1

Active cigarette smoking is associated with an exacerbation

of genetic susceptibility to diabetes

Wan-Yu Lin 1,2

*, Yu-Li Liu 3, Albert C. Yang

4,5, Shih-Jen Tsai

5,6,7, Po-Hsiu Kuo

1,2*

1

Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan

University, Taipei, Taiwan

2 Department of Public Health, College of Public Health, National Taiwan University, Taipei,

Taiwan 3 Center for Neuropsychiatric Research, National Health Research Institutes, Miaoli, Taiwan

4 Division of Interdisciplinary Medicine and Biotechnology, Beth Israel Deaconess Medical

Center/Harvard Medical School, Boston, MA, USA 5 Institute of Brain Science, National Yang-Ming University, Taipei, Taiwan

6 Division of Psychiatry, National Yang-Ming University, Taipei, Taiwan

7 Department of Psychiatry, Taipei Veterans General Hospital, Taipei, Taiwan

Short title: Smoking and genetic risk of diabetes

* Corresponding authors: Po-Hsiu Kuo, Ph.D. and Wan-Yu Lin, Ph.D.

Po-Hsiu Kuo, Ph.D. (http://orcid.org/0000-0003-0365-3587)

Room 521, No. 17, Xu-Zhou Road, Taipei 100, Taiwan

Phone: +886-2-33668015; Fax: +886-2-23511955; E-mail: phkuo@ntu.edu.tw

Wan-Yu Lin, Ph.D. (http://orcid.org/0000-0002-3385-4702)

Room 501, No. 17, Xu-Zhou Road, Taipei 100, Taiwan

Phone/Fax: +886-2-33668106; E-mail: linwy@ntu.edu.tw

2

Tweet (Figure 1 can be included in Tweet)

Active cigarette smoking is associated with an exacerbation of genetic susceptibility to

diabetes. This study included a discovery cohort (TWB1) of 25,460 and a replication cohort

(TWB2) of 58,774 unrelated Taiwan Biobank (TWB) subjects. Genetic risk score (GRS) of each

TWB2 subject was calculated with weights retrieved from TWB1 analyses. We then assessed the

significance of GRS-smoking interactions on FG/HbA1c/diabetes while adjusting for covariates.

A total of 5 smoking measurements were investigated respectively, including “active smoking

status”, “pack-years”, “years as a smoker”, “packs smoked per day”, and “hours as a passive

smoker per week”. Except passive smoking, all smoking measurements were associated with

FG/HbA1c/diabetes (P < 0.0033) and were associated with an exacerbation of the genetic risk of

FG/HbA1c (𝑃𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 < 0.0033). For example, each 1 standard deviation increase in GRS is

associated with a 1.68% higher FG in subjects consuming one more pack of cigarettes per day

(𝑃𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 = 1.9 × 10−7). Figure 1 shows that the GRS effects on FG/HbA1c/diabetes were

larger in smokers than in non-smokers. One of the mechanisms behind gene-smoking interactions

is epigenetics. When one is smoking, carbon monoxide displaces oxygen from his/her

hemoglobin. This will decrease the amount of oxygen available for delivery to many tissues.

Smoking can therefore affect one’s entire body including cardiovascular and circulatory systems.

We analyzed the blood DNA methylation data of 2,091 TWB subjects, and found cigarette

consumption is linked to differential DNA methylation of some diabetes susceptibility genes,

such as the KCNQ1 gene (potassium voltage-gated channel subfamily Q member 1). Through its

link with differential DNA methylation, smoking can regulate gene expressions and lead to more

damage to people who are more genetically predisposed to diabetes.

3

Abstract The heritability levels of two traits for diabetes diagnosis, fasting serum glucose (FG) and

glycated hemoglobin (HbA1c), were estimated to be 51% ~ 62%. Studies have shown that

cigarette smoking is a modifiable risk factor for diabetes. It is important to uncover whether

smoking may modify the genetic risk of diabetes. This study included a discovery cohort (TWB1)

of 25,460 and a replication cohort (TWB2) of 58,774 unrelated Taiwan Biobank subjects. Genetic

risk score (GRS) of each TWB2 subject was calculated with weights retrieved from TWB1

analyses. We then assessed the significance of GRS-smoking interactions on FG/HbA1c/diabetes

while adjusting for covariates. A total of 5 smoking measurements were investigated respectively,

including “active smoking status”, “pack-years”, “years as a smoker”, “packs smoked per day”,

and “hours as a passive smoker per week”. Except passive smoking, all smoking measurements

were associated with FG/HbA1c/diabetes (P < 0.0033) and were associated with an exacerbation

of the genetic risk of FG/HbA1c (𝑃𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 < 0.0033). For example, each 1 standard deviation

increase in GRS is associated with a 1.68% higher FG in subjects consuming one more pack of

cigarettes per day (𝑃𝐼𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 = 1.9 × 10−7). Smoking cessation is especially important for

people who are more genetically predisposed to diabetes.

Keywords: Gene-smoking interactions, Gene-environment interactions, Genetic risk score,

Polygenic risk score.

Introduction Diabetes have influenced hundreds of millions people around the world, and its prevalence

is gradually increasing (1). Genetic factors play an important role in the development of diabetes.

4

Fasting glucose (FG) and glycated hemoglobin (HbA1c) are commonly used for the diagnosis of

diabetes. An FG level > 126 mg/dL (2) and an HbA1c level > 6.5% (48 mmol/mol) (3) have been

proposed as diagnostic indicators for diabetes. The heritability levels of FG and HbA1c were

estimated to be 51% and 62%, respectively (4). The remaining variability in FG and HbA1c

could be explained by lifestyle factors.

Smoking has been found to be associated with a higher risk of insulin resistance and

diabetes (5). Nicotine is thought to be responsible for the association between cigarette smoking

and the development of diabetes (6). However, smoking has not been uniformly regarded as a

risk factor of diabetes (7). Moreover, it is important to uncover whether smoking may modify the

genetic risk of diabetes.

To know whether the genetic risk of diabetes may vary with cigarette smoking, we here

investigated “gene-smoking interactions” (G×S) on FG, HbA1c, and the dichotomous diabetes

status, respectively. Our Taiwan Biobank (TWB) data included a discovery cohort (TWB1) and a

replication cohort (TWB2). Genetic risk score (GRS) of each TWB2 subject was calculated with

weights retrieved from TWB1 analyses. We tested whether the GRS effects could be modified by

“active smoking status”, “the number of pack-years”, “years as a smoker”, “packs smoked per

day”, or “hours as a passive smoker per week”. This study aims to uncover whether smoking can

modulate the genetic predisposition to diabetes, and which smoking measurement is the most

critical effect modifier.

Research Design and Methods

Taiwan Biobank

The TWB keeps collecting genomic and lifestyle information from Taiwan residents aged

5

30 to 70 years (8). To participate in the TWB, community-based volunteers signed informed

consent, took physical examinations, and provided blood and urine samples. Their lifestyle

factors were further collected through a face-to-face interview with TWB researchers. TWB

received ethical approval from the Ethics and Governance Council of Taiwan Biobank, Taiwan,

and also from the Institutional Review Board on Biomedical Science Research/IRB-BM,

Academia Sinica, Taiwan. Our study was approved by the Research Ethics Committee of

National Taiwan University Hospital (NTUH-REC no. 201805050RINB).

This study included 27,737 and 67,512 subjects who have been whole-genome genotyped by

the TWB1 and TWB2 genotyping arrays until February, 2020, respectively. To explore cryptic

relatedness, we used PLINK 1.9 (9) to estimate PI-HAT = Probability(IBD = 2) + 0.5

Probability(IBD = 1), where IBD is the genome-wide identity by descent (IBD) sharing

coefficients between any two TWB individuals. Similar to many genetic studies (10; 11), we also

excluded relatives by removing one subject from every pair with PI-HAT 0.2. A total of

25,460 unrelated TWB1 subjects and 58,774 unrelated TWB2 subjects remained after this quality

control process.

Most TWB subjects were of Han Chinese ancestry (8). Released in April 2013, the TWB1

genotyping array was designed for Taiwan’s Han Chinese and run on the Axiom Genome-Wide

Array Plate System (Affymetrix, Santa Clara, CA, USA). Based on user experience in TWB1 and

next-generation sequencing of ~1,000 TWB individuals, the TWB2 genotyping array was

released in August 2018 to further cover specific SNPs in Taiwan population. A total of 632,172

and 648,611 autosomal SNPs were genotyped in TWB1 and TWB2 arrays, respectively. In

TWB1, we removed 27,628 SNPs with genotyping rates < 95%, and 6,900 SNPs with

Hardy-Weinberg test P-values < 5.7 × 10−7 (12). In TWB2, we removed 27,859 SNPs with

6

genotyping rates < 95%, and 14,656 SNPs with Hardy-Weinberg test P-values < 5.7 × 10−7

(12). Finally, 597,644 TWB1 SNPs and 606,096 TWB2 SNPs were kept in our analysis and were

used to construct ancestry principal components (PCs). A total of 92,052 SNPs were overlapped

between these two arrays.

We imputed the genotypes of autosomal SNPs using the Michigan Imputation Server

(https://imputationserver.sph.umich.edu/index.html), with the reference panel based on the East

Asian (EAS) population from the 1000 Genomes Phase 3 v5. After removing SNPs with

Hardy-Weinberg test P-values < 5.7 × 10−7 (12) and SNPs with low imputation information

score (Rsq < 0.8), 7,433,014 and 6,347,468 SNPs remained in TWB1 and TWB2, respectively.

Three diabetes-related traits

We here analyzed two continuous traits related to the diagnosis of diabetes, FG and HbA1c.

After a minimum 6-hour fast (no calorie intake for at least 6 hours), serum glucose was measured

using a Hitachi LST008 analyzer (Hitachi High-Technologies, Tokyo, Japan), whereas HbA1c

was measured with the Trinity Biotech Premier Hb9210 analyzer (Bray, Ireland/Kansas City, MO,

USA). Same as the criterion used worldwide (2; 3), the Ministry of Health and Welfare in Taiwan

defined an FG level higher than 126 mg/dL or an HbA1c level higher than 6.5% (48 mmol/mol)

as an indicator of diabetes.

Moreover, diabetes status was also analyzed as a dichotomous trait, where subjects with

diabetes included those with physician-diagnosed diabetes, or those having FG > 126 mg/dL or

HbA1c > 6.5 % (48 mmol/mol) according to the TWB test results.

Definition of smoking, drinking, regular exercise, and educational attainment

In TWB, “active smoker” was defined as a subject who had actively smoked for at least 6

7

months and had not quit smoking at the time his/her FG and HbA1c were measured. Moreover,

TWB researchers also asked every active smoker “how many packs of cigarettes smoked per day”

and “how many years as an active smoker”. “The number of pack-years” was then calculated as

multiplying “the number of packs smoked per day” by “years as a smoker”. Furthermore, TWB

researchers asked each individual “the number of hours as a passive smoker per week”. In total, 5

smoking measurements will be investigated in the following analyses.

Sex, age, body mass index (BMI), alcohol drinking status, regular exercise, and educational

attainment were regarded as covariates and adjusted in all of our regression analyses. Drinking

was defined as a subject having a weekly intake of more than 150 cc of alcohol for at least 6

months and having not stopped drinking at the time his/her FG and HbA1c were assessed.

Regular exercise was defined as engaging in 30 minutes of “exercise” three times a week.

“Exercise” indicates leisure-time activities such as jogging, mountain climbing, yoga, etc (13).

Educational attainment was recorded through a face-to-face interview with TWB researchers.

It was represented by a number ranging from 1 to 7, where 1 indicates “illiterate”, 2 means “no

formal education but literate”, 3 denotes “primary school graduate”, 4 represents “junior high

school graduate”, 5 indicates “senior high school graduate”, 6 means “college graduate”, and 7

denotes “Master’s or higher degree”. Similar to a recent study for diabetes (14), we also

considered “educational attainment” as an covariate in all analyses.

Association between smoking and FG/HbA1c/diabetes

Initially, to test whether smoking is significantly associated with FG/HbA1c/diabetes, we

regressed each trait according to the following model:

𝑔[𝐸(𝑌)] = 𝛽0 + 𝛽𝑆𝑚𝑜𝑘𝑖𝑛𝑔𝑆𝑚𝑜𝑘𝑖𝑛𝑔 + ∑ 𝛽𝐶𝑣𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣16𝑣=1 , (1)

where Y is natural log transformed FG (or HbA1c) with an identity link 𝑔[∙], or the diabetes

8

status with a logit link 𝑔[∙]. Smoking is coded as 1 for active smokers and 0 otherwise.

Sex, age, BMI, educational attainment, and ancestry PCs are commonly adjusted in models

for FG/HbA1c/diabetes (14; 15). Regular physical exercise has been shown to reduce the risk of

insulin resistance, metabolic syndrome and diabetes (16). Effects of cigarette smoking and

alcohol consumption on diabetes are usually investigated (5-7; 17). Therefore, a total of 16

covariates were adjusted in model (1), including sex, age (in years), BMI, drinking status (yes vs.

no), regular exercise (yes vs. no), educational attainment (a value ranging from 1 to 7), and the

first 10 ancestry PCs.

Because FG and HbA1c were skewed to the right, both traits were natural log transformed to

improve the R-square of model (1). Such a transformation was commonly used in analyses for

FG and HbA1c (18; 19). Based on model (1), we later replaced “active smoking status” with the

4 continuous smoking measurements, respectively.

Genetic risk score (GRS)

To build trait-specific GRS, we first identified FG-associated SNPs, HbA1c-associated

SNPs, and diabetes-associated SNPs based on the 25,460 unrelated TWB1 subjects. Each trait

(denoted by Y) was regressed on every SNP while adjusting for the 16 covariates, as follows:

𝑔[𝐸(𝑌)] = 𝛽0 + 𝛽𝑆𝑁𝑃,𝑗𝑆𝑁𝑃𝑗 +∑ 𝛽𝐶𝑣𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣16𝑣=1 , 𝑗 = 1,⋯ , 7433014. (2)

𝑃𝑆𝑁𝑃,𝑗, the p-value of testing 𝐻0: 𝛽𝑆𝑁𝑃,𝑗 = 0 𝑣𝑠. 𝐻1: 𝛽𝑆𝑁𝑃,𝑗 ≠ 0 from model (2), calibrates the

significance of marginal association of SNP j with Y. Because of ~8 million SNPs in TWB1, SNP

j is significantly associated with a trait if 𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10−9 =

0.05

8×106. We calculated GRS for

each of the 58,774 unrelated TWB2 subjects by

𝐺𝑅𝑆′ = ∑ 𝐼(𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10−9)�̂�𝑆𝑁𝑃,𝑗𝑆𝑁𝑃𝑗

7433014𝑗=1 , (3)

9

where �̂�𝑆𝑁𝑃,𝑗 (𝑗 = 1,⋯ , 7433014) had been estimated from model (2), and 𝑆𝑁𝑃𝑗 is the

number of minor alleles at the jth

SNP (0, 1, or 2) for the TWB2 individuals. The indicator

function 𝐼(𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10−9) is 1 if 𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10

−9 and 0 otherwise.

Because 𝐺𝑅𝑆′ is composed of SNPs that are more associated with FG/HbA1c/diabetes

(satisfying 𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10−9), this is the so-called “marginal-association filtering” in

gene-environment interaction analyses (20-23). To avoid collinearity in a GRS, we used the

default setting of the clumping procedure in PLINK 1.9 (9). To be specific, if there were multiple

variants with 𝑃𝑆𝑁𝑃,𝑗 < 6.25 × 10−9 within a 250-kb region, we kept the most significant one

(so-called “index variant”) and removed others in linkage disequilibrium with the index variant

(r2 > 0.5). To quantify how many standard deviations (sds) a 𝐺𝑅𝑆′ is from the mean, we

performed the z-score transformation on 𝐺𝑅𝑆′ and then obtained 𝐺𝑅𝑆. By fitting model (2) for

FG/HbA1c/diabetes separately, we obtained 𝐺𝑅𝑆 for FG/HbA1c/diabetes, respectively.

𝑆𝑁𝑃𝑗 is the number of minor alleles at the jth

SNP (0, 1, or 2). A positive �̂�𝑆𝑁𝑃,𝑗 indicates

that the minor allele is trait-increasing, and a subject with more copies of the minor allele (more

trait-increasing alleles) will receive an increment in his/her 𝐺𝑅𝑆. In contrast, a negative �̂�𝑆𝑁𝑃,𝑗

represents that the minor allele is trait-decreasing, and a subject with more copies of the minor

allele (more trait-decreasing alleles) will obtain a decrement in his/her 𝐺𝑅𝑆. Finally, a higher

𝐺𝑅𝑆 is associated with a larger FG/HbA1c/diabetes.

European-derived GRS

In addition to the “TWB1-derived GRS” (for simplicity, abbreviated as “GRS”), we also

calculated a “European-derived GRS” (abbreviated as “EuGRS”), based on 38

diabetes-associated SNPs that were previously identified from genome-wide association studies

10

(GWASs) (39-42). Most of these 38 SNPs were identified at the commonly-used genome-wide

significance level (𝑃𝑆𝑁𝑃 < 5 × 10−8). Individuals of those GWASs (39-42) were overwhelmingly

of European descent. The European-derived GRS was calculated as 𝐸𝑢𝐺𝑅𝑆 = ∑ 𝑤𝑗𝑆𝑁𝑃𝑗38𝑗=1 ,

where the weights (𝑤𝑗 , 𝑗 = 1,⋯ ,38) were the effect sizes reported by previous GWASs (39-42)

and summarized by Said et al. (14).

To present an analysis parallel to GRS-smoking interaction analysis (where TWB1 was used

to find trait-associated SNPs and TWB2 was used to test for interactions), EuGRS-smoking

interaction analysis was also performed on the TWB2 cohort with 58,774 subjects. There is one

more reason why we did not include the TWB1 cohort into the EuGRS-smoking interaction

analysis. While 32 out of the 38 SNPs were genotyped by the TWB2 array, only 27 were

genotyped by the TWB1 array. After imputation, still 2 SNPs in TWB1 failed to achieve the

criterion of imputation information score (i.e., their Rsq < 0.8). Therefore, we did not compute

EuGRS for the 25,460 TWB1 subjects. To sum up, both GRS-smoking interaction and

EuGRS-smoking interaction were evaluated in the TWB2 cohort with 58,774 subjects.

GRS-smoking interactions

We then used 𝐺𝑅𝑆 to test for the presence of G×S. The following model was considered

for each trait, respectively:

𝑔[𝐸(𝑌)] =

𝜙0 + 𝜙𝐺𝑅𝑆𝐺𝑅𝑆 + 𝜙𝑆𝑚𝑜𝑘𝑖𝑛𝑔𝑆𝑚𝑜𝑘𝑖𝑛𝑔 + 𝜙𝐼𝑁𝑇𝐺𝑅𝑆 × 𝑆𝑚𝑜𝑘𝑖𝑛𝑔 + ∑ 𝜙𝐶𝑣𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣16𝑣=1 +

∑ 𝜙𝐺𝐶𝑣16𝑣=1 𝐺𝑅𝑆 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣 + ∑ 𝜙𝑆𝐶𝑣

16𝑣=1 𝑆𝑚𝑜𝑘𝑖𝑛𝑔 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣, (4)

where Y is natural log transformed FG/HbA1c or the diabetes status, and GRS is the standardized

GRS z-score of the TWB2 subjects. Covariates adjusted for FG/HbA1c/diabetes have been

11

described under model (1). To further control confounding (24), 𝐺𝑅𝑆 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣 and

𝑆𝑚𝑜𝑘𝑖𝑛𝑔 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣 (𝑣 = 1,⋯ ,16) were also included in model (4). To interpret main

effects in model (4), we centered all explanatory variables at their means (25).

Let the p-value of testing 𝐻0: 𝜙𝐼𝑁𝑇 = 0 𝑣𝑠. 𝐻1: 𝜙𝐼𝑁𝑇 ≠ 0 be 𝑃𝐼𝑁𝑇. The presence of G×S

will be declared if 𝑃𝐼𝑁𝑇 <0.05

3×5= 0.0033, because 3 diabetes-related traits and 5 smoking

measurements were investigated. This significance level for GRS analyses was determined a

priori and would be used throughout this study.

DNA methylation data

One of the mechanisms behind G×S is epigenetics (33). Cigarette consumption has been

found to be linked to differential DNA methylation in some diabetes susceptibility genes, such as

KCNQ1 (potassium voltage-gated channel subfamily Q member 1) (34).

When one is smoking, carbon monoxide displaces oxygen from his/her hemoglobin. This

will decrease the amount of oxygen available for delivery to many tissues. Smoking can therefore

affect one’s entire body including cardiovascular and circulatory systems (35). TWB also

released blood DNA methylation data of 2,091 TWB subjects. These individuals were randomly

selected from TWB1 (833 subjects) and TWB2 (1,258 subjects). After GRS-smoking interaction

analysis, we also explored the associations between smoking and DNA methylation of diabetes

susceptibility genes.

Data and Resource Availability

Individual-level TWB data are available upon application to TWB

(https://www.twbiobank.org.tw/new_web/). Our application for the TWB data was approved in

12

February, 2020, with the accession number “TWBR10810-07”.

[Table 1 is approximately here]

Results

Association between smoking and FG/HbA1c/diabetes

Our discovery cohort contained 25,460 TWB1 subjects, where 3,005 (11.8%) were (active)

smokers and 22,455 were non-smokers. The replication cohort included 58,774 TWB2 subjects,

5,173 (8.8%) were (active) smokers and 53,601 were non-smokers. Table 1 presents basic

characteristics of these two cohorts. The average FG and HbA1c levels are higher in smokers

than in non-smokers. Model (1) was fitted to assess whether smoking is significantly associated

with FG/HbA1c/diabetes, while considering the covariates. As shown by Table 2, male, aging,

and a larger BMI, are associated with increased FG/HbA1c and risk of diabetes.

[Table 2 is approximately here]

Because FG/HbA1c was natural log transformed, (𝑒𝑥𝑝(�̂�𝑆𝑚𝑜𝑘𝑖𝑛𝑔) 1) × 100 is shown

to represent the percent change in FG/HbA1c that is associated with “active smoking status”,

while adjusting for sex, age, BMI, drinking status, regular exercise, educational attainment, and

the first 10 PCs. Both our discovery cohort (TWB1) and replication cohort (TWB2) showed,

compared with non-smokers, active smokers have an odds ratio (OR) of 1.36 for diabetes (95%

confidence interval [C.I.] by TWB1: 1.19-1.57; TWB2: 1.23-1.51).

[Table 3 is approximately here]

Based on model (1), we further analyzed the 4 continuous smoking measurements,

respectively. For example, TWB1 (TWB2) showed that, subjects consuming one more pack of

cigarettes per day have an OR of 1.51 (1.41) for diabetes (95% C.I. by TWB1: 1.32-1.73; TWB2:

13

1.27-1.56). The detailed results of model (1) for the 4 continuous smoking measurements can be

found in Supplemental Tables S1-S4.

GRS-smoking interactions in FG/HbA1c/diabetes

Analyzing TWB1 according to model (2), a total of 16, 12, and 6 SNPs were identified to

have 𝑃𝑆𝑁𝑃 < 6.25 × 10−9 in FG, HbA1c, and diabetes analyses, respectively. The information

of these SNPs was listed in Supplemental Tables S5-S7. A total of 5 out of the 16 FG-associated

SNPs, 4 out of the 12 HbA1c-associated SNPs, and 2 out of the 6 diabetes-associated SNPs were

also identified by another trait. The GRS of each TWB2 individual was then calculated based on

these 16/12/6 SNPs, with weights obtained from TWB1 analysis (i.e., �̂�𝑆𝑁𝑃𝑠 from model 2).

[Table 4 is approximately here]

Table 4 presents the results of model (4) when the smoking measurement is “active smoking

status”. FG (or HbA1c) was natural log transformed, and therefore (𝑒𝑥𝑝(�̂�𝐼𝑁𝑇) 1) × 100 =

1.05 represents that each 1 sd increase in GRS was associated with a 1.05% higher FG (or

0.51% higher HbA1c) in active smokers than in non-smokers (𝑃𝐼𝑁𝑇−𝐹𝐺 = 2.8 × 10−5;

𝑃𝐼𝑁𝑇−𝐻𝑏𝐴1𝑐 = 0.002). Each 1 sd increase in GRS was associated with a 1.16 times OR (95% C.I.:

1.05-1.28) for diabetes in active smokers than in non-smokers (𝑃𝐼𝑁𝑇−𝑑𝑖𝑎𝑏𝑒𝑡𝑒𝑠 = 0.0046). After

analyzing the 4 continuous smoking measurements sequentially according to model (4), we found

all smoking measurements, except “hours as a passive smoker per week”, are associated with the

exacerbation of the genetic risk of diabetes (Table 5).

[Table 5 is approximately here]

For example, each 1 sd increase in GRS is associated with a 1.68% higher FG (or 0.69%

higher HbA1c) in subjects with one more pack of cigarettes per day (𝑃𝐼𝑁𝑇−𝐹𝐺 = 1.9 × 10−7;

14

𝑃𝐼𝑁𝑇−𝐻𝑏𝐴1𝑐 = 4.5 × 10−4). The detailed results of model (4) for the 4 continuous smoking

measurements can be found in Supplemental Tables S8-S11.

Figure 1 shows the average of FG/HbA1c and the prevalence of diabetes stratified by

smoking status and the quintiles of the FG/HbA1c/diabetes GRS. The GRS effects on

FG/HbA1c/diabetes were larger in smokers than in non-smokers. Supplemental Figures S1-S4

present similar illustrations for the other 4 smoking measurements. Except “passive smoking”

(Figure S4), all smoking measurements exhibit interactions with GRS on FG/HbA1c/diabetes.

This is in line with the result shown in Table 5.

[Figure 1 is approximately here]

Among the 5 smoking measurements, only “active smoking status” and “years as a smoker”

presented interactions with GRS on diabetes under 𝛼 = 0.05 (Table 5). The power to identify

diabetes-associated SNPs was low, because only 2,207 (8.7%) among the 25,460 TWB1 subjects

had diabetes. As a result, we merely detected 6 diabetes-associated SNPs at the significance level

of 6.25 × 10−9 (Supplemental Table S7). The GRS constructed by these 6 SNPs may not be

able to well represent the genetic susceptibility to diabetes. Moreover, only 5,308 (9.0%) among

the 58,774 TWB2 subjects had diabetes, the power to detect GRS-smoking interactions was

limited. Therefore, none of the five smoking measurements were found to be associated with the

exacerbation of the genetic risk of diabetes under 𝛼 =0.05

15= 0.0033 (the last column of Table

5).

Although the response variable in model (4) is natural log transformed FG/HbA1c, these

significant findings in GRS-smoking interactions are not scale dependent. Supplemental Table

S12 shows that these significant results are still replicable when the response variable is

FG/HbA1c without natural log transformation.

15

Multicollinearity between variables in model (4) has been evaluated via variance inflation

factor (VIF), computed by the “car” package (https://cran.r-project.org/web/packages/car/). The

VIFs under all models were acceptable (smaller than 10), except the model to assess the

interaction between GRS and “hours as a passive smoker per week”. To reduce the

multicollinearity between variables in model (4), we further assessed GRS-smoking interactions

without controlling for the other interaction terms (i.e., no 𝐺𝑅𝑆 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣 or 𝑆𝑚𝑜𝑘𝑖𝑛𝑔 ×

𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣, 𝑣 = 1,⋯ ,16). The VIFs under these simpler models were all smaller than 5.

Moreover, Supplemental Table S13 shows that the results from simpler models were similar to

Table 5.

We then assessed GRS-smoking interactions within each gender group. Supplemental Table

S14 shows that the significant GRS-smoking interactions were mostly driven by the male group.

This is because 88.1% and 77.4% smokers were males in TWB1 and TWB2, respectively. The

detection of GRS-smoking interactions in females could be hampered by the small sample size of

female smokers.

SNP-smoking interactions in FG/HbA1c/diabetes

Four SNPs (rs2399794, rs7896600, rs1174605899, and rs11257655) in/near the CDC123

(cell division cycle 123) gene were identified to be associated with FG/HbA1c/diabetes. They

presented interactions with most smoking measurements except passive smoking (most

𝑃𝐼𝑁𝑇𝑠 < 0.05, Tables S5-S7). The CDC123 gene has been found to be associated with the

dysfunction of pancreatic β-cells (26). The inability of pancreatic β-cells to secrete adequate

levels of insulin is a major cause of diabetes (27).

Two SNPs (rs2233580 and rs61342118) in/near the PAX4 (paired box 4) gene were

identified to be associated with FG/HbA1c/diabetes. They presented interactions with some

16

smoking measurements except passive smoking (some 𝑃𝐼𝑁𝑇𝑠 < 0.05, Tables S5-S7). The PAX4

gene is necessary for the survival and proliferation of pancreatic β-cells (28). This gene has been

found to be hypermethylated in patients with diabetes (29).

Four SNPs (rs163177, rs60808706, rs11024175, and rs163184) in KCNQ1 were identified to

be associated with FG/HbA1c/diabetes. SNP rs11024175 was found to interact with all the 5

smoking measurements on HbA1c (𝑃𝐼𝑁𝑇𝑠 < 0.05, Table S6). KCNQ1 mRNAs were highly

expressed in adrenal tissues (30). Disorders of the adrenal cortex can result in glucose intolerance

and diabetes (31). Consistent with our finding, KCNQ1-by-smoking interaction has been reported

by a study in Han Chinese (32).

Smoking is associated with DNA methylation of diabetes susceptibility genes

The 2,091 individuals with methylation measures were randomly selected from TWB1 (833

subjects) and TWB2 (1,258 subjects). Their basic characteristics (Supplemental Table S15) are

similar to those of the whole TWB (Table 1). We annotated CpG sites available on the Illumina

Infinium MethylationEPIC BeadChip to the abovementioned three genes. A total of 72, 84, and

629 CpG sites are within/near CDC123, PAX4, and KCNQ1, respectively (“near” indicates 50 kb

in the 3’ and 5’ regions outside the gene boundary). The methylation percentage of a CpG site

was reported as a β-value ranging from 0 (no methylation) to 1 (full methylation). The β-value of

each CpG site was regressed on “packs smoked per day”, while adjusting for age, sex, and BMI

(these three covariates were also adjusted in a related study (34)). Although “packs smoked per

day” was served as the smoking factor, “active smoking status” provided similar results.

“Packs smoked per day” is associated with differential DNA methylation of CDC123

(cg06335123, p = 0.00025 < 0.05/72) and KCNQ1 (cg26963277, 𝑝 = 3.0 × 10−17; cg01744331,

𝑝 = 4.5 × 10−10; cg16556677, 𝑝 = 2.7 × 10−5< 0.05/629). More packs smoked per day, aging,

17

and a larger BMI are associated with decreased levels of DNA methylation at these four sites

(Supplemental Table S16). The three CpG sites in KCNQ1 (cg26963277, cg01744331, and

cg16556677) have been reported to be associated with smoking in the Rotterdam study (34). Our

finding from Han Chinese is in line with that study (34).

Through its link with differential DNA methylation, smoking can modulate gene expressions

of diabetes susceptibility genes and lead to more damage to people who are more genetically

predisposed to diabetes. However, we have not observed the association between smoking and

DNA methylation of PAX4. The mechanism behind PAX4-by-smoking interaction needs further

investigation.

A limitation is that our DNA methylation measures were made in peripheral blood, and the

more relevant tissue is probably pancreatic β-cell. Although we here found that active smoking is

associated with methylation at CpG sites in some diabetes susceptibility genes, additional work is

necessary to determine if this mediates the smoking-genotype interaction.

Discussion

Smoking is associated with insulin resistance, inflammation, and dyslipidemia (36).

Consistent with our finding, KCNQ1-by-smoking interaction has been reported by a study in Han

Chinese (32). Moreover, Wu et al. recently identified interactions between smoking status and

five SNPs at or near four genes (TCF7L2, CUBN, C2orf63, and FBN1) on the risk of diabetes, in

subjects of European and African ancestry (15). Two out of the five SNP-smoking interactions

can be replicated in our TWB2 HbA1c/diabetes analysis, at the nominal significance level of 0.05

(Supplemental Table S17). Both of the variants locate at the TCF7L2 (transcription factor 7 like 2)

gene. TCF7L2 is a diabetes-associated gene identified in subjects of European ancestry (37), but

the variants in this gene are not associated with FG/HbA1c/diabetes in TWB1 and so were not

18

selected to form our GRSs. Variants in TCF7L2 are associated with pancreatic β-cell function

(38).

For diabetes, Langenberg et al. have shown no interactions between genetic risk and

physical activity or dietary habits (1). Said et al.’s recent study based on the UK Biobank further

found no significant interactions between lifestyle and GRS on diabetes (14). However, their

lifestyle is a composite measure of active smoking status, BMI, and physical activity. It remains

unclear whether smoking alone may modify the genetic predisposition to diabetes. Moreover,

their GRS for diabetes was calculated by 38 diabetes-associated SNPs with effect sizes retrieved

from previously published GWASs (39-42). Most of these 38 SNPs were identified at the

commonly-used genome-wide significance level (𝑃𝑆𝑁𝑃 < 5 × 10−8).

Individuals of those GWASs (39-42) were overwhelmingly of European descent. A SNP in

the above-mentioned TCF7L2 gene, rs7903146, is also among the list of the 38 SNPs

(Supplemental Tables S18-S20). One of the 38 SNPs, rs13266634 (in the SLC30A8 gene), is also

among our 6 diabetes-associated SNPs (Table S7). Moreover, the 38 SNPs included variants

in/near CDC123 and KCNQ1, which were also identified to be associated with

FG/HbA1c/diabetes according to our discovery cohort (TWB1), as listed in Tables S5-S7.

We then found three smoking factors (“active smoking status”, “the number of pack-years”,

and “years as a smoker”) were associated with an exacerbation of 𝐸𝑢𝐺𝑅𝑆 on both FG and

HbA1c (Supplemental Tables S21-S22), at the nominal significance of 0.05 (𝑃𝐼𝑁𝑇 < 0.05).

“𝐸𝑢𝐺𝑅𝑆 × Packs smoked per day” interaction presented a borderline significance on FG

(𝑃𝐼𝑁𝑇 = 0.0582). Similar with the results based on TWB1-GRS, “𝐸𝑢𝐺𝑅𝑆 × Hours as a passive

smoker per week” interaction was not significant on FG or HbA1c (Supplemental Table S22).

We also analyzed diabetes as a dichotomous trait. Through logistic regression models, we

19

did not detect any significant 𝐸𝑢𝐺𝑅𝑆-smoking interactions. In addition to low statistical power

(< 10% subjects with diabetes), this negative finding may also result from the lack of

transferability of a European-derived GRS to TWB. Comparing Table 4 with Table S21, we can

see that TWB1-GRSs are more significant (smaller p-values) and more predictive (larger

R-squares) than 𝐸𝑢𝐺𝑅𝑆 for all the 3 traits, despite fewer SNPs used for TWB1-GRSs. The

detection of EuGRS-smoking interactions could be hampered by the inferior transferability of

𝐸𝑢𝐺𝑅𝑆 to TWB.

Supplemental Table S23 shows that the results of 𝐸𝑢𝐺𝑅𝑆-smoking interactions are still

replicable when the response variable is FG/HbA1c without natural log transformation. When

assessing 𝐸𝑢𝐺𝑅𝑆-smoking interactions without controlling for the other interaction terms (i.e.,

no 𝐸𝑢𝐺𝑅𝑆 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣 or 𝑆𝑚𝑜𝑘𝑖𝑛𝑔 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑣, 𝑣 = 1,⋯ ,16), Supplemental Table S24

shows that the results from simpler models were similar to Table S22. Moreover, in line with the

results from TWB1-GRS, Supplemental Table S25 also shows that the significant

𝐸𝑢𝐺𝑅𝑆-smoking interactions were mostly driven by the male group.

Diabetes is a growing health crisis around the world. According to the Centers for Disease

Control and Prevention and many previous studies, smoking increases inflammation in the body

(43) and also causes oxidative stress (44). Both inflammation and “oxidative stress” have been

shown to be linked to an increased risk of diabetes (45). “Oxidative stress” is an imbalance

between oxidants and antioxidants in favor of the oxidants (46), potentially modulating gene

expressions (47) and leading to more damage to people who are more genetically predisposed to

diabetes. Indeed, we here found smoking is associated with DNA methylation at KCNQ1 and

CDC123, implying that smoking can play a role in regulating gene expressions of these diabetes

susceptibility genes.

20

According to previous studies (48), the amount of nicotine (the main chemical in cigarettes)

absorbed by a passive smoker was between 1/10 and 1/3 of the amount in a cigarette. Therefore,

the concentration of nicotine in passive smokers is smaller than that in active smokers (48). This

may explain why passive smoking does not significantly modulate the genetic predisposition to

diabetes.

The harm of smoking is more impactful in subjects who are more genetically predisposed to

diabetes. This study shows that active cigarette smoking is associated with an exacerbation of

genetic risk of diabetes. Through constructing a GRS based on diabetes associated SNPs, it is

worthwhile to investigate whether smoking may exacerbate the genetic susceptibility to diabetes,

even for populations where smoking has not been found to be associated with diabetes.

Acknowledgements

The authors would like to thank the editors and three anonymous reviewers for their insightful

and constructive comments; Mr. Tsung-Hao Lee and Mr. Yu-Sheng Lee for imputation of the

TWB data; Mr. Ya-Chin Lee for assisting with the acquisition of the TWB data.

Funding

This study was supported by the Ministry of Science and Technology of Taiwan (grant number

MOST 107-2314-B-002-195-MY3 to Wan-Yu Lin). The acquisition of TWB data was supported

by two MOST grants (MOST 107-2314-B-002-195-MY3 to Wan-Yu Lin; MOST

102-2314-B-002-117-MY3 to Po-Hsiu Kuo) and a collaboration grant (National Taiwan

University Hospital: grant number UN106-050 to Shyr-Chyr Chen and Po-Hsiu Kuo).

21

Duality of Interest

No potential conflicts of interest relevant to this article were reported.

Author Contributions

W.-Y.L. conceived the study design, developed the analysis tool, analyzed the TWB data,

and wrote the manuscript. W.-Y.L. is the guarantor of this work and, as such, had full access to all

the data in the study and takes responsibility for the integrity of the data and the accuracy of the

data analysis. P.-H.K., S.-J.T., Y.-L.L., and A.C.Y. contributed to the writing of the manuscript.

All authors provided the TWB data and reviewed the manuscript.

References

1. Langenberg C, Sharp SJ, Franks PW, Scott RA, Deloukas P, Forouhi NG, Froguel P, Groop LC,

Hansen T, Palla L, Pedersen O, Schulze MB, Tormo MJ, Wheeler E, Agnoli C, Arriola L, Barricarte A,

Boeing H, Clarke GM, Clavel-Chapelon F, Duell EJ, Fagherazzi G, Kaaks R, Kerrison ND, Key TJ,

Khaw KT, Kroger J, Lajous M, Morris AP, Navarro C, Nilsson PM, Overvad K, Palli D, Panico S,

Quiros JR, Rolandsson O, Sacerdote C, Sanchez MJ, Slimani N, Spijkerman AM, Tumino R, van der

AD, van der Schouw YT, Barroso I, McCarthy MI, Riboli E, Wareham NJ: Gene-lifestyle interaction

and type 2 diabetes: the EPIC interact case-cohort study. Plos Med 2014;11:e1001647

2. American Diabetes Association: Diagnosis and classification of diabetes mellitus. Diabetes

Care 2006;29 Suppl 1:S43-48

3. International Expert C: International Expert Committee report on the role of the A1C assay in

the diagnosis of diabetes. Diabetes Care 2009;32:1327-1334

4. Leslie RD, Cohen RM: Biologic variability in plasma glucose, hemoglobin A1c, and advanced

glycation end products associated with diabetes complications. J Diabetes Sci Technol

2009;3:635-643

5. Willi C, Bodenmann P, Ghali WA, Faris PD, Cornuz J: Active smoking and the risk of type 2

diabetes: a systematic review and meta-analysis. JAMA 2007;298:2654-2664

6. Xie XT, Liu Q, Wu J, Wakui M: Impact of cigarette smoking in type 2 diabetes development.

Acta Pharmacol Sin 2009;30:784-787

7. Maddatu J, Anderson-Baucum E, Evans-Molina C: Smoking and the risk of type 2 diabetes.

22

Transl Res 2017;184:101-107

8. Chen CH, Yang JH, Chiang CWK, Hsiung CN, Wu PE, Chang LC, Chu HW, Chang J, Song IW, Yang

SL, Chen YT, Liu FT, Shen CY: Population structure of Han Chinese in the modern Taiwanese

population based on 10,000 participants in the Taiwan Biobank project. Human Molecular

Genetics 2016;25:5321-5331

9. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de

Bakker PI, Daly MJ, Sham PC: PLINK: a tool set for whole-genome association and

population-based linkage analyses. American journal of human genetics 2007;81:559-575

10. An JY, Gharahkhani P, Law MH, Ong JS, Han XK, Olsen CM, Neale RE, Lai J, Vaughan TL,

Bohmer AC, Jankowski J, Fitzgerald RC, Schumacher J, Palles C, Whiteman DC, MacGregor S,

Gammon MD, Corley DA, Shaheen NJ, Bird NC, Hardie LJ, Murray LJ, Reid BJ, Chow WH, Risch HA,

Ye WM, Liu G, Romero Y, Bernstein L, Wu AH, Agee M, Alipanahi B, Auton A, Bell RK, Bryc K,

Elson SL, Fontanillas P, Furlotte NA, Hinds DA, Huber KE, Kleinman A, Litterman NK, McIntyre MH,

Mountain JL, Noblin ES, Northover CAM, Pitts SJ, Sathirapongsasuti JF, Sazonova OV, Shelton JF,

Shringarpure S, Tian C, Tung JY, Vacic V, Wilson CH, Team BMR: Gastroesophageal reflux GWAS

identifies risk loci that also associate with subsequent severe esophageal diseases. Nature

Communications 2019;10

11. Calabro M, Drago A, Sidoti A, Serretti A, Crisafulli C: Genes involved in pruning and

inflammation are enriched in a large mega-sample of patients affected by Schizophrenia and

Bipolar Disorder and controls. Psychiatry Res 2015;228:945-949

12. WTCCC: Genome-wide association study of 14,000 cases of seven common diseases and

3,000 shared controls. Nature 2007;447:661-678

13. Lin WY, Chan CC, Liu YL, Yang AC, Tsai SJ, Kuo PH: Performing different kinds of physical

exercise differentially attenuates the genetic effects on obesity measures: Evidence from 18,424

Taiwan Biobank participants. PLoS genetics 2019;15:e1008277

14. Said MA, Verweij N, van der Harst P: Associations of Combined Genetic and Lifestyle Risks

With Incident Cardiovascular Disease and Diabetes in the UK Biobank Study. Jama Cardiol

2018;3:693-702

15. Wu P, Rybin D, Bielak LF, Feitosa MF, Franceschini N, Li Y, Lu Y, Marten J, Musani SK, Noordam

R, Raghavan S, Rose LM, Schwander K, Smith AV, Tajuddin SM, Vojinovic D, Amin N, Arnett DK,

Bottinger EP, Demirkan A, Florez JC, Ghanbari M, Harris TB, Launer LJ, Liu J, Liu J, Mook-Kanamori

DO, Murray AD, Nalls MA, Peyser PA, Uitterlinden AG, Voortman T, Bouchard C, Chasman D,

Correa A, de Mutsert R, Evans MK, Gudnason V, Hayward C, Kao L, Kardia SLR, Kooperberg C,

Loos RJF, Province MM, Rankinen T, Redline S, Ridker PM, Rotter JI, Siscovick D, Smith BH, van

Duijn C, Zonderman AB, Rao DC, Wilson JG, Dupuis J, Meigs JB, Liu CT, Vassy JL:

Smoking-by-genotype interaction in type 2 diabetes risk and fasting glucose. PLoS One

23

2020;15:e0230815

16. Bird SR, Hawley JA: Update on the effects of physical activity on insulin sensitivity in humans.

BMJ Open Sport Exerc Med 2016;2:e000143

17. Howard AA, Arnsten JH, Gourevitch MN: Effect of alcohol consumption on diabetes mellitus:

a systematic review. Ann Intern Med 2004;140:211-219

18. Webster RJ, Warrington NM, Beilby JP, Frayling TM, Palmer LJ: The longitudinal association of

common susceptibility variants for type 2 diabetes and obesity with fasting glucose level and

BMI. Bmc Med Genet 2010;11:140

19. Kim C, Bullard KM, Herman WH, Beckles GL: Association Between Iron Deficiency and A1C

Levels Among Adults Without Diabetes in the National Health and Nutrition Examination Survey,

1999-2006. Diabetes Care 2010;33:780-785

20. Dai JY, Kooperberg C, Leblanc M, Prentice RL: Two-stage testing procedures with

independent filtering for genome-wide gene-environment interaction. Biometrika

2012;99:929-944

21. Frost HR, Shen L, Saykin AJ, Williams SM, Moore JH, Alzheimer's Disease Neuroimaging I:

Identifying significant gene-environment interactions using a combination of screening testing

and hierarchical false discovery rate control. Genetic epidemiology 2016;40:544-557

22. Lin WY, Huang CC, Liu YL, Tsai SJ, Kuo PH: Genome-wide gene-environment interaction

analysis using set-based association tests. Frontiers in Genetics 2019;9:715

23. Lin WY, Huang CC, Liu YL, Tsai SJ, Kuo PH: Polygenic approaches to detect gene-environment

interactions when external information is unavailable. Brief Bioinform 2019;20:2236-2252

24. Keller MC: Gene x environment interaction studies have not properly controlled for potential

confounders: the problem and the (simple) solution. Biol Psychiatry 2014;75:18-24

25. Schielzeth H: Simple means to improve the interpretability of regression coefficients.

Methods Ecol Evol 2010;1:103-113

26. Simonis-Bik AM, Nijpels G, van Haeften TW, Houwing-Duistermaat JJ, Boomsma DI, Reiling E,

van Hove EC, Diamant M, Kramer MH, Heine RJ, Maassen JA, Slagboom PE, Willemsen G, Dekker

JM, Eekhoff EM, de Geus EJ, t Hart LM: Gene variants in the novel type 2 diabetes loci

CDC123/CAMK1D, THADA, ADAMTS9, BCL11A, and MTNR1B affect different aspects of

pancreatic beta-cell function. Diabetes 2010;59:293-301

27. Haythorne E, Rohm M, van de Bunt M, Brereton MF, Tarasov AI, Blacker TS, Sachse G, Silva

Dos Santos M, Terron Exposito R, Davis S, Baba O, Fischer R, Duchen MR, Rorsman P, MacRae JI,

Ashcroft FM: Diabetes causes marked inhibition of mitochondrial metabolism in pancreatic

beta-cells. Nat Commun 2019;10:2474

28. Brun T, Franklin I, St-Onge L, Biason-Louber A, Schoenle EJ, Wollheim CB, Gauthier BR: The

diabetes-linked transcription factor PAX4 promotes beta-cell proliferation and survival in rat and

24

human islets. J Cell Biol 2004;167:1123-1135

29. Sommese L, Zullo A, Mancini FP, Fabbricini R, Soricelli A, Napoli C: Clinical relevance of

epigenetics in the onset and management of type 2 diabetes mellitus. Epigenetics-Us

2017;12:401-415

30. Sarzani R, Pietrucci F, Francioni M, Salvi F, Letizia C, D'Erasmo E, Dessi Fulgheri P, Rappelli A:

Expression of potassium channel isoforms mRNA in normal human adrenals and

aldosterone-secreting adenomas. J Endocrinol Invest 2006;29:147-153

31. Nestler JE, McClanahan MA: Diabetes and adrenal disease. Baillieres Clin Endocrinol Metab

1992;6:829-847

32. Gao KP, Ren YC, Wang JJ, Liu ZC, Li JN, Li LL, Wang BY, Li H, Wang YX, Cao YK, Ohno K, Zhai RH,

Liang Z: Interactions between genetic polymorphisms of glucose metabolizing genes and

smoking and alcohol consumption in the risk of type 2 diabetes mellitus. Appl Physiol Nutr Me

2017;42:1316-1321

33. Kubota T, Miyake K, Hirasawa T: Epigenetic understanding of gene-environment interactions

in psychiatric disorders: a new concept of clinical genetics. Clinical Epigenetics 2012;4:1

34. Ligthart S, Steenaard RV, Peters MJ, van Meurs JBJ, Sijbrands EJG, Uitterlinden AG, Bonder MJ,

Hofman A, Franco OH, Dehghan A, Consortium B: Tobacco smoking is associated with DNA

methylation of diabetes susceptibility genes. Diabetologia 2016;59:998-1006

35. West R: Tobacco smoking: Health impact, prevalence, correlates and interventions. Psychol

Health 2017;32:1018-1036

36. Chang SA: Smoking and type 2 diabetes mellitus. Diabetes Metab J 2012;36:399-403

37. Scott RA, Scott LJ, Magi R, Marullo L, Gaulton KJ, Kaakinen M, Pervjakova N, Pers TH, Johnson

AD, Eicher JD, Jackson AU, Ferreira T, Lee Y, Ma C, Steinthorsdottir V, Thorleifsson G, Qi L, Van

Zuydam NR, Mahajan A, Chen H, Almgren P, Voight BF, Grallert H, Muller-Nurasyid M, Ried JS,

Rayner NW, Robertson N, Karssen LC, van Leeuwen EM, Willems SM, Fuchsberger C, Kwan P,

Teslovich TM, Chanda P, Li M, Lu Y, Dina C, Thuillier D, Yengo L, Jiang L, Sparso T, Kestler HA,

Chheda H, Eisele L, Gustafsson S, Franberg M, Strawbridge RJ, Benediktsson R, Hreidarsson AB,

Kong A, Sigurethsson G, Kerrison ND, Luan J, Liang L, Meitinger T, Roden M, Thorand B, Esko T,

Mihailov E, Fox C, Liu CT, Rybin D, Isomaa B, Lyssenko V, Tuomi T, Couper DJ, Pankow JS, Grarup N,

Have CT, Jorgensen ME, Jorgensen T, Linneberg A, Cornelis MC, van Dam RM, Hunter DJ, Kraft P,

Sun Q, Edkins S, Owen KR, Perry JRB, Wood AR, Zeggini E, Tajes-Fernandes J, Abecasis GR,

Bonnycastle LL, Chines PS, Stringham HM, Koistinen HA, Kinnunen L, Sennblad B, Muhleisen TW,

Nothen MM, Pechlivanis S, Baldassarre D, Gertow K, Humphries SE, Tremoli E, Klopp N, Meyer J,

Steinbach G, Wennauer R, Eriksson JG, Mnnisto S, Peltonen L, Tikkanen E, Charpentier G, Eury E,

Lobbens S, Gigante B, Leander K, McLeod O, Bottinger EP, Gottesman O, Ruderfer D, Bluher M,

Kovacs P, Tonjes A, Maruthur NM, Scapoli C, Erbel R, Jockel KH, Moebus S, de Faire U, Hamsten A,

25

Stumvoll M, Deloukas P, Donnelly PJ, Frayling TM, Hattersley AT, Ripatti S, Salomaa V, Pedersen

NL, Boehm BO, Bergman RN, Collins FS, Mohlke KL, Tuomilehto J, Hansen T, Pedersen O, Barroso

I, Lannfelt L, Ingelsson E, Lind L, Lindgren CM, Cauchi S, Froguel P, Loos RJF, Balkau B, Boeing H,

Franks PW, Barricarte Gurrea A, Palli D, van der Schouw YT, Altshuler D, Groop LC, Langenberg C,

Wareham NJ, Sijbrands E, van Duijn CM, Florez JC, Meigs JB, Boerwinkle E, Gieger C, Strauch K,

Metspalu A, Morris AD, Palmer CNA, Hu FB, Thorsteinsdottir U, Stefansson K, Dupuis J, Morris AP,

Boehnke M, McCarthy MI, Prokopenko I, Replication DIG, Meta-analysis C: An Expanded

Genome-Wide Association Study of Type 2 Diabetes in Europeans. Diabetes 2017;66:2888-2902

38. Florez JC, Jablonski KA, Bayley N, Pollin TI, de Bakker PI, Shuldiner AR, Knowler WC, Nathan

DM, Altshuler D, Diabetes Prevention Program Research G: TCF7L2 polymorphisms and

progression to diabetes in the Diabetes Prevention Program. N Engl J Med 2006;355:241-250

39. Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, Jackson AU, Wheeler E, Glazer

NL, Bouatia-Naji N, Gloyn AL, Lindgren CM, Magi R, Morris AP, Randall J, Johnson T, Elliott P,

Rybin D, Thorleifsson G, Steinthorsdottir V, Henneman P, Grallert H, Dehghan A, Hottenga JJ,

Franklin CS, Navarro P, Song KJ, Goel A, Perry JRB, Egan JM, Lajunen T, Grarup N, Sparso T, Doney

A, Voight BF, Stringham HM, Li M, Kanoni S, Shrader P, Cavalcanti-Proenca C, Kumari M, Qi L,

Timpson NJ, Gieger C, Zabena C, Rocheleau G, Ingelsson E, An P, O'Connell J, Luan J, Elliott A,

McCarroll SA, Payne F, Roccasecca RM, Pattou F, Sethupathy P, Ardlie K, Ariyurek Y, Balkau B,

Barter P, Beilby JP, Ben-Shlomo Y, Benediktsson R, Bennett AJ, Bergmann S, Bochud M,

Boerwinkle E, Bonnefond A, Bonnycastle LL, Borch-Johnsen K, Bottcher Y, Brunner E, Bumpstead

SJ, Charpentier G, Chen YDI, Chines P, Clarke R, Coin LJM, Cooper MN, Cornelis M, Crawford G,

Crisponi L, Day INM, de Geus EJC, Delplanque J, Dina C, Erdos MR, Fedson AC, Fischer-Rosinsky A,

Forouhi NG, Fox CS, Frants R, Franzosi MG, Galan P, Goodarzi MO, Graessler J, Groves CJ, Grundy

S, Gwilliam R, Gyllensten U, Hadjadj S, Hallmans G, Hammond N, Han XJ, Hartikainen AL,

Hassanali N, Hayward C, Heath SC, Hercberg S, Herder C, Hicks AA, Hillman DR, Hingorani AD,

Hofman A, Hui J, Hung J, Isomaa B, Johnson PRV, Jorgensen T, Jula A, Kaakinen M, Kaprio J,

Kesaniemi YA, Kivimaki M, Knight B, Koskinen S, Kovacs P, Kyvik KO, Lathrop GM, Lawlor DA, Le

Bacquer O, Lecoeur C, Li Y, Lyssenko V, Mahley R, Mangino M, Manning AK, Martinez-Larrad MT,

McAteer JB, McCulloch LJ, McPherson R, Meisinger C, Melzer D, Meyre D, Mitchell BD, Morken

MA, Mukherjee S, Naitza S, Narisu N, Neville MJ, Oostra BA, Orru M, Pakyz R, Palmer CNA,

Paolisso G, Pattaro C, Pearson D, Peden JF, Pedersen NL, Perola M, Pfeiffer AFH, Pichler I, Polasek

O, Posthuma D, Potter SC, Pouta A, Province MA, Psaty BM, Rathmann W, Rayner NW, Rice K,

Ripatti S, Rivadeneira F, Roden M, Rolandsson O, Sandbaek A, Sandhu M, Sanna S, Sayer AA,

Scheet P, Scott LJ, Seedorf U, Sharp SJ, Shields B, Sigurosson G, Sijbrands EJG, Silveira A, Simpson

L, Singleton A, Smith NL, Sovio U, Swift A, Syddall H, Syvanen AC, Tanaka T, Thorand B, Tichet J,

Tonjes A, Tuomi T, Uitterlinden AG, van Dijk KW, van Hoek M, Varma D, Visvikis-Siest S, Vitart V,

26

Vogelzangs N, Waeber G, Wagner PJ, Walley A, Walters GB, Ward KL, Watkins H, Weedon MN,

Wild SH, Willemsen G, Witteman JCM, Yarnell JWG, Zeggini E, Zelenika D, Zethelius B, Zhai GJ,

Zhao JH, Zillikens MC, Borecki IB, Loos RJF, Meneton P, Magnusson PKE, Nathan DM, Williams GH,

Hattersley AT, Silander K, Salomaa V, Smith GD, Bornstein SR, Schwarz P, Spranger J, Karpe F,

Shuldiner AR, Cooper C, Dedoussis GV, Serrano-Rios M, Morris AD, Lind L, Palmer LJ, Hu FB,

Franks PW, Ebrahim S, Marmot M, Kao WHL, Pankow JS, Sampson MJ, Kuusisto J, Laakso M,

Hansen T, Pedersen O, Pramstaller PP, Wichmann HE, Illig T, Rudan I, Wright AF, Stumvoll M,

Campbell H, Wilson JF, Hamsten A, Bergman RN, Buchanan TA, Collins FS, Mohlke KL, Tuomilehto

J, Valle TT, Altshuler D, Rotter JI, Siscovick DS, Penninx BWJH, Boomsma DI, Deloukas P, Spector

TD, Frayling TM, Ferrucci L, Kong A, Thorsteinsdottir U, Stefansson K, van Duijn CM, Aulchenko

YS, Cao A, Scuteri A, Schlessinger D, Uda M, Ruokonen A, Jarvelin MR, Waterworth DM,

Vollenweider P, Peltonen L, Mooser V, Abecasis GR, Wareham NJ, Sladek R, Froguel P, Watanabe

RM, Meigs JB, Groop L, Boehnke M, McCarthy MI, Florez JC, Consortium D, Consortium G,

Consortium GB, Consortium P, Investigators IBM: New genetic loci implicated in fasting glucose

homeostasis and their impact on type 2 diabetes risk (vol 42, pg 105, 2010). Nat Genet

2010;42:464-464

40. Morris AP, Voight BF, Teslovich TM, Ferreira T, Segre AV, Steinthorsdottir V, Strawbridge RJ,

Khan H, Grallert H, Mahajan A, Prokopenko I, Kang HM, Dina C, Esko T, Fraser RM, Kanoni S,

Kumar A, Lagou V, Langenberg C, Luan JA, Lindgren CM, Muller-Nurasyid M, Pechlivanis S,

Rayner NW, Scott LJ, Wiltshire S, Yengo L, Kinnunen L, Rossin EJ, Raychaudhuri S, Johnson AD,

Dimas AS, Loos RJF, Vedantam S, Chen H, Florez JC, Fox C, Liu CT, Rybin D, Couper DJ, Kao WHL, Li

M, Cornelis MC, Kraft P, Sun Q, van Dam RM, Stringham HM, Chines PS, Fischer K, Fontanillas P,

Holmen OL, Hunt SE, Jackson AU, Kong A, Lawrence R, Meyer J, Perry JRB, Platou CGP, Potter S,

Rehnberg E, Robertson N, Sivapalaratnam S, Stancakova A, Stirrups K, Thorleifsson G, Tikkanen E,

Wood AR, Almgren P, Atalay M, Benediktsson R, Bonnycastle LL, Burtt N, Carey J, Charpentier G,

Crenshaw AT, Doney ASF, Dorkhan M, Edkins S, Emilsson V, Eury E, Forsen T, Gertow K, Gigante B,

Grant GB, Groves CJ, Guiducci C, Herder C, Hreidarsson AB, Hui JN, James A, Jonsson A,

Rathmann W, Klopp N, Kravic J, Krjutskov K, Langford C, Leander K, Lindholm E, Lobbens S,

Mannisto S, Mirza G, Muhleisen TW, Musk B, Parkin M, Rallidis L, Saramies J, Sennblad B, Shah S,

Sigurdsson G, Silveira A, Steinbach G, Thorand B, Trakalo J, Veglia F, Wennauer R, Winckler W,

Zabaneh D, Campbell H, van Duijn C, Uitterlinden AG, Hofman A, Sijbrands E, Abecasis GR, Owen

KR, Zeggini E, Trip MD, Forouhi NG, Syvanen AC, Eriksson JG, Peltonen L, Nothen MM, Balkau B,

Palmer CNA, Lyssenko V, Tuomi T, Isomaa B, Hunter DJ, Qi L, Shuldiner AR, Roden M, Barroso I,

Wilsgaard T, Beilby J, Hovingh K, Price JF, Wilson JF, Rauramaa R, Lakka TA, Lind L, Dedoussis G,

Njolstad I, Pedersen NL, Khaw KT, Wareham NJ, Keinanen-Kiukaanniemi SM, Saaristo TE,

Korpi-Hyovalti E, Saltevo J, Laakso M, Kuusisto J, Metspalu A, Collins FS, Mohlke KL, Bergman RN,

27

Tuomilehto J, Boehm BO, Gieger C, Hveem K, Cauchi S, Froguel P, Baldassarre D, Tremoli E,

Humphries SE, Saleheen D, Danesh J, Ingelsson E, Ripatti S, Salomaa V, Erbel R, Jockel KH,

Moebus S, Peters A, Illig T, de Faire U, Hamsten A, Morris AD, Donnelly PJ, Frayling TM,

Hattersley AT, Boerwinkle E, Melander O, Kathiresan S, Nilsson PM, Deloukas P, Thorsteinsdottir

U, Groop LC, Stefansson K, Hu F, Pankow JS, Dupuis J, Meigs JB, Altshuler D, Boehnke M,

McCarthy MI, Control WTC, Insulin-Related M-AG, ANthropometric GI, Epidemiology AG, SAT2D

SATD, Replication DG: Large-scale association analysis provides insights into the genetic

architecture and pathophysiology of type 2 diabetes. Nat Genet 2012;44:981-+

41. Voight BF, Scott LJ, Steinthorsdottir V, Morris AP, Dina C, Welch RP, Zeggini E, Huth C,

Aulchenko YS, Thorleifsson G, McCulloch LJ, Ferreira T, Grallert H, Amin N, Wu GM, Willer CJ,

Raychaudhuri S, McCarroll SA, Langenberg C, Hofmann OM, Dupuis J, Qi L, Segre AV, van Hoek M,

Navarro P, Ardlie K, Balkau B, Benediktsson R, Bennett AJ, Blagieva R, Boerwinkle E, Bonnycastle

LL, Bostrom KB, Bravenboer B, Bumpstead S, Burtt NP, Charpentier G, Chines PS, Cornelis M,

Couper DJ, Crawford G, Doney ASF, Elliott KS, Elliott AL, Erdos MR, Fox CS, Franklin CS, Ganser M,

Gieger C, Grarup N, Green T, Griffin S, Groves CJ, Guiducci C, Hadjadj S, Hassanali N, Herder C,

Isomaa B, Jackson AU, Johnson PRV, Jorgensen T, Kao WHL, Klopp N, Kong A, Kraft P, Kuusisto J,

Lauritzen T, Li M, Lieverse A, Lindgren CM, Lyssenko V, Marre M, Meitinger T, Midthjell K,

Morken MA, Narisu N, Nilsson P, Owen KR, Payne F, Perry JRB, Petersen AK, Platou C, Proenca C,

Prokopenko I, Rathmann W, Rayner NW, Robertson NR, Rocheleau G, Roden M, Sampson MJ,

Saxena R, Shields BM, Shrader P, Sigurdsson G, Sparso T, Strassburger K, Stringham HM, Sun Q,

Swift AJ, Thorand B, Tichet J, Tuomi T, van Dam RM, van Haeften TW, van Herpt T, van

Vliet-Ostaptchouk JV, Walters GB, Weedon MN, Wijmenga C, Witteman J, Bergman RN, Cauchi S,

Collins FS, Gloyn AL, Gyllensten U, Hansen T, Hide WA, Hitman GA, Hofman A, Hunter DJ, Hveem

K, Laakso M, Mohlke KL, Morris AD, Palmer CNA, Pramstaller PP, Rudan I, Sijbrands E, Stein LD,

Jaakko T, Uitterlinden A, Walker M, Wareham NJ, Watanabe RM, Abecasis GR, Boehm BO,

Campbell H, Daly MJ, Hattersley AT, Hu FB, Meigs JB, Pankow JS, Pedersen O, Wichmann HE,

Barroso I, Florez JC, Frayling TM, Groop L, Sladek R, Thorsteinsdottir U, Wilson JF, Illig T, Froguel P,

van Duijn CM, Stefansson K, Altshuler D, Boehnke M, McCarthy MI, Investigators M, Consortium

G: Twelve type 2 diabetes susceptibility loci identified through large-scale association analysis.

Nat Genet 2010;42:579-U155

42. Zeggini E, Scott LJ, Saxena R, Voight BF, Marchini JL, Hu T, de Bakker PIW, Abecasis GR,

Almgren P, Andersen G, Ardlie K, Bostroem KB, Bergman RN, Bonnycastle LL, Borch-Johnsen K,

Burtt NP, Chen H, Chines PS, Daly MJ, Deodhar P, Ding CJ, Doney ASF, Duren WL, Elliott KS, Erdos

MR, Frayling TM, Freathy RM, Gianniny L, Grallert H, Grarup N, Groves CJ, Guiducci C, Hansen T,

Herder C, Hitman GA, Hughes TE, Isomaa B, Jackson AU, Jorgensen T, Kong A, Kubalanza K,

Kuruvilla FG, Kuusisto J, Langenberg C, Lango H, Lauritzen T, Li Y, Lindgren CM, Lyssenko V,

28

Marvelle AF, Meisinger C, Midthjell K, Mohlke KL, Morken MA, Morris AD, Narisu N, Nilsson P,

Owen KR, Palmer CNA, Payne F, Perry JRB, Pettersen E, Platou C, Prokopenko I, Qi L, Qin L,

Rayner NW, Rees M, Roix JJ, Sandbaek A, Shields B, Sjogren M, Steinthorsdottir V, Stringham HM,

Swift AJ, Thorleifsson G, Thorsteinsdottir U, Timpson NJ, Tuomi T, Tuomilehto J, Walker M,

Watanabe RM, Weedon MN, Willer CJ, Illig T, Hveem K, Hu FB, Laakso M, Stefansson K, Pedersen

O, Wareham NJ, Barroso I, Hattersley AT, Collins FS, Groop L, McCarthy MI, Boehnke M, Altshuler

D, Consortium D, Consor WTCC: Meta-analysis of genome-wide association data and large-scale

replication identifies additional susceptibility loci for type 2 diabetes. Nat Genet

2008;40:638-645

43. Lee J, Taneja V, Vassallo R: Cigarette smoking and inflammation: cellular and molecular

mechanisms. J Dent Res 2012;91:142-149

44. Isik B, Ceylan A, Isik R: Oxidative stress in smokers and non-smokers. Inhal Toxicol

2007;19:767-769

45. Halim M, Halim A: The effects of inflammation, aging and oxidative stress on the

pathogenesis of diabetes mellitus (type 2 diabetes). Diabetes Metab Synd 2019;13:1165-1172

46. Sies H: Oxidative stress: oxidants and antioxidants. Exp Physiol 1997;82:291-295

47. Allen RG, Tresini M: Oxidative stress and gene regulation. Free Radic Biol Med

2000;28:463-499

48. Jarvis MJ, Russell MA, Feyerabend C: Absorption of nicotine and carbon monoxide from

passive smoking under natural conditions of exposure. Thorax 1983;38:829-833

29

Tables

(Discovery cohort / Replication cohort) Overall Smokers Non-smokers

Total, n 25,460 / 58,774 3,005 / 5,173 22,455 / 53,601

Males, n (%) 12,800 (50.3) / 19,017 (32.4) 2,647 (88.1) / 4,002 (77.4) 10,153 (45.2) / 15,015 (28.0)

Age (years), mean (s.d.) 48.9 (11.1) / 50.4 (10.6) 46.4 (9.9) / 47.6 (10.3) 49.2 (11.2) / 50.7 (10.6)

BMI (kg/m2), mean (s.d.) 24.3 (3.7) / 24.2 (3.8) 25.4 (3.9) / 25.3 (4.1) 24.2 (3.7) / 24.1 (3.7)

Drinking, n (%) 1,799 (7.1) / 3,229 (5.5) 728 (24.2) / 1,218 (23.5) 1,071 (4.8) / 2,011 (3.8)

Regular exercise, n (%) 10,423 (40.9) / 24,325 (41.4) 858 (28.6) / 1,473 (28.5) 9,565 (42.6) / 22,852 (42.6)

Educational attainment, mean (s.d.) 5.5 (1.0) / 5.5 (1.0) 5.4 (0.9) / 5.4 (0.9) 5.5 (1.0) / 5.5 (1.0)

Fasting glucose (mg/dL), mean (s.d.) 96.3 (21.0) / 95.8 (20.7) 99.4 (26.2) / 99.3 (27.6) 95.9 (20.1) / 95.5 (19.9)

HbA1c (%), mean (s.d.) 5.73 (0.80) / 5.80 (0.81) 5.80 (0.93) / 5.91 (1.07) 5.72 (0.78) / 5.79 (0.78)

HbA1c (mmol/mol), mean (s.d.) 39 (8.7) / 40 (8.9) 40 (10.2) / 41 (11.7) 39 (8.5) / 40 (8.5)

Fasting glucose > 126 mg/dL, n (%) 1,114 (4.4) / 2,478 (4.2) 201 (6.7) / 344 (6.6) 913 (4.1) / 2,134 (4.0)

HbA1c > 6.5 % (48 mmol/mol), n (%) 1,752 (6.9) / 4,334 (7.4) 266 (8.9) / 530 (10.2) 1,486 (6.6) / 3,804 (7.1)

Subjects with physician-diagnosed diabetes, n

(%)

1,260 (4.9) / 3,099 (5.3) 183 (6.1) / 330 (6.4) 1,077 (4.8) / 2,769 (5.2)

Subjects with diabetes, n (%) 1 2,207 (8.7) / 5,308 (9.0) 335 (11.1) / 635 (12.3) 1,872 (8.3) / 4,673 (8.7)

30

Table 1. Basic characteristics stratified by “active smoking status”

1. Subjects with diabetes included those with physician-diagnosed diabetes, or those having FG > 126 mg/dL or HbA1c > 6.5 % (48

mmol/mol) according to the TWB test results.

31

(Discovery cohort / Replication

cohort) Fasting glucose (mg/dL) HbA1c (%) Diabetes (dichotomous trait)

Explanatory variables in

regression model (1) 1

Percent change

(%) p-value

Percent change

(%) p-value Odds ratio p-value

Sex

(1: female vs. 0: male) -3.79

2 (-3.37) 8.4E-79 (1.5E-124) -0.68 (-0.82) 4.2E-6 (5.9E-15) 0.72 (0.71) 6.9E-10 (1.8E-23)

Age

(in years, continuous variable) 0.34 (0.36) 1.3E-273 (0

3) 0.29 (0.29) 0

3 (0

3) 1.08 (1.08) 9.0E-180 (0

3)

Body mass index

(in kg/m2, continuous variable)

0.87 (0.85) 1.2E-239 (0 3) 0.70 (0.73) 4.0E-297 (0

3) 1.18 (1.18) 1.2E-155 (0

3)

Active smoking status

(1: yes vs. 0: no) 0.81 (0.88) 0.009 (1.3E-4) 0.83 (1.32) 2.1E-4 (5.7E-15) 1.36 (1.36) 1.3E-5 (2.4E-9)

Drinking status

(1: yes vs. 0: no) 1.32 (1.55) 4.9E-4 (2.7E-8) -0.96 (-1.28) 4.2E-4 (2.2E-10) 0.987 (0.89) 0.88 (0.06)

Regular exercise

(1: yes vs. 0: no) -0.67 (-0.90) 7.5E-4 (5.6E-12) -0.93 (-0.98) 1.3E-10 (6.0E-25) 0.88 (0.87) 0.011 (6.7E-6)

Educational attainment

(a value ranging from 1 to 7) -0.60 (-0.51) 1.4E-8 (4.1E-14) -0.51 (-0.19) 4.0E-11 (1.1E-4) 0.91 (0.89) 1.5E-4 (2.6E-15)

R-square 4

13.7 % (13.1%) 14.1 % (13.9%) 13.7 % (13.2 %)

Table 2. Results of regression model (1) (prior to GRS analysis)

1. Natural log transformed fasting glucose (or HbA1c) was regressed on sex, age, body mass index, active smoking status, drinking

status, regular exercise, educational attainment, and the first 10 PCs. To save space, we here omit the results of the 10 PCs.

2. Because fasting glucose (or HbA1c) was natural log transformed, (𝑒𝑥𝑝(�̂�𝑆𝑒 ) 1) × 100 is shown to represent the change in

32

fasting glucose (or HbA1c) between females and males, while adjusting for age, body mass index, active smoking status, drinking

status, regular exercise, educational attainment, and the first 10 PCs. For example, women on average have lower fasting glucose

than men by 3.79% (𝑃𝑆𝑒 −𝐹𝐺 = 8.4 × 10−79).

3. A p-value of 0 means that the test is extremely significant.

4. For continuous traits, R-square is the proportion of variance in natural log transformed fasting glucose (or HbA1c) that can be

explained by sex, age, body mass index, active smoking status, drinking status, regular exercise, educational attainment, and the

first 10 PCs. For the dichotomous trait (diabetes status), we present pseudo R-square, defined as one minus the ratio of the log

likelihood with intercepts only, and the log likelihood with all predictors.

33

(Discovery cohort / Replication cohort)

(D/R) Fasting glucose (mg/dL) HbA1c (%) Diabetes (dichotomous trait)

Smoking measurements in regression

model (1)

Percent change

(%) p-value

1

Percent change

(%) p-value

1 Odds ratio p-value

1

Active smoking status

(1: yes vs. 0: no) 0.81 (0.88) 0.009 (1.3E-4) 0.83 (1.32) 2.1E-4 (5.7E-15) 1.36 (1.36) 1.3E-5 (2.4E-9)

The number of pack-years

(mean±sd, D: 19.6±17.0; R: 19.5±17.6) 0.06 (0.05) 2.3E-7 (8.8E-9) 0.06 (0.06) 4.0E-12 (5.8E-22) 1.01 (1.01) 5.9E-9 (6.5E-12)

Years as a smoker

(mean±sd, D: 25.6±10.1; R: 26.3±10.8) 0.04 (0.04) 8.6E-4 (3.2E-6) 0.04 (0.05) 2.9E-6 (3.4E-20) 1.01 (1.01) 1.1E-6 (3.5E-11)

Packs smoked per day

(mean±sd, D: 0.72±0.51; R: 0.69±0.52) 1.70 (1.33) 4.9E-7 (3.6E-7) 1.63

2 (1.70) 3.4E-11 (7.0E-19) 1.51 (1.41) 1.1E-9 (2.3E-11)

Hours as a passive smoker per week

(mean±sd, D: 5.6±11.0; R: 5.4±10.6) 0.03 (0.04) 0.18 (0.02) 0.01 (0.03) 0.52 (0.03) 1.01 (1.01) 0.04 (0.01)

Table 3. Results of regression model (1) when replacing “active smoking status” with other smoking measurements (prior to

GRS analysis)

1 A total of 15 trait-smoking associations were tested here. A trait-smoking association is significant if 𝒑 < 𝟎.𝟎𝟓

𝟏𝟓= 𝟎. 𝟎𝟎𝟑𝟑.

2. Because HbA1c (or FG) was natural log transformed, (𝑒𝑥𝑝(�̂�𝑝𝑎𝑐𝑘𝑠) 1) × 100 = 1.63 (𝑃𝑝𝑎𝑐𝑘𝑠−𝐻𝑏𝐴1𝑐 = 3.4 × 10−11) is the

change in HbA1c that is associated with a pack increase of active cigarette smoking per day, while adjusting for sex, age, body

mass index, drinking status, regular exercise, educational attainment, and the first 10 PCs. The detailed result of regression model

(1) can be found in Supplemental Table S3.

34

Fasting glucose (mg/dL) HbA1c (%) Diabetes (dichotomous trait)

Explanatory variables in regression

model (4) 1

Percent

change (%) p-value

Percent

change (%) p-value Odds ratio p-value

Genetic risk score (GRS)

(z-score standardized) 1.72 2.2E-170 1.30 1.9E-190 1.27 4.4E-35

Sex

(1: female vs. 0: male) -3.32 1.3E-121 -0.81 1.0E-14 0.71 5.6E-23

Age

(in years, continuous variable) 0.36 0

2 0.29 0

2 1.08 0

2

Body mass index

(in kg/m2, continuous variable)

0.85 0 2 0.74 0

2 1.18 0

2

Active smoking status

(1: yes vs. 0: no) 0.60 0.08 0.88 3.7E-4 1.19 0.09

GRS × Active smoking status

(continuous variable) 1.05

3 2.8E-5 0.51 0.002 1.16 0.0046

Drinking status

(1: yes vs. 0: no) 2.09 2.4E-11 -0.93 3.6E-5 0.90 0.13

Regular exercise

(1: yes vs. 0: no) -0.92 1.1E-12 -1.02 4.2E-27 0.86 3.4E-6

Educational attainment

(a value ranging from 1 to 7) -0.50 8.7E-14 -0.21 2.5E-5 0.89 1.3E-15

R-square 4

14.3 % 15.4 % 14.0 %

Table 4. Results of regression model (4) when the smoking measurement is “active smoking status” (including GRS and

GRS-smoking interaction)

35

1. TWB1 was used to find trait-associated SNPs and TWB2 was used to test for interactions. Natural log transformed fasting glucose

(or HbA1c), or diabetes status, was regressed by model (4). To save space, we here omit the results of the 10 PCs, 𝐺𝑅𝑆 ×

𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑠, and 𝑆𝑚𝑜𝑘𝑖𝑛𝑔 × 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑠.

2. A p-value of 0 means that the test is extremely significant.

3. Because fasting glucose (or HbA1c) was natural log transformed, (𝑒𝑥𝑝(�̂�𝐼𝑁𝑇) 1) × 100 = 1.05 represents that each 1 sd

increase in GRS is associated with a 1.05% higher fasting glucose in (active) smokers than in non-smokers (𝑃𝐼𝑁𝑇 = 2.8 × 10−5).

4. For continuous traits, R-square is the proportion of variance in natural log transformed fasting glucose (or HbA1c) that can be

explained by the explanatory variables shown in model (4). For the dichotomous trait (diabetes status), we present pseudo

R-square, defined as one minus the ratio of the log likelihood with intercepts only, and the log likelihood with all predictors.

36

Fasting glucose (mg/dL) HbA1c (%) Diabetes (dichotomous trait)

Smoking measurements in regression

model (4)

Percent change

(%) 𝑷𝑰𝑵𝑻

Percent change

(%) 𝑷𝑰𝑵𝑻 Odds ratio 𝑷𝑰𝑵𝑻

GRS × Active smoking status 1.05 2.8E-5 * 0.51 0.002 * 1.16 0.0046

GRS × The number of pack-years 0.06 1.6E-7 * 0.02 7.3E-4 * 1.002 0.15

GRS × Years as a smoker 0.04 8.5E-6 * 0.02 2.5E-3 * 1.004 0.0048

GRS × Packs smoked per day 1.68 1.9E-7 * 0.69 4.5E-4 * 1.09 0.12

GRS × Hours as a passive smoker per

week 0.02 0.25 -0.007 0.58 0.996 0.30

Table 5. Results of regression model (4) when replacing “active smoking status” with the 4 continuous smoking measurements

(including GRS and GRS-smoking interaction)

* TWB1 was used to find trait-associated SNPs and TWB2 was used to test for interactions. A total of 15 GRS-smoking interactions

were tested here. A GRS-smoking interaction is significant if 𝑷𝑰𝑵𝑻 <𝟎.𝟎𝟓

𝟏𝟓= 𝟎. 𝟎𝟎𝟑𝟑.

37

Figure legends

Figure 1 Average of FG/HbA1c and the prevalence of diabetes stratified by smoking status and

the quintiles of the FG/HbA1c/diabetes genetic risk score

The solid lines are for smokers, whereas the dotted lines are for non-smokers. The black lines

depict predicted mean FG/HbA1c or predicted prevalence of diabetes based on model (4). Only

subjects without any missing in covariates can be predicted. The blue lines mark crude mean

FG/HbA1c or crude prevalence of diabetes, without adjusting for any covariates. The number

shown around each point represents the sample size of that category.