Thank you Pablito. That's a good read with a lot of information.RC

Thanks for this Pablo, I'm not in a trial but still find it very helpful in understanding a bit about how clinical trials work and how meaningful data can be derived from them.

As Wendy said, I too must read this a few times to take it all in but greatly appreciate the outline.

Thanks for taking the time and for sharing this with us, well done.

Dave

Thank you Pablo! A good read!

SF

Wow thank you Pablo. I need to read that a few more times to take it all in. Fabulous info. 

Thank you Pablo! Outstanding work, it will be very helpful for the members and guests to better understand the clinical trial process.

A lot of us are either offered places on trials or have to make decisions as to what direct acting antiviral (DAA) medication combo to choose for our treatment and, as such, we end up having to read complicated scientific papers to inform our decision-making process.  I did a masters degree in science a few years ago and had to learn research methods as part of the process so I thought it would be useful to explain some of the concepts and terminology one encounters in relation to HCV studies.   I haven't referenced any specific papers here but I did go through my university notes first and have - like many of us - read many papers around DAAs before starting treatment. 

This is not meant to be definitive or extensive in nature.  And if I have made any mistakes please feel free to comment below.  The other thing to say is that the science around DAAs is moving quickly so if you are reading this in the future - it's June 2016 at time of writing - the research will not doubt have changed, but the fundamentals behind analysing it wont.

RCTs

Randomised control trials (RCTs) are the gold standard studies in science, the best way proving if a new drug works.  Numerous variables can unduly influence the results of a study and an RCT is the best way to try and reduce these factors.  

A lecturer once gave me this example: if I get influenza on a Monday and then for the next 5 days I paint my face red by the Saturday my flu will be gone.  I may think that I have found a new cure for flu.  But if this new treatment was put to an RCT my theory would be disproved, as there would be a control arm in the study with no treatment and the results would be the same between the face-painting arm and the no treatment arm, i.e. both got better from flu.

Concepts

As above, the whole point of doing RCTs is to reduce bias.  Bias is any factor that could unduly influence the outcome the study results.  For example, generally researchers want to prove that their new drug works and without randomisation they could be tempted to assign patients with no cirrhosis, for example, to the experimental drug arm and those with cirrhosis to the control arm in order to give the experimental drug a better chance of performing better.

The process of randomisation is a computer-generated method of assigning trial participants to one of two (or more) arms in a study.  Randomisation takes the decision-making out of the researchers' hands; it is essentially a throw of a die as to which arm you get allocated to.

The arm of a study means which treatment option you get allocated to.  A common scenario in DAA trails would be to have one arm as drug X for 12 weeks, another as drug Y for 12 weeks, and a third arm as drug X for 24 weeks.

Having one arm of a study as a control group ensures that the researchers have a comparison group to compare the results of the experimental drug to.  Without this one wouldnt truly know if the experimental drug works.  

In the early phases of the development of a new drug it is usual to use a placebo (an inactive tablet that looks like the experimental drug) as the control arm; but then in later phases - if the early phases have shown the new drug to be effective - a gold standard comparator is used to compare the new drug against.  The gold standard being the most effective other drug on the market.  For a number of different reasons gold standard comparison studies (head-to-head trials of new DAAs vs. the best DAA known) have not been done yet, but they will be done in the near future.

There is an ethical issue at hand with placebo arms in the DAA studies, especially so in phase 3 studies, given that HCV is a lethal disease and the DAAs are very effective so giving no treatment could be deemed unethical.  A work-around you will encounter from time to time is the concept of a deferred treatment arm wherein people unlucky enough to be assigned to the placebo arm are entered into another part of the trial at the end of the initial study and are given the experimental drug.

New drugs generally go through 3 phases of studies before they get approved by the relevant regulatory bodies, e.g. the FDA in the US.  A phase one study is generally a lab-based or animal-based study with no human subjects.  They are proof-of-concept studies designed to test a scientific theory.  For example, DAAs are thought to block enzymes involved in the HCV replication process; this is fine in principal but does this theory stand up when tested in a lab?   Phase two studies will involve humans but they tend to include healthy subjects or a small number of HCV sufferers.  Their purpose is to test if the drug is safe and to observe its metabolism by the body.  Phase three studies are about rolling the new drug out and testing it in larger studies against a comparison drug or placebo.

Statistics

This is where it gets a bit complicated so Ill try and keep it simple.  Why use statistics at all?  Well, the only way to truly know that a new DAA is effective across the board would be to give it to everybody in the world who has HCV and check the results.  Obviously this is impossible so the work-around is to study smaller numbers of patients and then use statistics to see if the results in the study can be generalised to the whole population (i.e. would one expect to get the same results if this drug was given to other people, not in the study, with HCV).

The main thing to look for here is called the p-value, or the probability, i.e. what are the chances of the study's results being replicated when the drug is given to other people?  The scientifically agreed figure for a study having a positive result is a p-value of <0.5.  For reasons I wont go into the standard in science is present this as a double negative.  By this I mean if we use SVR12 (i.e. having no detectable viral load at 12 weeks after the end of treatment) as the desired outcome for a new DAA study then a p-value of <0.5 means that the chances of not getting a positive result when the drug is given to someone in the real world is less than 5%; or to flip this around: there is a 95% chance the drug would work if given to a patient outside of the study.

Sometimes confidence intervals are added to the p-value and look something like this: new DAA x 12 weeks gives 95% SVR p<0.5 (CI 91-98%).  Translated, this means that, as above, there is less than a 5% chance that the DAA would not provide an SVR when given to patients outside of the trial, and that the range of real world outcome lies between a 91% chance of SVR and a 98% chance.

What should I look for when considering entering a study?

1. Does the study have a placebo arm?  If so you risk the chance of receiving no treatment.  But, as above, the study may have a deferred treatment arm. 

2. If there is a placebo arm what is the randomisation ratio?  Because of the ethical issues mentioned previously it is rare to find a DAA study with a 50% chance of being in a placebo arm.  It's more likely that that there will be a randomisation ratio of 4:1 or 5:1 meaning your chance of getting no treatment is 1 in 4 or 1 in 5.

3. How long is the treatment in the study?  In real world clinical practice a doctor will - notwithstanding insurance approval/cost issues - weigh up your prognostic factors (i.e likelihood of successful treatment based on genotype, presence of cirrhosis etc.) and decide on which combination of DAAs will work best for you and for how long should you be treated.  A downside of being in a trial is that there is rarely any flexibility in terms of how long you will be treated for.  There are a range of different treatment lengths, but broadly-speaking standard treatment nowadays is for 12 weeks, or for 24 weeks if cirrhosis is present or if one has failed DAA treatment before.  If you enter a trial you may or may not get this length of treatment.

4. Resistance testing.  One of the main theories around why a small percentage of people fail DAAs is that they either have pre-existing Resistance Associated Variations (RAVs) to one or more of the DAA drugs being given in treatment or that RAVs develop during treatment.  RAVs are mutations of the HCV virus against which the DAA at hand may not be effective, and many studies gather data around these.  If you were unlucky enough to fail treatment then this would be very useful information for your doctor to have when guiding which re-treatment DAAs combo you should have; so ask the trial doctor if they are allowed to release RAV data to you.

What should I look for when reading a study to try and inform your decision about which DAA is best for your?

1. Impact factor.  Medical journals are ranked by their importance by a thing called impact factor.  The higher the impact factor the better the journal with better quality studies in it.  You will generally find the impact factor on the home page of the journal.  Below 2 is a low impact factor, 5 middling and above 10 very high.   Many studies are presented as posters at academic meetings these days.  Posters in of themselves are not a bad thing and many posters will go on the be published in journals, but be aware that posters are not yet published and, as such, have not gone through the review system most journals use to critique a paper before deciding whether it is good enough to publish.  

2. The results.  Obviously, this is the most important factor to judge and, as above, SVR12 or sometimes SVR24 is the main outcome variable of most DAA studies.

3. Number of participants.  The higher the number of participants in a study the more likely that the results are accurate.  If a study only has 30 patients in it the chances of the results being inaccurate are much higher than in a large study with, say, over 300 people in it.

4. The p-value. 

5. Did the study include patients like me?  For example, if you have cirrhosis it is important that the study you are using to guide your treatment choice contained participants with cirrhosis.  Initial studies, especially ones funded by pharmaceutical companies, tend to 'cherry pick' the best patients for inclusion, i.e. the ones for whom the DAA is most likely to work (e.g. patients with no cirrhosis or a low viral load). So if you have cirrhosis and the study you are reading only included patients without cirrhosis you need to bear this in mind.  As time passes, as is now the case with DAA therapies, more 'real world' studies occur that include more difficult-to-treat patients.

6. Funding.  It's worthwhile checking who funded the study. Pharmaceutical companies have to adhere to the same standards of research governance as independent researchers but it's always reassuring if the results from a pharma-funded study are replicated by an independent research group.  But often due to the cost of running large trials truly independent studies cannot be done, or when they are they tend to be smaller scale studies.