ANDREW HATTERSLEY: There has been an awful lot said about precision medicine and precision diabetes. But the key thing is that at the moment, there have been no clinical trials that have assessed whether such an approach works and whether it's any better than a one-size-fits-all prescribing. And so the TriMaster study, which was carried out in the UK with funding from the Medical Research Council, was really the first trial that has tried to assess directly in a trial as its primary endpoint, does precision medicine work?
So the rationale for our study was to try and improve the prescribing of treatment to lower blood glucose after metformin in type 2 diabetes. And our basic hypothesis was that we wanted to try and use clinical information to improve that choice of therapy for individual patients. The idea being that a precision medicine approach might be better than having the same tablet prescribed to all patients.
And so really what we are looking for was typical type 2 patients, either on metformin or metformin and a sulfonylurea, who required a third agent. And then we are offering them what at the time were the commonest drugs in the different classes. So that was a thiazolidinedione, pioglitazone, and DPP-4 inhibitors, sitagliptin, and in SGLT-2 inhibitors, canagliflozin. And these drugs were offered as a four-month trial of each one in random order in a double-blind trial.
What we were looking for was clinical characteristics that would be associated with a good response to one drug and a poor response to a second drug. The idea being that by separating people by that clinical characteristic, we could then match them to the right therapy. And there were two particular strata we used. The first one was BMI, and the idea of being that if you had a higher BMI, you responded better to pioglitazone. And if you had a low BMI, you responded better to sitagliptin.
The second strata that we looked at was renal function. And I would emphasize that this was EGFRs within the normal range, with the idea being that if you had a poorer renal functions, so an EGFR that was less than between 90 and 60, that those patients would respond better to sitagliptin. Well, if you had a higher EGFR, greater than 90, those patients would respond better to canagliflozin.
Stretching result from this study was that both primary endpoints were confirmed so that we established that, overall, there was no difference in how patients responded to these drugs, but there were differences when we looked in strata. So that meant that a stratified medicine approach resulted in a better drug selection for individual patients.
So therefore, we actually confirmed that when there was a BMI above 30, that these patients responded better to pioglitazone than to sitagliptin, and when there was a BMI less than 30, the patients responded better to sitagliptin than to pioglitazone. And similarly, when we look at the EGFR strata, that when the EGFR was greater than 90, the patients responded better to canagliflozin than to sitagliptin. And when it was between 90 and 60, these patients responded better to sitagliptin than to canagliflozin.
But it wasn't just the primary endpoints that were important. I think one of the striking things we found was that our patients, for the first time, were exposed to three different drugs in rapid succession. And they received these in a blinded manner. And that meant we could look at the side effects that they had, and we could really assess. And they, at the end of it, were able to choose their own preference.
So knowing what their weight was, what their HbA1c, and what the side effects they had with each drug, they were able to choose the drug that was best for them. And the striking thing was it was approximately a third chose each of the different drugs. So in other words, really fitting with the idea that they felt there was a drug that was best for them. And in a bit of exploratory analysis of this data, we found that patients' drug choice typically was the drug on which they had the lowest HbA1c and the least side effects. And so therefore it was that combination that really determine which drug a patient preferred, having tried all three.
One striking thing was that actually, it didn't appear to be weight, that the weight did not go with the choice of drug as the first preference, or, indeed, the weight gain seen. So overall, really we have established that precision medicine can improve glucose lowering. And we've also established that for an individual patient, there is a best drug, and it's most likely to be different drugs for different patients.
How did these results alter clinical practice? If your primary aim is to lower glucose, than we have been able to show that particular characteristics like BMI or EGFR should influence which drug will lower the glucose the most. Now, we accept that there are situations where you'll be choosing drugs for other reasons, such as secondary endpoints. But in many situations, those aren't -- aren't involved. And in those cases, this would be something that could help you decide which drug is likely to be the most successful in lowering glucose. And hence, it would be the longest time before that person needs a further prescription.
The other thing I would say is that we should all think about using what we did in the trials for patients. The patients found it incredibly helpful to try the three different drugs and felt very empowered by at the end of the trial, when they took that information to their primary care doctor, that then they would help be involved in the decision about which was the best drug for them.
And I think that idea of if we've got two drugs and we're not sure of them, we could give the patient a four-month trial on each of them and then at the end of that, ask them which one they would like to take. Certainly, it's not the case that people may or may not get side effects and may worry more about different side effects. And it's only by trying it that people really find out.