a therapeutic index (TI) can be determined. Establishing a large TI for PL over the clinical efficacious exposure is desirable, but doing so usually is not essential for continued development. One approach to determining the TI is to calculate the highest no-effect exposure in the most sensitive animal species divided by the anticipated human efficacious exposure (Cavagnaro 2008). Even for PL without concurrent toxicities, a minimally acceptable TI for PL could be used as a decision criterion. Criteria for an acceptable TI may depend on multiple factors such as the novelty of the compound or class; the clinical indication (Alzheimer’s disease vs. attention deficit disorder); treatment duration; intended population (adult vs. pediatric patients); tissue(s) involved (liver vs. heart); and competitive market. For compounds intended to treat life-threatening diseases, the potential benefit usually exceeds the risk of PL to human health, and PL (without concurrent toxicity) is unlikely to influence registration or treatment decisions. Phospholipidosis may also be of less concern for compounds that are administered for a short period (i.e., acute IV therapy) to a few weeks (antibiotics for most acute infections, treatments for acute trauma, etc.). When compounds are intended to be administered chronically or indefinitely in relatively healthy individuals, children, or women of childbearing potential, regulatory agencies may take a conservative approach and expect compounds to have a TI for PL above a specified threshold to build confidence that PL is unlikely to occur at therapeutic doses. Also, the TI and the no-observed-effect levels may decrease and the number of affected tissues may increase as longer-duration studies are conducted.
Managing Target Organs
Safety and efficacy are the highest priorities for any new drug. The target organs affected by PL in animal models and humans influence the level of safety concerns. In the authors’ experience, phospholipid accumulation in the lung or liver in the absence of other toxicities usually generates minimal to mild concern among regulatory officials. Phospholipidosis in neurons, the heart, or the eyes might generate more safety concerns. A drug that has the potential to induce PL in terminally differentiated organs such as the eye, the heart, and neurons may require additional monitoring in clinical settings to assess organ function and potential organ damage. Organs in which PL and toxicity may occur concurrently most commonly include the liver, lungs, and kidneys. This distribution probably reflects metabolism and excretion pathways for chemicals. When toxicity occurs in a PL-related target organ, standard clinical pathology parameters, organ-specific function tests, and novel biomarkers may be used to monitor concurrent toxicity and reversibility of toxicity. The TIs for toxicity and PL may be different. The TI for target organ toxicity, rather than for PL, should be used to make GO/NO GO decisions.
Biomarkers for Phospholipidosis
Some scientists consider demonstration of lysosomal lamellar bodies (LLB) by TEM as a biomarker of PL. However, there is a need for noninvasive clinical biomarkers, since PL in human tissues often cannot be detected during clinical trials. Recent literature describes bis(monoglycerol) phosphate (BMP), derived primarily from the liver and isolated from urine, as a potential biomarker for PL (Baronas et al. 2006). However, no clinical biomarkers have yet been found that reliably detect PL induced by multiple PL-producing chemicals in most tissues. This is yet another peculiarity of PL—each compound must be evaluated independently.
Benchmarking in Risk Management
Benchmarking marketed drugs of similar indication and/or chemical structure is recommended when evaluating the benefit versus risk of a compound. Information on marketed compounds with PL can be found in peer-reviewed literature and through the Summary Basis of Approval documents prepared by the U.S. Food and Drug Administration (http://www.access-data.fda.gov/scripts/cder/drugsatfda/index.cfm). Review of this information will provide added levels of confidence for a new drug submission package, particularly if the new drug compares equally or favorably to approved marketed drugs with similar indications. Benchmarking will also help determine differentiation and marketability of a new drug.
Regulatory Considerations in Risk Management of PL
As expectations for evaluation of human safety have increased in recent years, regulatory interest in PL has also intensified. One cannot assume that meeting the standards applied to compounds approved in the last decade will ensure registration in the future. In an expert opinion paper (Reasor, Hastings, and Ulrich 2006), it was stated that “from a regulatory perspective, and consistent with the task of determining drug safety, PL has been considered an adverse finding, whether justified of not” (p. 576). One regulatory concern is the phenotypic or mechanistic similarity of drug-induced PL to some genetic diseases (e.g., Niemann-Pick Disease) with long-term effects (Reasor, Hastings, and Ulrich 2006; Willard 2008). A few compounds that induce PL, most notably amiodarone (Jessurun, Boersma, and Crijns 1998), gentamicin (Tulkens 1986), chloroquine (Siddiqui et al. 2007), 4,4′-diethylaminoethoxyhexestrol (DHEA) (Shikata et al. 1972), and more recently telithromycin (Turner, Corey, and Abruty 2006) cause concurrent toxicity in humans. In these examples, the relationship between PL and toxicity is unknown, and a direct cause–effect relationship has not been established. At present, however, there are no predictable, relevant, noninvasive biomarkers of PL in human tissues. Hence, there may be concern regarding the potential for lack of detection in clinical settings and the possibility that PL might lead to functional consequences. Although PL ultimately may not preclude final registration, it may adversely affect labeling and marketing (Figure 4). In addition, approval may be contingent on the development of risk management strategies, such as the need for TEM confirmation and proof of reversibility in nonclinical studies.
In 2004, as a response to the Critical Path Initiative, the FDA formed the Phospholipidosis Working Group to address PL concerns and to improve consistency in regulatory review and interpretation. The FDA working group is creating a database using a large repository of data from INDs and NDAs with clinical information as well as nonclinical data. This database will facilitate the building of quantitative structure–activity relationship models for identification of characteristics associated with PL (Willard 2008). The overall plans of the FDA working group are to develop a general guidance document for use in future interpretation of PL (Berridge et al. 2007). Food and Drug Administration interests also include determining the incidence/prevalence of PL in clinical and preclinical settings, assessing any potential toxic implications of PL, and identifying biomarkers of PL (Berridge et al. 2007). Of recent interest to the FDA PL Working Group is a potential relationship between PL and QT prolongation, since it has been noted that the physiochemical properties of some marketed drugs that cause PL may also cause QT prolongation (Willard 2008).
If a compound in development produces PL in animals, discussion of the toxicology risk management strategy with the appropriate U.S. and/or E.U. regulatory agencies should be considered early in the development program to ensure acceptability of the strategy, particularly if there is no precedence for PL in a particular indication.
Conclusion
Drug-induced PL refers to an excessive, reversible accumulation of PL and associated drug in lysosomes. In many cases, PL in animal studies is not associated with evidence of concurrent toxicity or dysfunction in the affected organ at the same dose level. Despite the lack of conclusive evidence that PL produces adverse consequences, a compound that produces PL may face regulatory scrutiny. Without additional evidence of concurrent toxicity or adverse functional changes, the authors and others believe that PL should not be considered adverse. This article presents a rational, integrated, multi-tier, iterative risk management strategy for selection and development of compounds with potential to produce PL. The first tier uses in silico and in vitro assays to select compounds with reduced probability of producing PL. In the second tier, representative compounds with an acceptable combination of limited PL liabilities, desirable pharmaceutical properties, and evidence of efficacy are tested in exploratory in vivo studies (preferably fourteen-day studies in the rodent and nonrodent species to be used for GLP studies) to determine if the compounds actually cause PL and to develop structure–activity relationships. If a compound shows sufficient promise to develop despite evidence of PL in animals, a scientifically based risk management strategy incorporating standard nonclinical studies (including assessment of reversibility) and targeted clinical monitoring may be used to evaluate the human safety risks. Until regulatory guidelines are available, a well-thought-out risk management strategy would demonstrate due diligence for a drug development program and provide greater confidence in the safety of a new drug.
Footnotes
Acknowledgements
The authors would like to acknowledge Michelle Cholewa for administrative support of the manuscript; and express appreciation to Dominique Brees and Eugenia Floyd for contributions to photomicrographs.