Learn about Clinical Data
This section is my guide to teach you about some of the key elements of the clinical trials. I go over the phases of the clinical trials and what they are designed to evaluate. I explain some of the statistical measures used in the clinical data. I even go over a few of the key endpoints used in Oncology. This section is about helping you understand more about the clinical trials themselves before we try to actually read them.
Phases of Clinical Trials
First will be years of preclinical work. This is all about the chemistry, formulation, toxicology and manufacturing of the drug. They will have to do extensive testing and submit all that data to the FDA for the Investigational New Drug (IND) application. The first studies will be done in human cells. This will be to establish a proof of concept of how the biology works in the cells. Then they will move to animal models. This is often mice and NonHuman Primates (NHP). Since the systems in animals are not always exactly the same as humans, not every drug will be tested in both mice and monkeys. Some concepts like immunology might not be applicable to one of these animals. Once all the animal models are done, the company will submit all its data for an IND with the FDA. The FDA will review the data and determine if human trials can proceed. Whenever the FDA finds an issue in any stage of clinical development, they will often issue a hold or partial hold on that program along with a Complete Response Letter (CRL). This letter will request more information on topics of concern.
Then the drug will proceed into phase 1 clinical trials. The sole purpose of this stage of development is to prove the safety and pharmacokinetics (PK) of the drug. The PK goes by the acronym ADME for Absorption, Distribution, Metabolism and Excretion. Phase 1 is all about learning how the drug works in the human body and any side effects it might cause. In non life threatening diseases, this is done in healthy volunteers. In diseases like cancer, they will test in patients with the disease. The first 2 studies are the SAD (Single Ascending Dose) and MAD (Multiple Ascending Dose) studies. This is where they test multiple doses across groups of patients to see each dose response. You might hear of the phase 1a and phase1 b studies. Phase 1a is all about the dose escalation and safety. Phase 1b is about finding the right dose to move to phase 2. This might include expanding the patients tested at some doses that seem to have the best responses.
Then the drug moves to phase 2. This is all about testing the efficacy of the drug in patients with the disease. Sometimes you will hear about phase 2a and 2b. The phase 2a might be more dose testing in patients with the disease to see which ones respond best. Then they can expand to phase 2b where they focus on the dose where the best efficacy was seen. There can be many phase 2 trials for the same drug. Once safety has been established in phase 1, they can test the drug in many different diseases with phase 2 trials. This is where they will expand the trials into all indications where they think a good signal of response was seen. Some phase 2 programs will start with a basket trial. This is where they test the drug with patients from different diseases to see which diseases show a good response. Then they expand those indications into a complete phase 2 study to test the efficacy in that specific disease. The key to good phase 2 data is understanding exactly what that study drug contributes to the efficacy toward a disease. Often companies will design confusing trials with multiple drugs that mask the efficacy of any single drug.
After a successful efficacy study in phase 2, the drug will proceed to phase 3. The phase 3 is designed to test this drug vs the current standard of care. The phase 3 is the hardest as it has to prove the drug actually improves outcomes for patients vs what they might be taking now. If the standard of care is Chemo + PD1, then the study might be Study Drug + PD1 + Chemo vs PD1 + Chemo alone. Phase 3 has to prove the drug can improve the standard of care. This might only be for a small population of patients with a disease. Sometimes a drug can improve outcomes for patients with a specific form of the disease or a specific mutation of a cancer. If phase 3 is successful, then the company will file for a New Drug Application (NDA).
The FDA will take 60 days to look over the package and then accept or decline it. A normal review takes 10 months. If the drug gets an expedited review, then it only takes 6 months. Some statuses issued by the FDA can speed up the process for review to the 6 months time point like fast track designation. Once accepted, they will issue a PDUFA (Prescription Drug User Fee Act) date by which they will issue a ruling. Sometimes, a drug will need what they call a Post Marketing Study or a phase 4 study. This is usually a long term follow up study to ensure long term safety or efficacy. It is not very often you will see these studies, but it is important to know they exist if you see one.
The main key to seeing a good trial design is that it has only 1 variable between the group with the study drug and the control group. Some studies do not use a control group. This usually happens in very bad diseases where it would be unethical to use a control group. That is usually when they compare data to historical data they know from the history of the disease. A blinded study means the doctors and patients don't get to know who is on the study drug and who is on placebo. Open label means both the patients and doctors get to know who is on the study drug. These can be used to explore an indication. Some studies will undergo a Blinded Independent Central Review (BICR) by a panel of doctors who don't get to see anything but the data.
p values and HR values
The p value is a measure of statistical probability. We use this in clinical trials to determine if the data we are seeing is from the trial we designed or if it could have happened by random chance. This is not a measure of how well a drug works. That is called clinical significance. The statistical significance is a measure of the quality of the data. I can demonstrate this by using an example as the math behind it can be complicated.
If you have a quarter, and you flip it 5 times, you might expect to get 50% heads and 50% tails. Let us assume you flip it and get 4 heads and 1 tail. Is this a valid outcome? First thing we will realize is that flipping the coin 5 times probably isn't a good sample size. Just one flip accounts for 20% of the count. If we do it again and flip the coin 1,000 times, we might expect 500 heads and 500 tails. Now let us say we get 495 heads and 505 tails. Is this within the realm of statistical probability? That is where the p value comes in. This tells us if the data we collected was valid or if it could have been just random chance. For example, let us say we got 400 heads and 600 tails. Now, we might suspect the data we got was somehow flawed. Maybe our coin was bent and threw off the outcomes. The p value is about measuring the quality of the data in a clinical trial.
The p value works off a percentage. If you see a p = .01, that means there is a 1% chance the data was from random chance. In clinical trials, we insist the accuracy of the data be within 5%. If you see a p value = .051, that trial will be considered a failure as the data is untrustworthy and could have been due to random chance. Many times you will see a p value like p = .0001 which is just a faction of a percent chance that the data was by random chance. A lot of investors focus on this p value in a clinical trial as it measures the quality of the trial design. It is a good indicator, but it is not the only measure of clinical success. You can be 100% sure of the data you see, but the data is still a failure because it is not clinically meaningful.
Another measure used in some clinical trials is a Hazard Ratio (HR). This is the measure of risk between two events happening. This is often used to compare the possible risk reduction between two therapies. It is also read as a percentage. A HR ration of 1 means that the outcomes of both events were equal. If the HR ratio is above 1 then the study drug is doing worse than what it is being compared to. If the HR ratio is less than 1, then the study drug has lowered the risk. You might see HR = .7 in a clinical trial. That means the risk was lowered by 30% from 1 to .7. Not every trial will have a HR ratio, but it is important to understand how to read it when you see one in trial data.
RECIST Response Rates
When we deal with solid tumors, we use the Response Evaluation Criteria In Solid Tumors (RECIST 1.1) standard to evaluate responses. The data shows that 90% of tumors are solid tumors and 10% are liquid tumors which use a different set of criteria. It is important to understand the RECIST criteria for solid tumors.
They measure the response of a tumor for any therapy using a scan. This is usually a CT scan or something similar. They will actually measure the size of the tumor in centimeters. On the next scan, they will measure again and the change will be the response. Let us say they scan the tumor and its 5cm. Then they do a regimen of treatment and scan again. The tumor is now 4.8cm. That means that the tumor shrank .2 cm. The math would be 5 cm minus 4.8 cm / 5 cm = or .2 / 5 cm = .04 or 4% decrease in the tumor size. One thing to know about RECIST responses is they tend to be backward. When the tumor shrinks, the response is positive so a 4% smaller tumor is actually a +4% response. When the tumor gets bigger, you have a negative response. So if the tumor grew from 5 cm to 5.2 cm, the response would be -4%. The one exception to these criteria is the appearance of new lesions or tumors. It won't matter how much the measured tumor shrinks if a new lesion or tumor pops up on the scan. That is immediately treated as Progressive Disease.
There are 4 different responses a patient can have. The first is a Complete Response (CR) which means no detectable tumor or lesion can be seen on the scan. That doesn't mean it is 100% gone. It just means it is now too small to be seen on the scans. We consider that a complete response to that therapy. The next response is a Partial Response (PR). That is where the tumor has shrunk +30% or more. That means our 5 cm tumor would have to shrink to about 3.5 cm to reach a 30% reduction. The next response is called Stable Disease (SD). Anything from a +30% response to a -20% growth in the tumor is considered stable disease. Progressive Disease (PD) is anything over -20% where the tumor has grown bigger than 20%.
Objective Response Rate (ORR) measures only the Complete Responses and the Partial Responses. This is what we use in cancer trials. We only want to know what percent of patients had a good response to that therapy. Some companies will use a Clinical Benefit Rate (CBR), but we don't use them as investors. They tend to do that when the ORR would look too poorly on their therapy. As an investor, you only care about ORR.
You can calculate an ORR from the data if it is not presented for you. Let us say you see a trial where they enrolled 40 patients. They only had 20 patients eligible for a readout at the time of the data collection. Of those 20 patients, we see 2 CR, 5 PR, 5 SD and 8 PD. We know how each of these rates are calculated so we just plug in the numbers. The CR is 2 + the PR is 5 = 7 total patients had a response out of 20 which is 7/20 = 35% ORR. If a company tries to report a CBR rate which looks impressive, we can adjust the CBR, for example, the the CBR of 2 CR + 5 PR + 5 SD = 12 out of 20 patients for a CBR rate of 60%. That looks much more impressive, but if we do the math ourselves and calculate the ORR we will see the real response was 35% ORR. In my experience an ORR rate in the low to mid 30's is about average cancer therapies.
* I am not a doctor. This is not designed to be Medical Advice. Please refer to your doctor for Medical Decisions