Drug-Induced Phospholipidosis in FDA and EMA Reviews: Nonclinical Assessment, Development Impact and Labeling

Drug-induced phospholipidosis in FDA and EMA reviews: assessment, development impact and labeling

Drug-induced phospholipidosis (PLD), the lysosomal accumulation of phospholipid seen as foamy macrophages, vacuolated cells or lamellar bodies, is a common repeat-dose toxicology finding for cationic amphiphilic drugs. FDA and EMA review documents show a consistent pattern. On its own, PLD is rarely treated as a barrier to approval. Reviewers judge it on the company it keeps: associated tissue injury or functional effects, exposure margins, reversibility, the clinical dosing regimen, and whether a plausible clinical correlate exists. When those factors are unfavorable, the finding has led to a partial clinical hold, additional animal studies, targeted clinical monitoring, postmarketing commitments, EU risk management plan (RMP) entries and, most commonly, nonclinical labeling text.

Key takeaways

Default position: "human significance unknown." Both agencies usually conclude that the relevance of animal PLD to humans is unknown or unclear. Examples include adagrasib ¹¹⁵⁰¹¹⁴⁸, pimavanserin ¹²³⁹, vorapaxar in the EU ¹⁰⁷⁷, relugolix ⁷¹⁷¹⁰⁶² and bedaquiline in the EU SmPC ⁸⁹⁷.
The adverse/nonadverse line is drawn on associated pathology, not on PLD itself. For cariprazine, FDA pathology experts considered foamy macrophages alone adaptive and nonadverse at about 2.2 times human exposure, but active lung inflammation adverse at about 4.2 times ¹¹²³. For adagrasib, FDA considered PLD adverse only where it was associated with deaths and lung, cardiac, splenic or reproductive toxicity ¹¹⁵⁰¹¹⁵¹.
Labeling is the most frequent consequence. In the US, PLD appears in Section 13.2 (Animal Toxicology and/or Pharmacology) for products such as Nuplazid, Brexafemme, Krazati, Orgovyx, Myfembree, Sirturo and Vraylar ¹²³⁸⁷⁰¹¹¹⁵³⁷¹⁸¹⁰⁶²¹³⁴⁶¹²⁵⁵. In the EU, it appears in SmPC section 5.3 (Preclinical safety data) ⁸⁴⁷⁷⁹⁵⁸⁹⁷¹⁰⁸²⁵⁶³⁷⁸³⁸⁸¹.
Development requirements followed when margins were thin, the finding was progressive or lung function was at stake. Lomitapide is the clearest example: a partial clinical hold from 2007 to 2010 on trials longer than six months ¹⁰¹⁶¹³⁷¹.
Clinical PLD is labeled as a clinical risk only where it has been documented in patients. Hydroxychloroquine's Warnings and Precautions link biopsy-proven PLD to cardiomyopathy, myopathy/neuropathy and renal injury ¹²⁹⁴¹²⁹⁶¹²⁹⁷. Amiodarone labeling treats foamy macrophages as "markers of therapy" rather than evidence of toxicity ¹²¹⁸¹²¹⁹.

How reviewers characterize a phospholipidosis finding

Across both agencies, review documents assess PLD along the same set of variables.

Species, tissue distribution and confirmation

Reviewers record which species and organs are affected and whether PLD was confirmed ultrastructurally. Pimavanserin produced multi-system PLD in mice, rats and monkeys, prominently in rat lung and kidney ⁷⁶⁷⁷⁶⁸⁷⁶⁹. Bedaquiline produced PLD in mice, rats and dogs at all tested doses, even after a single dose, mainly in the monocytic-phagocytic system; its M2 and M3 metabolites were stronger inducers than the parent ⁵³⁹.

Confirmation matters. For lomitapide, early electron microscopy called the rat lung finding pulmonary PLD, but later examinations found mainly neutral-lipid vacuoles without the lysosomal inclusions characteristic of classical PLD, and the reviewer described the interpretation as uncertain ¹⁰¹²¹⁰¹³. For dalbavancin (Xydalba), CHMP rejected the applicant's attribution of pigmented, vacuolated cells to PLD as unsupported because no histological analysis had identified it. CHMP nonetheless required no further studies ⁶⁵⁹⁶⁶⁰. For osimertinib, FDA and CHMP both noted rat lung ultrastructure consistent with early PLD in females but also with lipofuscinosis or multivesicular bodies ⁹⁷⁸¹¹⁵⁷¹¹⁵⁸.

In vitro data can also enter the assessment. For netarsudil, PLD in Chinese hamster ovary cells (EC50 1.1 µM) was considered attainable in the cornea. Both agencies treated PLD as a plausible, though not established, mechanism for monkey corneal haze and patient corneal verticillata ⁹¹⁵⁹¹⁸⁹³⁰⁹³¹. In the reverse case, daprodustat and nine metabolites showed no significant activity in PLD assays ⁸⁴¹.

Exposure margins

Margins are central, and reviewers present them in PLD-specific tables where the data allow. The FDA pimavanserin review tabulated no-effect margins for lung PLD of 13-fold in mice, 0.7- to 2.0-fold in rats and 3.8- to 4.1-fold in monkeys ⁷⁷⁰. For bedaquiline, the FDA safety-margin table listed no NOAEL and no AUC margin for PLD in any species, and one rat study found PLD at about half of clinical exposure ⁵⁴⁰⁵⁴¹.

High margins weigh heavily in favor of the product. For vorapaxar, FDA's clinical reviewer gave most weight to the monkey finding at more than 100 times human exposure and said PLD should not bar approval ⁹⁵⁶⁹⁵⁸⁹⁵⁹. For linagliptin, reviewers judged human clinical risk minimal given that the chronic rat lung finding occurred at about 418 times the maximum recommended human dose (MRHD) ¹²⁰⁰¹²⁰². For brensocatib, rat lung PLD occurred at 152 to 202 times MRHD and the FDA review called its human relevance "not clear" ¹⁰²². CHMP accepted that the finding was not expected to be clinically significant ⁴⁶⁹.

Low margins do not automatically block approval, but they shift the burden. For berotralstat, FDA called AUC margins of 0.1 to 0.8 "generally irrelevant" because the limiting findings were attributable to PLD, which had not been seen in humans in clinical trials ⁸⁶⁴⁸⁶⁵. For netupitant, animal-to-human AUC multiples were 0.4 to 1.8, but daily animal dosing contrasted with human dosing no more often than every three weeks ⁸¹⁶⁸¹⁷⁸²¹.

Reversibility

Reviewers distinguish complete, partial and slow reversal. Bedaquiline PLD was "slowly reversible" ⁵⁴²⁵⁴³¹³⁴⁶. Cariprazine lung PLD did not reverse after one to two drug-free months ¹¹¹⁷¹²⁵⁵. Pimavanserin findings resolved in monkeys after four months but persisted in several rat tissues after six months ⁷⁶⁹⁷⁷¹. Berotralstat PLD in the nine-month monkey study was "not readily reversible", with most findings remaining after 13 weeks ⁸⁶³. In contrast, CHMP accepted brensocatib lung PLD as completely reversible ⁴⁶⁸.

Associated toxicity and functional consequence

This is where most adverse calls are made.

Lung. In 26-week rat studies of ibrexafungerp, marked alveolar foamy-histiocyte accumulation correlated with labored breathing and impaired respiration at about 10 times recommended human exposure. The FDA reviewer noted that awareness could help if related events emerged with longer clinical use ⁶⁹⁷⁶⁹⁸⁶⁹⁹. In rats given pimavanserin, lung inflammation and fibrosis were judged secondary to pulmonary PLD, and severe involvement was associated with morbidity or death ⁷⁷²⁷⁷³⁷⁷⁴⁷⁷⁵. For posaconazole, CHMP noted that PLD generally lacked functional effects except in the two-year rat study, where severe alveolar obstruction affected respiration ⁴²⁴⁴²⁵. For camizestrant (Etcamah), the female rat lung finding after six months was adverse because of an inflammatory infiltrate. It occurred at 60 times human exposure and was considered unlikely to be meaningful in humans ⁴⁵³⁴⁵⁴.
Kidney and liver. Rat PLD with pimavanserin was linked to kidney degeneration and raised BUN/creatinine ⁷⁷². FDA linked liver-enzyme and histologic findings with telithromycin to PLD, while the label concluded its significance for humans was unknown ⁶⁵²⁶⁵³.
Nervous system. With posaconazole, dogs alone showed neuronal vacuolation. Neurotoxicity testing at about three times human exposure found no functional deficit ⁴²⁶⁴²⁷. For rezafungin, CHMP assessed Schwann-cell inclusions in rats and monkeys as PLD without degeneration or nerve-conduction change ¹⁰²⁹¹⁰³⁰.
Muscle. For bedaquiline, CHMP noted muscle changes with raised CK/ALP. However, the proposed mechanism of myopathy was not PLD itself, and CHMP concluded that significant PLD was unlikely at human exposures ⁸⁹²⁸⁹³⁸⁹⁴.

When phospholipidosis changed development requirements

Product (INN)	Agency	Development consequence tied to PLD
Juxtapid / Lojuxta (lomitapide)	FDA	Partial clinical hold (June 2007) on trials longer than six months; request for a three-month rat study with recovery, an established pulmonary PLD NOAEL and electron microscopy. Hold kept until February 2010 after deficient responses ¹⁰¹⁶¹³⁷¹. Spirometry and DLCO used in trials ¹⁰¹⁴¹⁰¹⁸.
Akynzeo (netupitant/palonosetron)	FDA, EMA	FDA requested chronic rat and dog studies to assess PLD ⁸²²⁸²³. IRT-QT consult recommended cardiac electron microscopy and suggested phase 3 LVEF/troponin monitoring be considered ⁸²⁴⁸²⁵. In the EU, clinical development included liver-function tests and examination for lamellar bodies ⁵⁸⁴⁵⁸⁶.
Nuplazid (pimavanserin)	FDA	Postmarketing commitment 3069-4: special-stain re-evaluation of rat and monkey lung tissue for collagen and identification of a NOEL for lung inflammation ¹³⁶⁴¹³⁶⁵.
Orladeyo (berotralstat)	FDA, EMA	FDA agreed urinary BMP was not clinically useful for phase 3 monitoring but considered the proposed safety database inadequate given PLD potential and reliance on end-organ toxicity for detection ⁸⁶⁶. Pulmonary diffusion measured in two studies ⁶²³. EU RMP: required NI-PASS and paediatric study ⁶²⁵⁶²⁶¹³²⁸.
Orgovyx / Myfembree / Ryeqo (relugolix)	FDA, EMA	Phase 2 ophthalmologic exams and biomarker studies, then visual acuity in phase 3 ⁷²⁰. A development-stage request asked for visual-acuity results to monitor for ocular PLD ¹⁰⁶⁰. FDA said mechanistic biomarker studies were not needed for approval ⁷²¹. PLD was among clinical safety parameters in the EU dossier ⁸³⁹.
Rhopressa / Rhokiinsa (netarsudil)	FDA, EMA	FDA requested short- and long-term visual-function evaluations for corneal deposits ⁹¹⁹. CHMP supported a post-authorisation safety study of longer-term safety, including the deposits ⁹³⁶⁹³⁵.
Vraylar / Reagila (cariprazine)	FDA, EMA	FDA requested an explanation of the PLD-toxicity relationship and a review of dog lung histopathology, answered with an expert report ¹¹²⁵¹¹²⁶. PLD/fibrosis tracked as an adverse event of special interest in trials ⁸⁸⁰.
Rezzayo (rezafungin)	EMA	Tremor monitoring as a potential adverse event ¹⁰³². PIP modification advice to discuss a possible PLD and monkey tremor association in scientific advice ¹⁰³³.
Brinsupri (brensocatib)	EMA	Under PRIME, the signal prompted a call for careful safety characterisation ⁴⁷⁰⁴⁷¹.
Solithromycin	EMA	CHMP requested further discussion of the toxicological meaning and potential human implications of PLD ⁴⁹⁹.

Three patterns stand out.

Pulmonary PLD with uncertain long-term reversibility draws the strongest action. Lomitapide's hold was imposed because the long-term risk of pulmonary PLD could not be assessed ¹⁰¹⁶. It was lifted partly because similar animal findings with approved products had not translated into clinical risk ¹³⁷².
Monitoring targets the plausible end organ, not PLD itself. There is no validated clinical biomarker. The EU bedaquiline RMP states this directly ⁸⁹⁵. FDA declined urinary BMP as a phase 3 monitoring tool for berotralstat ⁸⁶⁶. Monitoring therefore uses functional proxies such as DLCO, visual acuity, ophthalmologic exams and organ-specific laboratory tests ¹⁰¹⁴⁷²⁰⁵⁸⁴.
Organ-toxicity monitoring is often broader than PLD. For bedaquiline, FDA's medical officer tied close trial monitoring to animal organ toxicities generally, not to PLD specifically ⁵⁵⁰. The EU SmPC notes that effects on the monocytic phagocytic system were not monitored clinically ⁸⁹⁷.

EU risk management plans

Under the EU RMP framework, CHMP has handled PLD in different ways.

Important potential risk with routine risk minimisation. Vorapaxar (Zontivity) listed PLD as an important potential risk, with SmPC section 5.3 as the only risk-minimisation measure ¹⁰⁷⁸¹⁰⁷⁹¹⁰⁸².
Important potential risk later removed. Netupitant PLD (Akynzeo) was an important potential risk in the initial RMP and was later removed from the list of safety concerns ⁵⁸⁹⁵⁹⁰¹³²⁷.
Important potential risk with additional pharmacovigilance. Berotralstat PLD is addressed by a required non-interventional PASS and paediatric study, because human relevance is unknown, particularly with lifelong treatment ⁶²⁴⁶²⁵⁶²⁶.
Enhanced pharmacovigilance. For posaconazole, the plan called for evaluating potential PLD signals, with pulmonary events monitored postmarketing ⁴²⁹⁴³⁰¹³²⁹.
Key nonclinical finding only. Bedaquiline, adagrasib and nirogacestat list PLD in the nonclinical safety specification without making it a separate risk item ⁸⁹⁵⁷⁹⁴⁹⁰⁴.

Labeling practice

United States

FDA places animal PLD in Section 13.2. Typical text describes species, tissues, exposure multiples, reversibility and a statement that human relevance is unclear. Examples:

Nuplazid: PLD most affected lungs and kidneys. In rats at 10 times or more MRHD exposure it was associated with rales, labored breathing and renal tubular degeneration, with deaths at 16 times or more. The label states that "the relevance of these findings to human risk is not clear" ¹²³⁸¹²³⁹. Labeling language was proposed in the nonclinical review ⁷⁷⁶.
Krazati: Findings at 150 mg/kg or more in rats (2 times or more human AUC) and 25 mg/kg in dogs (about equal to human AUC). Reproductive findings are also captured under 13.1 ¹¹⁴⁸¹¹⁵³¹¹⁵⁴.
Sirturo: PLD in all species "at almost all doses, even after very short exposures", slowly reversible ¹³⁴⁶¹³⁴⁷.
Orgovyx/Myfembree: Evidence of reversibility, significance in humans unknown ⁷¹⁸¹⁰⁶².
Vraylar: Not reversible after one to two drug-free months, at clinically relevant exposures, including juvenile rat lung findings suggestive of PLD ¹²⁵⁵¹²⁵⁶.
Ketek: Older-format labeling described PLD under ANIMAL PHARMACOLOGY with dose multiples ⁶⁴⁹.

FDA did not always add label text. The retrieved Zontivity prescribing information excerpts do not mention PLD ¹³¹⁹¹³²⁰¹³²¹. For safinamide, FDA's cross-discipline reviewer concluded only that the importance of PLD was unknown ⁴⁸⁷.

European Union

CHMP places animal PLD in SmPC section 5.3. Examples include artenimol/piperaquine (Eurartesim), where repeat-dose findings at exposures similar to clinical exposure were considered possibly relevant to clinical use ⁸⁴⁷⁸⁴⁹. Others include dronedarone (described as slight, reversible, species-specific and not relevant to humans) ⁵⁶³, safinamide ⁷⁸³, gilteritinib ¹¹⁸¹, cariprazine ⁸⁸¹, camizestrant ⁴⁵⁷, nirogacestat ⁹⁰⁵, imlunestrant ⁹⁸², crizotinib ⁶⁷² and memantine ⁹⁶⁶. For telithromycin, section 5.3 was updated after a re-evaluation of PLD-associated changes in five pivotal repeat-dose studies ¹²⁶³. CHMP also accepts omission where justified: for rezafungin, it agreed that no PLD text was needed in the SmPC ¹⁰²⁹¹⁰³⁰.

Clinical PLD in patients

Two older cationic amphiphilic drugs show how PLD is labeled once documented in humans.

Hydroxychloroquine (Plaquenil, revised 12/2023): Section 5.1 links cardiomyopathy to PLD on endomyocardial biopsy and notes PLD may occur in other organs. Sections 5.8 and 5.11 describe PLD on muscle, nerve and renal biopsy. The label advises cardiac monitoring as clinically indicated, periodic muscle strength and reflex assessment, and discontinuation if toxicity is suspected or shown by biopsy ¹²⁹⁴¹²⁹⁵¹²⁹⁶¹²⁹⁷.
Amiodarone (Cordarone): Under Pulmonary Toxicity, PLD (foamy cells, foamy macrophages) is present in most cases of drug-induced pulmonary toxicity but also in about 50% of all treated patients. The label therefore treats these cells as markers of therapy, not toxicity, and relies on baseline and periodic chest X-ray and pulmonary function testing ¹²¹⁸¹²¹⁹¹²²⁰.

These precedents help explain why reviewers compare new findings to amiodarone. For pimavanserin, toxicologists considered the animal lung findings likely clinically relevant, but not suggestive of the irreversible fibrosis associated with amiodarone ¹²⁷⁶.

Same product, different emphasis: FDA versus EMA

Product	FDA	EMA
Bedaquiline	No PLD NOAEL or AUC margin in any species. PLD not shown to cause functional change. Section 13.2 text ⁵⁴⁰⁵⁴⁵¹³⁴⁶	Significant PLD unlikely at human exposures, yet RMP says relevance unknown and no validated biomarker. SmPC 5.3 ⁸⁹⁴⁸⁹⁵⁸⁹⁷
Vorapaxar	Principal toxicology finding. High monkey margin considered most relevant. Should not bar approval ⁹⁵²⁹⁵⁸⁹⁵⁹	Important potential risk in RMP. SmPC 5.3 states margins acceptable and clinical significance unknown ¹⁰⁷⁸¹⁰⁸²
Safinamide	Importance unknown. Rat lung PLD possibly contributing to respiratory distress ⁴⁸⁷⁴⁸⁹	Unlikely to be clinically significant. Described in SmPC 5.3 ⁷⁸²⁷⁸³
Dronedarone	Foamy macrophages in rat lungs and other tissues noted ⁵⁹³⁵⁹⁴	Species-specific and not relevant to humans. A later pulmonary safety review could not establish whether PLD explained postmarketing pulmonary toxicity ⁵⁵⁹⁵⁶¹⁵⁶²
Adagrasib	Adverse at high rat doses with deaths, otherwise not. Section 13.2 ¹¹⁵⁰¹¹⁵³	Most margins below 1. RMP notes trial laboratory monitoring across affected organ systems. SmPC 5.3 ⁷⁹¹⁷⁹⁴⁷⁹⁵
Netupitant	Chronic studies requested. Uncertain clinical significance ⁸²²⁸²¹	Important potential risk later removed. SmPC 5.3 ⁵⁸⁹¹³²⁷⁵⁸⁴

The differences reflect emphasis more than conflicting science. CHMP more often formalizes the finding in the RMP and SmPC 5.3. FDA reviewers more often deal with it through pre-submission requests, study-specific risk discussions and Section 13.2 text.

Practical implications for sponsors

The following points are drawn from the review precedents above.

Characterize early and specifically. Confirm PLD by electron microscopy or equivalent, separate it from neutral-lipid or lipofuscin accumulation, and include metabolites. Ambiguous characterization prolonged the lomitapide hold ¹⁰¹²¹³⁷¹ and drew an "unsupported" finding for dalbavancin ⁶⁵⁹.
Present PLD-specific margins and recovery data. Reviewers built PLD-specific margin tables when they could ⁷⁷⁰⁵⁴⁰. Long recovery arms address the persistence questions raised for cariprazine, pimavanserin and berotralstat ¹²⁵⁵⁷⁷¹⁸⁶³.
Separate adaptive from adverse. Pathology peer review that distinguishes foamy macrophages alone from inflammation, fibrosis or functional effect was decisive for cariprazine and adagrasib ¹¹²³¹¹⁵⁰.
Plan functional clinical monitoring for the target organ. DLCO, visual acuity and organ-specific laboratory tests have been accepted in place of a PLD biomarker ¹⁰¹⁴⁷²⁰⁸⁶⁶.
Expect labeling. In both regions, a confirmed multi-organ PLD finding near clinical exposure will most likely appear in Section 13.2 or SmPC 5.3, usually with a statement that human significance is unknown ¹¹⁴⁸¹⁰⁶²⁸⁹⁷¹⁰⁸².

Across these reviews, phospholipidosis changed the development path when it was tied to functional injury, thin margins, or an uncertain long-term pulmonary risk; otherwise it was handled mainly through nonclinical labeling and, in the EU, the risk-management plan.