Summary | Novel PCSK9 outcomes in perspective: Lessons from FOURIER & ODYSSEY
To start her presentation, prof. Jennifer Robinson showed the classic CTT regression line. She noted that the ODYSSEY and FOURIER trial findings did not quite fall on the CTT regression line. This was surprising, because earlier trials with only 1-1.5 years of follow-up had shown 50% relative risk reduction. The explanation may lie in that little risk reduction was seen in the first year of FOURIER. In the second year this was more, showing 25% relative risk reduction of atherosclerotic CVD (ASCVD), which is lower than the anticipated 35% risk reduction based on 1.6 mmol/L decrease in LDL-c on the CTT regression line .
Robinson presented another CTT analysis of high versus low statin trials, showing an identical 28% reduction in the relative risk of CVD each year . This risk reduction emerged in the first year. Even if the trial had lasted longer, we would not have expected a 50-percent reduction in FOURIER.
When comparing the PCSK9 inhibitor trials, the ODYSSEY trial showed 15% relative risk reduction in major CV events after a median follow-up of 2.8 years , which is equivalent to the risk reduction observed in the FOURIER trial after a median follow-up of 2.2 years . In the SPIRE-2 trial another PCSK9 inhibitor, bococizumab, showed a much higher relative risk reduction of 21 % after a median follow-up of only one year . However, bococizumab did not move forward in development, because of allergenic potential due to mouse antigen.
The big difference in risk reduction can be explained by baseline LDL-c level, which was 134 mg/dL (3.5 mmol/L) in the SPIRE-2 trial versus 87 mg/dL (2.25 mmol/L) and 92 mg/dL (2.4 mmol/L) in the ODYSSEY and FOURIER trial, respectively. The importance of baseline LDL-c levels is demonstrated in the ODYSSEY trial with only benefit of all outcomes emerging in those patients whose LDL-c levels were over 100 mg/dL (2.6 mmol/L) at baseline [18,19]. Moreover, a meta-analysis demonstrated that baseline LDL-c drives all endpoints for CV risk reduction. There was no total mortality benefit with statins, PCSK9 inhibitors and ezetimibe when LDL-c levels <100 mg/dL (2.6 mmol/L) . With LDL-c levels >100 mg/dL (2.6 mmol/L) a ~10-14% reduction was seen in total mortality for each additional mmol reduction in LDL-c, indicating that the full effect of PCSK9 inhibitors is only observed at higher LDL-c levels. The reduction in total mortality was due to reduction in CV mortality.
Robinson showed a plot in which all studies came together, which demonstrates a clear log-linear relationship between LDL lowering and absolute CV reduction in patients with coronary heart disease and diabetes. This relationship is attenuated at lower LDL-c levels as seen in ODYSSEY and FOURIER [18,21,22]. Thus, the greatest benefit of PCSK9 antibodies can be obtained at higher LDL-c levels, as seen in high-risk populations.
Other evidence comes from the IVUS trial , in which lower achieved LDL-c reduction was associated with greater regression of plaque atheroma. Interestingly, the benefit seems to appear around LDL-c level of 100 mg/dL (2.6 mmol/L) [23-25]. The best value for money is probably when LDL-c levels are ≥100 mg/dL (2.6 mmol/L) and this may reflect the underlying pathophysiology. With very low LDL-c levels due to chronic statin therapy, it is more likely to observe regression in plaques, indicating stable plaques. However, patients with long-term statin therapy still suffer from events and the plaque does not disappear completely. More recent pathophysiologic evidence suggests that it is plaque erosion on these stable plaques that causes acute coronary syndromes. On the other hand, patients with LDL-c levels >100 mg/dL (2.6 mmol/L) show active plaques that are less likely to be stable, increasing the chance of large occlusive thrombi, STEMIs and even events like ASCVD that can be fatal. These results again show a big difference in response to LDL-c lowering therapies depending on baseline LDL-c levels.
Next, Robinson set out to translate these insights into clinical guidance. PCSK9 inhibitors are quite expensive and it is therefore not affordable to give those drugs to all patients with LDL-c levels >70 mg/dL (1.8 mmol/L). The benefit and net benefit from adding therapy can be quantified by calculating the number needed to treat to prevent one event (NNT) . If the absolute risk is high and the relative risk is reduced by ~22%, the absolute risk reduction is actually 1 divided by the NNT. Someone with lower risk but higher LDL-c levels and a greater relative risk reduction may have a very similar NNT to somebody whose risk is higher, even though LDL-c levels are lower.
Based on a systematic review and updated with subgroup information of the FOURIER trials, three different absolute ASCVD risk categories were identified: extremely high risk (CVD++; ≥45% 10-years ASCVD risk), very high risk (CVD+ risk factors/FH+ risk factors; 30-40% 10-years ASCVD risk) and high risk (CVD or FH, no risk factors; ~20% 10-years ASCVD risk [26,27].
A next step is to phenotype patients. Those with extremely high risk have a poorly controlled cardio-metabolic milieu and an extensive burden of atherosclerosis, and these patients will benefit greatly from further LDL-c lowering. The very high-risk patients still have an adverse cardio-metabolic milieu, with a less extensive burden of active plaque. Finally, high-risk patients have quite well-controlled risk factors and are maybe younger of age with single vessel coronary disease; this may represent primary prevention of FH .
When the phenotype of the patient is known, NNT can be estimated, showing cost-effectiveness at NNT of 20-25 with more acceptable cost-effectiveness at greater discounting [26,29].
To illustrate the selection of patients who benefit from expensive PCSK9 inhibition therapy, Robinson showed an extremely high-risk CVD FH patient with LDL-c levels of 100 mg/dL (2.6 mmol/L). Treatment with ezetimibe will likely give a ~20% reduction in LDL-c levels. The NNT is 39 and LDL-c levels may go down to ~70-80 mg/dL (1.8-2.1 mmol/L). It makes sense to add a PCSK9 inhibitor, because the NNTs are still quite reasonable. On the other hand, a PCSK9 inhibitor could have been added directly, because the NNT is in the cost-effectiveness range. When LDL-c levels are >100 mg/dL (2.6 mmol/L), the NNTs for adding ezetimibe are not very good, whereas PCSK9 inhibitors are still in the range of cost-effectiveness with these LDL-c levels .
In conclusion, prof. Robinson suggests to add LDL thresholds to various risk phenotypes, at which it is reasonable to add a PCSK9 inhibitor. For instance, treatment starts to approach cost-effectiveness when LDL-c levels >70 mg/dL (1.8 mmol/L) in extremely high-risk patients and >100 mg/dL (2.6 mmol/L) in very high-risk patients. In the high-risk patients with LDL-c levels ≥130 mg/dL (3.4 mmol/L), particularly primary prevention of FH is cost-effective. Obviously, PCSK9 inhibitors would likely be used more widely if they were much less expensive, but the model based on NNT can be used as a rule of thumb to select the right patients for use of these monoclonal antibodies.
This is a summary of the presentation given by Jennifer Robinson, during the PACE symposium entitled 'PCSK9 inhibition: Science, outcomes & guidance', held during ESC in Munich, Germany, on August 25, 2018.