Genetically instrumented LDL‐cholesterol lowering and multiple disease outcomes: A Mendelian randomization phenome‐wide association study in the UK Biobank

Abstract

<div xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="section"> <a class="named-anchor" id="bcp15793-sec-0001"> <!-- named anchor --> </a> <h5 class="section-title" id="d5424635e279">Aims</h5> <p dir="auto" id="d5424635e281">Lipid‐lowering medications are widely used to control blood cholesterol levels and manage a range of cardiovascular and lipid disorders. We aimed to explore the possible associations between LDL lowering and multiple disease outcomes or biomarkers. </p> </div><div xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="section"> <a class="named-anchor" id="bcp15793-sec-0002"> <!-- named anchor --> </a> <h5 class="section-title" id="d5424635e284">Methods</h5> <p dir="auto" id="d5424635e286">We performed a Mendelian randomization phenome‐wide association study (MR‐PheWAS) in 337 475 UK Biobank participants to test for associations between four proposed LDL‐C‐lowering genetic risk scores ( <i>PCSK9</i>, <i>HMGCR</i>, <i>NPC1L1</i> and <i>LDLR</i>) and 1135 disease outcomes, with follow‐up MR analyses in 52 serum, urine, imaging and clinical biomarkers. We used inverse‐variance weighted MR in the main analyses and complementary MR methods (weighted median, weighted mode, MR‐Egger and MR‐PRESSO) as sensitivity analyses. We accounted for multiple testing with false discovery rate correction ( <i>P</i> &lt; 2.0 × 10 ⁻⁴ for phecodes, <i>P</i> &lt; 1.3 × 10 ⁻² for biomarkers). </p> </div><div xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="section"> <a class="named-anchor" id="bcp15793-sec-0003"> <!-- named anchor --> </a> <h5 class="section-title" id="d5424635e314">Results</h5> <p dir="auto" id="d5424635e316">We found evidence for an association between genetically instrumented LDL lowering and 10 distinct disease outcomes, suggesting potential causality. All genetic instruments were associated with hyperlipidaemias and cardiovascular diseases in the expected directions. Biomarker analyses supported an effect of LDL‐C lowering through <i>PCSK9</i> on lung function (FEV [beta per 1 mg/dL lower LDL‐C −1.49, 95% CI −2.21, −0.78]; FVC [−1.42, 95% CI −2.29, −0.54]) and through <i>HMGCR</i> on hippocampal volume (beta per 1 mg/dL lower LDL‐C 6.09, 95% CI 1.74, 10.44). </p> </div><div xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="section"> <a class="named-anchor" id="bcp15793-sec-0004"> <!-- named anchor --> </a> <h5 class="section-title" id="d5424635e325">Conclusions</h5> <p dir="auto" id="d5424635e327">We found genetic evidence to support both positive and negative effects of LDL‐C lowering through all four LDL‐C‐lowering pathways. Future studies should further explore the effects of LDL‐C lowering on lung function and changes in brain volume. </p> </div>