LDL-C & heart disease – definitive proof? – Zoë Harcombe

Executive summary

* PCSK9s are relatively recent additions to the array of cholesterol lowering drugs. They are delivered by injection – fortnightly or monthly. One PCSK9 is called evolocumab.

* A 2023 paper reported on an update to the FOURIER trial (which was called FOURIER-OLE). FOURIER looked at cardiovascular outcomes in evolocumab vs placebo. FOURIER-OLE gave evolocumab to approximately 6,500 people from the FOURIER trial and then reviewed whether those in the lowest LDL-cholesterol group fared better than those whose LDL-C didn’t fall as much.

* FOURIER-OLE concluded that endpoints were better in the lower LDL-C group and adverse events were no worse. That’s the drug benefit proven, yes?

* Not necessarily. I reviewed several points as to why this may not be the case:

1) The characteristics of the patients who achieved the lowest LDL-C were very different to those who didn’t.

2) Some of the cardiac endpoints were subjective and these skewed the results in favour of those with lower LDL-C.

3) There were some serious concerns over deaths reported. Another academic research team fundamentally challenged deaths assigned to each group.

4) Serious adverse events were high across the study.

5) Lp(a) was strikingly different in different groups, and this could have made the difference.

6) The study had many other limitations.

7) The (drug company) conflicts of interest in the paper were vast.

Introduction

Professor Tim Noakes was sent an email by a fan of the diet-heart hypothesis. The email was brief. It asked, “Do you still resist the idea that LDL cholesterol is a very important factor in the generation of ASCVD?” (ASCVD stands for atherosclerotic cardiovascular disease). Attached to the email was a paper purporting to provide evidence for the question. The paper was called “Association between achieved Low-Density Lipoprotein cholesterol levels and long-term cardiovascular and safety outcomes: An analysis of FOURIER-OLE” and it was by Gaba et al (Ref 1).

Tim asked if I might like to take a look at the paper, so I did. It was about PCSK9 inhibitors. We have reviewed these previously (Ref 2).

PCSK9 inhibitors

PCSK9 stands for proprotein convertase subtilisin/kexin type 9. It is important to know the normal function of the PCSK9 gene and protein. The PCSK9 gene provides instructions for making a protein to help regulate the amount of cholesterol in the bloodstream. The PCSK9 protein controls the number of low-density lipoprotein (LDL) receptors, which are proteins on the surface of cells. These receptors play a vital role in regulating blood cholesterol levels. The receptors bind to particles called low-density lipoproteins (LDLs), which are the main carriers of cholesterol in the blood.

LDL receptors are particularly abundant in the liver, the organ responsible for removing most excess cholesterol from the body. The number of LDL receptors on the surface of liver cells determines how quickly cholesterol is removed from the bloodstream. The PCSK9 protein breaks down LDL receptors before they reach the cell surface, so that more cholesterol can remain in the bloodstream (Ref 3). That’s what PCSK9 is designed to do. PCSK9 inhibitors are designed to stop this, so that less cholesterol remains in the bloodstream. They are delivered by injection – every two weeks, or monthly.

The FOURIER Trial

FOURIER stands for “Further cardiovascular OUtcomes Research with PCSK9 Inhibition in subjects with Elevated Risk.” The seminal paper for the FOURIER trial was Sabatine et al, “Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease” (Ref 4). I reviewed that paper at the time (Ref 5). I found that it undermined the UK National Institute for Care and Health Excellence (NICE) guidelines. My review culminated in correspondence with NICE and an article in the UK national newspaper, The Telegraph (Ref 6).

The FOURIER trial originally recruited 27,564 patients. The patients were between 40 and 85 years of age; with cardiovascular disease; with a fasting LDL cholesterol (LDL-C) level of 70 mg/dL (1.8 mmol/l) or higher; and they were taking statins (at least atorvastatin at a dose of 20 mg daily or its equivalent). Any findings are thus only applicable to this population.

The trial randomised participants to receive either the PCSK9 (evolocumab) or a matching placebo injection. The main endpoint was the composite of cardiovascular death, myocardial infarction (heart attack), stroke, hospitalisation for unstable angina, or coronary revascularisation (having a stent fitted). The average (median) duration of follow-up was 2.2 years. That’s not long.

The study claimed that “Relative to placebo, evolocumab treatment significantly reduced the risk of the primary end point” (Ref 7).The study also claimed, “The results were consistent across key subgroups.” This was not true. I found key information in the Supplementary Appendix to disprove this (Ref 8). Evolocumab made no significance difference when baseline LDL-C was greater than 92 mg/dL (2.38 mmol/l).

My challenge to NICE was as follows. Given that the lowest LDL-C concentration at which NICE recommends the use of evolocumab is 3.5 mmol/l and given that evolocumab makes no difference when baseline LDL-C is greater than 2.38 mmol/l, there is no evidence of benefit for evolocumab in the circumstance for which NICE recommends it. Please can NICE confirm that NHS/taxpayer money will not be spent on this drug as a result. The NICE reply was unsatisfactory to say the least (Ref 9).

This study also raised concerns about death rates. Cardiovascular deaths and deaths from any cause were higher in the evolocumab intervention than the placebo. The risk ratios were not statistically significant, but the trial was stopped early, so we don’t know if these would have soon reached statistical significance and ended any prospects for this new drug.

FOURIER-OLE

FOURIER-OLE added the words “Open Label Extension” to FOURIER. In the FOURIER-OLE trial, 6,635 of the original 27,564 patients were given evolocumab, regardless of whether they were receiving evolocumab or placebo previously. These 6,635 people were then followed up for an average (median) of 5 years. (Data were available for 6,559 patients).

The purpose of the Gaba et al paper was to review the optimal level of LDL-C with regard to efficacy and safety. The first two LDL-C measurements taken during the trial were averaged. These were taken after 12 weeks and 24 weeks of receiving evolocumab. The patients were grouped according to this averaged LDL-C level. The cardiovascular incidents and safety outcomes were then compared across the groups.

That’s a simple study concept – treat everyone with the same drug (evolocumab). Group them by LDL-cholesterol level achieved. Review the incident levels per group to see which LDL-C level was ‘best’.

Efficacy in this study was measured as the composite of cardiovascular death, myocardial infarction, stroke, or hospital admission for unstable angina or coronary revascularisation. This was the same as for the original FOURIER trial. The secondary efficacy end point was measured as the composite of cardiovascular death, myocardial infarction, or stroke.

The 6,559 patients ended up in the following groups for LDL-C levels (I’ve put mg/dL and mmol/L for those using different measurements):

The results

Regarding efficacy, Gaba et al reported that the lower the LDL-C, the lower the risk of the endpoints of interest (cardiovascular death, myocardial infarction, stroke, or hospital admission for unstable angina or coronary revascularisation).

Regarding safety, Gaba et al reported that there were no significant differences between levels of LDL-C achieved and serious adverse events.

Better efficacy and no difference in safety, that’s a slam dunk, yes?

Not so fast…

Some concerns

1) The characteristics of the patients.

There were many significant differences between those who achieved the lowest level of LDL-C and those who remained at the highest level. Those who achieved the astonishingly low levels of LDL-C (below 20mg/dL) were more likely to be older, far more likely to be male, and far more likely to be white. They were less likely to be current smokers and less likely to have diabetes. Those differences would have been adjusted for. Some would have acted in favour of the lower group and some against. Adjustment is a black hole that we are unable to examine.

2) The endpoints.

In previous trials, which I have reviewed for cholesterol lowering medications (statins or PCSK9s), two of the parts of the composite end point can be subjective. These are hospital admission for unstable angina and coronary revascularisation. Admitting someone to hospital when they complain of a pain in the chest area and fitting a stent are decisions made between the physician and the patient. A patient is more likely to be admitted/operated on if the physician is more concerned about their condition. The physician will be more concerned about their condition if their LDL-C is higher. This inevitably means that the outcomes recorded for unstable angina and coronary revascularisation are higher in those with higher LDL-C.

This can be seen in the incidents table. Invariably, there are no significant differences in the objective incidents (death, heart attack and stroke) but there are significant differences in the subjective incidents. The subjective incidents have swayed the overall result in previous papers I have examined.

In this Gaba et al study, the annualised incident rate for the first event was more than double in the highest LDL-C group than the lowest for hospitalisation for unstable angina. The annualised incident rate was almost double in the highest LDL-C group than the lowest for coronary revascularisation. Strokes were not significantly different, but myocardial infarctions were. The overall (composite) endpoint was significantly different.

3) Deaths.

Table 2 reported annualised incident rates for the first event only. This showed that the LDL-C level achieved made no significant difference to coronary heart disease deaths or cardiovascular disease deaths. That was for the first event only.

The supplemental file reported all cardiovascular outcomes according to LDL-C levels achieved in FOURIER and FOURIER-OLE. This also showed that the LDL-C level achieved made no significant difference to coronary heart disease deaths or cardiovascular disease deaths.

Table 4 reported “all-cause mortality”, which showed a significant difference for those in the lowest LDL-C group compared to the highest. To the best of my memory, that is the first time I’ve seen a difference claimed for all-cause mortality in a PCSK9 trial.

Putting those findings together, it is being claimed that the lower the LDL-C achieved, the lower the death from any cause, but that coronary heart disease deaths and cardiovascular disease deaths were not lower. This means that non-CVD deaths must have been lower. Does that mean that PCSK9s don’t help the conditions that they are claimed to help – cardiovascular and coronary heart disease – but they do help something else? I don’t know the answer to that question.

A 2022 paper by Erviti et al raised serious questions about death records in the study (Ref 10). It found discrepancies between information on deaths in the clinical study report and in the 2017 trial results publication. The paper concluded, “After readjudication, deaths of cardiac origin were numerically higher in the evolocumab group than in the placebo group in the FOURIER trial, suggesting possible cardiac harm.”

4) Serious adverse events.

Table 4 in the Gaba et al paper presented safety outcomes. The annualised incident rate for serious adverse events was 12-13% across all groups. That’s one in eight people in the trial who suffered a serious adverse event, whatever LDL-C level they reached. And people unable to tolerate cholesterol lowering medications would have been removed from the trial long before the FOURIER-OLE stage.

5) Lp(a).

In the characteristics table, there was one measurement that stood out among all numbers. Lipoprotein(a) was reported. This is often referred to as Lp(a). In the lowest LDL-C group, this was 8 nmol/L. In the highest LDL-C group it was 40 nmol/L. That’s a five-fold difference.

This raises two questions – i) could Lp(a) rather than LDL-C be impacting the results? And ii) do PCSK9s affect Lp(a)? The answer to both questions is yes.

i) Dr Malcolm Kendrick has tried to understand the cause of heart disease for longer and more objectively than anyone I know. I recall his 2007 book, “The Great Cholesterol Con”, raising Lp(a) as something that probably does have something to do with heart disease. In his most recent book, “The Clot Thickens”, Malcolm summarises the important attributes of Lp(a):

– Humans have Lp(a) in the blood stream, in some cases at higher levels than LDL (although it is normally about a quarter of the LDL level).

– LDL and Lp(a) are identical in structure, other than the attachment of the protein Apo(a) to Lp(a).

– Lp(a) is designed to protect against the arterial damage caused by vitamin C deficiency (and other forms of arterial damage).

– Lp(a) is incorporated into blood clots that form on damaged artery walls.

– Lp(a) makes blood clots far more difficult to remove.

– Lp(a) can be found in high concentrations in atherosclerotic plaques.

– A raised Lp(a) level can, at least, triple the risk of cardiovascular disease.

It is clearly the case that the Lp(a) level could be impacting the results.

ii) It was not difficult to find evidence for the impact of PCSK9s on Lp(a). “Inhibitors of PCSK9 produce large reductions of LDL-C as monotherapy or when added to statins. An additional effect of PCSK9 inhibitors is to reduce lipoprotein(a) concentration by 20% to 25%” (Ref 11).

This study further reported “In patients with recent acute coronary syndromes and LDL-C near 70 mg/dL on optimized statin therapy, PCSK9 inhibition provides incremental clinical benefit only when lipoprotein(a) concentration is at least mildly elevated.” i.e., PCSK9s only help when Lp(a) was higher (and thus would be impacted by reduction).

6) Limitations.

There were a number of limitations of the study noted in the paper. First, LDL-C changes over time and the measurements in the first 12-24 weeks were used (despite this being a 5-year study). Second, imbalances in baseline characteristics may have confounded the results. Third, “Patients enrolled in FOURIER-OLE had to survive the parent study to then enroll in the open-label extension and thus may represent a lower-risk population compared with the original FOURIER population.” Finally, the FOURIER patients had stable CVD despite LDL-C levels of 70 mg/dL or higher, while being on cholesterol lowering regimes.

Summary

The question asked of Tim was “Do you still resist the idea that LDL cholesterol is a very important factor in the generation of ASCVD?”

Tim’s reply was “Absolutely. If blood cholesterol causes CHD then its removal must prevent the disease. Not reduce the incidence – REMOVE the disease. Dr Libby showed years ago that 70% of persons in statin trials develop CHD whilst taking the drug as part of the intervention arm of the trial. He called them the Forgotten Majority and blamed them for not taking their statins. What he really showed is that cholesterol cannot be the CAUSE of CHD. May be a REASON but is not the cause. Time to find the real cause.

Type 2 diabetes is the main cause of CHD. That is absolutely clear.

Treat metabolic syndrome not ‘cholesterol’.”

I would add the following points…

The trial offered as evidence was relevant to a specific group of people and does not apply beyond these.

In other papers I have reviewed, with many end points, the overall outcome has been swayed by two interventions (hospitalisation for angina and revascularisation), which are subjectively made in people with higher LDL-C. This paper was the same.

I don’t recall previous PCSK9 trials claiming a reduction in all-cause mortality. This one did, but there were no differences in CHD or CVD deaths, implying that lowering LDL-C was doing something to health, but not ASCVD. There was also an accuracy/integrity issue over the death numbers.

Lp(a) could have provided an explanation. The Lp(a) findings alone would allow Tim to challenge the question asked of him.

Just because events are lower when LDL-C is lower doesn’t mean that the mechanism is LDL-C. PCSK9s could lower both events and LDL-C and have a different mechanism of effectiveness.

It should also be noted that FOURIER and FOURIER-OLE were funded by Amgen – the maker of evolocumab. The conflicts of interest declared were long and pharma dominated. They are in the p.s. below.

p.s.
Sources of Funding
The study was funded by Amgen.

Disclosures
Dr Gaba reports no disclosures. Dr O’Donoghue has received grant funding through Brigham and Women’s Hospital from Amgen, Novartis, AstraZeneca, Janssen, Intarcia, and GlaxoSmithKline and consulting fees from Amgen, Novartis, AstraZeneca, and Janssen. Dr Park, J.F. Kuder, Dr Im, and S.A. Murphy are members of the TIMI (Thrombolysis in Myocardial Infarction) Study Group, which has received institutional research grant support through Brigham and Women’s Hospital from Abbott, Amgen, Anthos Therapeutics, AstraZeneca, Bayer Health-Care Pharmaceuticals, Inc, Daiichi-Sankyo, Eisai, Intarcia, MedImmune, Merck, Novartis, Pfizer, Quark Pharmaceuticals, Regeneron Pharmaceuticals, Inc, Roche, Siemens Healthcare Diagnostics, Inc, The Medicines Company, and Zora Biosciences. Dr Wiviott is a member of the TIMI Study Group, which has research grant or contract support from Amgen, AstraZeneca, Daiichi Sankyo, Eisai, Janssen, Merck, and Pfizer; consults for AstraZeneca, Boston Clinical Research Institute, Icon Clinical, and Novo Nordisk; and his spouse is an employee of Merck. Dr Atar reports speaker’s honoraria and consultancy fees from Amgen, AstraZeneca, Bayer, Boehringer-Ingelheim, Bristol Myers Squibb, Pfizer, Merck, Novartis, and Sanofi-Regeneron and research funding to his institution from Bristol Myers Squibb, Pfizer, Medtronic, and Roche Diagnostics. Dr De Ferrari is a steering committee member for Amgen, Novartis, and Livanova; consultant for Sanofi, Amarin, and UCB; and involved in trial participation with fees to the hospital for Amgen, Novartis, Merck, and Livanova. Dr Gaciong receives lectures and consultancy fees from Amgen, Behringer Ingelheim, Sanofi, and Bayer. Dr Toth receives research grants from Pfizer and speaker and consultancy fees from Amgen, AstraZeneca, Bayer, Berlin-Chemie, Boehringer Ingelheim, EGIS, KRKA, MSD, Novartis, Pfizer, Richter, Roche, Sandoz, Sanofi, Servier, Solvay, Takeda, TEVA, and Wörwag. Dr Gouni-Berthold receives research grants from Amgen, Akcea, Novartis, and Sanofi and speaker and consultancy fees from Amgen, Aegereon, Amarin, Akcea, Daiichi-Sankyo, Novartis, Pfizer, Regeneron, and Sanofi. Dr Lopez-Miranda is a speaker and receives consultancy fees from Amgen, Sanofi, Laboratorios Ferrer, Laboratorios Dr Esteve, Novartis, and MSD. Dr Schiele receives honoraria for lectures and consulting from Amgen, AstraZeneca, Bayer, Mylan, Novo Nordisk, Novartis, Pfizer, Recordati, Sanofi, and Servier. Dr Mach reports no personal conflicts of interest. The Cardiology Department at Geneva University Hospital has received unrestricted grants in the lipid field from Amgen, Daiichi-Sankyo, MSD, Novartis, and Sanofi. Drs Flores-Arredondo, López, Elliott-Davey, Wang, Monsalvo, and Abbasi are employees of Amgen Inc or own Amgen stock or stock options, or both. Dr Giugliano receives clinical trial and research support from Amgen, Anthos Therapeutics, Daiichi-Sankyo, and Ionis; honoraria for lectures or the continuing medical education program from Amgen, Centrix, Daiichi-Sankyo, Dr Reddy’s Laboratories, Medical Education Resources, Medscape, Menarini, Merck, Pfizer, SAJA Pharmaceuticals, Servier, Shanghai Medical Telescope, and Voxmedia; and is a consultant for Amarin, Amgen, Bayer, Boston Scientific, Caladrius, CryoLife, CSL Behring, CVS Caremark, Daiichi Sankyo, Esperion, Gilead, Hengrui, Inari, Janssen, Novartis, Paratek, Pfizer, PhaseBio Pharmaceuticals, and Samsung. Dr Sabatine has research grant support through Brigham and Women’s Hospital from Abbott, Amgen, Anthos Therapeutics, AstraZeneca, Daiichi-Sankyo, Eisai, Intarcia, Ionis, Medicines Company, MedImmune, Merck, Novartis, Pfizer; consults for Althera, Amgen, Anthos Therapeutics, AstraZeneca, Beren Therapeutics, Bristol Myers Squibb, Fibrogen, Intarcia, Merck, Moderna, Novo Nordisk, and Silence Therapeutics; and is a member of the TIMI Study Group, which has also received institutional research grant support through Brigham and Women’s Hospital from ARCA Biopharma, Inc, Janssen Research and Development, LLC, Siemens Healthcare Diagnostics, Inc, Softcell Medical Limited, Regeneron, Roche, and Zora Biosciences.

References
Ref 1: Gaba et al. Association between achieved Low-Density Lipoprotein cholesterol levels and long-term cardiovascular and safety outcomes: An analysis of FOURIER-OLE. Circulation. 2023. https://pubmed.ncbi.nlm.nih.gov/36779348/
Ref 2:https://www.dietandhealthtoday.com/?s=Pcsk9
Ref 3: https://medlineplus.gov/genetics/gene/pcsk9/
Ref 4: Sabatine et al. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. NEMJ. 2017.
Ref 5: https://www.zoeharcombe.com/2017/04/evolocumab-nice-guidelines/
Ref 6: https://www.telegraph.co.uk/news/2017/05/06/nhs-wasting-tens-thousands-year-wonder-drug-stroke-heart-attacks/
Ref 7: The results were “(1344 patients [9.8%] vs. 1563 patients [11.3%]; hazard ratio, 0.85; 95% confidence interval [CI], 0.79 to 0.92; P<0.001.)”
Ref 8: Supplementary Figure S5 – Efficacy in Key Subgroups – on page 55 of 57 https://tinyurl.com/aetwzv5w
Ref 9: https://www.zoeharcombe.com/2017/06/evolocumab-nice-guidelines-a-reply/
Ref 10: Erviti et al. Restoring mortality data in the FOURIER cardiovascular outcomes trial of evolocumab in patients with cardiovascular disease: a reanalysis based on regulatory data. BMJ Open. 2022. https://bmjopen.bmj.com/content/12/12/e060172
Ref 11: Schwartz et al. Lipoprotein(a) and Benefit of PCSK9 Inhibition in Patients With Nominally Controlled LDL Cholesterol. J Am Coll Cardiol. 2021. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8822604/