Reply to “Commentary: ‘No Genetic Causality between Tobacco Smoking and Venous Thromboembolism: A Two-Sample Mendelian Randomization Study’ ”

We recently received comments on our findings[1] from Liu et al for which we are grateful, and we appreciate their careful scholarship and thoughtfulness. We would like to respond to the comments of Liu et al as follows:

First, Liu et al argue that a major issue in our study is the failure to account for sample overlap. However, as pointed out by Liu et al, the bias due to sample overlap is mainly due to false-positive errors, whereas our findings are not significant. However, we also replaced venous thromboembolism (VTE) data from different data sources ([Table 1])[2] for validation in this response, and the results showed that our conclusions were stable ([Table 2]).

Table 1 Summary information on the included exposure/outcome phenotypes

Phenotypic name	No. of cases	No. of controls	Sample size	Ancestry	Sex	Consortium (cohort)/author	Public release year	Study reference or description URL
Age of initiation of regular smoking	NA	NA	728,826	Mixed (84.35%European)	Both sexes	GSCAN/Gretchen R. B. Saunders et al[3]	2022	Nature. 2022 Dec;612(7941):720–724. doi: 10.1038/s41586-022-05477-4.
Ever smoked regularly	NA	NA	3,383,199	Mixed (79.11%European)	Both sexes	GSCAN/Gretchen R. B. Saunders et al[3]	2022	Nature. 2022 Dec;612(7941):720–724. doi: 10.1038/s41586-022-05477-4.
Cigarettes per day	NA	NA	784,353	Mixed (79.03%European)	Both sexes	GSCAN/Gretchen R. B. Saunders et al[3]	2022	Nature. 2022 Dec;612(7941):720–724. doi: 10.1038/s41586-022-05477-4.
Smoking cessation	NA	NA	1,400,535	Mixed (82.26%European)	Both sexes	GSCAN/Gretchen R. B. Saunders et al[3]	2022	Nature. 2022 Dec;612(7941):720–724. doi: 10.1038/s41586-022-05477-4.
Venous thromboembolism	21,021	391,160	412,181	European	Both sexes	FinnGen.R10/Mitja I. Kurki et al[2]	2023	https://www.finngen.fi/en
Deep venous thrombosis	6,501	357,111	363,612	European	Both sexes	FinnGen.R10/Mitja I. Kurki et al[2]	2023	https://www.finngen.fi/en
Pulmonary embolism	10,046	401,128	411,174	European	Both sexes	FinnGen.R10/Mitja I. Kurki et al[2]	2023	https://www.finngen.fi/en

Table 2 Associations of smoking-related phenotypes with VTE (including its subtypes) estimated by five MR methods

Outcome	Exposure	Method	No. of SNPs	Beta	SE	p-Value	OR (95% CI)
Venous thromboembolism	Age of initiation of regular smoking	Inverse variance weighted	6	0.332	0.253	0.190	1.393 (0.848, 2.289)
		MR Egger	6	[minus]0.043	0.630	0.949	0.958 (0.279, 3.291)
		Weighted median	6	0.124	0.285	0.663	1.132 (0.648, 1.979)
		Weighted mode	6	0.054	0.383	0.894	1.055 (0.498, 2.235)
		Simple mode	6	[minus]0.051	0.449	0.913	0.950 (0.394, 2.290)
	Ever smoked regularly	Inverse variance weighted	254	0.099	0.053	0.064	1.104 (0.994, 1.225)
		MR Egger	254	0.360	0.193	0.063	1.433 (0.982, 2.092)
		Weighted median	254	0.108	0.070	0.122	1.114 (0.971, 1.278)
		Weighted mode	254	0.136	0.210	0.517	1.146 (0.759, 1.729)
		Simple mode	254	0.167	0.221	0.451	1.182 (0.766, 1.822)
	Cigarettes per day	Inverse variance weighted	61	0.116	0.065	0.076	1.123 (0.988, 1.276)
		MR Egger	61	0.065	0.108	0.548	1.067 (0.864, 1.319)
		Weighted median	61	0.022	0.076	0.769	1.022 (0.881, 1.186)
		Weighted mode	61	0.047	0.067	0.481	1.048 (0.920, 1.194)
		Simple mode	61	0.225	0.155	0.152	1.252 (0.924, 1.696)
	Smoking cessation	Inverse variance weighted	18	[minus]0.030	0.108	0.785	0.971 (0.785, 1.201)
		MR Egger	18	0.151	0.316	0.639	1.163 (0.627, 2.159)
		Weighted median	18	[minus]0.073	0.157	0.641	0.930 (0.684, 1.263)
		Weighted mode	18	0.003	0.191	0.989	1.003 (0.690, 1.457)
		Simple mode	18	0.024	0.245	0.922	1.025 (0.633, 1.657)
	Past tobacco smoking	Inverse variance weighted	95	[minus]0.144	0.084	0.089	0.866 (0.734, 1.022)
		MR Egger	95	[minus]0.450	0.348	0.200	0.638 (0.322, 1.263)
		Weighted median	95	[minus]0.091	0.097	0.345	0.913 (0.755, 1.103)
		Weighted mode	95	0.041	0.224	0.857	1.041 (0.671, 1.616)
		Simple mode	95	0.008	0.231	0.973	1.008 (0.641, 1.585)
	Current tobacco smoking	Inverse variance weighted	35	0.188	0.299	0.530	1.206 (0.671, 2.168)
		MR Egger	35	[minus]0.593	1.114	0.598	0.553 (0.062 ,4.902)
		Weighted median	35	0.054	0.375	0.885	1.056 (0.507, 2.200)
		Weighted mode	35	[minus]0.087	0.589	0.883	0.916 (0.289, 2.908)
		Simple mode	35	0.248	0.709	0.729	1.281 (0.319, 5.143)
Deep venous thrombosis	Age of initiation of regular smoking	Inverse variance weighted	6	0.988	0.482	0.041	2.685 (1.043, 6.910)
		MR Egger	6	1.759	1.188	0.213	5.809 (0.566, 59.662)
		Weighted median	6	1.485	0.469	0.002	4.417 (1.762, 11.073)
		Weighted mode	6	1.572	0.574	0.041	4.817 (1.562, 14.848)
		Simple mode	6	1.499	0.693	0.083	4.478 (1.151, 17.416)
	Ever smoked regularly	Inverse variance weighted	254	0.204	0.093	0.028	1.226 (1.023, 1.470)
		MR Egger	254	0.440	0.337	0.193	1.553 (0.802, 3.006)
		Weighted median	254	0.214	0.125	0.087	1.239 (0.969, 1.584)
		Weighted mode	254	0.483	0.325	0.139	1.620 (0.857, 3.064)
		Simple mode	254	0.539	0.389	0.167	1.714 (0.800, 3.671)
	Smoking cessation	Inverse variance weighted	18	[minus]0.104	0.190	0.586	0.901 (0.621, 1.309)
		MR Egger	18	0.679	0.546	0.232	1.972 (0.676, 5.755)
		Weighted median	18	0.049	0.250	0.844	1.050 (0.643, 1.715)
		Weighted mode	18	0.120	0.348	0.733	1.128 ( 0.571, 2.229)
		Simple mode	18	0.047	0.428	0.915	1.048 (0.453, 2.425)
	Cigarettes per day	Inverse variance weighted	61	0.065	0.085	0.441	1.067 (0.904, 1.260)
		MR Egger	61	[minus]0.062	0.139	0.657	0.940 (0.715, 1.235)
		Weighted median	61	[minus]0.099	0.125	0.430	0.906 (0.709, 1.158)
		Weighted mode	61	[minus]0.036	0.118	0.762	0.965 (0.766, 1.216)
		Simple mode	61	0.389	0.275	0.161	1.476 (0.862, 2.529)
	Past tobacco smoking	Inverse variance weighted	95	[minus]0.159	0.120	0.184	0.853 (0.675, 1.078)
		MR Egger	95	[minus]0.307	0.496	0.537	0.736 (0.278, 1.945)
		Weighted median	95	[minus]0.150	0.165	0.363	0.861 (0.623, 1.189)
		Weighted mode	95	[minus]0.173	0.360	0.631	0.841 (0.415 ,1.704)
		Simple mode	95	[minus]0.447	0.442	0.314	0.640 (0.269 ,1.520)
	Current tobacco smoking	Inverse variance weighted	35	[minus]0.182	0.418	0.664	0.834 (0.367 ,1.892)
		MR Egger	35	[minus]0.523	1.543	0.737	0.593 (0.029 ,12.211)
		Weighted median	35	0.405	0.580	0.485	1.500 (0.481 ,4.677)
		Weighted mode	35	0.747	1.133	0.514	2.111 (0.229 ,19.464)
		Simple mode	35	0.888	1.279	0.492	2.430 (0.198 ,29.813)
Pulmonary embolism	Age of initiation of regular smoking	Inverse variance weighted	6	0.307	0.292	0.292	1.360 (0.768 ,2.408)
		MR Egger	6	0.265	0.762	0.745	1.304 (0.293 ,5.805)
		Weighted median	6	0.238	0.386	0.538	1.269 (0.595 ,2.706)
		Weighted mode	6	0.273	0.476	0.591	1.314 (0.517 ,3.340)
		Simple mode	6	0.157	0.562	0.792	1.170 (0.388 ,3.521)
	Ever smoked regularly	Inverse variance weighted	254	0.169	0.071	0.017	1.184 (1.031 ,1.359)
		MR Egger	254	0.460	0.256	0.073	1.584 (0.959 ,2.617)
		Weighted median	254	0.167	0.100	0.095	1.181 (0.972 ,1.436)
		Weighted mode	254	0.454	0.281	0.108	1.574 (0.907 ,2.733)
		Simple mode	254	0.454	0.318	0.155	1.574 (0.843 ,2.939)
	Smoking cessation	Inverse variance weighted	18	0.094	0.154	0.543	1.098 (0.812 ,1.486)
		MR Egger	18	0.115	0.442	0.799	1.121 (0.472 ,2.667)
		Weighted median	18	0.218	0.211	0.303	1.243 (0.822 ,1.880)
		Weighted mode	18	0.209	0.244	0.404	1.232 (0.763 ,1.990)
		Simple mode	18	0.144	0.305	0.642	1.155 (0.635 ,2.102)
	Cigarettes per day	Inverse variance weighted	61	0.119	0.093	0.200	1.127 (0.939 ,1.352)
		MR Egger	61	0.137	0.154	0.377	1.147 (0.848 ,1.552)
		Weighted median	61	0.038	0.104	0.714	1.039 (0.848 ,1.273)
		Weighted mode	61	0.014	0.096	0.885	1.014 (0.840 ,1.223)
		Simple mode	61	[minus]0.015	0.253	0.953	0.985 (0.599 ,1.619)
	Past tobacco smoking	Inverse variance weighted	95	[minus]0.051	0.113	0.652	0.950 (0.761 ,1.186)
		MR Egger	95	[minus]0.061	0.469	0.898	0.941 (0.375 ,2.361)
		Weighted median	95	0.125	0.140	0.374	1.133 (0.861 ,1.490)
		Weighted mode	95	0.343	0.332	0.303	1.410 (0.736 ,2.700)
		Simple mode	95	0.225	0.389	0.564	1.252 (0.584 ,2.685)
	Current tobacco smoking	Inverse variance weighted	35	0.675	0.421	0.109	1.963 (0.860 ,4.480)
		MR Egger	35	[minus]0.490	1.566	0.757	0.613 (0.028 ,13.205)
		Weighted median	35	0.258	0.502	0.607	1.295 (0.484 ,3.464)
		Weighted mode	35	[minus]0.086	0.868	0.921	0.917 (0.167 ,5.031)
		Simple mode	35	1.307	1.038	0.217	3.694 (0.483 ,28.267)

Abbreviation: MR, Mendelian randomization.

Second, Liu et al identified a second major issue in this study is the choice of data, including the use of outdated datasets and an incomplete consideration of the exposure phenotype. In response to their point about outdated datasets, we employed the Round 10 VTE GWAS (genome-wide association study) summary statistics released in 2023 by FinnGen in our current response, and it is clear that the results ([Table 2]) are also consistent with our previous findings.

Incomplete consideration of smoking phenotypes, which we must recognize as correct, does not affect our results. Please note that we also included this time four smoking phenotypes ([Table 1])[3] published by the GWAS and Sequencing Consortium of Alcohol and Nicotine use (GSCAN) in 2022, and the results showed that the association of all six smoking phenotypes with VTE was not statistically significant ([Table 2]). As for their subtypes, we note that the inverse variance-weighted (IVW) method suggests possible age of initiation of regular smoking–DVT (deep vein thrombosis), ever smoked regularly–DVT, and ever smoked regularly–PE (pulmonary embolism) associations ([Table 2]), but we must point out that the age of initiation of regular smoking–DVT association suffers from too small several instrumental variables and large differences in the estimates of the five methods. As for the ever smoked regularly–DVT and ever smoked regularly–PE associations, none of the methods were significant except for the IVW method. Moreover, considering that the associations of all six smoking phenotypes with VTE were not statistically significant, the significant results in their subtypes should be treated with caution.

In addition, and also due to the third question posed to us by Liu et al, the instrumental variables addressed in our current response did not undergo the exclusion of single nucleotide polymorphisms (SNPs) associated with confounders to control for horizontal pleiotropy as described in our previous studies.[1] For the conflicting association results in VTE and its subtypes, it is likely that bias due to SNPs associated with confounders was attributed.

Publication History

Received: 30 July 2024

Accepted: 15 September 2024

Article published online:
03 October 2024

• References

• 1 Du HC, Zheng YF, Shen MQ, Deng BY. No genetic causality between tobacco smoking and venous thromboembolism: a two-sample Mendelian randomization study. Thromb Haemost 2024; 124 (08) 795-802
• 2 Kurki MI, Karjalainen J, Palta P. et al; FinnGen. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023; 613 (7944) 508-518
• 3 Saunders GRB, Wang X, Chen F. et al; 23andMe Research Team, Biobank Japan Project. Genetic diversity fuels gene discovery for tobacco and alcohol use. Nature 2022; 612 (7941) 720-724