What Is a Multiple Ascending Dose Study?

A multiple ascending dose (MAD) study is an early-phase clinical trial in which small groups of volunteers receive repeated doses of an experimental drug at progressively higher levels, allowing researchers to learn how the body handles the compound over time and at what point side effects become unacceptable. These studies typically fall within Phase 1 of drug development and are conducted in healthy volunteers, though some involve patients with the target disease. The MAD study is one of the first real-world tests of how a drug behaves when taken more than once, bridging the gap between a single dose in a lab setting and the kind of repeated dosing that patients would eventually use in practice.

How a MAD Study Differs From a Single Ascending Dose Study

Most first-in-human drug trials begin with a single ascending dose (SAD) stage, where each volunteer receives just one dose of the drug and is monitored for safety and drug levels in the blood. Once each dose level clears safety review, the next group of volunteers gets a slightly higher single dose. The MAD stage comes after, and its logic is similar but with a critical twist: volunteers take the drug repeatedly over days or weeks, mimicking the way someone would actually use the medication in real life. A trial of the drug etrasimod, for example, tested single doses ranging from 0.1 to 5 mg in the SAD portion, then had volunteers take daily doses of 0.35 to 3 mg in the MAD portion to see what happened with sustained exposure.¹ This distinction matters because drugs can behave quite differently after repeated dosing. Some accumulate in the body. Others trigger the body to break them down faster over time. A single dose simply cannot reveal these patterns.

The Cohort Structure

MAD studies are organized into cohorts, each assigned a different dose level. Within each cohort, volunteers are randomly assigned to receive either the active drug or a placebo, and usually neither the volunteer nor the research staff knows who got which. This blinding helps ensure that subjective side effects like headaches or nausea are evaluated fairly. A common randomization ratio is three active participants for every one on placebo, though this varies. In a MAD study of the blood-pressure drug baxdrostat, for instance, five cohorts received doses ranging from 0.5 mg to 5.0 mg once daily for ten days, with each cohort randomizing either six or nine participants to the drug and two or three to placebo.²

The dosing duration also varies by compound. Some MAD studies dose participants for just five days; others extend to several weeks. A study of the potential antiepileptic ACT-709478 used an up-titration design in its later cohorts, where the same volunteers moved through escalating dose levels of 10, 30, 60, and eventually 100 mg over the course of weeks.³ This kind of within-subject escalation is less common but allows researchers to explore higher doses with fewer total participants.

Safety Measures That Protect Participants

Because these are among the first human exposures to a new drug, MAD studies are layered with safety precautions. One widely used approach is sentinel dosing. Before giving the drug to an entire cohort, researchers dose just two volunteers first, one on the active drug and one on placebo, then wait a set period to watch for any alarming reactions. The waiting window depends on what is known about how quickly the drug acts and how long its effects last. Only after a safety review clears those initial participants does the rest of the cohort begin dosing.⁴

Beyond sentinel dosing, MAD studies use predefined stopping rules. Before the trial starts, the protocol specifies exactly what kinds of side effects, at what severity and frequency, would pause or halt dose escalation. These rules cover decisions about individual participants (should this person stop taking the drug?), the current dose level (should the whole cohort stop?), and whether the next higher dose should proceed at all. The goal is to keep the process systematic rather than relying on subjective judgment in the moment.⁵

Heart rhythm monitoring is another standard feature. Drug regulators care deeply about whether a new compound affects the electrical timing of the heartbeat, because certain changes can be dangerous. MAD studies are well-suited to investigate this because they test a range of doses up to the maximum tolerated level, making it possible to detect heart-rhythm effects early and inform how much monitoring future trials will need.⁶

What Researchers Learn About How the Body Processes the Drug

A central purpose of the MAD study is understanding pharmacokinetics with repeated dosing: how blood levels of the drug rise and fall, whether the drug piles up in the body over time, and how long it takes to reach a stable (“steady-state”) concentration. Researchers draw blood samples at numerous time points and measure drug concentrations to build a detailed picture.

One key question is how quickly steady state is reached. For the antiviral entecavir, steady-state blood levels were achieved by about ten days after the first dose, and the amount of drug in the bloodstream scaled roughly in proportion to the dose.⁷ Cenobamate, a drug for epilepsy, took about two weeks to reach steady state and accumulated to roughly five times the single-dose level at doses between 50 and 300 mg per day.⁸ These numbers directly inform how doctors would eventually prescribe the drug, including how long to wait before adjusting a dose.

Not every drug behaves predictably with repeated dosing. Some show time-dependent changes that a single dose would never reveal. Midostaurin, a cancer drug, initially showed increasing blood levels over the first few days, but then concentrations actually declined by 30 to 50 percent before settling at a steady state, with an average accumulation factor of about 1.7.⁹ This pattern suggests the body ramps up its ability to break down the drug over time, something that only becomes visible with multiple dosing. Another compound, SCR430, showed even more dramatic changes: repeated dosing led to roughly a four-fold drop in drug exposure compared to a single dose, driven by the body clearing the drug much faster after repeated exposure.¹⁰

Measuring the Drug’s Biological Effects

Alongside safety and blood-level data, MAD studies often include pharmacodynamic measurements, meaning they track whether the drug is actually doing something biologically at each dose level. For a drug intended to lower a specific protein in the brain, researchers might measure that protein in spinal fluid. For a drug targeting the immune system, they might track markers of immune activation in the blood.

In a MAD study of RO7020531, a drug designed to stimulate part of the innate immune system, researchers found that the biological response increased in a dose-dependent manner once the dose hit a certain threshold.¹¹ A MAD study of avagacestat, a drug studied for Alzheimer’s disease, showed that daily doses above a certain level reduced key proteins in spinal fluid in a dose-dependent fashion over 28 days.¹² These pharmacodynamic readouts are crucial because they help identify the dose range that is not just safe but also likely to be effective, shaping the doses that move forward into larger trials.

When MAD Studies Enroll Patients Instead of Healthy Volunteers

Most MAD studies recruit healthy people, since the primary aim is to characterize safety and drug behavior without the confounding effects of illness. But some drugs are too potent or too targeted to test safely in healthy individuals. Cancer therapies, for instance, often go straight into patients. Similarly, biologic drugs aimed at specific diseases may be studied in the patient population from the start. A multiple ascending dose trial of bapineuzumab, an antibody tested for Alzheimer’s disease, enrolled 234 patients with mild to moderate Alzheimer’s and tested four escalating intravenous dose levels.¹³ A study of sirukumab for lupus enrolled patients with confirmed disease, testing escalating intravenous doses every two weeks.¹⁴ In these situations, the MAD design remains the same in principle, with escalating dose cohorts and placebo controls, but researchers also get a preliminary look at whether the drug has any effect on the disease itself.

Food, Genetics, and Other Variables

MAD studies sometimes layer on additional questions. One of the most common is whether food changes how the body absorbs the drug. In a study of oxfendazole, an antiparasitic drug, eating a fatty breakfast before taking the drug nearly doubled the total drug exposure and increased peak blood levels by about half compared to taking it on an empty stomach.¹⁵ Another study, of the compound SAL0133, found that drug exposure roughly doubled at steady state when the dose was doubled under fed conditions.¹⁶ These findings directly affect dosing instructions that eventually appear on a drug’s label.

Genetic variation is another factor that surfaces in MAD studies. People carry different versions of the enzymes that metabolize drugs, and these differences can dramatically alter drug exposure. A MAD study of DA-8031 found that volunteers with a particular genetic variation in the CYP2D6 enzyme had significantly different drug levels than those with other variants.¹⁷ Identifying these genetic influences early can flag populations that might need dose adjustments and head off safety problems before the drug reaches a broader group of patients.

Immunogenicity of Biologic Drugs

For biologic drugs, which are typically large protein molecules given by injection or infusion, MAD studies serve an additional purpose: checking whether the body’s immune system mounts a response against the drug itself. When antibodies form against a therapeutic protein, they can neutralize the drug, reduce its effectiveness, or in some cases trigger allergic reactions. A MAD study of astegolimab, an antibody targeting a protein involved in inflammation, detected antidrug antibodies in 14 to 23 percent of participants who received subcutaneous injections and in 33 to 50 percent of those who received intravenous infusions.¹⁸

Predicting immunogenicity before a trial starts remains difficult. Computational models have been developed to estimate how many people will develop these antibodies, but the models tend to overshoot. In one evaluation, observed antidrug antibody rates ranged from about a third to two-thirds of participants after single doses, while the model captured those rates reasonably well. After multiple doses, however, observed rates were somewhat lower, and the model overpredicted the incidence.¹⁹ Repeated exposure can sometimes induce immune tolerance, where the body stops reacting to the drug. This is another phenomenon that only becomes visible in the MAD stage.

How Starting Doses Are Chosen

A natural question is how researchers decide what dose to start with, since giving too much of an untested drug could be dangerous and giving too little wastes time. Both the U.S. Food and Drug Administration and the European Medicines Agency have published guidance on this. The FDA’s approach starts with the highest dose in animal studies that produced no observable adverse effects, then applies safety margins to translate that into a human starting dose. The EMA’s guidance takes a broader view, addressing many of the risks inherent in first-in-human trials and outlining strategies to manage them.²⁰ In practice, the SAD study typically identifies a safe dose range first, and the MAD stage then explores repeated dosing within that range, often starting below the highest tolerated single dose to account for accumulation.

Using Models to Speed Up Development

Running a MAD study is expensive and time-consuming. Each cohort must complete its dosing and safety review before the next can begin, and the whole trial can take months. Increasingly, drug developers use mathematical modeling to predict MAD results from SAD data, potentially allowing some dose levels to be skipped or trial designs to be streamlined. For cofrogliptin, a long-acting diabetes drug, researchers built models from single-dose data that successfully predicted how the drug would behave in the MAD stage and even forecasted long-term efficacy in patients.²¹ For the hemophilia drug Mim8, modeling was used to transform early trial data into a dosing strategy that simplified treatment for both children and adults.²²

These model-informed approaches do not replace the MAD study, but they can make it leaner. If a model confidently predicts steady-state levels and biological activity for an intermediate dose, researchers might skip that cohort and test only the doses where uncertainty remains high. Regulators have become increasingly open to this kind of efficiency, especially when the modeling is transparent and well-validated.

What Happens After the MAD Study

The data package from a MAD study feeds directly into the design of Phase 2 trials, where the drug is tested in patients for the first time (or in larger patient groups if the MAD already enrolled patients). The dose range selected for Phase 2 is shaped almost entirely by what the MAD study revealed about safety thresholds, drug accumulation, time to steady state, and biological activity. A drug that showed strong pharmacodynamic effects at a well-tolerated dose in the MAD will move forward with confidence. A drug that required near-toxic doses to produce any biological effect is likely to be shelved.

MAD studies also generate data that regulators require in the drug’s eventual application for approval. The pharmacokinetic profiles, food-effect data, and safety findings all become part of the permanent record. Even years later, when a drug is on the market, doctors and pharmacists can trace the dosing recommendations on the label back to what was learned in those earliest human studies. For a process that typically involves only a few dozen volunteers, the MAD study punches well above its weight in shaping a drug’s entire future.