I promised I’d write about the Tidelusib in PSP trial published last week, so here we are…
Reading and making sense of clinical trials can be a little tricky, so I’ll try and take you through it bit by bit. The headlines are:
Did Tideglusib work? No
Does that mean it didn’t work? Hard to be absolutely certain
Was it safe to take? Probably yes (although a lot of people didn’t complete the study)
Tideglusib is a drug designed to stop the build of of tau protein in the brain, the tau protein that builds up in the brains of people with Progressive Supranuclear Palsy (but also in Alzheimer’s disease, Corticobasal Degeneration and some forms of Frontotemporal Dementia). Tideglusib works by stopping an important step that allows proteins to form clumps, a step where a phosphate ion is added to the tau protein by an enzyme called GSK-3. The hope was that stopping this enzyme working would halt the build up of tau, thereby slow down the disease.
To test whether any drug works, it needs to go through different phases of testing. This study covered phase II “is it safe in humans?” and phase III “does it work?”. An impressive 146 people with PSP signed up to the study, all with ‘mild’ or ‘moderate’ disease. Each person with PSP was given either 600mg of Tideglusib, 800mg of Tideglusib or a sugar pill.
PAPER 1 – clinical measures
Unfortunately, Tideglusib did not improve any of the clinical measurements in people with PSP. In particular, it did not improve the main clinical measure of disease severity on the PSP rating scale. Let’s look at a few issues to see how confident we can be that this is truly a negative result.
Making sure you have enough people in a study is a challenge, too few and you may not have enough people to find a difference in scores between the treated and placebo groups. I take my hat off to these researchers for their efforts to get together 146 people with PSP willing to undergo the rigorous testing this study required. PSP is relatively rare, and this study was only achieved by combining efforts across countries and across continents. I’m sure this signals things to come, for example with studies run through the PSP research network in the UK.
Unfortunately people with PSP are often frail, tend to be older and therefore suffer with other diseases. We’re told some of the reasons why people with PSP pulled out of the study – not sticking to the schedule, abnormal blood tests, side-effects, death, and withdrawl of consent. A larger number withdrew from the study without explanation. So, when it came to the final count, only 64% of the 146 people with PSP who started were still taking the drug. That means that there may not have been enough people to tell whether there was a difference between those taking the drug and those taking the placebo. The researchers used some statistical wizardry to guess the missing values, but it’s not the same as having everyone finish the study.
The next thing to check when a study doesn’t show positive results is whether the measures were correct. In this study, the main measure was the PSP rating scale. It’s a very good scale that we use in the Cambridge clinic, and it captures a wide variety of the challenge faced by people with PSP. In previous studies, it seems to track disease severity pretty well. That doesn’t, though, make it a perfect test and it may just not have been sensitive enough to show a difference with this drug.
I would have chosen differently when picking some of the other measures used to capture specific aspects of the disease. For example, we’ve found that people with PSP struggle with the tests used for verbal fluency and the cognitive test in the Frontal Assessment Battery, but these tests don’t particularly change with time. So, it’s going to be hard to prove that a drug can shift these scores. Having said that, there are not many tests around so I may complain about the clinical test used, but I would be struggling to suggest sensible alternatives.
To summarise, the numbers of people in the trial are quite small. But the clinical measures were reasonable if not perfect. So, if the drug works at all, it’s too subtle a change to produce a noticeable effect on sensible disease measures.
PAPER 2 – imaging measures
37 of the people who took part in the study above went on to have MRI scans at the start of the study, and again after one year. The researchers looked at whether shrinkage in the brain can be slowed down with Tideglusib. This is a good measure to use for this study, since Progressive Supranuclear Palsy causes relatively rapid shrinkage in very specific areas in the brain, particularly the midbrain that sits at the bottom of the brain.
The whole brain seemed to shrink a little less with Tideglusib than without. But, there were regions of the brain that were seemed to shrink less than other regions, namely the parietal lobe (put your hands on the side of your head where they meet in the middle at your crown, and your fingers will lie over your parietal lobes) and the occipital lobe (put your hand on the back of your head and your palm will lie roughly over your occipital lobe).
And therein lies my confusion when reading the results of this study. The parietal lobe and the occipital lobes are not where we see the most marked changes under the microscope when we look at the brains of people with PSP at post-mortem. Shrinkage of the parietal and occipital lobes do not explain the key clinical features of PSP.
The researchers offer a plausible explanation for this. The drug targets the GSK-3 enzyme (the enzyme I mentioned above that adds a phosphate to the tau protein). There is more of this enzyme in the parietal and occipital regions. So, perhaps the drug worked, but just not in the places that PSP affects.
A second explanation is more technical, relating to the technique used to find shrinkage being more sensitive to changes in the parietal and occipital lobes.
The researchers also suggest they saw a whiff of a change in areas known to be affected by Progressive Supranuclear Palsy, but this remains no more than an unsubstantiated whiff and the statistics did not convincingly prove a change in any of those regions.
Unfortunately there were only 9 people in the placebo group, comparing with 28 in the drug groups. These unbalanced numbers can cause problems with the analysis, and I would really have liked to have seen more people having completed MRIs with more equal numbers across groups, but these are challenging demands in this sort of a trial.
We clinicians are a sceptical bunch. And I’m not yet convinced that Tideglusib is worth taking. However, it has done something that no other drug tested in dementia has done before – it showed a difference in brain shrinkage. If this can be explored further, then maybe, (and it’s only a maybe) we might see something develop from the Tideglusib story. But for the moment, I will not be advising anyone who comes to my clinic to get excited just yet.
Finally, I would really like to applaud all the Tauros researchers. This is a pretty well designed trial, with lots of people for such an uncommon disease. They have attempted to tackle a challenging disease head on. Together with the Davunetide trial, they’ve shown the way, so I look forward to writing about more similar clinical trials in the future (hopefully with more positive results).