Skip to main content
Download PDF
  • Research article
  • Open access
  • Published:

Assessing the safety of lipid-modifying medications among Chinese adolescents: a drug-target Mendelian randomization study

BMC Medicine volume 21, Article number: 410 (2023) Cite this article

Abstract

Background

With increasing hypercholesterolemia prevalence in East Asian adolescents, pharmacologic interventions (e.g., HMGCR inhibitors (statins) and PCSK9 inhibitors) may have to be considered although their longer-term safety in the general adolescent population is unclear. This study aims to investigate the longer-term safety of HMGCR inhibitors and PCSK9 inhibitors among East Asian adolescents using genetics.

Methods

A drug-target Mendelian randomization study leveraging the Global Lipid Genetics Consortium (East Asian, n = 146,492) and individual-level data from Chinese participants in the Biobank clinical follow-up of Hong Kong’s “Children of 1997” birth cohort (n = 3443, aged ~ 17.6 years). Safety outcomes (n = 100) included anthropometric and hematological traits, renal, liver, lung function, and other nuclear magnetic resonance metabolomics. Positive control outcomes were cholesterol markers from the “Children of 1997” birth cohort and coronary artery disease from Biobank Japan.

Results

Genetic inhibition of HMGCR and PCSK9 were associated with reduction in cholesterol-related NMR metabolomics, e.g., apolipoprotein B (HMGCR: beta [95% CI], − 1.06 [− 1.52 to − 0.60]; PCSK9: − 0.93 [− 1.56 to − 0.31]) and had the expected effect on the positive control outcomes. After correcting for multiple comparisons (p-value < 0.006), genetic inhibition of HMGCR was associated with lower linoleic acid − 0.79 [− 1.25 to − 0.35]. Genetic inhibition of PCSK9 was not associated with the safety outcomes assessed.

Conclusions

Statins and PCSK9 inhibitors in East Asian adolescents appeared to be safe based on the outcomes concerned. Larger studies were warranted to verify these findings. This study serves as a proof of principle study to inform the medication safety among adolescents via genetics.

Peer Review reports

Background

Three-hydroxy-3-methylglutaryl coenzyme A (HMGCR) reductase inhibitors, collectively referred to as statins, are the first-line pharmacologic treatments for children with familial hypercholesterolemia (FH), which demonstrated a reduction in cardiovascular risk in adulthood [1]. However, with increasing rates of obesity in the adolescent populations, in particular East Asians [2], prescriptions of lipid-modifying medications may have to be considered for non-FH adolescents who have elevated, uncontrolled lipid profiles to mitigate the subsequent risk of atherosclerotic cardiovascular disease. Unfortunately, a recent review commented a lack of safety studies on these medications in adolescents [3]. As such, it would be of public health interest to evaluate any possible side effects of the use of these medications in the general adolescent population [4].

Real-world data, such as pharmaco-epidemiologic studies of medication side effects, are susceptible to confounding by indication or immortal time bias [5]. The use of drug-target Mendelian randomization studies has gained popularity in recent years, although only a few studies have been conducted to study the side effects of medications in European adult populations, such as anti-hypercholesteremia and antihypertensives [6,7,8,9]. The extent to which these side effects can be extrapolated to the adolescent population remains unclear. To better evaluate possible safety issues associated with statin use in Chinese adolescents, who constitute a significant proportion of the global population, we evaluated the association of genetically proxied inhibition of HMGCR with a wide range of phenotypes (n = 100) in an originally population-representative birth cohort in Hong Kong. For completeness, we also considered genetic inhibition of proprotein convertase subtilisin/kexin type 9 (PCSK9) and Niemann-Pick C1-Like 1 (NPC1L1) as these are emerging classes of medications for the treatment of hypercholesteremia in adolescents [10].

Methods

Study design

This is a drug-target Mendelian randomization study using individual and summary data, which relies on three instrumental variable assumptions. First to satisfy relevance, genetic variants within the protein-encoding genes (HMGCR, PCSK9, and NPC1L1) and associated with LDL-C were used to proxy the use of statins, PCSK9 inhibitors, and ezetimibe respectively. Second, the use of genetic variants not associated with confounders of exposure-outcome association addresses independence. Third, genetic variants are independent of the outcome given the exposure thereby addressing exclusion restriction [11]. The schematic diagram of the study design is depicted in Fig. 1. This study is reported according to Strengthening the reporting of observational studies in epidemiology using Mendelian randomization (STROBE-MR) statement (https://www.strobemr.org/).

Fig. 1
figure 1

Schematic diagram of drug target Mendelian randomization study design

Full size image

“Children of 1997” birth cohort

The Hong Kong “Children of 1997” birth cohort is a population-representative Chinese birth cohort, which recruited 88% of all ethnic Chinese births (n = 8327) in Hong Kong in April and May 1997 [12]. Recruitment was conducted at all the Maternal and Children Health Centres in Hong Kong at their first postnatal visit for free preventive care and immunizations. Infant and family characteristics were ascertained using a self-administered questionnaire [13]. In 2005, record linkage was established to obtain routine information and clinical measurements (96% successful matching, n = 7999). During the Biobank clinical follow-up phase 1 in 2013–2016, at 17.5 years (n = 3460), with a supplementary Biobank clinical follow-up phase 2 in 2017 (n = 158), participants provided biospecimen (blood, saliva, urine, stool, hair, and nails) and completed comprehensive measurements (e.g., anthropometrics and lung function). Fasting blood samples and their derivatives (e.g., buffy coat and plasma) were used for biochemical assays such as liver function tests and metabolomic profiling using high-throughput nuclear magnetic resonance (NMR) spectroscopy metabolomics (Nightingale Health Ltd., Helsinki, Finland, https://nightingalehealth.com/) [14], which assessed routine lipids, 14 lipoprotein subclasses, esterified fatty acid composition, and a broad range of low-molecular-weight metabolites, such as amino acids, ketone bodies, and glycolysis and gluconeogenesis-related metabolites [14, 15].

DNA extraction from either blood, buffy coat, or saliva samples, where appropriate, and genotyping were performed for 3582 participants by the Centre for PanorOmic Sciences, The University of Hong Kong, with Infinium Asian Screening Array BeadChip Kit (v1.0). Phasing was performed using Eagle v2.4, and genotyping imputation was performed with 1000 Genomes phase 3 (Version 5, GRCh37/hg19) in the Michigan Imputation Server using Minimac4 software (https://imputationserver.sph.umich.edu/index.html#!). For quality control, samples with call rate < 0.98, recorded sex mismatched with genetic-inferred sex, second degree of relatedness or above, high heterozygosity (deviation > 3 standard deviation (SD)), or variants with call rate < 0.98, and imputation score < 0.3 were excluded.

Instrument selection

We identified genetic instruments to proxy the effects of each lipid-modifying target using the Global Lipids Genetics Consortium (GLGC), the latest genome-wide association study (GWAS), which included 146,492 middle-aged (mean aged of 55.8 years) participants of East Asian (EAS) ancestry [16]. LDL-C was measured or calculated using the Friedewald equation from either fasting or non-fasting blood samples. LDL-C was corrected by dividing the LDL-C by 0.7 if a person was taking lipid-modifying medications [16]. To reduce the likelihood of false positives, we only considered genetic variants located within 100 kilobase (kb) pairs on either side of the protein-encoding genes (HMGCR, PCSK9, and NPC1L1), and those associated with LDL-C at genome-wide significance (p-value < 5 × 10−8) in GLGC were considered. None of the genetic variants for NPC1L1 passed through the selection process and hence was not considered further. A recent study found that genetic variants of LDL-C identified in middle-aged adults of EAS ancestry are highly predictive of LDL-C measurement at different stages of the life-course including adolescents [17]. Nevertheless, we cross-verified the variant-LDL-C associations in the “Children of 1997” birth cohort, adjusted for age at sample collection, sex, and the top 6 principal components (PC) of ancestry. We selected variants associated with LDL-C in two independent cohorts (GLGC and “Children of 1997” birth cohort, p-value < 0.05) and in low linkage disequilibrium (LD) (clumping r2 < 0.3, clumping window of 10,000 kb) using a reference panel of EAS ancestry from 1000 Genomes Project (phase 3). To assess the robustness of the results, we also used a single index variant (the smallest p-value).

Genetic risk scores for HMGCR and PCSK9 in the “Children of 1997” birth cohort

To maximize statistical power, [18] an externally weighted genetic risk score (GRS) for HMGCR and PCSK9 was constructed for each participant in the “Children of 1997” birth cohort. Specifically, the GRS for HMGCR and PCSK9 were constructed by summing the number of all LDL-C lowering alleles for each variant in the HMGCR and PCSK9 gene regions, weighted by the effect of that variant on LDL-C in EAS ancestry reported by GLGC [16].

Positive control outcomes

To verify the validity of the GRS used to proxy the effect of HMGCR and PCSK9 inhibition, we considered cholesterol-related metrics and coronary artery disease (CAD) as the positive control outcomes, as lowering cholesterol levels and reducing CAD risk are well-established effects of HMGCR and PCSK9 inhibition [19]. We used cholesterol-related metrics from the NMR metabolomic panel of the “Children of 1997” birth cohort, including cholesterol, triglycerides, apolipoproteins, phospholipids, cholesteryl esters, free cholesterol, total lipids, other lipids, lipoprotein particle sizes, lipoprotein particle concentrations, lipoprotein subclasses, and relative lipoprotein lipid concentrations at 17.6 years which were used as efficacy outcomes (Additional file 1: Table S1). Summary genetic associations with CAD (29,319 cases and 183,134 controls) were obtained from Biobank Japan in middle-aged people of EAS ancestry [20].

Safety outcomes

To systematically evaluate the safety of genetic inhibition of HMGCR and PCSK9, we considered extensive health outcomes assessed at ~ 17.6 years during the Biobank clinical follow-up. We included 20 hematologic traits, 12 anthropometric traits, 18 body composition traits, renal function (urea), liver function (alkaline phosphatase (ALP), alanine transaminase (ALT), bilirubin and total protein), and lung function (forced expiratory volume in the first second (FEV1), forced vital capacity (FVC), FEV1/FVC, peak expiratory flow (PEF), the mid-forced expiratory flow at 25–75% of the pulmonary volume (FEF 25–75%)). We also included other metrics from NMR metabolomic panel, i.e., amino acids, fatty acids, fluid balance, glycolysis-related metabolites, inflammation, and ketone bodies as safety outcomes given some of these metabolites (e.g., branched chain amino acid) may be related to cardiometabolic diseases [21, 22] (Additional file 1: Table S2).

Statistical analysis

Instrument strength was assessed using approximated F statistics \(\left(\frac{\mathrm{Beta}^2}{\mathrm{Standarderror}^2}\right)\), where a higher F statistic suggests weak instrument bias is less likely. Each continuous outcome was standardized (mean 0, SD 1) to facilitate comparison. With individual level “Children of 1997” birth cohort data, the associations of the HMGCR and PCSK9 GRS with these outcomes were obtained using a multivariable linear regression model adjusted for age at sample collection, sex, and the top 6 PCs of ancestry. The variance explained by the HMGCR and PCSK9 GRS were assessed as the proportion of variance remaining after covariates fitted in a linear model [23]. The effects of genetic inhibition of HMGCR and PCSK9 on the outcomes were assessed using the Wald ratio \(\left(\frac{\mathrm{GRSonoutcome}}{\mathrm{GRSonLDL}-\mathrm C}\right)\). Effect sizes and 95% confidence intervals (95% CI) were presented per SD change in the outcome per SD decrease LDL-C induced by genetic inhibition of HMGCR and PCSK9. A combination of multiple imputation and inverse probability of participation/attribution weighting was used to account for potential attrition [24]. Briefly, multiple imputation based on a flexible additive regression model with predictive mean matching was used to impute the missing data for variables related to response [25]. Inverse probability weights were then obtained using a response model so as to minimize the probability of selection bias from attrition. Estimates from the analysis model were pooled from 50 computed datasets using Rubin’s rules [24]. A complete case analysis was used as the sensitivity analysis.

With summarized data from Biobank Japan, the effects of genetic inhibition of HMGCR and PCSK9 on the positive control outcome CAD were assessed using weighted generalized linear regression, i.e., a modified version of inverse-variance weighted (IVW) method that accounts for pairwise correlation between variants [26]. We obtained the correlation matrix between variants in participants of EAS ancestry using the 1000 Genomes Project (phase 3) and assessed heterogeneity from Cochran’s Q statistic [27].

To correct for the multiple testing of these biologically correlated outcomes, we used PC analysis to obtain the number of PC that explained 99% of the variance in the cholesterol related and safety outcomes. The p value threshold after correction for multiple testing was set as 6 × 10−4 (α = 0.05/75). Post hoc power calculation using type I error α of 0.05 and power of 0.8 was performed using an online tool (https://shiny.cnsgenomics.com/mRnd/). Data analyses and visualization were obtained using the R packages “ggforestplot” and “EpiViz,” R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Study population

There were 3443 participants (1760 males and 1683 females) with valid genotype information and outcomes. Characteristics of these participants by tertile of HMGCR score and PCSK9 score are shown in Table 1. Compared with participants with the first tertile of HMGCR score and PCSK9 score, participants in the 3rd tertile had higher lipids.

Table 1 Characteristics of participants in “Children of 1997” birth cohort Biobank follow-up
Full size table

The validity of genetic inhibition of HMGCR and PCSK9

In total, 5 variants (rs111913932, rs6453131, rs191835914, rs13356670, and rs12173076) were selected for targeting HMGCR and 3 variants (rs12739979, rs199717562, and rs2094470) for PCSK9. Their associations with LDL-C in GLGC and “Children of 1997” are summarized in Additional file 1: Table S3. There was little evidence of weak instrumental bias as the F-statistics were greater than 10 (HMGCR: 38 to 266 and PCSK9: 34 to 42). HMGCR and PCSK9 GRS explained 0.54% and 0.27% of LDL-C, and thus, our study was adequately powered to detect an odds ratio of 0.77 and 0.67 per SD reduction in LDL-C for CAD and a beta coefficient of ± 0.46 and ± 0.60 per SD reduction in LDL-C for the continuous outcomes. Genetic inhibition of each lipid-modifying target was associated with lower CAD risk (HMGCR: 0.66 per 1 SD reduction in LDL-C, 95% CI 0.54 to 0.81 and PCSK9: 0.55, 95% CI 0.34 to 0.90). Suggestive evidence of heterogeneity was found for the effects of genetic inhibition of HMGCR on CAD, which may be due to the missense variant of rs19183514 (Additional file 1: Tables S3-S4).

The effect of genetic inhibition of HMGCR and PCSK9 on efficacy outcomes

Genetic inhibition of HMGCR was associated with substantial reduction in several measures of cholesterol (total cholesterol, non-high density lipoprotein cholesterol (non-HDL-C), remnant cholesterol, very low lipid density lipoprotein (VLDL-C), clinical LDL-C and LDL-C) and apolipoproteins (ApoB and the ratio of apolipoprotein B to apolipoprotein A1 (ApoB/ApoA1)); p-value ≤ 6 × 10−4 (Fig. 2 and Additional file 1: Table S5). In addition, genetic inhibition of HMGCR was also associated with lower lipoprotein subclasses, including cholesterol (-C), cholesterol esters (-CE), free cholesterol (-FC), total lipids (-L), particle concentration (-P), and phospholipids (-PL) in intermediate-density lipoprotein (IDL), large LDL (L-LDL), median LDL (M-LDL), medium VLDL (M-VLDL), small LDL (S-LDL), small VLDL (S-VLDL), and very small VLDL (XS-VLDL). The metabolic profiling of genetic inhibition of PCSK9 was similar to genetic inhibition of HMGCR, for example, genetic inhibition of HMGCR and PCSK9 led to reduction of ApoB (HMGCR beta: − 1.06 SD, − 1.52 to − 0.60, p-value = 5 × 10−6 and PCSK9 beta: − 0.93 SD, − 1.56 to − 0.31, p-value = 3 × 10−3) (Fig. 2 and Additional file 1: Table S5). Similar association of genetic inhibition of HMGCR and PCSK9 and efficacy outcomes were found in sensitivity analyses using complete case and index variant analyses (Additional file 1: Table S6-7).

Fig. 2
figure 2

Genetic inhibition of HMGCR and PCSK9 (per lower LDL-C in SD) on efficacy outcomes in “Children of 1997” birth cohort Biobank follow-up

Full size image

The effect of genetic inhibition of HMGCR and PCSK9 on safety outcomes

Genetic inhibition of HMGCR was associated with lower linoleic acid (LA) (beta: − 0.79 SD, 95% CI − 1.25 to − 0.34, p-value = 6 × 10−4). In addition, genetic inhibition of HMGCR was nominally associated with lower omega-6, polyunsaturated, total, saturated, and monounsaturated fatty acids, pyruvate, FEV1, FVC, head circumference, height, and higher glycerol (Fig. 3 and Additional file 1: Table S8). There was no association of genetic inhibition of HMGCR with other fatty acids or their ratios, amino acids, aromatic amino acids, branched-chain amino acids, fluid balance, glycolysis-related metabolites, inflammation, ketone bodies, anthropometric traits, body composition, hematological traits, markers of liver function, lung function, or renal function. Similarly, no association of genetic inhibition of PCSK9 with these safety outcomes was observed (Fig. 3 and Additional file 1: Table S8). Sensitivity analyses from complete case analyses and index variant analyses gave similar conclusions (Additional file 1: Table S9-10).

Fig. 3
figure 3

Genetic inhibition of HMGCR and PCSK9 (per lower LDL-C in SD) on safety outcomes in “Children of 1997” birth cohort Biobank follow-up

Full size image

Discussion

This is one of the first drug-target Mendelian randomization studies to assess the efficacy and safety of HMGCR inhibitors and PCSK9 inhibitors among Chinese adolescents. We found metabolic profiling was similar for genetic mimics of HMGCR inhibitors and PCSK9 inhibitors, and their efficacy in lowering apolipoproteins and cholesterol were similar to previous RCTs and Mendelian randomization studies in European and EAS adults [28,29,30,31]. A systematic review of RCTs showed relatively short-term safety of statins (no more than 48- week) and PCSK9 inhibitors (24-week) in pediatric patients with FH [32, 33]. Our study extends by providing genetic evidence that longer-term on-target effect of statins and PCSK9 inhibitors did not appear to have clear adverse effects on a wide range of health outcomes in the general Chinese adolescents.

Selecting the appropriate genetic instruments to proxy effects of medication has been challenging and not as standardized as compared to conventional Mendelian randomization studies [34]. Nevertheless, our findings that both genetic inhibition of HMGCR and PCSK9 was associated with lower risk of CAD is consistent with Mendelian randomization studies in EAS and Europeans [19, 31, 35] and with RCTs [36], and so provides some validation. Consistent with earlier RCTs in children and adolescents with FH, [37,38,39] we did not find an association of genetic inhibition of HMGCR with liver function, possibly because of lack of statistical power. The infrequent reported statin-associated asymptomatic increase in transaminases (> 3 times the upper limit of normal) and hepatotoxicity reported in trials are likely due to idiosyncratic or immune allergic reactions [40]. However, in 2012, the US Food and Drug Administration recommended removal of routine periodic monitoring of liver enzymes among statins users [41].

Majority of the trials on safety among adolescents were restricted to those with FH. However, with increasing rates of sedentary behaviors and obesity, it is likely that statins and/or PCSK9 inhibitors would be potential pharmaceutical interventions for the overall pediatric population since the safety profiles of these interventions are not well documented in long-term studies [42]. Using genetics, this study did not provide strong evidence for possible adverse effects regarding the use of these medications. More importantly, this study provides a proof of concept in using genetics to infer possible side effects of medications in adolescents where large-scale trials with long duration are more difficult to conduct [43]. With increasing global trends in other cardiometabolic risk factors such as hypertension and hyperglycemia among adolescents [44, 45], drug-target Mendelian randomization leverages genetics, and appropriate surrogate and safety endpoints in adolescents provide a potentially more cost-effective approach to identify adverse effects prior to the conduct of observational studies and RCTs. Establishing possible safety profile can also help inform corresponding clinical management [6,7,8,9].

This drug-target Mendelian randomization study has some limitations. First, the study used genetic variants to mimic the on-target effect of life-long (including developmental) exposure to a relatively modest LDL-C reduction induced by genetic inhibition of HMGCR and PCSK9 among Chinese adolescents. This may not be equivalent to the effects of pharmacological interventions administered at specific time points, higher dosages, for shorter periods, or to individuals with particular indications such as FH [46]. Second, we cannot assert that the observed associations in this study are solely attributable to the LDL-C reduction properties of the medication. For example, previous studies have suggested that statins can have pleiotropic effects related to cardiovascular outcomes [47, 48]. A head-to-head comparison with a conventional multivariable Mendelian randomization study of LDL-C, while adjusting for other lipid traits, could help address these research gaps. Third, although we have included rare to common variants to mimic HMGCR and PCSK9 inhibition, a recent study suggests that the inclusion of rare and loss-of-function (LoF) variant may provide further insights into the efficacy and safety of therapeutic targets [49]. However, LoF variants tend to maintain at very low frequencies in populations, [50] and studies exploring the health effects of LoF variants typically require extensive biobanks, such as the UK Biobank. Fourth, genetic variants for genetic inhibition of HMGCR and PCSK9 were selected and validated in people of EAS descent only. While there is a minor imbalance in age distribution across tertiles of HMGCR and PCSK9 scores, age was adjusted for in the study, making it unlikely to explain our findings. We were also precluded from using colocalization to assess the suitability of the chosen instruments, because of the lack of relevant expression or protein quantitative trait loci data in EAS. Nevertheless, the variants used for statins and PCSK9 inhibitors are quite highly correlated with those previously used in Europeans and EAS [31]. The diversity in human genetics highlights the need for ethnicity-specific GWAS and corresponding Mendelian randomization studies, with appropriate controls for population stratification, to generate contextually specific evidence [51]. It would be valuable to replicate this study in the adolescents of other ethnicities. Fifth, constrained by sample size and hence statistical power, we cannot rule out the possibility of false negatives. Similarly, due to statistical power issues, we were unable to consider dichotomized outcomes (e.g., hypertension and prediabetes) or sex-specific associations [52]. Furthermore, statistical power may reduce the capability to detect false negatives when accounting for multiple corrections. In this context, results should be interpreted with caution, as any nominally significant associations (e.g., lung function and height) or associations with large effect sizes (e.g., acetate) may suggest potential safety concerns and require additional replications in larger studies [53]. Further research to validate specific side effects could facilitate a more targeted approach that circumvents the challenges of multiple testing when corresponding birth cohort studies are not extensive. Sixth, although we have included a wide range of health outcomes (n = 100), such as metabolomics, we did not cover all possible side effects, including muscle symptoms and related biomarkers (e.g., creatine kinase and myoglobin) [54] and testosterone reduction [55]. Lastly, our study is more likely to detect milder side effects, as lethal variants may have led to reduced participation in GLGC (i.e., selection bias) due to the presence of more severe side effects and developmental disorders. This could, in turn, impact the likelihood of identifying and including these variants in our study. A broader spectrum of potential side effects should be assessed in larger studies, such as linking electronic health records of clinical outcomes to biobank studies, e.g., the All of Us study, which recruited over 40,000 participants aged 18–29, [56] and utilizing birth cohorts with whole genome sequencing data.

Conclusions

This study did not find evidence that raise substantial concerns about the longer-term safety of statins and PCSK9 inhibitor in this Chinese adolescent population; these findings require verification in larger studies. Our study serves as a proof-of-concept study showing the use of well-characterized birth cohorts with multi-omics data (genetics and metabolomics) and phenotypes to facilitate the assessment of the efficacy and safety of drug targets among adolescents.

Availability of data and materials

Information on two lipid-modifying drug targets (HMGCR and PCSK9) were sourced from the druggable genome. The genome-wide association summary statistics for LDL-C, restricted to population of East Asian ancestry, was downloaded from Global Lipids Genetics Consortium (http://csg.sph.umich.edu/willer/public/glgc-lipids2021/). The genome-wide association summary statistics for coronary artery disease was downloaded from Biobank Japan (https://pheweb.jp/downloads). Data from the “Children of 1997” birth cohort are available upon reasonable request from the “Children of 1997” data access committee: aprmay97@hku.hk. R code for Mendelian randomization analysis, principal component analysis, and visualization are in Additional file 2.

Abbreviations

ALP:

Alkaline phosphatase

ALT:

Alanine transaminase

ApoB:

Apolipoprotein B

ApoB/ApoA1:

The ratio of apolipoprotein B to apolipoprotein A1

CAD:

Coronary artery disease

EAS:

East Asian

FEF 25–75%:

Mid-forced expiratory flow at 25–75% of the pulmonary volume

FH:

Familial hyperchloremia

FEV1:

Forced expiratory volume in the first second

FVC:

Forced vital capacity

GLGC:

Global Lipids Genetics Consortium

GRS:

Genetic risk score

GWAS:

Genome-wide association study

HMGCR:

Three-hydroxy-3-methylglutaryl coenzyme A

IVW:

Inverse-variance weighted

LD:

Linkage disequilibrium

LDL-C:

Low-density lipoprotein cholesterol

NMR:

Nuclear magnetic resonance

Non-HDL-C:

Non-high density lipoprotein cholesterol

PC:

Principal components

PCSK9:

Proprotein convertase subtilisin/kexin type 9

PEF:

Peak expiratory flow

RCT:

Randomized controlled trials

SD:

Standard deviation

VLDL-C:

Very low lipid density lipoprotein cholesterol

References

  1. Luirink IK, Wiegman A, Kusters DM, et al. 20-year follow-up of statins in children with familial hypercholesterolemia. N Engl J Med. 2019;381(16):1547–56.

    Article  CAS  PubMed  Google Scholar 

  2. Collaboration NCDRF. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet. 2017;390(10113):2627–42.

    Article  Google Scholar 

  3. Vuorio A, Kuoppala J, Kovanen PT, Humphries SE, Tonstad S, Wiegman A, Drogari E, Ramaswami U. Statins for children with familial hypercholesterolemia. Cochrane Database Syst Rev. 2019;2019(11):CD006401.

  4. Thompson PD, Panza G, Zaleski A, Taylor B. Statin-associated side effects. J Am Coll Cardiol. 2016;67(20):2395–410.

    Article  CAS  PubMed  Google Scholar 

  5. Suissa S. Immortal time bias in pharmaco-epidemiology. Am J Epidemiol. 2008;167(4):492–9.

    Article  PubMed  Google Scholar 

  6. Gill D, Georgakis MK, Koskeridis F, et al. Use of genetic variants related to antihypertensive drugs to inform on efficacy and side effects. Circulation. 2019;140(4):270–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Luo S, Schooling CM, Wong ICK, Au Yeung SL. Evaluating the impact of AMPK activation, a target of metformin, on risk of cardiovascular diseases and cancer in the UK Biobank: a Mendelian randomisation study. Diabetologia. 2020;63(11):2349–58.

    Article  CAS  PubMed  Google Scholar 

  8. Ardissino M, Slob EAW, Rajasundaram S, et al. Safety of beta-blocker and calcium channel blocker antihypertensive drugs in pregnancy: a Mendelian randomization study. BMC Med. 2022;20(1):288.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Rosoff DB, Bell AS, Jung J, Wagner J, Mavromatis LA, Lohoff FW. Mendelian randomization study of PCSK9 and HMG-CoA reductase inhibition and cognitive function. J Am Coll Cardiol. 2022;80(7):653–62.

    Article  CAS  PubMed  Google Scholar 

  10. Santos RD, Ruzza A, Hovingh GK, et al. Evolocumab in pediatric heterozygous familial hypercholesterolemia. N Engl J Med. 2020;383(14):1317–27.

    Article  CAS  PubMed  Google Scholar 

  11. Schmidt AF, Finan C, Gordillo-Maranon M, et al. Genetic drug target validation using Mendelian randomisation. Nat Commun. 2020;11(1):3255.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Schooling CM, Hui LL, Ho LM, Lam TH, Leung GM. Cohort profile: ‘children of 1997’: a Hong Kong Chinese birth cohort. Int J Epidemiol. 2012;41(3):611–20.

    Article  PubMed  Google Scholar 

  13. Lam TH, Leung GM, Ho LM. The effects of environmental tobacco smoke on health services utilization in the first eighteen months of life. Pediatrics. 2001;107(6):E91.

    Article  CAS  PubMed  Google Scholar 

  14. Wurtz P, Kangas AJ, Soininen P, Lawlor DA, Davey Smith G, Ala-Korpela M. Quantitative serum nuclear magnetic resonance metabolomics in large-scale epidemiology: a primer on -omic technologies. Am J Epidemiol. 2017;186(9):1084–96.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Soininen P, Kangas AJ, Würtz P, Suna T, Ala-Korpela M. Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics. Circ Cardiovasc Genet. 2015;8(1):192–206.

    Article  CAS  PubMed  Google Scholar 

  16. Graham SE, Clarke SL, Wu KH, et al. The power of genetic diversity in genome-wide association studies of lipids. Nature. 2021;600(7890):675–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Tam CHT, Lim CKP, Luk AOY, et al. Development of genome-wide polygenic risk scores for lipid traits and clinical applications for dyslipidemia, subclinical atherosclerosis, and diabetes cardiovascular complications among East Asians. Genome Med. 2021;13(1):29.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Davies NM, Holmes MV, Davey SG. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Ference BA, Robinson JG, Brook RD, et al. Variation in PCSK9 and HMGCR and risk of cardiovascular disease and diabetes. N Engl J Med. 2016;375(22):2144–53.

    Article  CAS  PubMed  Google Scholar 

  20. Ishigaki K, Akiyama M, Kanai M, et al. Large-scale genome-wide association study in a Japanese population identifies novel susceptibility loci across different diseases. Nat Genet. 2020;52(7):669–79.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Lotta LA, Scott RA, Sharp SJ, et al. Genetic predisposition to an impaired metabolism of the branched-chain amino acids and risk of type 2 diabetes: a Mendelian randomisation analysis. PLoS Med. 2016;13(11):e1002179.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Tobias DK, Lawler PR, Harada PH, et al. Circulating branched-chain amino acids and incident cardiovascular disease in a prospective cohort of US women. Circ Genom Precis Med. 2018;11(4):e002157.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Stoffel MA, Nakagawa S, Schielzeth H. partR2: partitioning R(2) in generalized linear mixed models. PeerJ. 2021;9:e11414.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Seaman SR, White IR, Copas AJ, Li L. Combining multiple imputation and inverse-probability weighting. Biometrics. 2012;68(1):129–37.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Metten MA, Costet N, Multigner L, Viel JF, Chauvet G. Inverse probability weighting to handle attrition in cohort studies: some guidance and a call for caution. BMC Med Res Methodol. 2022;22(1):45.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Burgess S, Dudbridge F, Thompson SG. Combining information on multiple instrumental variables in Mendelian randomization: comparison of allele score and summarized data methods. Stat Med. 2016;35(11):1880–906.

    Article  PubMed  Google Scholar 

  27. Greco MF, Minelli C, Sheehan NA, Thompson JR. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med. 2015;34(21):2926–40.

    Article  Google Scholar 

  28. Wurtz P, Wang Q, Soininen P, et al. Metabolomic profiling of statin use and genetic inhibition of HMG-CoA reductase. J Am Coll Cardiol. 2016;67(10):1200–10.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Sliz E, Kettunen J, Holmes MV, et al. Metabolomic consequences of genetic inhibition of PCSK9 compared with statin treatment. Circulation. 2018;138(22):2499–512.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Richardson TG, Leyden GM, Wang Q, et al. Characterising metabolomic signatures of lipid-modifying therapies through drug target mendelian randomisation. PLoS Biol. 2022;20(2):e3001547.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Schooling CM, Au Yeung SL, Zhao JV. Exploring pleiotropic effects of lipid modifiers and targets on measures of the coagulation system with genetics. Thromb Haemost. 2022;122(8):1296–303.

    Article  PubMed  Google Scholar 

  32. Dombalis S, Nash A. The effect of statins in children and adolescents with familial hypercholesterolemia: a systematic review. J Pediatr Health Care. 2021;35(3):292–303.

    Article  PubMed  Google Scholar 

  33. Singh A, Mittal S, Kazimuddin M. Evolocumab in pediatric heterozygous familial hypercholesterolemia. N Engl J Med. 2021;384(1):84.

    Article  PubMed  Google Scholar 

  34. Gill D, Georgakis MK, Walker VM, et al. Mendelian randomization for studying the effects of perturbing drug targets. Wellcome Open Res. 2021;6:16.

    Article  PubMed  PubMed Central  Google Scholar 

  35. Swerdlow DI, Preiss D, Kuchenbaecker KB, et al. HMG-coenzyme A reductase inhibition, type 2 diabetes, and bodyweight: evidence from genetic analysis and randomised trials. Lancet. 2015;385(9965):351–61.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Mills EJ, Wu P, Chong G, et al. Efficacy and safety of statin treatment for cardiovascular disease: a network meta-analysis of 170,255 patients from 76 randomized trials. QJM. 2011;104(2):109–24.

    Article  CAS  PubMed  Google Scholar 

  37. Stein EA, Illingworth DR, Kwiterovich PO Jr, et al. Efficacy and safety of lovastatin in adolescent males with heterozygous familial hypercholesterolemia: a randomized controlled trial. JAMA. 1999;281(2):137–44.

    Article  CAS  PubMed  Google Scholar 

  38. de Jongh S, Ose L, Szamosi T, et al. Efficacy and safety of statin therapy in children with familial hypercholesterolemia: a randomized, double-blind, placebo-controlled trial with simvastatin. Circulation. 2002;106(17):2231–7.

    Article  PubMed  Google Scholar 

  39. Wiegman A, Hutten BA, de Groot E, et al. Efficacy and safety of statin therapy in children with familial hypercholesterolemia: a randomized controlled trial. JAMA. 2004;292(3):331–7.

    Article  CAS  PubMed  Google Scholar 

  40. Ward NC, Watts GF, Eckel RH. Statin toxicity. Circ Res. 2019;124(2):328–50.

    Article  CAS  PubMed  Google Scholar 

  41. The US Food and Drug Administration. Cholesterol-lowering drugs get labeling changes. Available: https://www.fda.gov/drugs/special-features/cholesterol-lowering-drugs-get-labeling-changes.

  42. Stewart J, McCallin T, Martinez J, Chacko S, Yusuf S. Hyperlipidemia. Pediatr Rev. 2020;41(8):393–402.

    Article  PubMed  Google Scholar 

  43. Lagler FB, Hirschfeld S, Kindblom JM. Challenges in clinical trials for children and young people. Arch Dis Child. 2021;106(4):321–5.

    Article  PubMed  Google Scholar 

  44. Lascar N, Brown J, Pattison H, Barnett AH, Bailey CJ, Bellary S. Type 2 diabetes in adolescents and young adults. Lancet Diabetes Endocrinol. 2018;6(1):69–80.

    Article  PubMed  Google Scholar 

  45. Khoury M, Urbina EM. Hypertension in adolescents: diagnosis, treatment, and implications. Lancet Child Adolesc Health. 2021;5(5):357–66.

    Article  PubMed  Google Scholar 

  46. Swanson SA, Tiemeier H, Ikram MA, Hernán MA. Nature as a trialist?: Deconstructing the analogy between Mendelian randomization and randomized trials. Epidemiology. 2017;28(5):653–9.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Yu Z, Zhang L, Zhang G, et al. Lipids, apolipoproteins, statins, and intracerebral hemorrhage: a Mendelian randomization study. Ann Neurol. 2022;92(3):390–9.

    Article  CAS  PubMed  Google Scholar 

  48. Schooling CM, Zhao JV, Au Yeung SL, Leung GM. Investigating pleiotropic effects of statins on ischemic heart disease in the UK Biobank using Mendelian randomisation. Elife. 2020;9:e58567.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Minikel EV, Karczewski KJ, Martin HC, et al. Evaluating drug targets through human loss-of-function genetic variation. Nature. 2020;581(7809):459-+.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Karczewski KJ, Francioli LC, Tiao G, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature. 2020;581(7809):434–43.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Uitterlinden AG. Diversity in human genetics studies accelerates discovery and improves health care. Nat Rev Cardiol. 2022;19(5):289–90.

    Article  CAS  PubMed  Google Scholar 

  52. Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ. 2006;332(7549):1080.

    Article  PubMed  PubMed Central  Google Scholar 

  53. Burgess S, Butterworth AS, Thompson JR. Beyond Mendelian randomization: how to interpret evidence of shared genetic predictors. J Clin Epidemiol. 2016;69:208–16.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Cholesterol Treatment Trialists C. Effect of statin therapy on muscle symptoms: an individual participant data meta-analysis of large-scale, randomised, double-blind trials. Lancet. 2022;400(10355):832–45.

    Article  Google Scholar 

  55. Schooling CM, Au Yeung SL, Freeman G, Cowling BJ. The effect of statins on testosterone in men and women, a systematic review and meta-analysis of randomized controlled trials. BMC Med. 2013;11:57.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. National Institutes of Health. Data snapshots of All of Us research program. Available https://www.researchallofus.org/data-tools/data-snapshots/. Accessed May 2023.

Download references

Acknowledgements

We would like to thank Global Lipids Genetics Consortium and Biobank Japan for making their genome-wide association summary statistics available for the benefit of this work. We thank the participants and their parents of the “Children of 1997” birth cohort.

Authors’ Twitter handles

SL: @ShanLuo7; BH: @AnnabelHE3; SLAY: @Ryan_Au_Yeung

Funding

The “Children of 1997” birth cohort was initially supported by the Health Care and Promotion Fund, Health and Welfare Bureau, Government of the Hong Kong SAR [HCPF Grant # 216106], and re-established in 2005 funded by the Health and Health Services Research Fund [HHSRF Grant # 03040771, 05060671, 07080751, 07080841] and the Research Fund for the Control of Infectious Diseases in Hong Kong (RFCID Grant # 04050172, 06060592), Government of the Hong Kong SAR. The birth cohort has also received funding from the University Research Committee Strategic Research Theme (SRT) of Public Health Granted Research, The University of Hong Kong. The most recent follow-up was partly funded by the WYNG Foundation. The GWAS analyses was funded by the Health and Medical Research Fund, Government of the Hong Kong SAR (CFS-HKU1). The NMR metabolomic profiling was funded by Preemptive retention/Start up fund, Li Ka Shing Faculty of Medicine, The University of Hong Kong, and by Health and Medical Research Fund Government of the Hong Kong SAR (17181271). This work is a sub-study of the “Children of 1997” birth cohort funded by Health and Medical Research Fund, Government of the Hong Kong SAR (07181916). Publication was made possible in part by support from the HKU Libraries Open Access Author Fund sponsored by the HKU Libraries. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Author information

Authors and Affiliations

  1. School of Public Health, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 1/F, Patrick Manson Building, 7 Sassoon Road, Hong Kong SAR, China

    Shan Luo, Baoting He, Man Ki Kwok, Gabriel M. Leung, C Mary Schooling & Shiu Lun Au Yeung

  2. Department of Paediatrics, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong SAR, China

    Hugh Simon Lam

  3. Division of Cardiology, Queen Mary Hospital, The University of Hong Kong, Hong Kong SAR, China

    Yap Hang Chan

  4. Department of Surgery, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China

    Clara Sze Man Tang

  5. Dr. Li Dak-Sum Research Centre, The University of Hong Kong, Hong Kong SAR, China

    Clara Sze Man Tang

  6. School of Public Health and Health Policy, City University of New York, New York, NY, USA

    C Mary Schooling

Authors
  1. Shan Luo
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Hugh Simon Lam
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Yap Hang Chan
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Clara Sze Man Tang
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Baoting He
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Man Ki Kwok
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Gabriel M. Leung
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. C Mary Schooling
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Shiu Lun Au Yeung
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

SL: conceptualization, methodology, data curation, formal analysis, writing—original draft, writing—review and editing and visualization. HSL: funding acquisition and writing—review and editing. YHC: funding acquisition and writing—review and editing. CSMT: validation, supervision, funding acquisition, and writing—review and editing. BH: methodology, data curation, validation and writing—review and editing. MKK: data curation, and writing—review and editing. GML: data curation, methodology, funding acquisition, and writing—review and editing. CMS: data curation, supervision, funding acquisition and writing—review and editing. SLAY: conceptualization, methodology, project administration, data curation, funding acquisition, supervision, and writing—review and editing. SL and SLAY had directly accessed and verified the underlying data reported in the manuscript. SL is the guarantor of the findings. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Shiu Lun Au Yeung.

Ethics declarations

Ethics approval and consent to participate

Ethical approval for the study, including comprehensive health-related analyses, was obtained from the University of Hong Kong, Hospital Authority Hong Kong West Cluster Joint Institutional Review Board (UW19-829). Informed written consent was obtained from the parents/guardians, or from the participant if 18 years or older, before participation in the Biobank clinical follow-up. No ethics approval was required for the analyses using summary statistics from GLGC and Biobank Japan. The contributing studies to these consortia received ethical approval from their specific institutional review boards, and written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Completing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1. Table S1.

The primary efficacy outcomes from the “Children of 1997” birth cohort. Table S2. The safety outcomes from the “Children of 1997” birth cohort. Table S3. The genetic variants selected for the HMGCR score and PCSK9 score and their association in the Global Lipids Genetics Consortium and the “Children of 1997” birth cohort. Table S4. The effect of genetic inhibition of HMGCR and PCSK9 on coronary artery disease in East Asian. Table S5. The effect of genetic inhibition of HMGCR and PCSK9 on efficacy outcomes in the “Children of 1997” birth cohort. Table S6. The effect of genetic inhibition of HMGCR and PCSK9 on efficacy outcomes using complete case analysis in the “Children of 1997” birth cohort. Table S7. The effect of genetic inhibition of HMGCR and PCSK9 (using index variant) on efficacy outcomes in the “Children of 1997” birth cohort. Table S8. The effect of genetic inhibition of HMGCR and PCSK9 on safety outcomes in the “Children of 1997” birth cohort. Table S9. The effect of genetic inhibition of HMGCR and PCSK9 on safety outcomes using complete case analysis in the “Children of 1997” birth cohort. Table S10. The effect of genetic inhibition of HMGCR and PCSK9 (using index variant) on safety outcomes in the “Children of 1997” birth cohort. Supplementary Note. R code for Mendelian randomization analysis, principal component analysis, and visualization.

Additional file 2.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Luo, S., Lam, H.S., Chan, Y.H. et al. Assessing the safety of lipid-modifying medications among Chinese adolescents: a drug-target Mendelian randomization study. BMC Med 21, 410 (2023). https://doi.org/10.1186/s12916-023-03115-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12916-023-03115-y

Keywords

BMC Medicine

ISSN: 1741-7015

Contact us