The Translational Pipeline From Laboratory to Approval

Why this path matters when you read about "breakthroughs"

News about a new compound, device or therapy usually reports one step on a long road. Knowing the steps tells you how far a finding has actually traveled. This page describes the path for a new drug under the U.S. Food and Drug Administration (FDA), with a short note on Europe. Supplements, medical devices, lab tests and procedures follow different rules, so do not assume the same steps apply to them. Evidence labels used on this site: preclinical (cells/animals only), promising but limited, approved medicine. This page is for education and is not medical advice.

The steps at a glance

Step 1 and 2: from thousands of compounds to a few

The FDA says that at the discovery stage thousands of compounds may be potential candidates, but after early testing only a small number look promising [1]. Developers then study how the body absorbs and clears the compound, how it appears to work, side effects and how it compares with similar drugs [1].

Before human testing, researchers must find out whether a product could cause serious harm. The two kinds of preclinical research are laboratory studies and animal studies, and the FDA requires good laboratory practices for the laboratory studies [2]. These studies are usually not large [2]. See Animal Models and Their Limits for what animal results can and cannot predict.

Step 3: the three phases of human trials

Developers must file an Investigational New Drug application before human research begins; it includes animal and toxicity data, manufacturing information and the study plans [3]. The FDA describes the phases this way [3]:

• Phase 1: about 20 to 100 healthy volunteers or people with the condition, for safety and dosage. Roughly 70% of drugs move on.
• Phase 2: up to several hundred people with the condition, for early evidence of effect and side effects. These studies "aren't large enough to show whether the drug will be beneficial." Roughly 33% move on.
• Phase 3: about 300 to 3,000 people with the condition, to test whether the product offers a treatment benefit and to watch for adverse reactions. Roughly 25 to 30% move on.

Because many drugs drop out at each phase, a promising phase 1 or 2 result is far from proof of benefit. A separate analysis of more than 400,000 trial records found that success rates differ by disease area; for cancer drugs it estimated 3.4% [6]. It also found that trials using biomarkers to select patients had higher overall success probabilities [6]. For more on each phase see Clinical-Trial Phases Explained.

Step 4: review and approval

The sponsor submits a New Drug Application containing everything about the drug, from preclinical data to phase 3 results [4]. If the application is complete, the review team has 6 to 10 months to decide; statisticians review clinical data, pharmacologists review the animal data, and inspectors visit trial sites looking for fabrication, manipulation or withholding of data [4]. Approval is for a stated use, and the approved labeling "accurately and objectively describes the basis for approval" [4].

In Europe, the European Medicines Agency (EMA) scientifically evaluates centralised marketing authorisation applications, and the European Commission grants the authorisation that is valid across EU Member States, Iceland, Norway and Liechtenstein [7]. Laws change; check the regulator for the current process.

Step 5: learning continues after approval

The FDA states that it is impossible to have complete information about a drug's safety at approval, so the true safety picture evolves over months and years on the market [5]. It reviews reports of problems and can add cautions to the labeling [5]. By law, developers are prohibited from advertising unapproved uses of their product [5].

What this means for claims you meet

• "Studied in the lab" is step 1 or 2. It is preclinical (cells/animals only) evidence, not evidence of benefit in people.
• "In clinical trials" covers a wide range. A phase 1 trial mainly asks about safety and dose.
• A clinic offering something is not approval. Approval is a regulator's decision for a named product and use.
• A published review has argued that about 89% of novel drugs fail in human trials, with roughly half of those failures due to unexpected human toxicity [8]. Treat it as a reminder that most candidates never arrive, not as a precise rate.

Related guides

Discovery vs Clinical Application separates a lab finding from a usable therapy, and How We Label Evidence in Specialty Topics explains the labels used here.