LDL-c lowering PCSK9 gene variants associated with lower CV mortality in general population
Low LDL Cholesterol by PCSK9 Variation Reduces Cardiovascular MortalityLiterature - Benn M, Typbjaerg-Hansen A and Nordestgaard BG - J Am Coll Cardiol. 2019. 73(24). DOI: 10.1016/j.jacc.2019.03.517
Introduction and methods
Randomized trials evaluating LDL-lowering have consistently shown that this reduces the risk of CV disease proportional to the absolute reduction in LDL-c , irrespective of the method of LDL-c lowering . Reductions in CV mortality with statin therapy have only been demonstrated in PROSPER , and 20-year follow-up of WOSCOPS , and lower all-cause mortality in JUPITER  and 20-year follow-up of WOSCOPS . Inhibition of PCSK9 protein levels with monoclonal antibodies also lowers LDL-c levels and the risk of atherosclerotic CV events , but it is unknown whether such LDL-c reductions may translate into reductions in CV and all-cause mortality.
This study therefore tested the hypothesis that genetically low LDL-c due to variation in the PCSK9 gene is causally associated with low CV and all-cause mortality in a general population of Northern European ancestry. Data of 109,566 individuals of the Copenhagen General Population Study and the Copenhagen City Heart Study were examined to assess the association of PCSK9 genetic variation identified in populations of North European ancestry with LDL-c levels, as well as the association between the genetic variants and CV and all-cause mortality. Moreover, the association of 0.5 mmol/L (19.4 mg/dL) lower LDL-c and mortality endpoints was also assessed. The results were validated in individual genetic data on PCSK9 variants and information on all-cause and cause-specific death from 429,749 individuals from the UK Biobank , combined with effect sizes of the same genetic variants on LDL-c from the Global Lipids Genetics Consortium .
The following PCSK9 variants were selected: R46L (rs11591147; minor allele frequency [MAF]): 0.03), R237W (rs148195424; MAF: 0.001), I474V (rs562556; MAF: 0.32), and E670G (rs505151; MAF:
0.08), because they are robustly associated with PCSK9 and lower LDL-c concentrations in different populations of European ancestry. A weighted PCSK9 allele score was constructed by summation of number of alleles weighted by the allele effect on LDL-c and allele frequency.
- A higher number of weighted PCSK9 alleles were associated with stepwise lower LDL-c and apoB, but not with lp(a) and triglyceride concentrations. For instance, carrying 17.8 to 31.6 vs 0 to 2.6 weighted PCSK9 LDL-c lowering alleles was associated with 0.61 mmol/L lower LDL-c and 12.6 mg/dL lower apoB.
- The weighted PCSK9 allele score explained 0.6% of the variation in LDL-c (F-statistic: 159).
- Median follow-up was 10 years (range: 0 to 42 years). Compared with individuals with 0 to 2.6 weighted PCSK9 alleles, a higher number of weighted alleles were associated with stepwise lower LDL-c of up to 0.61 mmol/L and with lower CV mortality (P for trend: 0.001). Cancer mortality was not affected by the number of weighted alleles (P=0.68), nor was other mortality (P-0.21) or all-cause mortality (P=0.11).
- In the two Copenhagen studies, causal, genetic analyses suggested that an extrapolated 0.5 mmol/L lower LDL-c was associated with risk ratios of 0.79 (95%CI: 0.63 to 0.99, P=0.04) for CV mortality and 1.02 (95%CI: 0.91 to 1.12, P=0.63) for all-cause mortality. In data of the UK Biobank, the risk ratio for CV mortality was the same, but non-significant, and the one for all-cause mortality was similarly close to 1/no effect. When the studies were combined, CV mortality was reduced (0.79, 95%CI: 0.65-0.95, P=0.01).
- Risk of CV mortality as a function of lower LDL-c matched the known linear associations. A 0.5 mmol/L (19.4 mg/dL) lower LDL-c associated with an OR for CV mortality of 0.93 (95%CI: 0.88-0.98, P=0.01) in primary prevention statin trials, and an OR of 0.89 (95%CI: 0.71-1.03, P=0.39) in secondary prevention trials of PCSK9 inhibitors, and an extrapolated 0.79 (95%CI: 0.65-0.95, P=0.01) for genetic variation in PCSK9 in the Copenhagen studies and UK Biobank combined.
This study examined the causal effect of low LDL-c as a consequence of genetic variants in PCSK9 on mortality, and observed a lower risk of CV mortality in over 100,000 individuals from the Danish general population. Cancer mortality, other mortality and all-cause mortality were not significantly altered by the PCSK9 variants. A higher number of weighted PCSK9 alleles was associated with lower LDL-c and lower apoB levels, but lp(a) and triglycerides were not affected in these genetic studies. These data suggest that long-term LDL-c lowering treatment, for instance with PCSK9 inhibitors, may translate into reductions in CV mortality. But, RCTs with long-term follow-up should establish such effects.
Schwartz and Taylor note how uncommon it is in biomedical research to move from discovery of a target of therapy to an approved treatment within a dozen years, as happened with PCSK9. Identification of both loss-of-function and gain-of-function genetic variants that were associated with respectively low and high CV event rates, contributed to the understanding of the role of the protein and rapid development of PCSK9 inhibitors. It was, however, unclear how these variants affected death in the general population, because they are rare.
Placebo-controlled trials of monoclonal antibodies directed against PCSK9 have demonstrated reductions in composite primary endpoints that comprised fatal and nonfatal CV events, but none of the trials thus far have shown an effect of PCSK9 inhibition on CV death. In trials with relatively brief exposure to a treatment, death is too infrequent to serve as a reliable primary endpoint. Another limitation of such trials is that achieved lipid levels may be influenced and confounded by baseline characteristics or adherence.
Genetic analyses can avoid these limitations. The study by Benn et al. showed that ‘In the 5 categories of progressively increasing PCSK9 allele scores and decreasing LDL-C, cardiovascular mortality rates per 1,000 patient-years of observation were 3.8%, 3.1%, 3.0%, 2.7%, and 0.9%, respectively.’ These data provide insight into the effect of decades-long exposure to modestly lower levels of LDL-c. Schwartz and Taylor think the absence of a relation between the allele score and total mortality may be explained by the expectation that LDL-c levels have a primary impact on CV death, which was only 23% of all deaths in the Copenhagen cohort. Alternatively, lowering the risk of CV death may be accompanied by an increase in long-term competing risks of non-CV death.
They note that the slopes of the pharmacological and genetic relations of LDL-c lowering and CV mortality are not the same, as should be expected, due to differences in absolute LDL-c level, the magnitude of LDL-c reduction, exposure time and the prevalence of CVD in the studied cohorts. Moreover, statins also lower TG-rich lipoproteins, and PCSK9 inhibitors lower lp(a) levels, which may also affect CV mortality.
‘The modest effects on LDL-C for 3 lower allele score categories, comprising over 97% of all persons in the Copenhagen cohort, means that for most individuals these particular PSCK9 variants are of limited clinical importance.’ Nevertheless, this study adds to the evidence that points at long-term pharmacological inhibition of PCSK9 having a CV mortality-lowering effect in selected patients with elevated LDL-c, as well as reduce morbidity in primary and secondary prevention settings.