Summary | PCSK9 as target for treatment: The genetic validation

Prof. Brian Ference started his presentation by illustrating that PCSK9 is the prime example of how genomics can be used to guide drug discovery and development. Just 15 years ago, gain-of-function mutations in the PCSK9 gene were discovered to be associated with markedly elevated plasma LDL cholesterol (LDL-c) levels in the phenotype of familial hypercholesterolemia (FH) [1]. Just a few years later, rare loss-of-function mutations in the PCSK9 gene were shown to be associated with remarkably lower plasma LDL-c levels and lower risk of cardiovascular disease (CVD). Indeed, in African Americans rare loss-of-function mutations in the PCSK9 gene were associated with a lifetime exposure to lower LDL and 88% lower lifetime risk of CVD, as shown in the Atherosclerotic Risk and Community (ARIC) Study and in the Dallas Heart Study. In European individuals, more common partial loss-of-function mutations were associated with lifetime exposure to intermediate LDL-c levels and a more modest, but still substantial 50% reduction in lifetime risk of CVD [2]. These data established PCSK9 as a genetically validated therapeutic target in reducing lifetime risk of CVD.

In the same period, the mechanism by which PCSK9 inhibition lowers plasma LDL-c levels was elucidated. Normally, expression of both the LDL receptor (LDL-R) on liver cells and PCSK9 is upregulated in response to decreased intracellular concentrations of cholesterol. Circulating PCSK9 protein binds to the LDL-R and marks the LDL-LDL-R complex for degradation in the lysosome. PCSK9 inhibition prevents this degradation, thereby allowing the receptor to be re-used. The resulting increased density of LDL-R on the hepatocytes leads to lower plasma LDL-c levels, and thereby reduced risk of CVD [3-6].

These data gave rise to the development of a number of different therapies directed against PCSK9, with the monoclonal antibodies against PCSK9 being the first therapy in clinical development. Several studies demonstrated that treatment with a monoclonal antibody directed against PCSK9 consistently reduces plasma LDL-c levels by 50-60% in numerous different patient populations, regardless of the background therapy. A meta-analysis on the cardiovascular (CV) outcomes of these initial studies suggested that 50% reduction in LDL-c could potentially translate into 50% reduction in lifetime risk of CVD [7,8]. The suggestion that 50% reduction in plasma LDL-c could result in 50% reduction in CV risk creates perhaps an irrational exuberance about what we might expect therapeutically from inhibition of PCSK9, as compared with the effect of a rare loss-of-function mutation in PCSK9.

To more precisely anticipate what we might expect from PCSK9 inhibition, a Mendelian randomization study, framed as a naturally randomized trial, was conducted, using a number of common variants in the PCSK9 gene. The study demonstrated that lower LDL-c levels due to genetically lower PCSK9 levels was indeed causally associated with risk of CVD and that this effect appeared to be consistent across numerous different outcomes and composite outcomes. A dose-response between genetically reduced LDL-c due to PCSK9 polymorphisms and lower risk of CVD was observed [9].

The study also showed that the effect of lower LDL-c due to PCSK9 polymorphisms appeared to be similar to the effect of lower LDL-c due to other genetic variants associated with risk of CVD. When each of these variants is plotted, a log-linear association between the absolute magnitude of LDL-c reduction and the corresponding reduction in lifetime risk of CVD appears. This log-linear association is similar to the log-linear association observed in statin trials between the absolute reduction in LDL-c and the corresponding proportional reduction in the risk of CV events. A difference of course lies in that long-term exposure to lower LDL-c is associated with a greater reduction in risk of CV events per unit change in LDL-c, as compared to short-term exposure to lower LDL-c levels as achieved in randomized trials [10,11]. This implies that LDL has both a causal and a cumulative effect on risk of CVD. Because of this cumulative effect, one can therefore not use genetics or Mendelian randomization to directly estimate the effect of therapeutic PCSK9 inhibition in a trial.

To translate genetic effects into an expected therapeutic benefit in a short-term trial, one has to compare PCSK9 variants with a reference standard, which is the HMG-CoA reductase (HMGCR) gene coding for the target of statins. Lower LDL-c due to HMGCR gene variants that mimic the effect of statins are clearly associated with a lower risk of CVD, which is consistent across multiple different composite CV outcomes and in all subgroups studied, similar to treatment with a statin in randomized trials [10,12]. The effects of PCSK9 and HMGCR variants appear to be nearly identical on the risk of CVD per unit change in LDL-c. This implies that lower LDL-c due to PCSK9 inhibition and HMG-CoA reductase inhibition have biologically equivalent effects on the risk of CVD. It is therefore reasonable to anticipate that treatment with a PCSK9 inhibitor and statins should have therapeutically equivalent effects on the risk of CV events, implicating that roughly a 20% reduction in LDL-c per mmol/L, which is seen with statins, should be expected with a PCSK9 inhibitor, rather than a 50% reduction [9].

When examining the results of three major PCSK9 inhibition trials (FOURIER, ODYSSEY and SPIRE-2) and plotting the results on the Cholesterol Treatment Trialists (CTT) regression line, it appears that PCSK9 inhibition may result in a smaller reduction in the risk of CVD per unit change in LDL-c, as compared to statins [12,13]. However, it should be noted that the CTT regression line represents the average expected benefit after five years of treatment with a statin. Importantly, treatment with a statin only lowers LDL-c by about 10% during the first year and ~20% during each subsequent year of treatment [12]. The CTT regression line was calculated as a simple meta-analysis of effects of statins per mmol/L reduction in LDL-c each year of therapy. But we can now calculate a CCT regression line for any duration of therapy. Doing that shows that statins should lower CVD risk by ~10% after one year of therapy, ~15% after two years of therapy, ~18% after three years of therapy and only after four of five years an LDL-c reduction of 20-22% per mmol/L is seen. When plotting the effect of PCSK9 inhibition in the ODYSSEY, FOURIER and SPIRE-2 trials on CTT lines based on duration of therapy, the effect was nearly exactly what one would have anticipated from the observed absolute reduction in LDL-c and the duration of therapy. Combination of all PCSK9 inhibition trials, including >54,700 participants with >5,000 events, resulted in a 15% relative risk reduction after a median follow-up of 2.5 years of therapy. This is precisely what would have been anticipated from the same absolute magnitude of LDL-c reduction and the same duration of statin therapy. Indeed, PCSK9 inhibitors and statins appear to have equivalent effects on CV outcomes during each year of therapy. During the first year both PCSK9 inhibitors and statins reduced the risk of CVD by ~10% per mmol/L reduction and ~20% during the second year, demonstrating that PCSK9 inhibitors and statins do have therapeutically equivalent effects on the risk of CVD per unit change in LDL-c, precisely as anticipated by the genetic studies [14].

Ference repeated that PCSK9 is the prime example of how we can use genomics to guide drug discovery and development. Rare gain-of-function mutations are associated with markedly elevated plasma LDL-c levels, while loss-of-function mutations are associated with remarkably lower plasma LDL-c levels and correspondingly lower risk of CVD. Mendelian randomization studies suggested that lower LDL-c due to PCSK9 inhibition and HMG-CoA reductase inhibition have biologically equivalent effects on the risk of CVD. Randomization trial evidence suggests that PCSK9 inhibitors and statins have therapeutically equivalent effects on the risk of CVD.

References

Educational information

This is a summary of the presentation given by Brian Ference, during the PACE symposium entitled 'PCSK9 inhibition: Science, outcomes & guidance', held during ESC in Munich, Germany, on August 25, 2018.

View the lecture by prof. Ference View slides