Q2. What were the Hot Topics at the ESC 2020 with regards to biomarkers for HF?
One of the fiercest studies now in cardiology, and in many disciplines, is the use of algorithms based on large data, artificial intelligence or machine learning. There are a lot of buzz words out there, but basically, it all boils down to the same principle; doctors, nurses, healthcare providers, in general, on a daily basis, collect a lot of data on an individual patient, and quite often nowadays, enter these data into electronic patient charts but not much is really being done with it.
What you see now is that technology buffs, but also doctors with an interest in technology, are trying to get these data out and form it in such a way that it teaches us what’s the phenotype a patient might suffer from, what the likelihood is that a certain patient might benefit from a certain drug, or may develop certain harmful side effects of certain drugs. I think biomarkers are probably very integral and important part of artificial intelligence algorithms because they are taken repeatedly, and they fluctuate with the severity of the disease. In other words, if your patient deteriorates, most likely one, two or three biomarkers will increase or decrease by whatever their behaviour is, and if a patient improves, the biomarker will bend back. So, I think using these patterns is very promising in finding and creating new algorithms and optimising drug regimen, the diagnostic algorithms and prognostic calculations for individual patients.
Q1d. Many HF trials have NT-proBNP as an endpoint and for treatment monitoring. What would be the role of biomarkers moving forward from clinical trials and clinical practice perspectives?
Most contemporary heart failure trials now have NT-proBNP as an entry criterion. In part, that is done because we know that an elevated NT-proBNP is clearly associated with worse prognosis, and that means if you design a trial in a way that patients have elevated NT-proBNP, as an investigator you are rather sure that there will be a sufficient number of endpoints, so it becomes easy to power a trial. Though that behaviour is understandable, to some extent, it is also a little bit complex because some people will say, “Well, if a trial used NT-proBNP as an inclusion criterion, then I will only use the drug if a patient has this certain level of NT-pro BNP.” However, I don’t agree with that. We all know that NT-proBNP fluctuates very strongly according to filling status and severity of disease and even day-to-day. So, there are patients with severe structural heart disease that at some point, may have a low NT-proBNP, but they would be still eligible for most of the heart failure drugs.
With regards to endpoints, there is very strong evidence that NT-proBNP is amongst the most straightforward and strongest surrogate endpoints of heart failure. If a drug decreases NT-proBNP, this is almost always translated into hard endpoints reduction. For guidelines and guideline committees and also regulatory authorities, there is still a gap between this knowledge and actual translation into guidelines. In other words, guideline committees and regulatory authorities still tend to value hard endpoints much harder, and basically, it is the only parameter based on which drugs can be recommended or can be registered for a certain disease such as heart failure. In smaller diseases such as paediatric heart failure or amyloidosis, that is if you have smaller subsets of the diseases, biomarkers may be acknowledged as bona fide surrogates for hard endpoints. But in the large subgroups or large sub-phenotypes such as HFrEF, HFpEF and post-myocardial infarction (MI) heart failure, I think both guideline committees and also regulatory authorities still demand hard endpoint trials to be conducted.
Q2a. What are your thoughts on PARALLAX study and the effects of ARNi in patients with HFpEF? What is the value of Biomarkers in HFpEF?
That was a very interesting additional trial being presented in one of the hotlines – the PARALLAX study. Last year at the ESC 2019, we saw PARAGON-HF which was a hard outcome trial with the ARNi sacubitril–valsartan compared to valsartan in patients with heart failure with preserved ejection fraction, and that trial narrowly missed the primary endpoints. The p-value with that was 0.06, but in reality, the investigators were just a few events short of a significant effect. So very promising, but at the same time also disappointing.
PARALLAX study was somewhat different, although the treatment allocation was comparable. So again, it was sacubitril–valsartan compared to RAAS blockade. It was not for valsartan exclusively, but it was either valsartan or enalapril, or other drugs. It was more like treatment as usual because although it was not evidence-based, patients with HFpEF, in reality, receive a very high percentage of RAAS blockade of up to 80% or so.
To the investigators, you can either give your patient whatever you would give him, and then the other half would receive sacubitril–valsartan. It was not a hard endpoint trial. The follow up was only 24 weeks, but the primary endpoint was NT-proBNP after 12 weeks, and there were a number of secondary endpoints that had to do with the quality of life and exercise performance – it was kind of a mixed trial.
The outcomes were not mixed; they were overall positive. The primary endpoint was a net decrease of NT-proBNP when patients received sacubitril–valsartan when compared to the usual care. There were some transient beneficial effects seen in the quality of life, so after 12 weeks, the patient felt better and after 24 weeks or so, this was kind of neutralised, and there were some exploratory endpoints – hard endpoints such as heart failure hospitalisation and cardiovascular death. The same endpoints were as seen in PARAGON, and again these were effectively and significantly reduced by sacubitril–valsartan but the numbers were small, so these were exploratory endpoints.
I think we probably have come to an end of testing sacubitril–valsartan in HFpEF. We have two big trials and we have to see how the heart failure community works it out. I think the problem is that the background therapy with RAAS blockers is pretty good, but in trials, it was always borderline significant with regards to outcome, so it never received a full recommendation, yet people use it, and now we have another treatment that again, is somewhat better. The real comparison, of course, would be sacubitril–valsartan versus placebo, because that’s the evidence-based therapy that we have for HFpEF. But I’m afraid that such a trial will never be launched again. So, this is what we have, and I think it is pretty convincing data with regards to several surrogate endpoints, that sacubitril–valsartan is superior to other agents.
The primary endpoint in PARALLAX trial was NT-proBNP reduction. NT-proBNP levels tend to not be too elevated in HFpEF, where nevertheless, they were importantly reduced. That is a strong signal that the disease severity, the rates of congestion and myocardial stretch was effectively reduced. Another question is, “where we are going with biomarkers in HFpEF?” I think heart failure with preserved ejection fraction is just coming of age. All the large trials that were either borderline significant or neutral have shown that this disease is really not one disease, but it’s a mixed bag – there are patients in there that have hypertensive heart disease, patients in there that have renal disease, there are patients in there maybe with concealed hypertrophic cardiomyopathy, or amyloids. Well, there may be all kinds of reasons and what we are learning now using biomarkers alongside with other genetic and clinical parameters is how to properly categorise heart failure with preserved ejection fraction into different diseases, and for each of these different diseases there may be a treatment, and I think the best example is amyloids. Five years ago, amyloid was still part of the heart failure with preserved ejection fraction spectrum, but now we all acknowledge that we should take it out because it’s a different disease and more importantly, there is a very effective therapy now for amyloid. I am convinced there are more sub-phenotypes in heart failure with preserved ejection fraction spectrum that we can take out and treat with targeted treatments, and at the end of the day, we probably are going to end up with a certain group that we really cannot categorise, that is really a mixed bag, where there’s no primary driver, and that group probably remains to be very difficult to diagnose and treat. But I’m pretty optimistic that there are other groups that we can tease out, that can be treated better, and I believe that biomarkers will play a very important role in that. We are blessed with a number of biomarkers – cardiomyocytes biomarkers, biomarkers of the extracellular matrix, biomarkers of comorbidities; I think that’s very important. It’s going to be mixing again – I believe using modern techniques, we will be able to tease out at least two, three, four sub phenotypes in the near future.
