This is one of those post (read: rants) where I want to put an idea out into the ether for someone to chew on. It starts with a very simple question:
Why is ‘the drug’ the focus of a clinical trial?
If our goal is to find beneficial therapies for people with Parkinson’s, then the way we currently clinically test drugs is utterly nonsensical.
And if we do not change our “we’ve always done it this way” mindset, then we are simply going to repeat the mistakes of the past. Others are changing, so why aren’t we?
In today’s post, we will consider one possible alternative approach.
Why is ‘the drug‘ the focus of a clinical trial?
The way we clinically test drugs makes absolutely no sense when you actually stop and think about it.
Other medical disciplines (such as oncology) have woken up to this fact, and it is time for the field of Parkinson’s research to do this same.
Let me explain:
The easiest way to test a novel treatment for a medical condition is to simply give it to your population of interest and see what happens.
For example, we have a drug (let’s call it ‘Curetide’ – à la a previous rant) and we would like to test it on a group of people with a particular medical condition (for the rest of this post we will use the example of Parkinson’s). So, we take a group of 20-30 people with Parkinson’s and we give them Curetide to take over a certain period of time, and we conduct assessments on each of them to see if the drug is having any effect. At the end of our study, we pool all the data together and look to see if Curetide had any beneficial effect on their Parkinson’s features.
This approach begins with the rather ridiculous assumption that everyone affected by Parkinson’s is the same, and that they are affected by the condition in the same way. If you have ever walked into a Parkinson’s support group meeting, you will know that reality is somewhat different from the theoretically ideal situation.
What if the drug only works for the red person?
Not only are people different, but the way their bodies process drugs varies significantly as well. Thus, in the schematic above, our Parkinson’s affected population should not be viewed as a equal population, but rather in different shades of different colours.
Now, to counter this problem of heterogeneity (individual differences) within our affected population of interest, our current clinical testing system takes the very prudent step of comparing our drug of interest with a neutral control (or placebo) treatment. A placebo is a treatment that looks identical to the drug of interest (in this case ‘Curetide’ – trademark pending by the way!) in every respect, except that it has no biological effect – in drug testing the placebo pill contains an ‘inactive substance’. This placebo is given to a control group in parallel to the treatment group, and it is usually given in a ‘blinded’ manner (meaning that the participants in the study are unaware of which treatment they are receiving – they do not know if they are being given the drug being tested or the placebo).
In an ideal world, this situation would look something like this:
And the only difference between the two groups would be the drug being tested – everything else would be exactly the same. The potential therapeutic benefit of the drug is being compared to the effect of a neutral placebo treatment between two identical groups.
But again, reality is not very accommodating:
As the image above suggests, both groups in any clinical trial will made up of a mixed bunch of individuals.
Now, some very sharp readers will point out that the two groups are still identical. Thanks to the variability between individual participants within each group, the two groups will balance out – they will shift to some kind of “Parkinson’s average” – and thus, the two groups can be considered equal. In theory, this would return us confidently to a situation where the drug is being compared to the placebo (between two almost equal groups).
There may be some truth to this.
And lest we forget, “we have always done it this way”.
But this is where the whole system begins to fail the patients.
Our current system requires a drug like Curetide to have a huge effect on the ‘treatment group’. A lot of people within the treatment group need to exhibit positive outcomes from taking Curetide. If the drug does not have a big enough beneficial effect on enough people in the treatment group (when compared to the control group), it will not succeed in passing through the various stages of the clinical trial process. It will therefore fail, and Curetide will be sent to the ever-growing scrapheap of failed drugs.
Under our current system, we basically have a filtering system for blockbuster drugs.
This approach may be considered great for the large pharmaceutical companies, whose business models are designed around developing drugs that can be sold to large proportions of the population.
But the whole system fails to appreciate an troublesome fact: the fact that we are all unique.
Let’s reconsider the situation from above once more:
Two identical groups. One experimental drug (Curetide) being tested. And let’s keep the two groups on the drug (or placebo) for 12 months and then see what effect the drug has had.
What could possibly go wrong?
Hypothetical question for you: what happens if I tell you that unbeknownst to us, Curetide only works in the ‘red’ individuals in the image above?
Will the drug be successful in the current clinical trial system? Will Curetide have a big enough impact on the results of the treatment group for it to pass the assessment process?
The average person in the treatment group will exhibit no difference when compared to the control group, and this will result in Curetide failing to demonstrate any effect in the context of the whole trial… even though it may have been very effective for one particular person. In fact, that one ‘responder’ in the treatment group would probably be considered an outlier, and the drug will be quickly shipped off to that ever-growing scrapheap of failed drugs I mentioned above.
Not looking good. Source: Pixabay
In every clinical trial of an experimental treatment, there will be responders and non-responders to the treatment.
As the label on the can suggests, ‘responders’ are individuals who exhibit a positive (or – in really bad situations – a negative) response to the treatment, while a ‘non-responder’ presents a response to the treatment that is no better than any of the members of the control group. The distribution of these groups can vary from treatment to treatment – some drugs will have a lot of responders, while others will have very few – but there are almost always responders and non-responders in the treatment group of a clinical trial.
The difference between responders and non-responders is based on our individual biology.
For example, in 2016, researchers discovered that there are hundreds of people living in Norway who are completely unresponsive to opioid-based pain killers (such as morphine) due to a genetic variation in their DNA (Source). In an opiate-based pain killer drug trial before 2016, these individuals would probably have been the non-responders in a clinical trial treatment group, potentially lessening the chances of success for the experimental pain killer.
So basically, our current system is a filter for treatments that have the most responders.
And while this may at first appear to be a useful approach for the pharmaceutical industry in their efforts to identify blockbuster drugs, in truth billions of dollars of investor money are being wasted on ‘failed’ trials where there were not enough positive responders in the treatment group. It is wasteful approach that does not serve the affected community nor society as a whole. And critically, it is reducing our chances of overall success by limiting the number of potentially useful drugs.
There is the very real possibility that effective drugs (for certain individuals) are being lost or missed out on because our current system is flawed.
If our goal is to truly find beneficial therapies for people with Parkinson’s, then the system needs to change.
We have to shift our approach from filtering for drugs that treat the maximum number of responders, to simply filtering patients.
Let me explain:
Take a group of people diagnosed with Parkinson’s. Give them an experimental drug (for example, Curetide) and then monitor them over a period of time. Maybe no one responds to that drug – fine, but after a period of time we move everyone to the next drug and repeat the process. Those who respond to that second drug stay on the second drug. Those who do not respond (the non-responders) get shifted to the next drug. And this process would continue thus:
One critical part of such a system would be feedback.
The characteristics that define a person who responds to a particular drug would be collected and fed back to the clinicians, thus allowing for better determination of who might respond to that drug in the future. New participants and new drugs could be added to such a study design as it proceeds, adding to the pool of information being collected and fed back to help better assign individuals.
Another critical part of such a system would be continuous measures of assessment which would be assessed continuously and acted on. That is to say, if it is apparent that a drug is having no impact, the individual would be shifted to another drug (as opposed to waiting till the end of a classical clinical trial to determine if a drug has actually had any impact). These assessments would require us to learn as much as we can about each person enrolled in such a study, from their genetic risk factors and biochemistry through to their environment.
And the adaptive nature of this system would allow for different combinations of drugs over time. For example, after a certain period of time the responders to drug no.2 in the image above, could be given drug no.3 to see if any additive benefits could be gained. If they respond well to the combination of drug no. 2+3, then they would stay on that regime. If they didn’t exhibit any additional benefits, they could be shifted to a combination of drug no.2+4 (as indicated in the image below):
Such an additive approach would work well for a condition like Parkinson’s where ‘a cure’ would require a multi-component treatment approach will be likely (Click here to read more about this).
Overall, when compared to the classical clinical trial, this new style of clinical trial is considered adaptive. An adaptive clinical trial “evaluates a medical device or treatment by observing participant outcomes (and possibly other measures, such as side-effects) on a prescribed schedule, and modifying parameters of the trial protocol in accord with those observations” (Source). To read more about adaptive clinical trials – click here for a good review.
And a lot of clinical trials are now incorporating aspects of the adaptive trial idea. In fact, in 2013 it was calculated that approximately 20% of Phase III trials utilise some elements of adaptive design (Source).
The most popular type of adaptive clinical trial is referred to as a multi-arm, multi-stage (MAMS) platform trial. These compare multiple drugs at the same time – stopping an drugs that fail to demonstrate benefits, while continuing with those that do. But thus far most of these trials have been conducted on the group level (rather than individual patients), and thus are still under that influence of the “maximum number of responders” effect we mentioned above. In addition, many of them have not included cross-over components (where participants are shifted to new drugs).
Multi-arm, multi-stage (MAMS) platform trial. Source: Eupati
One of the best examples of an adaptive clinical trial are the I-SPY studies – which are focused on cancer.
I-SPY (“Investigation of Serial studies to Predict Your therapeutic response with imaging and molecular analysis”) is an adaptive trial design that has enabled two experimental breast cancer drugs to deliver promising results after just six months of clinical testing (those drugs being Veliparib and Neratinib – Click here to read more about this). The value of the I-SPY trials, however, go well beyond the clinical results, as they have given proof of principle validation to the concept, resulting in a host of additional adaptive clinical trial projects for various medical conditions, such as Prostate cancer (the Stampede trials), Multiple Sclerosis (the MS-Smart trial) and Alzheimer’s (the EP-AD consortium).
Which begs the obvious question: why are Parkinson’s clinical trials lagging behind this trend? Where is the adaptive trial for Parkinson’s?
And rather than simply jumping on the band wagon, why shouldn’t an adaptive Parkinson’s clinical trial attempt something innovative and radical – raising the bar further for adaptive trials and providing the template for something new?
There are several good reasons for the lag.
- Cost: Adaptive clinical trials to date have been very expensive to set up and run. But if we compare that to the cost of all the failed clinical trials of the past, perhaps the cost is not so great.
- Definitions: Given the variety we see in individuals with the condition, it is tempting to ask ‘what exactly is Parkinson’s?’ While cancers have very specific biopsy-based definitions of certain tumors, Parkinson’s is only just coming up with some basic groupings of subgroups based on genetic risk factors. Hopefully, however, characterising responders to a particular drug in an adaptive clinical trial might highlight novel groupings.
- Assessment: My haters (loveable as I am, I do have a few) will say that we are not there yet with biomarkers and tools of assessment, particularly if we are going to measure benefits in individual patients. But this only brings into question the methods we use in our current clinical trials (if we can’t measure individuals accurately, how do we measure groups of individuals accurately?). In truth, however, I am inclined to agree on this matter in the case of Parkinson’s. But we have to start somewhere, right?
There are also other confounding issues regarding this type of study:
- Adherence: How does an investigator make sure an individual participant sticks to the treatment regime? If the patient hears about some amazing new herbal remedy (in some dark corner of the interweb) or reads a newspaper headline about some new re-purposed drug for Parkinson’s, what is to stop them from trying it in a desperate effort to slow the condition? Such an action would potentially mess up their trial results.
- Choosing an intermediate outcome measures: How and when do the investigators track individuals over time and decide to shift participants to or from a particular drug? This comes back to the issue of assessment mentioned above (Click here for a technical read on this topic)
- Lifestyle: How can investigators account for life style factors like exercise when assessing/measuring individuals?
You can probably tell that the patient filtering idea proposed here is not perfect in the real world setting of the research clinic.
And there are certainly aspects of it that are completely impractical.
I do not profess to know everything or to have any of the answers, and I certainly do not care if this post makes me the subject of mockery or ridicule with my research colleagues. But what I would like to see is constructive discussion (involving the Parkinson’s affected community) regarding the current state of our clinical trials. And critically, the sharing of some thoughts/ideas regarding innovation in this important area.
In particular, I would like to see a discussion that questions the dogma and “the way we’ve always done it” mindset that surrounds our current approach to clinical trials.
* * * * * * * * * * * * * * * * * *
Hopefully this post is food for thought for someone.
The biggest human temptation is to settle for too little – Thomas Merton
The banner for today’s post was sourced from Searchengineland