For orally administered products, whether a drug can be taken with or without food is not a labeling afterthought — it shapes dosing instructions, exposure-response interpretation, and the operational conduct of pivotal trials. Regulatory and clinical pharmacology teams need to know when in development the FDA expects food-effect data, what study design it will accept, and when an assessment can be waived, because getting the timing wrong can leave a gap the agency raises at review.
The analysis below traces the FDA's expectations across the development lifecycle, from the recommended high-fat/high-calorie study design and its labeling consequences to the conditions under which a food-effect study may be deferred or waived. It draws on the agency's bioavailability and food-effect guidance alongside Drugs@FDA clinical pharmacology reviews to show how these studies are actually run and how their results reach the label.
Want Rhizome's help on your own question? Try it for free.
When the FDA expects food-effect studies: timing, design, and review precedents
Food can change how much of an oral drug reaches systemic circulation and how fast it gets there. The FDA treats characterizing that change as a core clinical pharmacology deliverable for orally administered products, because the answer feeds directly into dosing instructions, exposure-response interpretation, and the DOSAGE AND ADMINISTRATION section of labeling. This overview sets out when in development the FDA expects a food-effect assessment, the study design it recommends, when a study can be waived, and what actual Drugs@FDA clinical pharmacology reviews show about how these studies are run and how their results land in labeling.
Why the FDA wants food-effect data
For orally administered drugs, the FDA uses food-effect (FE) studies during new drug development to determine whether food affects exposure, how large and variable that effect is, and which meal types matter 1136. The general bioavailability guidance frames food effects on bioavailability as an "essential early goal" of pharmacokinetic characterization, with PK work continuing throughout development 2. In practice, a food-effect result answers a labeling question: can the drug be taken with or without food, must it be taken on an empty stomach, or must it be taken with a meal.
The FDA judges the clinical relevance of any food effect using the totality of the clinical data, including exposure-response and PK-PD relationships. If the 90% confidence interval for the fed-versus-fasted comparison does not fall within bioequivalence limits, the sponsor is expected to interpret the clinical significance and propose corresponding label instructions rather than default to a mechanical rule 8.
When in development the FDA expects the study
The recurring theme across guidance and review precedent is early, and on the right formulation.
Early clinical development (Phase 1). Food-effect and relative bioavailability work belongs in early PK characterization, in Phase 1 or initial clinical studies, not deferred to the NDA 2. The FDA advises sponsors to assess the effect of a high-fat meal early in development and to test food effects before the pivotal safety and efficacy trials, so the results can guide dose and administration decisions and inform labeling 914. Preliminary assessments can be folded into pilot or first-in-human studies 914.
Review precedents bear this out. Fosamprenavir (Lexiva) had a "pivotal, Phase I, single dose" food-effect study in healthy adults covering both the tablet and suspension 67. Tofacitinib modified-release (Xeljanz XR) included a "Food Effect on proposed-commercial formulation" study within its Phase 1 healthy-volunteer program 81. Saxagliptin's definitive food-effect study was run at the 10 mg dose used in the Phase 3 program, on a formulation the FDA described as "almost identical" to the anticipated marketed formulation 768083.
On the to-be-marketed formulation, or formally bridged to it. A definitive FE study should use the final to-be-marketed oral formulation 914. The FDA repeatedly pressed sponsors on this point. For enzalutamide (Xtandi), the agency stated it is preferable to conduct the food-effect study with the to-be-marketed formulation 65. For vismodegib (Erivedge), the FDA asked the sponsor to confirm the food-effect study used the to-be-marketed formulation 6466. For dalfampridine (Ampyra), prior food-effect studies had used formulations different from the to-be-marketed 10 mg SR product, and the FDA said another food-effect study would have to be conducted if the existing one could not be justified for the final formulation 68. For palonosetron (Aloxi), the FDA said the food effect should be obtained for the to-be-marketed formulation and would accept a Phase 3-related result only if the formulations were sufficiently comparable 69.
Where the pivotal formulation already matched the market image, the study could double as both. For testosterone undecanoate (TLANDO), the to-be-marketed formulation was used in both the pivotal Phase 3 study and the definitive food-effect study 72. When an earlier study used a different formulation, the FDA accepted a bridging approach: afatinib (Gilotrif) bridged food-effect information from an earlier study to the final market-image Phase 3 product 62, and topiramate (Qudexy XR) had its definitive food-effect result, generated with an "MD" formulation, bridged to the to-be-marketed formulation 73. Aprepitant (Emend) reviews likewise directed that planned clinical pharmacology studies, including the food effect, be characterized with the proposed market image 63.
The practical sequence that emerges: run a preliminary assessment early, then lock in a definitive study on the final formulation (or bridge to it) before the results are relied on for approval and labeling.
The standard study design
FDA guidance recommends a consistent template for a definitive food-effect study, and reviews show it applied across therapeutic areas.
Design. A randomized, balanced, single-dose, two-treatment (fed versus fasted), two-period crossover 29. A three-way crossover is acceptable when the study is also assessing the bioavailability of two formulations 29. Some BE-specific scenarios use two separate two-way crossovers or a single four-way crossover, but that is a bioequivalence recommendation rather than the general FE design 33.
Treatment conditions. One period is fasted (empty stomach); the other is fed with the high-fat test meal 29.
Meal composition. The FDA recommends a high-fat, high-calorie meal because it produces the greatest effect on GI physiology and systemic drug availability 151822. The meal is specified as roughly 50% of calories from fat and about 800 to 1000 kcal in the food-effect guidance 30, and about 900 to 1000 kcal in the more recent immediate-release BE guidance 1822. The target breakdown is approximately 150 kcal from protein, 250 kcal from carbohydrate, and 500 to 600 kcal from fat 301822. For some lower-risk products a lower-fat, lower-calorie meal may be acceptable, and if the reference/comparator labeling specifies a meal, that meal should be used 1518.
Timing of the meal and fasting. In the fed arm, subjects begin the meal 30 minutes before dosing and finish it within 30 minutes, with the drug given 30 minutes after the meal starts 1833. Fasted studies are conducted after an overnight fast, with no food for at least 4 hours postdose 184043.
Reviews show these features in practice, frequently extending the design to a third arm to capture a low-fat meal. Deferasirox (Jadenu) used a single-center, open-label, randomized, three-period, six-sequence crossover comparing fasted, low-fat meal, and high-fat meal after a single dose 112. Abiraterone acetate (Zytiga) used a Phase 1, single-dose, open-label, three-period crossover across fasting, low-fat, and high-fat meals 125. Dexlansoprazole ODT (Dexilant SoluTab) used an open-label, randomized, single-dose, three-period crossover under high-fat conditions 117. Eltrombopag (Promacta) used a Phase 1, open-label, randomized, three-period balanced crossover of fasted versus high-fat breakfast 119.
When a food-effect study can be waived
The clearest waiver pathway is for BCS Class 1 immediate-release products with high bioavailability. The FDA may waive an FE study for BCS Class 1 immediate-release drug products that also have high bioavailability (F greater than or equal to 0.85), and sponsors are advised to consult the agency on feasibility 34.
The Biopharmaceutics Classification System also predicts where food effects are likely. For BCS Class I, food effects are least likely because absorption is generally pH- and site-independent; the FDA notes that more than 80% of BCS Class 1 immediate-release products are not affected by high-fat meals 1534. For BCS Classes II, III, and IV, food effects are most likely and their direction and magnitude are difficult or impossible to predict without a fed study 15. For immediate-release products submitted under an ANDA, a fed BE study is generally recommended except when both test and reference products are rapidly dissolving, have similar dissolution profiles, and are BCS Class I 14. A food-effect study may also be unnecessary where a biowaiver is accepted for a formulation change 29.
Modified-release and extended-release products
For modified-release (MR) and extended-release (ER) products, food-effect studies serve a second purpose beyond exposure: they test whether the product maintains its intended release behavior or whether food triggers rapid release or "dose dumping." Reviews used the food-effect result to decide (1) whether the product genuinely behaved as modified-release, (2) whether food caused dose dumping or a clinically important exposure change, and (3) what the label should say about meals 87919597102106110111.
Examples across the spectrum:
- No meaningful effect, label "with or without food." Diltiazem ER (Cardizem LA) showed small fed changes with 90% CIs within no-food-effect limits, and the reviewer concluded the label should state diltiazem can be administered with or without food 95. Desvenlafaxine ER was found administrable without regard to meals, with labeling stating the food-effect changes were not expected to be clinically significant 98102104.
- Food present but tolerated in labeling. For desvenlafaxine fumarate ER, fed-state AUC rose about 29% (a reviewer comment later cited 37%) but the change was still judged not clinically significant, supporting "without regard to meals" 102104.
- Food altered release, label toward fasting or warnings. Azithromycin ER (Zmax) showed faster release with food that negated the ER formulation's GI-tolerability advantage, so the review concluded it should be taken on an empty stomach 97. Oxcarbazepine ER (Oxtellar XR) showed a significant increase in parent-drug peak exposure with food and was directed to be administered under fasting conditions 106. Fluvastatin MR (Lescol) reviewers declined to allow "slow release" labeling and required that the large fed-state exposure increase and high variability be noted 91. Hydrocodone ER (Hysingla ER) had a definitive food-effect study showing higher fed Cmax and AUC, and the review framed food as materially changing PK 100.
- Dose-dumping and alcohol testing. MR oral products commonly triggered in vitro alcohol dose-dumping studies. Deutetrabenazine XR (Austedo XR) reported no in vitro alcohol dose-dumping potential 87. Sitagliptin/metformin XR (Zituvimet XR) showed lower release at 5%, 20%, and 40% alcohol and was judged to carry no additional alcohol dose-dumping risk 93. Clonidine ER was found not to dose dump in the presence of a high-fat meal, though in vitro data suggested alcohol could modify its release 110111.
How the result drives labeling
The direction and magnitude of the food effect, filtered through clinical relevance, determines the administration instruction. The precedents fall into three buckets.
Take with food (or with a meal). When food increases exposure, reduces variability, improves tolerability, or matches the conditions used in pivotal trials. Canagliflozin/metformin XR (Invokamet XR) is directed to be administered with a meal (specifically the morning meal); metformin AUClast rose about 72% fed while canagliflozin was minimally affected 115129. Venlafaxine ER absorption was unaffected by a high-fat meal, but nausea was more frequent on an empty stomach, so it is recommended with food 124. Lubiprostone (Amitiza) had a 55% fed-state Cmax decrease with unchanged AUC; because clinical relevance was unclear and pivotal trials dosed with food, the review supported labeling for administration with food 127. Amoxicillin (Moxatag) had unchanged AUC but a fed Cmax reduction and delayed Tmax, and the sponsor recommended administration with food 130.
Take on an empty stomach (or before meals / at bedtime). When food increases exposure to a degree that raises safety concerns, or reduces a peak that matters clinically. Efavirenz (Sustiva) fed-state exposure rose sharply (tablet AUC about 28% higher, Cmax about 79% higher), and the label directs dosing at bedtime on an empty stomach to reduce high concentrations and CNS effects 116118. Eltrombopag (Promacta) fed AUC fell about 59% and Cmax about 65%, a clinically significant reduction supporting empty-stomach administration 119. Abiraterone (Zytiga) showed a dramatic food effect (high-fat meal raised Cmax roughly 17-fold and AUC roughly 10-fold), driving very specific fasting instructions 125. Dexlansoprazole ODT lost 38% of Cmax with food and was directed to be taken at least 30 minutes before a meal 117.
Take without regard to food. When the food effect is small or not clinically significant. Azithromycin's two 250 mg tablets were not decreased by a high-fat meal (although one capsule formulation was food-affected) 113114. Fingolimod (Gilenya) showed similar fed and fasted AUC and Cmax with no meal restriction needed 121. Amlodipine/atorvastatin (Caduet) showed no meaningful effect on amlodipine and a modest atorvastatin reduction (Cmax about 32%, AUC about 11%) that did not drive a meal restriction 126128.
The through-line is consistent with the guidance: a high-fat, single-dose crossover generates the exposure comparison, and the label instruction follows from whether that fed-versus-fasted difference is clinically meaningful for safety, efficacy, or variability 811.
Practical takeaways
- Plan the food-effect assessment as an early Phase 1 activity, and design the definitive study on the final to-be-marketed formulation, or build a defensible bridge to it before relying on the result 9146568.
- Default to the randomized, single-dose, two-period fed-versus-fasted crossover with the high-fat, high-calorie meal (about 50% fat, roughly 800 to 1000 kcal), adding a low-fat arm when a graded meal effect is informative 293018.
- For BCS Class 1 immediate-release products with high bioavailability, discuss a possible waiver with the FDA rather than assuming one 34.
- For modified-release products, expect the food-effect study to be read alongside dose-dumping and alcohol-release data, and expect fasting or warning language if food accelerates release or sharply raises peak exposure 879397106.