Clinical Trials

Key Concepts in Clinical Trials

Controls and Placebos

When clinical trial participants recieve a drug and their diseases changes in some way (disease worsens, improves or new symptoms develop etc), how do we know that the change is due to the drug and not due to some other force? The way to control for these other forces is by using a control group—a group of trial participants who think they're being given the new drug but are actually getting a fake inactive placebo. Clinical trialists use placebos not because they believe that placebos are magical or that positive thoughts can cure disease but because they want to isolate the effects of their drug from the other forces that make diseases improve (eg many diseases eventually improve on their own, participants sometimes feel better just knowing they're being cared for).

Valid placebos should be as close as possible to the drug in aesthetics and formulation (the makeup of the tablet excluding the active drug component) to avoid unblinding. For example, clinical trials of vaccines include a placebo consisting of the entire formulation except the immunogenic component. This is because the vaccine formulation has components that cause some discomfort at the injection site—if plain saline was used instead, the control group would immediately know they have not recieved the vaccine. Some other qualities of drugs that could complicate placebo design include colour, odour, taste or distinctive side effects.

In some cases, an effective treatment already exists for the disease being studied. In such situations, it is unethical for the researchers to withold treatment from the control group (it will also be extremely hard to recruit participants if they have a 50:50 chance of getting no treatment whatsoever). In such cases, it is beneficial to use an "active comparator": the current best treatment for the disease.

Randomisation

All clinical trial participants (as humans) are highly variable in their age, sex, genetics, pre-existing diseases, lifestyle behaviours etc. Any of these factors can influence their body's response to a drug. If humans are involved in allocating participants (eg if participants allocate themselves to the group they want) there is always potential for systematic allocation bias. For example, what if males respond better to the drug BUT males are more likely to ask to join the control group. This will result in a systematic bias against the drug. Randomisation overcomes this problem by allocating participants to a group based on nothing but chance (in practise, usually done by a random number generator computer program).

Something important to be aware of is that randomisation does not guarantee that both groups will be perfectly matched for all variables (nor is that the goal of randomisation). Instead, randomisation aims to remove bias from the allocation process so that any differences between groups can be attributed to drug action rather than a systematic difference in the makeup of the groups. By convention, papers reporting clinical trial results will include a table (usually the first table in the results section) showing the baseline characteristics of each group (ie the percentage of females, average age, BMI, percentage of people who smoke etc) so that the reader can judge for themselves how well matched the groups are.

Blinding

Whether intentional or not, both patients and researchers can influence the results of clinical trials if they know who is recieving the drug and who is recieving the control: patients in describing their symptoms and researchers in recording/interpreting their observations. Blinding solves this problem.

In a double-blind trial, neither the participants nor the researchers know who is recieving drug or control. An alternative design is a single-blind trial in which only the participants are blinded to the intervention. Single-blind trials are less robust because the potential for researcher-related bias remains. However, double-blinding is not always possible due to logistical (eg financial, time, workforce limitation) or design reasons (eg trial of a surgical intervention compared to sham surgery). A third type of trial is an unblinded or "open-label" trial in which both researchers and participants know who is recieving drug and who is recieving control.

Four (five) Phases

Overview

The final stages of drug development—human clinical trials—are separated neatly into four distinct phases—1, 2, 3 and 4. These phases occur in sequential order and each represents a barrier to the next. Phases 1-3 occur before approval and phase 4 represents the post-approval monitoring. The journey of clinical testing is treacherous taking about a decade to complete, costing up to $100 million USD and failing most of the time. Yes, 80-90% of drugs entering phase 1 will not make it to approval.

An additional phase—phase 0—has earned popularity since the mid-2000s. As the name suggests, it is performed before phase 1 but this page will discuss phase 0 last as it is still seen as less important to the traditional 1, 2, 3, 4 paradigm.

Phase 1 is focussed on the safety of the drug in human subjects. Phase 2 is focussed on the feasibility of the drug as an efficacious treatment. Phase 3 is focussed on determining the efficacy in treating patients with the disease. Phase 4 is focussed on observing the success (or failure) of the drug during real-world use by clinicians and patients.

Phase 2 and 3 studies are mandated by medicines regulators and the results can be viewed in journal publications or sometimes in the public product monographs that are indexed in huge databases by regulators such as the ARTG (Australia), DailyMed (United States), EMA (Europe) and Health Canada (Canada). Phase 0, 1 and 4 studies are usually not mandated and the public doesn’t always get access to the results.

In general, the number of participants, cost, time and complexity of studies increases with each phase.

Comparison of clinical trial phases.

	Phase 0	Phase 1	Phase 2	Phase 3	Phase 4
Focus	Proof of translatability	Safety	Efficacy	Efficacy	Safety and effectiveness
Research questions	Does the drug behave as expected in humans?	What is the safe dosage range? What is the best route of administration?	Is the drug efficacious in treating the disease?	Is the drug more efficacious than current standard of care?	Is the drug useful/effective? Are there rare side effects?
Number of participants	10-20	10-100	50-500	100-10,000	1,000-1,000,000
Dosage	Microdose	Increasing dosage	Therapeutic dosage	Therapeutic dosage	Approved dosage
Types of participant	Healthy volunteers	Healthy volunteers	Healthy or sick volunteers	Sick volunteers	Real world patients
Duration	1 week	3-12 months	1-2 years	1-5 years	Ongoing
Approximate rate of progression to next phase	90%	70%	30%	60%
Risk to participants	Low (extremely low dose)	High (first humans to take the drug at a reasonable dose)	Moderate (first patients to take the drug for a reasonable amount of time)	Low (plenty of experience from phase 2 regarding safe dosing, side effects etc)	Low (approved only once rigourous safety standards are met)

Phase 0

Phase 0 studies are sometimes called "microdosing studies" because they involve administering extremely small doses of the drug—typically less than 1% of the dose expected to have a therapeutic effect. These miniscule doses are well below the threshold of toxicity, making phase 0 studies a great opportunity to safely study the drug’s behaviour in human bodies for the first time.

The research questions in phase 0 are related to confirming the assumptions from earlier preclinical work…

• Does the drug behave as expected (in terms of pharmacokinetics) in humans?

Phase 0 studies serve to bridge the gap between animal studies and phase 1 human trials; they whether the biological characteristics of the drug which were studied in animals prior are similar in humans. If the drug behaves similarly in humans, it can progress to phase 1. If not, the company can terminate the drug and potentially saves millions of dollars. The biological characteristics are primarily related to the drug pharmacokinetic parameters: bioavailability, volume of distribution, route of excretion etc. Drugs with a well-defined site of action (eg inside the joints or the brain) may be measured in that compartment (eg in joint aspirate or in cerebrospinal fluid) to confirm that the drug reaches its site of action.

Phase 0 trials were introduced by FDA recommendation in 2006. They are growing in popularity but are still relatively uncommon and mostly considered optional. Phase 1 participants take great personal risk (often with small financial compensation) to help advance drug development. They deserve the additional protection afforded by a phase 0 trial.

Phase 1

Phase 1 is primarily a safety and scoping study. The goal is to find a suitable method of administration, safe dosage range and begin to characterise the side effect profile. Design is usually dose escalation: patients are given a low starting dose which is slowly increased until toxicity (or another pre-defined outcome) is reached. Phase 1 may be further broken down into phases 1a and 1b where this dose escalation occurs with single doses (1a) or many cumulative doses (1b). Since this phase is focussed on dosing, there may be a small investigation of food effects to determine if the drug is absorbed differently when taken with a meal.

The research questions in phase 1 are related to characterising ideal dosing parameters…

• What is the safe dosage range?
• What is the ideal route of administration?

A tragic example of poorly-planned phase 1 trial is that of theralizumab.

Phase 2

Phase 2 is the first-in-patients phase; the drug is given to patients who have the disease of interest to see how they respond to treatment at the therapeutic dose which was determined in phase 1. This can be thought of as a scoping efficacy trial: phase 2 only looks to see if the drug can improve an aspect of disease (eg a sign or symptom). The goal is to see whether the drug is likely to be efficacious in the patient population.

The research questions in phase 2 are related to establishing therapeutic potential…

• Does the drug improve symptoms/markers of the condition?

Phase 3

Like phase 2, phase 3 is also primarily concerned with drug efficacy. Phase 3 differs from phase 2 in that the drug is compared to the current best/standard treatment. This comparison tells us if the drug could improve patient care and could be a valuable addition to the pharmacopoeia.

The research questions in phase 3 are related to the drug's performance against the current standard of care…

• Is the drug more efficacious than the current standard treatment?
• Is the drug better tolerated than the current standard treatment?

At first glance, it might look like good results in phase 2 should necessarily translate into good results in phase 3. This isn’t always true. An important difference between phase 2 and 3 is that phase 2 may look at markers of disease as an endpoint whereas phase 3 will have to look at a patient-relevant outcome. Biomarkers are indicators of disease (or indicators of disease severity) and correlate well with outcomes but may not be directly impactful to the patient’s life (eg blood pressure, blood sugar levels, tumour size). Outcomes are the effects experienced by the patient which do impact their life (eg heart attacks, pain, disability, death). For example, high blood pressure is a marker of cardiovascular disease because it increases risk of heart attacks. We give medication for high blood pressure because we want to lower the patient’s risk of heart attacks (not because high blood pressure itself is bad in its own right). A phase 2 trial would ask the question “Does the drug lower blood pressure?”; a phase 3 trial would ask the question “Does the drug prevent heart attacks?”.

An illustrative case study is that of semagacestat, an inhibitor of gamma-secretase, the enzyme responsible for producing amyloid-beta plaques which accumulate in the brains of Alzheimer’s disease patients. The phase 2 trial of semagacestat included 51 patients given semagacestat or placebo. When the investigators took samples from the patient’s blood and spinal fluid, they found levels of beta-amyloid where lowered as predicted by the drug’s mechanism of action. However, in semagacestat’s phase 3 trial which looked at clinical progression of Alzheimer’s disease in 2,009 patients, the patients who received semagacestat actually deteriorated more quickly than those receiving placebo!

Phase 4

Phase 4 generally uses bigger cohorts than all of the pre-marketing phases (phases 1-3) combined; even if your pre-market trials had 10,000 participants, you're still unlikely to observe a rare 1 in 100,000 side effect.

The research questions in phase 4 are related to effects of the drug in large, “real world” populations…

• What is the effectiveness in actual use?
• What are the rare side effects?

Phase 4 studies may also seek to answer questions related to cost-effectiveness of the drug (with respect to the current standard of care), monitor for potential drug interactions or compare effectiveness across different subpopulations (eg males vs females, high vs low disease severity, older vs younger patients).

You may have noticed that I changed from using the term “efficacy” when talking about phases 1-3 studies to “effectiveness” when talking about phase 4 studies. In this context, “efficacy” refers to the magnitude of benefit in a randomised controlled trial (RCT); “effectiveness” refers to the magnitude of benefit in real world use. In making the distinction, we raise an important issue about whether results of RCTs necessarily translate to clinical practise. You might be able to think of reasons why efficacy might not mirror effectiveness. I have listed a few possible scenarios below…

• Patients with more severe disease are more likely to get newer (ostensibly better) drugs. If you track mortality in your phase 4 study, effectiveness will seem lower than efficacy in phase 3 because the patients most likely to get the new drug are more likely to die due to their more severe disease. If you track a surrogate marker in your phase 4 study, effectiveness could seem higher because the patients with the most severe disease have more room for improvement.
• New-to-market drugs are more expensive than their older alternatives. Patients who are able to afford the expensive drug are also more likely to be able to afford other health-promoting things (eg healthy food, gym membership, rest and relaxation activities, regular doctor visits). This will make the drug effectiveness seem higher.
• Patients who use the drug in real life may have characteristics that would exclude them from a clinical trial (eg smokers, high BMI, other chronic health conditions, other chronic medications, documented anaphylaxis, advanced age).

Quiz