Taller height and risk of coronary heart disease and cancer: A within-sibship Mendelian randomization study

  1. Laurence J Howe  Is a corresponding author
  2. Ben Brumpton
  3. Humaira Rasheed
  4. Bjørn Olav Åsvold
  5. George Davey Smith
  6. Neil M Davies  Is a corresponding author
  1. Medical Research Council Integrative Epidemiology Unit, Population Health Sciences, University of Bristol, United Kingdom
  2. Population Health Sciences, Bristol Medical School, University of Bristol, Barley House, Oakfield Grove, United Kingdom
  3. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Norway
  4. HUNT Research Center, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Norway
  5. Clinic of Medicine, St. Olavs Hospital, Trondheim University Hospital, Norway
  6. Department of Endocrinology, Clinic of Medicine, St. Olavs Hospital, Trondheim University Hospital, Norway

Abstract

Background:

Taller people have a lower risk of coronary heart disease but a higher risk of many cancers. Mendelian randomization (MR) studies in unrelated individuals (population MR) have suggested that these relationships are potentially causal. However, population MR studies are sensitive to demography (population stratification, assortative mating) and familial (indirect genetic) effects.

Methods:

In this study, we performed within-sibship MR analyses using 78,988 siblings, a design robust against demography and indirect genetic effects of parents. For comparison, we also applied population MR and estimated associations with measured height.

Results:

Within-sibship MR estimated that 1 SD taller height lowers the odds of coronary heart disease by 14% (95% CI: 3–23%) but increases the odds of cancer by 18% (95% CI: 3–34%), highly consistent with population MR and height-disease association estimates. There was some evidence that taller height reduces systolic blood pressure and low-density lipoprotein cholesterol, which may mediate some of the protective effects of taller height on coronary heart disease risk.

Conclusions:

For the first time, we have demonstrated that the purported effects of height on adulthood disease risk are unlikely to be explained by demographic or familial factors, and so likely reflect an individual-level causal effect. Disentangling the mechanisms via which height affects disease risk may improve the understanding of the etiologies of atherosclerosis and carcinogenesis.

Funding:

This project was conducted by researchers at the MRC Integrative Epidemiology Unit (MC_UU_00011/1) and also supported by a Norwegian Research Council Grant number 295989.

Editor's evaluation

The authors examined the role of height in cancer, coronary heart disease and cardiovascular disease risk factors, using four different designs. They found that height increases risk of cancer and decreases risk of coronary heart disease, while the associations for the cardiovascular disease risk factors were largely null. This will be mainly of interest to epidemiologists.

https://doi.org/10.7554/eLife.72984.sa0

Introduction

Height is a classical complex trait influenced by genetic and early-life environmental factors. Despite the nonmodifiable nature of adult height, evaluating the effects of height on noncommunicable disease risk can give insights into the etiology of adulthood diseases (Emerging Risk Factors Collaboration, 2012; Davey Smith et al., 2000). Two major groupings of diseases, cardiovascular disease and cancer, have divergent associations with height (Emerging Risk Factors Collaboration, 2012; Davey Smith et al., 2000; Stefan et al., 2016). Taller people are less likely to develop cardiovascular disease, including coronary heart disease (CHD) (Emerging Risk Factors Collaboration, 2012; Nelson et al., 2015; Nüesch et al., 2016; Hebert et al., 1993; Batty et al., 2009; Marouli et al., 2019) and stroke (Njølstad et al., 1996), but more likely to be diagnosed with cancer (Emerging Risk Factors Collaboration, 2012; Green et al., 2011; Zhang et al., 2015; Thrift et al., 2015; Dixon-Suen et al., 2018; Batty et al., 2006; Gunnell et al., 2001; Perkins et al., 2016). The mechanisms via which height influences disease risks are unclear. The association between height and cardiovascular disease may be mediated via favorable lipid profiles (Emerging Risk Factors Collaboration, 2012; Nelson et al., 2015), lower systolic blood pressure (SBP) (Emerging Risk Factors Collaboration, 2012; Langenberg et al., 2003), lung function (Marouli et al., 2019; Gunnell et al., 2003), lower heart rate (Smulyan et al., 1998), and coronary artery vessel dimension (O’Connor et al., 1996). The increased cancer incidence among taller individuals could relate to early-life exposure to hormones such as insulin-like growth factor 1 (IGF-1) (Renehan et al., 2004; Clayton et al., 2011) or the increased number of cells in taller individuals (Stefan et al., 2016; Green et al., 2011; Albanes and Winick, 1988). However, although overall cancer risk is higher among taller individuals (Green et al., 2011; Batty et al., 2006; Gunnell et al., 2001), there is some evidence for heterogeneity across cancer subtypes with null or inverse associations observed between height and risk of stomach, oropharyngeal, and esophageal cancers (Green et al., 2011; Batty et al., 2006; Gunnell et al., 2001; Perkins et al., 2016).

Height is highly heritable, but the average height across the European populations has increased over the last hundred years (Hatton, 2013), illustrating the effects of early-life environmental factors such as nutrition and childhood infections. The associations of height with adulthood diseases and relevant biomarkers could reflect the biomechanical effects relating to increased stature (e.g., number of cells or larger arteries; O’Connor et al., 1996) or could reflect confounding by early-life environmental factors that influence both height and later-life health such as parental socioeconomic position. For example, wealthier parents may provide their offspring with better nutrition, leading to increased adult height, and a better education, potentially leading to improved health in adulthood (Perkins et al., 2016). Thus, it is unclear whether height has a causal effect on the risk of cardiovascular disease and cancer or if a confounding factor influences both height and disease risk.

Mendelian randomization (Smith and Ebrahim, 2003) analyses, using genetic variants associated with height as a proxy for observed height, have been used to strengthen the evidence for causal effects of height on adulthood diseases (Nelson et al., 2015; Nüesch et al., 2016; Zhang et al., 2015; Thrift et al., 2015; Dixon-Suen et al., 2018). The underlying premise being that genetic variants associated with height, unlike height itself, are unlikely to be associated with potential confounders such as childhood nutrition. However, there is growing evidence that estimates from genetic epidemiological studies using unrelated individuals may capture effects relating to demography (population stratification, assortative mating) and familial effects (e.g., indirect genetic effects of relatives where parental genotype influences offspring phenotypes) (Barton and Hermisson, 2019; Berg et al., 2019; Sohail et al., 2019; Ruby et al., 2018; Haworth et al., 2019; Brumpton et al., 2019; Lee et al., 2018; Kong et al., 2018; Young et al., 2018). Indeed, recent articles have illustrated the potential for genetic analyses of height to be affected by these biases (Berg et al., 2019; Sohail et al., 2019), including a Mendelian randomization study of height on education (Brumpton et al., 2019). One approach to overcome these potential biases is to use data from siblings (Brumpton et al., 2019; Davies et al., 2019) and exploit the shared early-life environment of siblings and the random segregation of alleles during meiosis (Smith and Ebrahim, 2003). Indeed, true Mendelian randomization was initially proposed as existing within a parent-offspring design (Smith and Ebrahim, 2003; Davey Smith et al., 2020; Figure 1).

Mendelian randomization within families.

The random allocation of alleles within a parent-offspring quad (two parents and two offspring), initially observed by Mendel, is illustrated. Consider a height influencing genetic variant H where on average individuals with the H+ allele are taller than individuals with the H- allele. From Mendel’s law of segregation, parent 1, who is heterozygous at this allele, has an equal chance of transmitting either an H+ or H- allele to offspring. Parent 2, homozygous at this allele, will always transmit a copy of the H- allele. It follows that 50% of this pair’s offspring will be heterozygous (as parent 1) and 50% will be homozygous for the H- allele (as parent 2). On average, the heterozygous offspring will be taller than the homozygous H- offspring, with this difference a consequence of random segregation of gametes.

Here, we used data from 40,275 siblings from UK Biobank (Bycroft et al., 2018) and 38,723 siblings from the Norwegian HUNT study (Krokstad et al., 2013) to estimate the effects of adulthood height on CHD, cancer risk, and relevant biomarkers. Study-level information is contained in Table 1. We report the estimates of the effects of height on CHD and cancer from both phenotypic models and Mendelian randomization, with and without accounting for family structure.

Table 1
UK Biobank and HUNT study characteristics.

Information on the UK Biobank and Norwegian HUNT studies, including descriptive of the sibling samples, is given.

UK BiobankHUNT
Sibling sample:N individuals (N sibships)40,275 (19,588)38,723 (15,179)
Recruitment period:years2006–2010HUNT2 (1995–97)HUNT3 (2006–08)
Year of birth:median (Q1, Q3)1950 (1945, 1956)1951 (1937, 1963)
Sex:male (%)42.248.7
Male height (cm):mean (SD)175.7 (6.7)177.6 (6.7)
Female height (cm):mean (SD)162.4 (6.2)164.4 (6.3)
Coronary heart disease:N cases (% of sample)3006 (7.5%)6447 (16.6%)
Cancer:N cases (% of sample)6724 (16.5%)2323 (6.0%)

Methods

UK Biobank

Overview

UK Biobank is a large-scale prospective cohort study, described in detail previously (Bycroft et al., 2018; Sudlow et al., 2015). In brief, 503,325 individuals aged between 38 and 73 years were recruited between 2006 and 2010 from across the United Kingdom. For the purpose of this study, we used a subsample of 40,275 siblings from 19,588 families (Brumpton et al., 2019). Full-siblings were derived using UK Biobank-provided estimates of pairwise identical by state (IBS) kinships (>0.5–21 * IBS0, <0.7) and IBS0 (>0.001, <0.008), the proportion of unshared loci (Hill and Weir, 2011). This research has been conducted using the UK Biobank Resource under Application Number 15825. UK Biobank has ethical approval from the North West Multi-centre Research Ethics Committee (MREC). All UK Biobank participants provided written informed consent.

Phenotype data

At baseline, study participants attended an assessment center where they completed a touch-screen questionnaire, were interviewed, and had various measurements and samples taken. Height (field ID: 12144-0.0) and sitting height (field ID: 20015-0.0) were measured using a Seca 202 device at the assessment center. Seated height is equivalent to trunk length, leg length was defined as height minus seated height, and the leg to trunk ratio was calculated by taking the ratio of leg and trunk length. SBP was measured using an automated reading from an Omron Digital blood pressure monitor (field ID: 4080-0.0). Biomarkers of interest, including direct low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), glucose, and IGF-1, were measured using blood samples and the Beckman Coulter AU5800 or the DiaSorin LIASON XL (IGF-1) analyzers.

International Classification of Disease (10th edition) (ICD10) codes and Office of Population Censuses and Surveys Classifications of Interventions and Procedures (OPCS) codes were used to identify CHD and cancer (all subtypes and a stratified analysis) cases using several data sources: (1) secondary care data from Hospital Episode Statistics (HES), (2) death register data, and (3) cancer registry data. The stratified analysis included a subset of cancer subtypes (lung, oropharyngeal, stomach, esophageal, pancreatic, bladder, and multiple myeloma). Relevant codes are given in Supplementary file 1A. Both prevalent and incident cases were included in the analyses.

Genotyping

The UK Biobank study participants (N = 488,377) were genotyped using the UK BiLEVE (N = 49,950) and the closely related UK Biobank Axiom Arrays (N = 438,427). Directly genotyped variants were pre-phased using SHAPEIT3 (O’Connell et al., 2016) and imputed using Impute4 and the UK10K (Walter et al., 2015), Haplotype Reference Consortium (McCarthy et al., 2016) and 1000 Genomes Phase 3 (Genomes Project Consortium, 2015) reference panels. More details are given in a previous publication (Bycroft et al., 2018).

HUNT

Overview

The Trøndelag Health Study (HUNT) is a series of general health surveys of the adult population of the demographically stable Nord-Trøndelag region, Norway, as detailed in a previous study (Holmen et al., 2003). The entire adult population of this region (~90,000 adults in 1995) is invited to attend a health survey (includes comprehensive questionnaires, an interview, clinical examination, and detailed phenotypic measurements) every 10 years. To date, four health surveys have been conducted, HUNT1 (1984–1986), HUNT2 (1995–1997), HUNT3 (2006–2008), and HUNT4 (2017–2019), and all surveys have a high participation rate (Krokstad et al., 2013). This study includes 38,723 siblings from 15,179 families who participated in the HUNT2 and HUNT3 surveys. Siblings were identified using KING software (Manichaikul et al., 2010), with pairs defined as follows: kinship coefficient between 0.177 and 0.355, the proportion of the genomes that share two alleles identical by descent (IBD) > 0.08, and the proportion of the genome that share zero alleles IBD > 0.04. The use of HUNT data in this study was approved by the Regional Committee for Ethics in Medical Research, Central Norway (2017/2479). All HUNT study participants provided written informed consent.

Phenotype data

Height was measured to the nearest 1.0 cm using standardized instruments with participants wearing light clothes without shoes. SBP was measured using automated oscillometry (Critikon Dinamap 845XT and XL9301, acquired by GE Medical Systems Information Technologies in 2000) on the right arm in a relaxed sitting position (Holmen et al., 2003; Krokstad et al., 2013). SBP was measured twice with a 1 min interval between measurement with the mean of both measurements used in this study.

All HUNT participants provided nonfasting blood samples when attending the screening site. Total cholesterol, HDL-C, and TG levels in HUNT2 were measured in serum samples using enzymatic colorimetric methods (Boehringer Mannheim, Mannheim, Germany). In HUNT3, participants’ total cholesterol was measured by enzymatic cholesterol esterase methodology; HDL-C was measured by accelerator selective detergent methodology; and TGs were measured by glycerol phosphate oxidase methodology (Abbott, Clinical Chemistry, USA). LDL-C levels were calculated using the Friedewald formula (Friedewald et al., 1972) in both surveys. Participants in HUNT with TG levels ≥ 4.5 mmol/L (n = 1349) were excluded for LDL-C calculation as the Friedewald formula is not valid at higher TG levels. For all these phenotypes, if the participant attended both HUNT2 and HUNT3 surveys, then the values from HUNT2 were used for the analysis presented here.

The unique 11-digit identification number of every Norwegian citizen was used to link the HUNT participant records with the hospital registry, which included the three hospitals in the area (up to March 2019). We used ICD-10 and ICD-9 codes 410–414 and I20–I25 to define CHD, including both prevalent and incident cases. Cancer status (yes/no) was self-reported in HUNT2, HUNT3, and HUNT4 questionnaires. Individuals with discordant responses across different questionnaires were excluded from analyses. Due to the nature of cancer data collection, only prevalent cancer cases were included in analyses.

Genotyping

DNA samples were available from 71,860 HUNT samples from HUNT2 and HUNT3 and were genotyped (Krokstad et al., 2013) using one of the three different Illumina HumanCoreExome arrays: HumanCoreExome12 v1.0 (n = 7570), HumanCoreExome12 v1.1 (n = 4960), and University of Michigan HUNT Biobank v1.0 (n = 58,041; HumanCoreExome-24 v1.0, with custom content). Quality control was performed separately for genotype data from different arrays. The call rate of genotyped samples was >99%. Imputation was performed on samples of recent European ancestry using Minimac3 (v2.0.1, http://genome.sph.umich.edu/wiki/Minimac3) (Das et al., 2016) from a merged reference panel constructed from (1) the Haplotype Reference Consortium panel (release version 1.1) (McCarthy et al., 2016) and (2) a local reference panel based on 2202 whole-genome sequenced HUNT participants (Zhou et al., 2017). The subjects included in the study were of European ancestry and had passed the quality control.

Statistical analysis

Population and within-sibship models

The population model is a conventional regression model where the outcome is regressed (linear or logistic) against the exposure (height or height polygenic score [PGS]) with the option to include covariates.

The within-sibship model is an extension to the population model that includes a family mean term, the average exposure value across each family (height or height PGS), with each individual exposure value centered about the family mean exposure. To account for relatedness between siblings, standard errors are clustered by family in both models. More information on these models is contained in previous publications (Brumpton et al., 2019; Howe et al., 2021) with statistical code available on GitHub (Howe, 2022).

Phenotypic and Mendelian randomization analyses

In phenotypic analyses, we used regression models (within-sibship and population) to estimate the association between measured height and all outcomes (CHD, cancer, SBP, LDL-C, HDL-C, TG, glucose, and IGF-1) using linear models for continuous outcomes and logistic models for binary disease outcomes. In both cohorts, we used a standardized measure of height after adjusting for age and sex and also standardized continuous outcomes after adjusting for age and sex.

In Mendelian randomization analyses, we fit regression models as above but used an age/sex-standardized height PGS instead of measured height. The height PGS was constructed in PLINK (Purcell et al., 2007) using 372 independent (LD clumping: 250 kb, r2 < 0.01, p<5 × 10–8) genetic variants from a previous height genome-wide association study (Wood et al., 2014) that did not include UK Biobank or HUNT. Again, we standardized and adjusted for age/sex for continuous outcomes. To estimate the effect of the PGS on height, we fit a model regressing measured standardized height against the height PGS. We then generated scaled Mendelian randomization estimates by taking the Wald ratio of the PGS-outcome associations and the PGS-height associations. All statistical analyses were conducted using R (v. 3.5.1).

There are three core instrumental variable assumptions for Mendelian randomization analyses. First, the genetic variants should be robustly associated with the exposure (relevance). Second, there should be no unmeasured confounders of the genetic variant-outcome association (independence). Third, the genetic variants should only influence the outcome via their effect on the exposure (the exclusion restriction) (Haycock et al., 2016; Didelez and Sheehan, 2007; Lawlor et al., 2008).

UK Biobank and HUNT meta-analyses

We performed phenotypic and Mendelian randomization analyses (using population and within-sibship models) in both UK Biobank and HUNT. For phenotypes measured in both studies (CHD, cancer, LDL-C, HDL-C, TG), we combined estimates across both studies using a fixed-effects model in the metafor R package for meta-analysis. We tested for heterogeneity between UK Biobank/HUNT estimates using the difference of two means test statistic (Altman and Bland, 2003).

Outcomes

Using the previously described models and meta-analysis procedure, we estimated the effects of height on CHD, cancer, LDL-C, HDL-C, TG, glucose, and IGF-1. As a sensitivity analysis, we used phenotypic models to evaluate the associations between dimensions of height (leg length, trunk length, and leg to trunk ratio) with CHD and cancer in UK Biobank. A further sensitivity analysis involved repeating cancer analyses in UK Biobank with a subset of cancers not phenotypically associated with height (described above).

Results

Adulthood height and risk of CHD and cancer

We found consistent evidence across population and within-sibship models, using both measured height and a height PGS, that taller adulthood height reduced CHD risk and increased the risk of cancer (Supplementary file 1B and C).

Within-sibship Mendelian randomization estimated that 1 SD taller height (approximately 6.8 cm for men and 6.2 cm for women) reduced the odds of CHD by 14% (95% CI 3–23%) but increased the odds of cancer by 18% (95% CI 3–34%). These estimates were consistent across analyses using measured height as well as with population Mendelian randomization estimates. For example, population Mendelian randomization analyses estimated that 1 SD taller height reduced the odds of CHD by 10% (95% CI 4–16%) and increased the odds of cancer by 9% (95% CI 2–16%) (Table 2, Figure 2).

Taller height and risk of coronary heart disease and cancer.

The meta-analysis results from four different models used to evaluate the effect of height on coronary heart disease (CHD) and cancer risk are displayed. First, a phenotypic population model with measured height as the exposure and age and sex included as covariates. Second, a within-sibship phenotypic model with the family mean height included as an additional covariate to account for family structure. Third, a population Mendelian randomization model with height polygenic score (PGS) as the exposure exploiting advantageous properties of genetic instruments. Fourth, a within-sibship Mendelian randomization model with the family mean PGS included as a covariate to control for parental genotypes. Across all four models, we found consistent evidence that taller height reduces the odds of CHD and increases the odds of cancer.

Table 2
Mendelian randomization (MR) results: change in outcome (SD units), per 1 SD increase in height.

Population and within-sibship MR estimates of height on the eight different outcomes are shown. The presented estimates are from UK Biobank, HUNT, and the combined fixed-effects meta-analysis. The heterogeneity p-value refers to the difference between the UK Biobank and HUNT estimates.

OutcomeModelUK BiobankHUNTCombinedStudy heterogeneityp-value
Systolic blood pressurePopulation–0.044 (–0.074, –0.015)–0.025 (–0.057, 0.008)–0.036 (–0.058, –0.014)0.38
Within-sibship–0.077 (–0.137, –0.017)0.010 (–0.040, 0.059)–0.025 (–0.063, 0.013)0.03
High-density lipoprotein cholesterolPopulation–0.039 (–0.070, –0.008)–0.010 (–0.043, 0.024)–0.025 (–0.048, –0.003)0.21
Within-sibship–0.038 (–0.096, 0.021)0.001 (–0.046, 0.047)–0.014 (–0.050, 0.022)0.31
Low-density lipoprotein cholesterolPopulation–0.066 (–0.095, –0.036)–0.065 (–0.098, –0.032)–0.065 (–0.087, –0.044)0.99
Within-sibship–0.083 (–0.141, –0.025)–0.014 (–0.061, 0.033)–0.041 (–0.078, –0.005)0.07
TriglyceridesPopulation0.011 (–0.018, 0.040)–0.006 (–0.038, 0.027)0.004 (–0.018, 0.025)0.43
Within-sibship0.024 (–0.034, 0.081)0.032 (–0.018, 0.082)0.028 (–0.009, 0.066)0.84
GlucosePopulation0.032 (0.005, 0.060)N/A0.032 (0.005, 0.060)N/A
Within-sibship0.023 (–0.030, 0.077)N/A0.023 (–0.030, 0.077)N/A
IGF-1Population–0.005 (–0.035, 0.025)N/A–0.005 (–0.035, 0.025)N/A
Within-sibship–0.045 (–0.093, 0.004)N/A–0.045 (–0.093, 0.004)N/A
Cancer (OR)Population1.12 (1.04, 1.20)0.99 (0.88, 1.12)1.09 (1.02, 1.16)0.089
Within-sibship1.21 (1.03, 1.42)1.12 (0.90, 1.39)1.18 (1.03, 1.34)0.57
Coronary heart disease (OR)Population0.94 (0.84, 1.04)0.88 (0.80, 0.96)0.90 (0.84, 0.96)0.33
Within-sibship0.81 (0.65, 1.02)0.88 (0.77, 1.00)0.86 (0.77, 0.97)0.55

We then evaluated the associations between dimensions of height (trunk length, leg length, and leg to trunk ratio) and risk of CHD/cancer in UK Biobank. We found little evidence of heterogeneity between estimates, although stronger conclusions are limited by statistical power (Supplementary file 1D). We also ran a sensitivity analysis in UK Biobank, rerunning height-cancer analyses including only cases with one of seven cancer subtypes (lung, oropharyngeal, stomach, esophageal, pancreatic, bladder, and multiple myeloma) for which a previous study found little evidence they associated with height (Green et al., 2011). These subtypes generally show very strong social patterning, which could explain the attenuated associations with height that is also often socially patterned. As expected, the association of measured height with this subset of cancers (population OR 0.99; 95% CI 0.92–1.06; within-sibship OR 1.01; 95% CI 0.88–1.15) was less strong than the association between height and the all-cancer outcome (population OR 1.05; 95% CI 1.02–1.07; within-sibship OR 1.05; 95% CI 1.01–1.09). Mendelian randomization estimates were imprecise because of the modest number of cases for these cancers (Supplementary file 1E).

Adulthood height and biomarkers

Using measured biomarkers, both population and within-sibship models found evidence for the association between taller height and lower SBP, lower circulating LDL-C, and higher circulating IGF-1 levels. There was some evidence for heterogeneity in phenotypic associations between height and biomarkers in UK Biobank and HUNT, such as for SBP, which was more strongly associated with height in UK Biobank (Supplementary file 1B).

Population Mendelian randomization results suggested that taller height reduced SBP (per 1 SD taller height, 0.036 SD decrease; 95% CI 0.014–0.058), LDL-C (per 1 SD taller height, 0.065 SD decrease; 95% CI 0.044–0.087), HDL-C (per 1 SD taller height, 0.025 SD decrease; 95% CI 0.003–0.048) but increased glucose (per 1 SD taller height, 0.032 SD increase; 95% CI 0.005–0.060). In contrast, we found little evidence that taller height affected TG or IGF-1 levels. Within-sibship Mendelian randomization estimates were consistent with population estimates; SBP (per 1 SD taller height, 0.025 SD decrease; 95% CI –0.013 to 0.063), LDL-C (per 1 SD taller height, 0.041 SD decrease; 95% CI 0.005–0.078), HDL-C (per 1 SD taller height, 0.014 SD decrease; 95% CI –0.022 to 0.050) and glucose (per 1 SD taller height, 0.023 SD increase; 95% CI –0.030 to 0.077) (Figure 3, Table 2).

Mendelian randomization estimates of the effects of taller height on biomarkers.

The meta-analysis results from population and within-sibship Mendelian randomization analyses estimating the effect of taller height on biomarkers across UK Biobank and HUNT are shown. The estimates were broadly similar between the two models, suggesting the modest effects of demography and indirect genetic effects.

There was some putative evidence for heterogeneity in the Mendelian randomization effect estimates between UK Biobank and HUNT. For example, within-sibship Mendelian randomization estimate suggested the effects of height on SBP in UK Biobank (0.077 SD decrease; 95% CI 0.017–0.137) but the effect estimate was in the opposite direction in HUNT (0.010 SD increase; 95% CI –0.040 to 0.059; heterogeneity p=0.03) (Table 2).

Discussion

In this study, we used sibling data from two large biobanks to estimate the effects of height on CHD, cancer, and relevant biomarkers. We found consistent evidence across all models, including within-sibship Mendelian randomization, that taller height is protective against CHD but increases the risk of cancers. We found less consistent evidence for the effects of height on biomarkers; population and within-sibship phenotypic models as well as population Mendelian randomization models suggested modest effects of taller height on SBP, LDL-C, and HDL-C. However, the confidence intervals for within-family Mendelian randomization of height and biomarkers were too wide to draw strong conclusions.

Our findings are largely consistent with previous studies (Emerging Risk Factors Collaboration, 2012; Nelson et al., 2015; Nüesch et al., 2016; Hebert et al., 1993; Marouli et al., 2019; Green et al., 2011; Zhang et al., 2015; Thrift et al., 2015; Dixon-Suen et al., 2018; Carslake et al., 2013) that used nonsibling designs, and with the hypothesis that height affects CHD and cancer risk. However, previous studies were potentially susceptible to bias relating to geographic and socioeconomic variation in height and height genetic variants (Barton and Hermisson, 2019; Sohail et al., 2019; Lee et al., 2018). Indeed, a recent within-sibship Mendelian randomization study found that the previously reported effects of height and body mass index on educational attainment were greatly attenuated when using siblings (Brumpton et al., 2019). Here, we provided robust evidence for individual-level effects of height by demonstrating that the previous evidence for effects of height on adulthood disease risk is unlikely to have been confounded by demography or indirect genetic effects. The major strengths of our work are the use of within-sibship Mendelian randomization (Davies et al., 2019) and the triangulation (Lawlor et al., 2016) of evidence from across phenotypic, genetic, and within-sibship models.

A limitation of our analyses is that because of limited sibling data and the statistical inefficiency of within-family models, we have limited statistical power to investigate the effects of height on disease subtypes,further explore the mechanisms using multivariable Mendelian randomization (Burgess and Thompson, 2015), and perform sensitivity analyses to evaluate horizontal pleiotropy. An additional limitation is that our study may have been susceptible to selection and survival biases relating to nonrandom participation in UK Biobank and/or HUNT and the requirement of at least two siblings to survive to be recruited. Indeed, cancer and CHD are both leading worldwide causes of mortality and so cases for one disease may have a reduced likelihood of developing the other disease due to increased mortality. Therefore, our study may have been susceptible to survival bias relating to competing risks. We mitigated this by defining cases for both diseases using both nonfatal and fatal events. Our study analyzed families with two or more siblings jointly participating in a cohort; nevertheless, further research is required to investigate the impact of selection bias on family studies.

Adulthood height is nonmodifiable, and the interpretation of causality is nuanced because it is unclear whether biological effects relate to stature itself, increased childhood growth, or to factors highly correlated with height such as lung function (Marouli et al., 2019; Gunnell et al., 2003) and artery length (Palmer et al., 1990). Previous studies Gunnell et al., 2001; Langenberg et al., 2003; Gunnell et al., 2003; Regnault et al., 2014 have explored the possibility that associations may relate to dimensions of height, with evidence that blood pressure is associated with trunk but not leg length (Regnault et al., 2014). Here, we found that the effects of height on disease risk due to leg or trunk length were similar. We found consistent effects of increased height across etiologically heterogeneous cancer subtypes, which implies that the mechanism could relate to the larger number of cells in taller individuals or a generalized growth phenotype. Notably there is minimal evidence of a correlation between the size of an organism and cancer risk (Peto’s paradox), suggesting that the number of cells hypothesis could influence cancer risk in humans but would not explain variation in cancer risk across different organisms (Caulin and Maley, 2011). Our Mendelian randomization estimates for the effects of height on a subset of cancers not strongly phenotypically associated with height (Green et al., 2011) were consistent with the combined cancer estimates, although we had limited power in this dataset because of the modest prevalence of the cancer subtypes.

The estimated effects of height on disease risk were relatively consistent between the Norwegian HUNT and UK Biobank studies. Contrastingly, the heterogeneity between UK Biobank and HUNT for analyses involving SBP and LDL-C suggests that some effects of height could be population specific. Alternatively, heterogeneity could relate to the variance in associations between adulthood height and early-life environmental confounders across countries (Perkins et al., 2016). Additional explanations could relate to the differences in biomarker measurement between studies (e.g., measuring LDL-C directly or using the Friedewald formula, differences in fasting level before samples were taken), selection bias (Munafò et al., 2016), or differences between the cohorts in terms of recruitment and participation. Further work is required to investigate if our findings generalize to non-European populations; biological mechanisms could be expected to be largely consistent across populations but context-specific (e.g., social) mechanisms could lead to geographic heterogeneity.

To conclude, using within-sibship Mendelian randomization, we showed that height has individual-level effects on risk of CHD and cancers as well as several biomarkers. Larger family datasets and additional analyses including two-step (Relton and Davey Smith, 2012) and multivariable Mendelian randomization (Burgess and Thompson, 2015) could be used to investigate the potential mediators of these relationships.

Data availability

We used individual level data from the UK Biobank and HUNT cohorts. Participants in these studies have consented to the use of their data in medical research and so these data are not publicly available. Data access can be applied for by qualified researchers. For access to UK Biobank individual level participant data, please send enquiries to access@ukbiobank.ac.uk and see information on the UK Biobank website http://www.ukbiobank.ac.uk. UK Biobank access generally involves submitting project proposals which are evaluated by the study data access committee. Researchers associated with Norwegian research institutes can apply for the use of HUNT data and samples with approval by the Regional Committee for Medical and Health Research Ethics. HUNT data is governed by Norwegian law, therefore researchers from other countries may apply if collaborating with a Norwegian Principal Investigator. Detailed information on the data access procedure of HUNT can be found at https://www.ntnu.edu/hunt/data. Statistical code for population and within-sibship models used in the manuscript is available on GitHub https://github.com/LaurenceHowe/WithinSibshipModels/ (copy archived at swh:1:rev:44d2435d841bc424b56eee2d8534d52dd4adf763).

References

    1. Dixon-Suen SC
    2. Nagle CM
    3. Thrift AP
    4. Pharoah PDP
    5. Ewing A
    6. Pearce CL
    7. Zheng W
    8. Australian Ovarian Cancer Study Group
    9. Chenevix-Trench G
    10. Fasching PA
    11. Beckmann MW
    12. Lambrechts D
    13. Vergote I
    14. Lambrechts S
    15. Van Nieuwenhuysen E
    16. Rossing MA
    17. Doherty JA
    18. Wicklund KG
    19. Chang-Claude J
    20. Jung AY
    21. Moysich KB
    22. Odunsi K
    23. Goodman MT
    24. Wilkens LR
    25. Thompson PJ
    26. Shvetsov YB
    27. Dörk T
    28. Park-Simon T-W
    29. Hillemanns P
    30. Bogdanova N
    31. Butzow R
    32. Nevanlinna H
    33. Pelttari LM
    34. Leminen A
    35. Modugno F
    36. Ness RB
    37. Edwards RP
    38. Kelley JL
    39. Heitz F
    40. du Bois A
    41. Harter P
    42. Schwaab I
    43. Karlan BY
    44. Lester J
    45. Orsulic S
    46. Rimel BJ
    47. Kjær SK
    48. Høgdall E
    49. Jensen A
    50. Goode EL
    51. Fridley BL
    52. Cunningham JM
    53. Winham SJ
    54. Giles GG
    55. Bruinsma F
    56. Milne RL
    57. Southey MC
    58. Hildebrandt MAT
    59. Wu X
    60. Lu KH
    61. Liang D
    62. Levine DA
    63. Bisogna M
    64. Schildkraut JM
    65. Berchuck A
    66. Cramer DW
    67. Terry KL
    68. Bandera EV
    69. Olson SH
    70. Salvesen HB
    71. Thomsen LCV
    72. Kopperud RK
    73. Bjorge L
    74. Kiemeney LA
    75. Massuger LFAG
    76. Pejovic T
    77. Bruegl A
    78. Cook LS
    79. Le ND
    80. Swenerton KD
    81. Brooks-Wilson A
    82. Kelemen LE
    83. Lubiński J
    84. Huzarski T
    85. Gronwald J
    86. Menkiszak J
    87. Wentzensen N
    88. Brinton L
    89. Yang H
    90. Lissowska J
    91. Høgdall CK
    92. Lundvall L
    93. Song H
    94. Tyrer JP
    95. Campbell I
    96. Eccles D
    97. Paul J
    98. Glasspool R
    99. Siddiqui N
    100. Whittemore AS
    101. Sieh W
    102. McGuire V
    103. Rothstein JH
    104. Narod SA
    105. Phelan C
    106. Risch HA
    107. McLaughlin JR
    108. Anton-Culver H
    109. Ziogas A
    110. Menon U
    111. Gayther SA
    112. Ramus SJ
    113. Gentry-Maharaj A
    114. Wu AH
    115. Pike MC
    116. Tseng C-C
    117. Kupryjanczyk J
    118. Dansonka-Mieszkowska A
    119. Budzilowska A
    120. Rzepecka IK
    121. Webb PM
    122. Ovarian Cancer Association Consortium
    (2018) Adult height is associated with increased risk of ovarian cancer: a Mendelian randomisation study
    British Journal of Cancer 118:1123–1129.
    https://doi.org/10.1038/s41416-018-0011-3
    1. Holmen J
    2. Midthjell K
    3. Krüger Ø
    (2003)
    The Nord-Trøndelag Health Study 1995–97 (HUNT 2): objectives, contents, methods and participation
    Norsk Epidemiologi 13:19–32.
    1. Walter K
    2. Min JL
    3. Huang J
    4. Crooks L
    5. Memari Y
    6. McCarthy S
    7. Perry JRB
    8. Xu C
    9. Futema M
    10. Lawson D
    11. Iotchkova V
    12. Schiffels S
    13. Hendricks AE
    14. Danecek P
    15. Li R
    16. Floyd J
    17. Wain LV
    18. Barroso I
    19. Humphries SE
    20. Hurles ME
    21. Zeggini E
    22. Barrett JC
    23. Plagnol V
    24. Brent Richards J
    25. Greenwood CMT
    26. Timpson NJ
    27. Durbin R
    28. Soranzo N
    29. Bala S
    30. Clapham P
    31. Coates G
    32. Cox T
    33. Daly A
    34. Danecek P
    35. Du Y
    36. Durbin R
    37. Edkins S
    38. Ellis P
    39. Flicek P
    40. Guo X
    41. Guo X
    42. Huang L
    43. Jackson DK
    44. Joyce C
    45. Keane T
    46. Kolb-Kokocinski A
    47. Langford C
    48. Li Y
    49. Liang J
    50. Lin H
    51. Liu R
    52. Maslen J
    53. McCarthy S
    54. Muddyman D
    55. Quail MA
    56. Stalker J
    57. Sun J
    58. Tian J
    59. Wang G
    60. Wang J
    61. Wang Y
    62. Wong K
    63. Zhang P
    64. Barroso I
    65. Birney E
    66. Boustred C
    67. Chen L
    68. Clement G
    69. Cocca M
    70. Danecek P
    71. Davey Smith G
    72. Day INM
    73. Day-Williams A
    74. Down T
    75. Dunham I
    76. Durbin R
    77. Evans DM
    78. Gaunt TR
    79. Geihs M
    80. Greenwood CMT
    81. Hart D
    82. Hendricks AE
    83. Howie B
    84. Huang J
    85. Hubbard T
    86. Hysi P
    87. Iotchkova V
    88. Jamshidi Y
    89. Karczewski KJ
    90. Kemp JP
    91. Lachance G
    92. Lawson D
    93. Lek M
    94. Lopes M
    95. MacArthur DG
    96. Marchini J
    97. Mangino M
    98. Mathieson I
    99. McCarthy S
    100. Memari Y
    101. Metrustry S
    102. Min JL
    103. Moayyeri A
    104. Muddyman D
    105. Northstone K
    106. Panoutsopoulou K
    107. Paternoster L
    108. Perry JRB
    109. Quaye L
    110. Brent Richards J
    111. Ring S
    112. Ritchie GRS
    113. Schiffels S
    114. Shihab HA
    115. Shin SY
    116. Small KS
    117. Soler Artigas M
    118. Soranzo N
    119. Southam L
    120. Spector TD
    121. St Pourcain B
    122. Surdulescu G
    123. Tachmazidou I
    124. Timpson NJ
    125. Tobin MD
    126. Valdes AM
    127. Visscher PM
    128. Wain LV
    129. Walter K
    130. Ward K
    131. Wilson SG
    132. Wong K
    133. Yang J
    134. Zeggini E
    135. Zhang F
    136. Zheng HF
    137. Anney R
    138. Ayub M
    139. Barrett JC
    140. Blackwood D
    141. Bolton PF
    142. Breen G
    143. Collier DA
    144. Craddock N
    145. Crooks L
    146. Curran S
    147. Curtis D
    148. Durbin R
    149. Gallagher L
    150. Geschwind D
    151. Gurling H
    152. Holmans P
    153. Lee I
    154. Lönnqvist J
    155. McCarthy S
    156. McGuffin P
    157. McIntosh AM
    158. McKechanie AG
    159. McQuillin A
    160. Morris J
    161. Muddyman D
    162. O’Donovan MC
    163. Owen MJ
    164. Palotie A
    165. Parr JR
    166. Paunio T
    167. Pietilainen O
    168. Rehnström K
    169. Sharp SI
    170. Skuse D
    171. St Clair D
    172. Suvisaari J
    173. Walters JTR
    174. Williams HJ
    175. Barroso I
    176. Bochukova E
    177. Bounds R
    178. Dominiczak A
    179. Durbin R
    180. Farooqi IS
    181. Hendricks AE
    182. Keogh J
    183. Marenne G
    184. McCarthy S
    185. Morris A
    186. Muddyman D
    187. O’Rahilly S
    188. Porteous DJ
    189. Smith BH
    190. Tachmazidou I
    191. Wheeler E
    192. Zeggini E
    193. Al Turki S
    194. Anderson CA
    195. Antony D
    196. Barroso I
    197. Beales P
    198. Bentham J
    199. Bhattacharya S
    200. Calissano M
    201. Carss K
    202. Chatterjee K
    203. Cirak S
    204. Cosgrove C
    205. Durbin R
    206. Fitzpatrick DR
    207. Floyd J
    208. Reghan Foley A
    209. Franklin CS
    210. Futema M
    211. Grozeva D
    212. Humphries SE
    213. Hurles ME
    214. McCarthy S
    215. Mitchison HM
    216. Muddyman D
    217. Muntoni F
    218. O’Rahilly S
    219. Onoufriadis A
    220. Parker V
    221. Payne F
    222. Plagnol V
    223. Lucy Raymond F
    224. Roberts N
    225. Savage DB
    226. Scambler P
    227. Schmidts M
    228. Schoenmakers N
    229. Semple RK
    230. Serra E
    231. Spasic-Boskovic O
    232. Stevens E
    233. van Kogelenberg M
    234. Vijayarangakannan P
    235. Walter K
    236. Williamson KA
    237. Wilson C
    238. Whyte T
    239. Ciampi A
    240. Greenwood CMT
    241. Hendricks AE
    242. Li R
    243. Metrustry S
    244. Oualkacha K
    245. Tachmazidou I
    246. Xu C
    247. Zeggini E
    248. Bobrow M
    249. Bolton PF
    250. Durbin R
    251. Fitzpatrick DR
    252. Griffin H
    253. Hurles ME
    254. Kaye J
    255. Kennedy K
    256. Kent A
    257. Muddyman D
    258. Muntoni F
    259. Lucy Raymond F
    260. Semple RK
    261. Smee C
    262. Spector TD
    263. Timpson NJ
    264. Charlton R
    265. Ekong R
    266. Futema M
    267. Humphries SE
    268. Khawaja F
    269. Lopes LR
    270. Migone N
    271. Payne SJ
    272. Plagnol V
    273. Pollitt RC
    274. Povey S
    275. Ridout CK
    276. Robinson RL
    277. Scott RH
    278. Shaw A
    279. Syrris P
    280. Taylor R
    281. Vandersteen AM
    282. Barrett JC
    283. Barroso I
    284. Davey Smith G
    285. Durbin R
    286. Farooqi IS
    287. Fitzpatrick DR
    288. Hurles ME
    289. Kaye J
    290. Kennedy K
    291. Langford C
    292. McCarthy S
    293. Muddyman D
    294. Owen MJ
    295. Palotie A
    296. Brent Richards J
    297. Soranzo N
    298. Spector TD
    299. Stalker J
    300. Timpson NJ
    301. Zeggini E
    302. Amuzu A
    303. Pablo Casas J
    304. Chambers JC
    305. Cocca M
    306. Dedoussis G
    307. Gambaro G
    308. Gasparini P
    309. Gaunt TR
    310. Huang J
    311. Iotchkova V
    312. Isaacs A
    313. Johnson J
    314. Kleber ME
    315. Kooner JS
    316. Langenberg C
    317. Luan J
    318. Malerba G
    319. März W
    320. Matchan A
    321. Min JL
    322. Morris R
    323. Nordestgaard BG
    324. Benn M
    325. Ring S
    326. Scott RA
    327. Soranzo N
    328. Southam L
    329. Timpson NJ
    330. Toniolo D
    331. Traglia M
    332. Tybjaerg-Hansen A
    333. van Duijn CM
    334. van Leeuwen EM
    335. Varbo A
    336. Whincup P
    337. Zaza G
    338. Zeggini E
    339. Zhang W
    340. The UK10K Consortium
    341. Writing group
    342. Production group
    343. Cohorts group
    344. Neurodevelopmental disorders group
    345. Obesity group
    346. Rare disease group
    347. Statistics group
    348. Ethics group
    349. Incidental findings group
    350. Management committee
    351. Lipid meta-analysis group
    352. The UCLEB Consortium
    (2015) The UK10K project identifies rare variants in health and disease
    Nature 526:82–90.
    https://doi.org/10.1038/nature14962
    1. Wood AR
    2. Esko T
    3. Yang J
    4. Vedantam S
    5. Pers TH
    6. Gustafsson S
    7. Chu AY
    8. Estrada K
    9. Luan J
    10. Kutalik Z
    11. Amin N
    12. Buchkovich ML
    13. Croteau-Chonka DC
    14. Day FR
    15. Duan Y
    16. Fall T
    17. Fehrmann R
    18. Ferreira T
    19. Jackson AU
    20. Karjalainen J
    21. Lo KS
    22. Locke AE
    23. Mägi R
    24. Mihailov E
    25. Porcu E
    26. Randall JC
    27. Scherag A
    28. Vinkhuyzen AAE
    29. Westra HJ
    30. Winkler TW
    31. Workalemahu T
    32. Zhao JH
    33. Absher D
    34. Albrecht E
    35. Anderson D
    36. Baron J
    37. Beekman M
    38. Demirkan A
    39. Ehret GB
    40. Feenstra B
    41. Feitosa MF
    42. Fischer K
    43. Fraser RM
    44. Goel A
    45. Gong J
    46. Justice AE
    47. Kanoni S
    48. Kleber ME
    49. Kristiansson K
    50. Lim U
    51. Lotay V
    52. Lui JC
    53. Mangino M
    54. Mateo Leach I
    55. Medina-Gomez C
    56. Nalls MA
    57. Nyholt DR
    58. Palmer CD
    59. Pasko D
    60. Pechlivanis S
    61. Prokopenko I
    62. Ried JS
    63. Ripke S
    64. Shungin D
    65. Stancáková A
    66. Strawbridge RJ
    67. Sung YJ
    68. Tanaka T
    69. Teumer A
    70. Trompet S
    71. van der Laan SW
    72. van Setten J
    73. Van Vliet-Ostaptchouk JV
    74. Wang Z
    75. Yengo L
    76. Zhang W
    77. Afzal U
    78. Arnlöv J
    79. Arscott GM
    80. Bandinelli S
    81. Barrett A
    82. Bellis C
    83. Bennett AJ
    84. Berne C
    85. Blüher M
    86. Bolton JL
    87. Böttcher Y
    88. Boyd HA
    89. Bruinenberg M
    90. Buckley BM
    91. Buyske S
    92. Caspersen IH
    93. Chines PS
    94. Clarke R
    95. Claudi-Boehm S
    96. Cooper M
    97. Daw EW
    98. De Jong PA
    99. Deelen J
    100. Delgado G
    101. Denny JC
    102. Dhonukshe-Rutten R
    103. Dimitriou M
    104. Doney ASF
    105. Dörr M
    106. Eklund N
    107. Eury E
    108. Folkersen L
    109. Garcia ME
    110. Geller F
    111. Giedraitis V
    112. Go AS
    113. Grallert H
    114. Grammer TB
    115. Gräßler J
    116. Grönberg H
    117. de Groot L
    118. Groves CJ
    119. Haessler J
    120. Hall P
    121. Haller T
    122. Hallmans G
    123. Hannemann A
    124. Hartman CA
    125. Hassinen M
    126. Hayward C
    127. Heard-Costa NL
    128. Helmer Q
    129. Hemani G
    130. Henders AK
    131. Hillege HL
    132. Hlatky MA
    133. Hoffmann W
    134. Hoffmann P
    135. Holmen O
    136. Houwing-Duistermaat JJ
    137. Illig T
    138. Isaacs A
    139. James AL
    140. Jeff J
    141. Johansen B
    142. Johansson Å
    143. Jolley J
    144. Juliusdottir T
    145. Junttila J
    146. Kho AN
    147. Kinnunen L
    148. Klopp N
    149. Kocher T
    150. Kratzer W
    151. Lichtner P
    152. Lind L
    153. Lindström J
    154. Lobbens S
    155. Lorentzon M
    156. Lu Y
    157. Lyssenko V
    158. Magnusson PKE
    159. Mahajan A
    160. Maillard M
    161. McArdle WL
    162. McKenzie CA
    163. McLachlan S
    164. McLaren PJ
    165. Menni C
    166. Merger S
    167. Milani L
    168. Moayyeri A
    169. Monda KL
    170. Morken MA
    171. Müller G
    172. Müller-Nurasyid M
    173. Musk AW
    174. Narisu N
    175. Nauck M
    176. Nolte IM
    177. Nöthen MM
    178. Oozageer L
    179. Pilz S
    180. Rayner NW
    181. Renstrom F
    182. Robertson NR
    183. Rose LM
    184. Roussel R
    185. Sanna S
    186. Scharnagl H
    187. Scholtens S
    188. Schumacher FR
    189. Schunkert H
    190. Scott RA
    191. Sehmi J
    192. Seufferlein T
    193. Shi J
    194. Silventoinen K
    195. Smit JH
    196. Smith AV
    197. Smolonska J
    198. Stanton AV
    199. Stirrups K
    200. Stott DJ
    201. Stringham HM
    202. Sundström J
    203. Swertz MA
    204. Syvänen AC
    205. Tayo BO
    206. Thorleifsson G
    207. Tyrer JP
    208. van Dijk S
    209. van Schoor NM
    210. van der Velde N
    211. van Heemst D
    212. van Oort FVA
    213. Vermeulen SH
    214. Verweij N
    215. Vonk JM
    216. Waite LL
    217. Waldenberger M
    218. Wennauer R
    219. Wilkens LR
    220. Willenborg C
    221. Wilsgaard T
    222. Wojczynski MK
    223. Wong A
    224. Wright AF
    225. Zhang Q
    226. Arveiler D
    227. Bakker SJL
    228. Beilby J
    229. Bergman RN
    230. Bergmann S
    231. Biffar R
    232. Blangero J
    233. Boomsma DI
    234. Bornstein SR
    235. Bovet P
    236. Brambilla P
    237. Brown MJ
    238. Campbell H
    239. Caulfield MJ
    240. Chakravarti A
    241. Collins R
    242. Collins FS
    243. Crawford DC
    244. Cupples LA
    245. Danesh J
    246. de Faire U
    247. den Ruijter HM
    248. Erbel R
    249. Erdmann J
    250. Eriksson JG
    251. Farrall M
    252. Ferrannini E
    253. Ferrières J
    254. Ford I
    255. Forouhi NG
    256. Forrester T
    257. Gansevoort RT
    258. Gejman PV
    259. Gieger C
    260. Golay A
    261. Gottesman O
    262. Gudnason V
    263. Gyllensten U
    264. Haas DW
    265. Hall AS
    266. Harris TB
    267. Hattersley AT
    268. Heath AC
    269. Hengstenberg C
    270. Hicks AA
    271. Hindorff LA
    272. Hingorani AD
    273. Hofman A
    274. Hovingh GK
    275. Humphries SE
    276. Hunt SC
    277. Hypponen E
    278. Jacobs KB
    279. Jarvelin MR
    280. Jousilahti P
    281. Jula AM
    282. Kaprio J
    283. Kastelein JJP
    284. Kayser M
    285. Kee F
    286. Keinanen-Kiukaanniemi SM
    287. Kiemeney LA
    288. Kooner JS
    289. Kooperberg C
    290. Koskinen S
    291. Kovacs P
    292. Kraja AT
    293. Kumari M
    294. Kuusisto J
    295. Lakka TA
    296. Langenberg C
    297. Le Marchand L
    298. Lehtimäki T
    299. Lupoli S
    300. Madden PAF
    301. Männistö S
    302. Manunta P
    303. Marette A
    304. Matise TC
    305. McKnight B
    306. Meitinger T
    307. Moll FL
    308. Montgomery GW
    309. Morris AD
    310. Morris AP
    311. Murray JC
    312. Nelis M
    313. Ohlsson C
    314. Oldehinkel AJ
    315. Ong KK
    316. Ouwehand WH
    317. Pasterkamp G
    318. Peters A
    319. Pramstaller PP
    320. Price JF
    321. Qi L
    322. Raitakari OT
    323. Rankinen T
    324. Rao DC
    325. Rice TK
    326. Ritchie M
    327. Rudan I
    328. Salomaa V
    329. Samani NJ
    330. Saramies J
    331. Sarzynski MA
    332. Schwarz PEH
    333. Sebert S
    334. Sever P
    335. Shuldiner AR
    336. Sinisalo J
    337. Steinthorsdottir V
    338. Stolk RP
    339. Tardif JC
    340. Tönjes A
    341. Tremblay A
    342. Tremoli E
    343. Virtamo J
    344. Vohl MC
    345. Amouyel P
    346. Asselbergs FW
    347. Assimes TL
    348. Bochud M
    349. Boehm BO
    350. Boerwinkle E
    351. Bottinger EP
    352. Bouchard C
    353. Cauchi S
    354. Chambers JC
    355. Chanock SJ
    356. Cooper RS
    357. de Bakker PIW
    358. Dedoussis G
    359. Ferrucci L
    360. Franks PW
    361. Froguel P
    362. Groop LC
    363. Haiman CA
    364. Hamsten A
    365. Hayes MG
    366. Hui J
    367. Hunter DJ
    368. Hveem K
    369. Jukema JW
    370. Kaplan RC
    371. Kivimaki M
    372. Kuh D
    373. Laakso M
    374. Liu Y
    375. Martin NG
    376. März W
    377. Melbye M
    378. Moebus S
    379. Munroe PB
    380. Njølstad I
    381. Oostra BA
    382. Palmer CNA
    383. Pedersen NL
    384. Perola M
    385. Pérusse L
    386. Peters U
    387. Powell JE
    388. Power C
    389. Quertermous T
    390. Rauramaa R
    391. Reinmaa E
    392. Ridker PM
    393. Rivadeneira F
    394. Rotter JI
    395. Saaristo TE
    396. Saleheen D
    397. Schlessinger D
    398. Slagboom PE
    399. Snieder H
    400. Spector TD
    401. Strauch K
    402. Stumvoll M
    403. Tuomilehto J
    404. Uusitupa M
    405. van der Harst P
    406. Völzke H
    407. Walker M
    408. Wareham NJ
    409. Watkins H
    410. Wichmann HE
    411. Wilson JF
    412. Zanen P
    413. Deloukas P
    414. Heid IM
    415. Lindgren CM
    416. Mohlke KL
    417. Speliotes EK
    418. Thorsteinsdottir U
    419. Barroso I
    420. Fox CS
    421. North KE
    422. Strachan DP
    423. Beckmann JS
    424. Berndt SI
    425. Boehnke M
    426. Borecki IB
    427. McCarthy MI
    428. Metspalu A
    429. Stefansson K
    430. Uitterlinden AG
    431. van Duijn CM
    432. Franke L
    433. Willer CJ
    434. Price AL
    435. Lettre G
    436. Loos RJF
    437. Weedon MN
    438. Ingelsson E
    439. O’Connell JR
    440. Abecasis GR
    441. Chasman DI
    442. Goddard ME
    443. Visscher PM
    444. Hirschhorn JN
    445. Frayling TM
    446. PAGEGE Consortium
    447. LifeLines Cohort Study
    (2014) Defining the role of common variation in the genomic and biological architecture of adult human height
    Nature Genetics 46:1173–1186.
    https://doi.org/10.1038/ng.3097
    1. Zhang B
    2. Shu X-O
    3. Delahanty RJ
    4. Zeng C
    5. Michailidou K
    6. Bolla MK
    7. Wang Q
    8. Dennis J
    9. Wen W
    10. Long J
    11. Li C
    12. Dunning AM
    13. Chang-Claude J
    14. Shah M
    15. Perkins BJ
    16. Czene K
    17. Darabi H
    18. Eriksson M
    19. Bojesen SE
    20. Nordestgaard BG
    21. Nielsen SF
    22. Flyger H
    23. Lambrechts D
    24. Neven P
    25. Wildiers H
    26. Floris G
    27. Schmidt MK
    28. Rookus MA
    29. van den Hurk K
    30. de Kort WLAM
    31. Couch FJ
    32. Olson JE
    33. Hallberg E
    34. Vachon C
    35. Rudolph A
    36. Seibold P
    37. Flesch-Janys D
    38. Peto J
    39. Dos-Santos-Silva I
    40. Fletcher O
    41. Johnson N
    42. Nevanlinna H
    43. Muranen TA
    44. Aittomäki K
    45. Blomqvist C
    46. Li J
    47. Humphreys K
    48. Brand J
    49. Guénel P
    50. Truong T
    51. Cordina-Duverger E
    52. Menegaux F
    53. Burwinkel B
    54. Marme F
    55. Yang R
    56. Surowy H
    57. Benitez J
    58. Zamora MP
    59. Perez JIA
    60. Cox A
    61. Cross SS
    62. Reed MWR
    63. Andrulis IL
    64. Knight JA
    65. Glendon G
    66. Tchatchou S
    67. Sawyer EJ
    68. Tomlinson I
    69. Kerin MJ
    70. Miller N
    71. Chenevix-Trench G
    72. kConFab Investigators, Australian Ovarian Study Group
    73. Haiman CA
    74. Henderson BE
    75. Schumacher F
    76. Marchand LL
    77. Lindblom A
    78. Margolin S
    79. Hooning MJ
    80. Martens JWM
    81. Tilanus-Linthorst MMA
    82. Collée JM
    83. Hopper JL
    84. Southey MC
    85. Tsimiklis H
    86. Apicella C
    87. Slager S
    88. Toland AE
    89. Ambrosone CB
    90. Yannoukakos D
    91. Giles GG
    92. Milne RL
    93. McLean C
    94. Fasching PA
    95. Haeberle L
    96. Ekici AB
    97. Beckmann MW
    98. Brenner H
    99. Dieffenbach AK
    100. Arndt V
    101. Stegmaier C
    102. Swerdlow AJ
    103. Ashworth A
    104. Orr N
    105. Jones M
    106. Figueroa J
    107. Garcia-Closas M
    108. Brinton L
    109. Lissowska J
    110. Dumont M
    111. Winqvist R
    112. Pylkäs K
    113. Jukkola-Vuorinen A
    114. Grip M
    115. Brauch H
    116. Brüning T
    117. Ko Y-D
    118. Peterlongo P
    119. Manoukian S
    120. Bonanni B
    121. Radice P
    122. Bogdanova N
    123. Antonenkova N
    124. Dörk T
    125. Mannermaa A
    126. Kataja V
    127. Kosma V-M
    128. Hartikainen JM
    129. Devilee P
    130. Seynaeve C
    131. Van Asperen CJ
    132. Jakubowska A
    133. Lubiński J
    134. Jaworska-Bieniek K
    135. Durda K
    136. Hamann U
    137. Torres D
    138. Schmutzler RK
    139. Neuhausen SL
    140. Anton-Culver H
    141. Kristensen VN
    142. Grenaker Alnæs GI
    143. DRIVE Project
    144. Pierce BL
    145. Kraft P
    146. Peters U
    147. Lindstrom S
    148. Seminara D
    149. Burgess S
    150. Ahsan H
    151. Whittemore AS
    152. John EM
    153. Gammon MD
    154. Malone KE
    155. Tessier DC
    156. Vincent D
    157. Bacot F
    158. Luccarini C
    159. Baynes C
    160. Ahmed S
    161. Maranian M
    162. Healey CS
    163. González-Neira A
    164. Pita G
    165. Alonso MR
    166. Álvarez N
    167. Herrero D
    168. Pharoah PDP
    169. Simard J
    170. Hall P
    171. Hunter DJ
    172. Easton DF
    173. Zheng W
    (2015) Height and Breast Cancer Risk: Evidence From Prospective Studies and Mendelian Randomization
    Journal of the National Cancer Institute 107:djv219.
    https://doi.org/10.1093/jnci/djv219

Decision letter

  1. Edward D Janus
    Reviewing Editor; University of Melbourne, Australia
  2. Matthias Barton
    Senior Editor; University of Zurich, Switzerland
  3. Sara Hägg
    Reviewer; Karolinska Institutet, Sweden

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Taller height and risk of coronary heart disease and cancer: a within-sibship Mendelian randomization study" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Sara Hägg (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this letter to help you prepare a revised submission.

Essential revisions:

1. A major source of bias in studies of cause and effect is selection (or collider) bias. The study also needs to consider possible selection bias from inevitably only selecting survivors to recruitment of their height (measured or genetically endowed), the disease of interest and any competing risk of the disease of interest. Whether, the sibling design, by requiring at least two siblings to survive to recruitment, is more open to selection bias than a population-based design could also be considered.

2. The study provides an interesting comparison of four different methods. Height increasing cancer risk is plausible because of the potential mechanism given, i.e., IGF1, the consistency with well accepted biological theories from evolutionary biology, and that cancer deaths tend to occur at younger ages than cardiovascular disease deaths. The results for coronary artery disease are possible but, could be overstated Greater consideration should be also be given to as to how "an individual-level causal effect" observed largely in Europeans generalizes.

3. Whether the findings for height decreasing the risk of coronary heart disease could be partly due to an artefact of survivorship bias (https://en.wikipedia.org/wiki/Survivorship_bias) arising from taller people dying of cancer before they have a chance to get coronary heart disease should be considered in more detail.

4. Please explain more clearly about adjustment for age, does it account for the fact that risk of disease goes up exponentially with age? Please also consider whether cohort is also relevant. Cohort may affect both height and disease risk. In the UK Biobank the recruitment period is quite short, so cohort should not be an issue. Please clarify over what time period the HUNT study was recruited and whether cohort should be considered as well as age.

5. Adhere to eLife title standards. Also, not clear that both observational and MR estimates are calculated on individual level data from HUNT and UKB. I think the title should be modified and make more justice to the study.

Same thing with abstract. Lots of background added but not the phenotypic associations. By the way, why call it phenotypic- the usual phrase is observational association? Biomarker results not described either.

6. Add more discussion on the CVD-cancer paradox that has been discussed in literature.

Reviewer #1 (Recommendations for the authors):

Methods

Please explain more clearly about adjustment for age, does it account for the fact that risk of disease goes up exponentially with age? Please also consider whether cohort is also relevant. Cohort may affect both height and disease risk. In the UK Biobank the recruitment period is quite short, so cohort should not be an issue. Please clarify over what time period the HUNT study was recruited and whether cohort should be considered as well as age.

Is a p-value cutoff given for SNP selection into the height PGS?

Please clarify whether prevalent cases were included as well as incident cases. Given, genetics are a lifelong exposure, it might be better to include prevalent cases, so as to avoid time related biases, as explained here[1].

Results

Please give a table describing the baseline characteristics of the two studies, particularly in terms of age and sex, and possibly recruitment period.

Discussion

Please consider the limitations of this study in more detail, specifically.

1. Using a PGS for the exposure increases power but means it is difficult to check for any pleotropic effects. How might this affect interpretation of the study?

2. This study is carefully designed to avoid use of overlapping samples for exposure and outcome. Is the study used for height unbiased?

3. The possibility of selection bias from survivorship. This study only includes people who have survived to recruitment, so it will be missing people who died before recruitment because of their height or height-related genetic endowment, the disease of interest or a competing risk of the disease of interest. Missing these people can attenuate or reverse estimates for exposures that affect survival. If cancer causes death at earlier ages than coronary heart disease then the results for coronary heart disease could be an artefact of the competing risk of previous death from cancer.

4. Please clarify whether the results presented are specific to the UK Biobank and HUNT studies or explain their generalizability and transportability.

5. Could differences between UK Biobank and HUNT be due to differences to design?

6. Are both the comparisons between the sibling and population estimates for cancer and coronary heart disease in the expected directions?

[1] Hernán MA, Sauer BC, Hernández-Díaz S, Platt R, Shrier I: Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. J Clin Epidemiol 2016, 79:70-75.

Reviewer #2 (Recommendations for the authors):

How likely is it that the earlier MR results are misleading? Biased by demographics etc? It relates to the novelty of the paper because the MR association has been shown before multiple times.

What is the rationale for doing the biomarker analysis? Would it not be more informative to investigate if they mediate the effect from height on CHD?

Table 1 should be sample characteristics in HUNT and UKB.

What about sex stratifications? Is it possible to do for same-sex sib pairs? Is it of importance in the population at large regardless of sib ship pairs? For the traits used here it should matter.

All figures and tables should contain information so that they are stand alone items.

https://doi.org/10.7554/eLife.72984.sa1

Author response

Essential revisions:

1. A major source of bias in studies of cause and effect is selection (or collider) bias. The study also needs to consider possible selection bias from inevitably only selecting survivors to recruitment of their height (measured or genetically endowed), the disease of interest and any competing risk of the disease of interest. Whether, the sibling design, by requiring at least two siblings to survive to recruitment, is more open to selection bias than a population-based design could also be considered.

We have added the possibility of selection (collider) bias to the limitations section of the discussion. We have also noted that further work is required to understand the susceptibility of family studies to selection bias.

“An additional limitation is that our study may have been susceptible to selection and survival biases relating to non-random participation in UK Biobank and/or HUNT and the requirement of at least two siblings to survive to be recruited. Indeed, cancer and coronary heart disease are both leading worldwide causes of mortality and so cases for one disease may have reduced likelihood of developing the other disease due to increased mortality. Therefore, our study may have been susceptible to survival bias relating to competing risks. We mitigated this by defining cases for both diseases using both non-fatal and fatal events. Our study analysed families with two or more siblings jointly participating in a cohort; nevertheless, further research is required to investigate the impact of selection bias on family studies.”

2. The study provides an interesting comparison of four different methods. Height increasing cancer risk is plausible because of the potential mechanism given, i.e., IGF1, the consistency with well accepted biological theories from evolutionary biology, and that cancer deaths tend to occur at younger ages than cardiovascular disease deaths. The results for coronary artery disease are possible but, could be overstated Greater consideration should be also be given to as to how "an individual-level causal effect" observed largely in Europeans generalizes.

We have added to the discussion how our conclusions are based on individuals of recent European descent and so may not generalise to non-European populations if the mechanisms are context-specific.

“Further work is required to investigate if our findings generalise to non-European populations; biological mechanisms could be expected to be largely consistent across populations but context-specific (e.g., social) mechanisms could lead to geographical heterogeneity.”

3. Whether the findings for height decreasing the risk of coronary heart disease could be partly due to an artefact of survivorship bias (https://en.wikipedia.org/wiki/Survivorship_bias) arising from taller people dying of cancer before they have a chance to get coronary heart disease should be considered in more detail.

We have added discussion of competing risks between the two diseases to the discussion.

“An additional limitation is that our study may have been susceptible to selection and survival biases relating to non-random participation in UK Biobank and/or HUNT and the requirement of at least two siblings to survive to be recruited. Indeed, cancer and coronary heart disease are both leading worldwide causes of mortality and so cases for one disease may have reduced likelihood of developing the other disease due to increased mortality. Therefore, our study may have been susceptible to survival bias relating to competing risks. We mitigated this by defining cases for both diseases using both non-fatal and fatal events. Our study analysed families with two or more siblings jointly participating in a cohort; nevertheless, further research is required to investigate the impact of selection bias on family studies.”

4. Please explain more clearly about adjustment for age, does it account for the fact that risk of disease goes up exponentially with age? Please also consider whether cohort is also relevant. Cohort may affect both height and disease risk. In the UK Biobank the recruitment period is quite short, so cohort should not be an issue. Please clarify over what time period the HUNT study was recruited and whether cohort should be considered as well as age.

We adjusted for age (birth year / birth cohort) to capture phenotypic variation in both height and the disease/biomarker outcomes (disease risk increasing with age). Age / birth year is a potential confounder of the relationship between measured height and an outcome. However, age/birth year is very unlikely to confound the relationship between height genetic variants and an outcome because these factors cannot plausibly influence genotype. Selection on height could influence allele frequencies over time leading to variation by birth year but any changes in the context of this study would be minimal.

Similarly, cohort (HUNT 2 or HUNT 3) could be a potential confounder of the phenotypic analyses but is very unlikely to confound the MR analyses. The primary focus of our manuscript is on the within-sibship MR analyses and the triangulation of evidence across different study designs. We generally found consistent effect estimates across both phenotypic and MR analyses suggesting that non-linear age and cohort effects are unlikely to have explained our results.

At the request of the reviewers, we have added Table 1 which contains information on time periods of recruitment and other study-level information.

5. Adhere to eLife title standards. Also, not clear that both observational and MR estimates are calculated on individual level data from HUNT and UKB. I think the title should be modified and make more justice to the study.

The eLife guide to authors requires titles to be less than 120 characters, gives the following example:

Predicting the effect of statins on cancer risk using genetic variants from a Mendelian randomization study in the UK Biobank (125 characters)

We have removed the colon from our title to read:

“Taller height and risk of coronary heart disease and cancer, a within-sibship Mendelian randomization study” (107 characters)

We cannot include the study names (UK Biobank and HUNT) because the title would become too long. However, we are happy to consider alternative titles.

Same thing with abstract. Lots of background added but not the phenotypic associations. By the way, why call it phenotypic- the usual phrase is observational association? Biomarker results not described either.

We previously did not mention the phenotypic associations in the abstract in the interests of word count and brevity. We have now shortened the background section and added information on the phenotypic associations.

We used “phenotypic” over “observational” to describe the non-genetic analyses because the term “observational” is somewhat misleading because genetic data is also observational. We believe “phenotypic” makes it clearer that the analysis is non-genetic and only using phenotype data. However, we are happy to modify the terminology to “observational” at the discretion of the editors.

6. Add more discussion on the CVD-cancer paradox that has been discussed in literature.

Peto’s paradox notes that cancer risk correlates with body size within a species but cancer risk between species does not correlate with body size. We have added discussion of this to the manuscript.

“Notably there is minimal evidence of a correlation between the size of an organism and cancer risk (Peto’s paradox) suggesting that the number of cells hypothesis could influence cancer risk in humans but would not explain variation in cancer risk across different organisms 63.”

Reviewer #1 (Recommendations for the authors):

Methods

Please explain more clearly about adjustment for age, does it account for the fact that risk of disease goes up exponentially with age? Please also consider whether cohort is also relevant. Cohort may affect both height and disease risk. In the UK Biobank the recruitment period is quite short, so cohort should not be an issue. Please clarify over what time period the HUNT study was recruited and whether cohort should be considered as well as age.

We adjusted for age (birth year) to capture phenotypic variation in both height and the disease/biomarker outcomes (disease risk increasing with age). Age is a potential confounder of the relationship between measured height and an outcome. However, age/birth year is very unlikely to confound the relationship between height genetic variants and an outcome because these factors cannot plausibly influence genotype. Selection on height could influence allele frequencies over time leading to variation by birth year but any changes in the context of this study would be minimal.

Similarly, cohort (HUNT 2 or HUNT 3) could be a potential confounder of the phenotypic analyses but is very unlikely to confound the MR analyses. The primary focus of our manuscript is on the within-sibship MR analyses and the triangulation of evidence across different study designs. We generally found consistent effect estimates across both phenotypic and MR analyses suggesting that non-linear age and cohort effects are unlikely to have influenced our results.

At the request of the reviewers, we have added Table 1 which contains information on time periods of recruitment and other study-level information.

Is a p-value cutoff given for SNP selection into the height PGS?

We have modified the text to clarify that the height PGS was constructed using genome-wide significant variants. “The height PGS was constructed in PLINK 52 using 372 independent (LD clumping: 250 kb, r2 < 0.01, P < 5x10-8)”

Please clarify whether prevalent cases were included as well as incident cases. Given, genetics are a lifelong exposure, it might be better to include prevalent cases, so as to avoid time related biases, as explained here[1].

We have clarified in the methods that both prevalent and incident cases were included in analyses in UK Biobank (cancer/CHD) and HUNT (CHD). Cancer in HUNT is an exception as this data was extracted from questionnaires at study baseline and so only prevalent cases were included, and there was no longitudinal follow-up.

Height is largely non-modifiable from early adulthood and height genetic variants are fixed at conception, while coronary heart disease and cancer cases (both prevalent and incident) are likely to arise decades later. Therefore, we believe that the inclusion of both prevalent and incident cases to maximise power is appropriate.

Results

Please give a table describing the baseline characteristics of the two studies, particularly in terms of age and sex, and possibly recruitment period.

We have added a table detailing the baseline characteristics of the UK Biobank and HUNT study populations to the manuscript (Table 1).

Discussion

Please consider the limitations of this study in more detail, specifically.

1. Using a PGS for the exposure increases power but means it is difficult to check for any pleotropic effects. How might this affect interpretation of the study?

Horizontal pleiotropy is a potential bias in MR studies. Here, we focused on tackling other potential biases in MR studies (population stratification, assortative mating, and indirect genetic effects) using a within-sibship design. In theory, pleiotropy can also be evaluated in within-sibship MR. However, within family analyses generally have far lower statistical power than MR using data from unrelated individuals. Other studies have investigated whether horizontal pleiotropy can explain this relationship using unrelated individuals, these studies have generally found little evidence that horizontal pleiotropy explains these relationships. For example, see: Marouli E, Del Greco MF, Astley CM, et al. Mendelian randomisation analyses find pulmonary factors mediate the effect of height on coronary artery disease. Communications biology. We have added to the limitations, that with limited power, we did not perform sensitivity analyses for horizontal pleiotropy.

“A limitation of our analyses is that because of limited sibling data and the statistical inefficiency of within-family models, we have limited statistical power to investigate effects of height on disease subtypes, to further explore mechanisms using multivariable Mendelian randomization 60 and to perform sensitivity analyses to evaluate horizontal pleiotropy.”

2. This study is carefully designed to avoid use of overlapping samples for exposure and outcome. Is the study used for height unbiased?

We used summary statistics from the GIANT-consortium height GWAS of unrelated individuals to extract information on height SNPs. These summary data are likely to be affected by population stratification, assortative and indirect genetic effects. We included only genome-wide significant variants which, although potentially affected by these sources of associations, are still likely to be valid genetic instruments for height. Indeed, we found that the height PGS constructed from these SNPs was strongly associated with height in the within-sibship model in both studies.

3. The possibility of selection bias from survivorship. This study only includes people who have survived to recruitment, so it will be missing people who died before recruitment because of their height or height-related genetic endowment, the disease of interest or a competing risk of the disease of interest. Missing these people can attenuate or reverse estimates for exposures that affect survival. If cancer causes death at earlier ages than coronary heart disease then the results for coronary heart disease could be an artefact of the competing risk of previous death from cancer.

We have added discussion of competing risks between the two diseases to the discussion.

“An additional limitation is that our study may have been susceptible to selection and survival biases relating to non-random participation in UK Biobank and/or HUNT and the requirement of at least two siblings to survive to be recruited. Indeed, cancer and coronary heart disease are both leading worldwide causes of mortality and so cases for one disease may have reduced likelihood of developing the other disease due to increased mortality. Therefore, our study may have been susceptible to survival bias relating to competing risks. We mitigated this by defining cases for both diseases using both non-fatal and fatal events. Our study analysed families with two or more siblings jointly participating in a cohort; nevertheless, further research is required to investigate the impact of selection bias on family studies.”

4. Please clarify whether the results presented are specific to the UK Biobank and HUNT studies or explain their generalizability and transportability.

We have added to the discussion how our conclusions are based on individuals of recent European descent and so may not generalise to non-European populations if the mechanisms are context-specific.

“Further work is required to investigate if our findings generalise to non-European populations; biological mechanisms could be expected to be largely consistent across populations but context-specific (e.g., social) mechanisms could lead to geographical heterogeneity.”

5. Could differences between UK Biobank and HUNT be due to differences to design?

We have added this as a possible explanation for the heterogeneity between the studies for the biomarker analyses in the discussion.

“Additional explanations could relate to differences in biomarker measurement between studies (e.g. measuring LDL-C directly or using the Friedewald formula, differences in fasting level before samples were taken), selection bias 64 or differences between the cohorts in terms of recruitment and participation.”

6. Are both the comparisons between the sibling and population estimates for cancer and coronary heart disease in the expected directions?

We believe the reviewer is referring to the estimates in Figure 2. If so, the sibling and population estimates were highly consistent in terms of directionality and magnitude for both cancer and coronary heart disease.

[1] Hernán MA, Sauer BC, Hernández-Díaz S, Platt R, Shrier I: Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. J Clin Epidemiol 2016, 79:70-75.

Reviewer #2 (Recommendations for the authors):

How likely is it that the earlier MR results are misleading? Biased by demographics etc? It relates to the novelty of the paper because the MR association has been shown before multiple times.

As the reviewer notes, several previous MR studies have provided evidence of the height cancer and height CHD relationships using data from unrelated individuals. Similarly previous MR studies of unrelated individuals also provided consistent evidence that taller height and lower BMI increase educational attainment. However, a recent within-sibship MR study (Brumpton et al. 2020 Nature Communications) found that the height/BMI on education effects completely attenuate in within-sibship models suggesting that the previous MR estimates from unrelated individuals were biased. Therefore, it was certainly possible that the previous height cancer/CHD MR estimates from unrelated individuals could also have been biased.

Furthermore, previous studies in eLife (Berg et al., Sohail et al. 2019) also reported evidence that population stratification inflated estimates of polygenic adaptation on taller height, suggesting that height may be a phenotype particularly susceptible to bias from demographic factors. This further highlights the importance of within-sibship studies for height analyses.

What is the rationale for doing the biomarker analysis? Would it not be more informative to investigate if they mediate the effect from height on CHD?

We included analyses with biomarkers relevant to cancer and coronary heart disease to determine if height-biomarker estimates were consistent with the height-disease estimates. Biomarkers are continuous so analyses would likely have higher statistical power than for the binary disease outcomes. Statistical power was limited to perform multivariable MR analyses to investigate if these biomarkers mediate the effects of height.

Table 1 should be sample characteristics in HUNT and UKB.

We have added a table of sample characteristics for UKB and HUNT as suggested in Table 1.

What about sex stratifications? Is it possible to do for same-sex sib pairs? Is it of importance in the population at large regardless of sib ship pairs? For the traits used here it should matter.

Sex stratification would be possible in theory but would substantially reduce the sample size and statistical power because it would remove all families without same-sex siblings (e.g. a sibling pair with 1 male and 1 female) and then split the remaining dataset into two. Therefore, in practice with available data, we believe that a sex-stratified analysis would be far too underpowered to detect heterogeneity.

All figures and tables should contain information so that they are stand alone items.

We have been through the manuscript and ensured that table and figure descriptions provide enough detail. In particular, we have added descriptions for Tables 1 (new) and 2 and all of the supplemental tables.

https://doi.org/10.7554/eLife.72984.sa2

Article and author information

Author details

  1. Laurence J Howe

    1. Medical Research Council Integrative Epidemiology Unit, Population Health Sciences, University of Bristol, Bristol, United Kingdom
    2. Population Health Sciences, Bristol Medical School, University of Bristol, Barley House, Oakfield Grove, Bristol, United Kingdom
    3. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Software, Writing – original draft, Writing – review and editing
    For correspondence
    laurence.howe@bristol.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2819-9686
  2. Ben Brumpton

    1. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Trondheim, Norway
    2. HUNT Research Center, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Levanger, Norway
    3. Clinic of Medicine, St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
    Contribution
    Data curation, Funding acquisition, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
  3. Humaira Rasheed

    1. Medical Research Council Integrative Epidemiology Unit, Population Health Sciences, University of Bristol, Bristol, United Kingdom
    2. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Data curation, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3331-5864
  4. Bjørn Olav Åsvold

    1. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Trondheim, Norway
    2. Department of Endocrinology, Clinic of Medicine, St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
    Contribution
    Data curation, Funding acquisition, Writing – review and editing
    Competing interests
    No competing interests declared
  5. George Davey Smith

    1. Medical Research Council Integrative Epidemiology Unit, Population Health Sciences, University of Bristol, Bristol, United Kingdom
    2. Population Health Sciences, Bristol Medical School, University of Bristol, Barley House, Oakfield Grove, Bristol, United Kingdom
    Contribution
    Funding acquisition, Methodology, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1407-8314
  6. Neil M Davies

    1. Medical Research Council Integrative Epidemiology Unit, Population Health Sciences, University of Bristol, Bristol, United Kingdom
    2. Population Health Sciences, Bristol Medical School, University of Bristol, Barley House, Oakfield Grove, Bristol, United Kingdom
    3. K.G. Jebsen Center for Genetic Epidemiology, Department of Public Health and Nursing, NTNU, Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Funding acquisition, Investigation, Methodology, Supervision, Writing – review and editing
    For correspondence
    neil.davies@bristol.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2460-0508

Funding

Norwegian Research Council (295989)

  • Neil Martin Davies

MRC Integrative Epidemiology Unit (MC_UU_00011/1)

  • Laurence J Howe
  • Neil M Davies
  • George Davey Smith

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Quality Control filtering of the UK Biobank data was conducted by R Mitchell, G Hemani, T Dudding, and L Paternoster as described in the published protocol (https://doi.org/10.5523/bris.3074krb6t2frj29yh2b03x3wxj). The University of Bristol support the MRC Integrative Epidemiology Unit (MC_UU_00011/1). NMD was supported by a Norwegian Research Council grant number 295989. The Trøndelag Health Study (The HUNT Study) is a collaboration between HUNT Research Centre (Faculty of Medicine and Health Sciences, NTNU, Norwegian University of Science and Technology), Trøndelag County Council, Central Norway Regional Health Authority, and the Norwegian Institute of Public Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This publication is the work of the authors, who serve as the guarantors for the contents of this paper.

Ethics

Human subjects: This research has been conducted using the UK Biobank Resource under Application Number 15825. UK Biobank has ethical approval from the North West Multi-centre Research Ethics Committee (MREC). All UK Biobank participants provided written informed consent. The use of HUNT data in this study was approved by the Regional Committee for Ethics in Medical Research, Central Norway (2017/2479). All HUNT study participants provided written informed consent.

Senior Editor

  1. Matthias Barton, University of Zurich, Switzerland

Reviewing Editor

  1. Edward D Janus, University of Melbourne, Australia

Reviewer

  1. Sara Hägg, Karolinska Institutet, Sweden

Publication history

  1. Preprint posted: July 19, 2021 (view preprint)
  2. Received: August 11, 2021
  3. Accepted: March 9, 2022
  4. Accepted Manuscript published: March 18, 2022 (version 1)
  5. Version of Record published: March 24, 2022 (version 2)

Copyright

© 2022, Howe et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 478
    Page views
  • 66
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Laurence J Howe
  2. Ben Brumpton
  3. Humaira Rasheed
  4. Bjørn Olav Åsvold
  5. George Davey Smith
  6. Neil M Davies
(2022)
Taller height and risk of coronary heart disease and cancer: A within-sibship Mendelian randomization study
eLife 11:e72984.
https://doi.org/10.7554/eLife.72984

Further reading

    1. Epidemiology and Global Health
    Catherine Meh, Prabhat Jha
    Research Article

    Preference for sons and smaller families and, in the case of China, a one-child policy, have contributed to missing girl births in India and China over the last few decades due to sex-selective abortions. Selective abortion occurs also among Indian and Chinese diaspora, but their variability and trends over time are unknown. We examined conditional sex ratio (CSR) of girl births per 1000 boy births among second or third births following earlier daughters or sons in India, China, and their diaspora in Australia, Canada, United Kingdom (UK), and United States (US) drawing upon 18.4 million birth records from census and nationally representative surveys from 1999 to 2019. Among Indian women, the CSR in 2016 for second births following a first daughter favoured boys in India (866), similar to those in diaspora in Australia (888) and Canada (882). For third births following two earlier daughters in 2016, CSRs favoured sons in Canada (520) and Australia (653) even more than in India (769). Among women in China outside the one-child restriction, CSRs in 2015 for second order births somewhat favoured more girls after a first son (1154) but more heavily favoured boys after a first daughter (561). Third-birth CSRs generally fell over time among diaspora, except among Chinese diaspora in the UK and US. In the UK, third-birth CSRs fell among Indian but not among other South Asian diasporas. Selective abortion of girls is notable among Indian diaspora, particularly at higher-order births.

    1. Epidemiology and Global Health
    Yupeng Liu, Siyu Le ... Shuran Wang
    Research Article

    Background:

    The effect of calcium supplementation on bone mineral accretion in people under 35 years old is inconclusive. To comprehensively summarize the evidence for the effect of calcium supplementation on bone mineral accretion in young populations (≤35 years).

    Methods:

    This is a systematic review and meta-analysis. The Pubmed, Embase, ProQuest, CENTRAL, WHO Global Index Medicus, Clinical Trials.gov, WHO ICTRP, China National Knowledge Infrastructure (CNKI), and Wanfang Data databases were systematically searched from database inception to April 25, 2021. Randomized clinical trials assessing the effects of calcium supplementation on bone mineral density (BMD) or bone mineral content (BMC) in people under 35 years old.

    Results:

    This systematic review and meta-analysis identified 43 studies involving 7,382 subjects. Moderate certainty of evidence showed that calcium supplementation was associated with the accretion of BMD and BMC, especially on femoral neck (standardized mean difference [SMD] 0.627, 95% confidence interval [CI] 0.338–0.915; SMD 0.364, 95% CI 0.134–0.595; respectively) and total body (SMD 0.330, 95% CI 0.163–0.496; SMD 0.149, 95% CI 0.006–0.291), also with a slight improvement effect on lumbar spine BMC (SMD 0.163, 95% CI 0.008–0.317), no effects on total hip BMD and BMC and lumbar spine BMD were observed. Very interestingly, subgroup analyses suggested that the improvement of bone at femoral neck was more pronounced in the peripeak bone mass (PBM) population (20–35 years) than the pre-PBM population (<20 years).

    Conclusions:

    Our findings provided novel insights and evidence in calcium supplementation, which showed that calcium supplementation significantly improves bone mass, implying that preventive calcium supplementation before or around achieving PBM may be a shift in the window of intervention for osteoporosis.

    Funding:

    This work was supported by Wenzhou Medical University grant [89219029].