An atlas on risk factors for type 2 diabetes: a wide-angled Mendelian randomisation study

Aims/hypothesis The aim of this study was to use Mendelian randomisation (MR) to identify the causal risk factors for type 2 diabetes. Methods We first conducted a review of meta-analyses and review articles to pinpoint possible risk factors for type 2 diabetes. Around 170 possible risk factors were identified of which 97 risk factors with available genetic instrumental variables were included in MR analyses. To reveal more risk factors that were not included in our MR analyses, we conducted a review of published MR studies of type 2 diabetes. For our MR analyses, we used summary-level data from the DIAbetes Genetics Replication And Meta-analysis consortium (74,124 type 2 diabetes cases and 824,006 controls of European ancestry). Potential causal associations were replicated using the FinnGen consortium (11,006 type 2 diabetes cases and 82,655 controls of European ancestry). The inverse-variance weighted method was used as the main analysis. Multivariable MR analysis was used to assess whether the observed associations with type 2 diabetes were mediated by BMI. We used the Benjamini–Hochberg method that controls false discovery rate for multiple testing. Results We found evidence of causal associations between 34 exposures (19 risk factors and 15 protective factors) and type 2 diabetes. Insomnia was identified as a novel risk factor (OR 1.17 [95% CI 1.11, 1.23]). The other 18 risk factors were depression, systolic BP, smoking initiation, lifetime smoking, coffee (caffeine) consumption, plasma isoleucine, valine and leucine, liver alanine aminotransferase, childhood and adulthood BMI, body fat percentage, visceral fat mass, resting heart rate, and four plasma fatty acids. The 15 exposures associated with a decreased risk of type 2 diabetes were plasma alanine, HDL- and total cholesterol, age at menarche, testosterone levels, sex hormone binding globulin levels (adjusted for BMI), birthweight, adulthood height, lean body mass (for women), four plasma fatty acids, circulating 25-hydroxyvitamin D and education years. Eight associations remained after adjustment for adulthood BMI. We additionally identified 21 suggestive risk factors (p < 0.05), such as alcohol consumption, breakfast skipping, daytime napping, short sleep, urinary sodium, and certain amino acids and inflammatory factors. Conclusions/interpretation The present study verified several previously reported risk factors and identified novel potential risk factors for type 2 diabetes. Prevention strategies for type 2 diabetes should be considered from multiple perspectives on obesity, mental health, sleep quality, education level, birthweight and smoking. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00125-020-05253-x) contains supplementary material, which is available to authorised users.


ESM Method
The review of risk factors in observational study Database: PubMed Time: Until January 10, 2020 In total, 1360 papers Search strategy: ((((((((((Meta- We firstly deleted all replicates with the same study materials. Since we aimed to pinpoint the possible risk factors for type 2 diabetes (not a quantitative metaanalysis), we merely screened the title and abstract of the studies. The studies with the same topic were included once regardless of used materials (study base). Results from this review are presented below: NA, not available; PMID, PubMed ID; SD indicates standard deviation; SNP, single-nucleotide polymorphism All biomarkers were measured in serum or blood levels, except for sodium and potassium in urinary levels. Source for Lifetime anxiety disorder: https://www.biorxiv.org/content/10.1101/203844v2.full *Not all these SNPs were included in the MR analyses because some were not associated with the exposure at the genome-wide significance level (p<5×10 -8 ), were in linkage disequilibrium, or not available in the type 2 diabetes dataset. Case definitions established a lifetime diagnosis of anorexia nervosa via hospital or register records, structured clinical interviews, or online questionnaires based on standardized criteria (Diagnostic and Statistical Manual of Mental Disorders (DSM) III-R, DSM-IV, International Classification of Diseases (ICD) 8, ICD-9 or ICD-10), whereas in the UK Biobank, cases self-reported a diagnosis of anorexia nervosa.

Lifetime anxiety disorder BioRxiv
Cases met one of two definitions. First was self-reporting a lifetime professional diagnosis of one of the core five anxiety disorders, (generalised anxiety disorder, social phobia, panic disorder, agoraphobia or specific phobia; n=21 108). Further case was defined as meeting criteria for a likely lifetime diagnosis of DSM-IV generalised anxiety disorder based on anxiety questions from the Composite International Diagnostic Interview (CIDI) Shortform questionnaire.
Post-traumatic stress disorder 31594949 PTSD assessment was based either on lifetime (where possible) or current PTSD (i.e. including participants with a potential lifetime PTSD diagnosis as controls), and PTSD diagnosis was established using various instruments and different versions of the DSM (DSM-III-R, DSM-IV, DSM-5). Schizophrenia 29483656 Unknown

Nutritional factor & Internal biomarker
Homocysteine 23824729 Homocysteine was measured in each cohort by using one of the following methods: isotope-dilution liquid chromatography-tandem mass spectrometry, gas chromatography-coupled mass spectrometry, HPLC, or enzymatic, immune, or chemiluminescence Isoleucine 27898682 measured using liquid chromatography coupled with tandem mass spectrometry; using gas chromatography mass spectrometry or ultra-performance liquid chromatography coupled with tandem mass spectrometry; or by nuclear magnetic resonance at 0 and 120 min during the course of an oral glucose tolerance test.

Lipoprotein(a) 20032323
Lp(a) lipoprotein was measured by means of one latex-enhanced immunoturbidimetric assay (Immuno)16 in samples from case subjects that had been obtained in study clinics. In addition, Lp(a) lipoprotein was measured in a random subgroup of case subjects and control subjects with the use of a second latex-enhanced immunoturbidimetric assay (Randox Laboratories) on an ADVIA 1800 autoanalyzer (Siemens).

Bilirubin levels 31801373
Serum total bilirubin levels (both direct and indirect bilirubin) were measured through automated biochemical profiling, and the unit of bilirubin concentration was milligram per deciliter. Coffee intake was collected using a 24-hour recall questionnaire (Oxford WebQ) in a subset of UK Biobank participants. Researchers used the mean intake from participants who completed at least two dietary recalls.

Caffeine intake 21490707
Intake was assessed using a validated semi-quantitative food frequency questionnaire (FFQ). For each item, participants were asked how often, on average, they had consumed a specified amount of each beverage or food over the past year. The participants could choose from nine frequency categories (never, 1-3 per month, 1 per week, 2-4 per week, 5-6 per week, 1 per day, 2-3 per day, 4-5 per day and 6 or more per day). Intakes of nutrients and caffeine were calculated using US Department of Agriculture food composition sources. In these calculations, researchers assumed that the content of caffeine was 137 mg per cup of coffee, 47 mg per cup of tea, 46 mg per can or bottle of cola or other caffeinated carbonated beverage, and 7 mg per 1 oz serving of chocolate candy.
Researchers assessed the total intake of caffeine by summing the caffeine content for the specified amount of each food multiplied by a weight proportional to the frequency of its use. Breakfast skipping 31190057 The breakfast cereal-skipping data from 24-h recalls from the UK Biobank

Lifetime smoking 31689377
Smoking measures available in the UK Biobank were self-reported and collected at initial assessment. They included: smoking status, age at initiation in years, age at cessation in years and number of cigarettes smoked per day. The lifetime smoking index was constructed based on the method outlined by Leffondré, Abrahamowicz, Xiao, and Siemiatyck (2006).

Daytime napping 31409809
Self-reported daytime sleepiness was ascertained in the UK Biobank using the question "How likely are you to dose off or fall asleep during the daytime when you don't mean to? (e.g. when working, reading or driving)" with the response options of "Never/rarely", "sometimes", "often", "all of the time", "do not know", and "prefer not to answer". Participants reporting "do not know" and "prefer not to answer" were set to missing. Other responses were coded continuously as 1 to 4 corresponding to the severity of daytime sleepiness. Sleep duration 30846698 Participants were asked: About how many hours sleep do you get in every 24 h? (please include naps), with responses in hour increments. Sleep duration was treated as a continuous variable and also categorized as either short (6 h or less), normal (7 or 8 h), or long (9 h or more) sleep duration. Extreme responses of less than 3 h or more than 18 h were excluded17 and Do not know or Prefer not to answer responses were set to missing. Participants who self-reported any sleep medication (see Supplementary Method 1) were excluded.

Short sleep (<7 hours) 30846698
Long sleep (>9 hours) 30846698 Apnea-hypopnea index 31786426 The AHI was defined as the total number of apneas and hypopneas per hour of sleep. The delta-AHI is the increase/decrease of AHI across eight years as shown in the formula below: delta-AHI = AHI follow-up -AHI baseline OSA was defined according to AHI. Researchers elected to define OSA as AHI≥ 15 events/hour, a cutoff point that presents more solid evidence of associated outcomes. Individuals with incident OSA were those with an AHI < 15/h at baseline, which presented an AH≥15h at follow-up. Individuals that did not develop OSA (controls) were those with an AHI ≤ 15h at both baseline and follow-up studies.

Insomnia 30804565
Insomnia complaints were measured using questionnaire data; an independent sample (the Netherlands Sleep Register)12, which gives access to similar question data, as well as clinical interviews assessing insomnia disorder, was used to validate the specific questions so that they were good proxies of insomnia disorder.

Morningness 30696823
Responses to two identical questions ("Are you naturally a night person or a morning person?") were used to define the dichotomous morning person phenotype in the 23andMe cohort, with one question having a wider selection of neutral options. For the first instance, the possible answers were "Night owl", "Early bird" and "Neither", and for the second "Night person", "Morning person", "Neither", "It depends" and "I'm not sure". Individuals with discordant or neutral responses to both were excluded. For those with one neutral and one nonneutral response, their non-neutral response was used to define their phenotype. Morning people were coded as 1 (cases; N = 120,478) and evening people were coded as 0 (controls; N = 127,622). The UK Biobank collected a single self-reported measure of Chronotype ("Morning/evening person (chronotype)"; data-field 1180). Participants were prompted to answer the question "Do you consider yourself to be?" with one of six possible answers: "Definitely a 'morning' person", "More a 'morning' than 'evening' person", "More an 'evening' than a 'morning' person", "Definitely an 'evening' person", "Do not know" or "Prefer not to answer", which we coded as 2, 1, −1, −2, 0 and missing, respectively.

Restless leg syndrome 29029846
People with restless legs syndrome were recruited in specialist outpatient clinics for movement disorders and in sleep units. Restless legs syndrome was diagnosed in face-to-face interviews by an expert neurologist, based on the International Restless Legs Syndrome Study Group diagnostic criteria

Moderate to vigorous physical activity 29899525
Moderate-to-vigorous PA (MVPA) was calculated by taking the sum of total minutes/week of MPA multiplied by four and the total number of VPA minutes/week multiplied by eight, corresponding to their metabolic equivalents.

Strenuous sports or other exercises 29899525
In the last 4 weeks did you spend any time doing the following?" and follow-up questions assessing the frequency and typical duration of "strenuous sports" and of "other exercises". The possible responses to the initial question were: 'walking for pleasure', 'other exercises', 'strenuous sports', 'light DIY', 'heavy DIY', 'none of the above', and 'prefer not to answer'. Researchers identified individuals spending 2-3 days/week or more doing strenuous sports or other exercises (SSOE), for a duration of 15-30 min or greater.

Vigorous physical activity 29899525
For vigorous PA (VPA), participants were asked: "In a typical WEEK, how many days did you do 10 min or more of vigorous physical activity? (These are activities that make you sweat or breathe hard such as fast cycling, aerobics, heavy lifting)". For each of these questions, those who indicated 1 or more such days were then asked "How many minutes did you usually spend doing moderate/vigorous activities on a typical DAY".
Participants were asked to include activities performed for work, leisure, travel and around the house. We excluded individuals who selected "prefer not to answer" or "do not know" on the above questions, those reporting not being able to walk, and individuals reporting more than 16 h of either MPA or VPA per day. Those reporting >3 h/day of VPA or MPA were recoded to 3 h, as recommended.

Sex-related factor
Age at menarche 28436984 Age at menarche in women. Age at voice breaking in men (23andme: How old were you when your voice began to crack/deepen? UKBB: When did your voice break?) Age at menopause 26414677 ANM was self-reported and defined as the age at last naturally occurring menstrual period followed by at least 12 consecutive months of amenorrhea. Testosterone levels 32042192 Unknown Sex hormone binding globulin 32042192 Unknown

Estradiol levels 29325096
In six discovery cohorts (FHS, GOOD, MrOS Sweden Gothenburg, MrOS Sweden Malmo¨ , andMrOs United States), measurements were performed using either GC-MS or liquid chromatography tandem mass spectrometry. In the remaining discovery cohorts (LURIC, Invecchiare in Chianti, Multi-Ethnic Study of Atherosclerosis, and RS1) measurements were performed using immunoassays. In the replication cohort (EMAS), E2 was measured using the GC-MS technique.

Obesity-related factor & education
Birthweight 23202124 Birth weight was measured by trained personnel in some studies but in others was self-reported in adulthood.

Childhood BMI 26604143
Childhood obesity cases were defined as having a BMI ≥ 95th percentile, whereas childhood normal weight controls were defined as having a BMI < 50th percentile. Adulthood BMI 30124842 Self-reported and measured values Adulthood height 30124842 Body fat percentage 30305743 Body composition estimation by impedance measurement.

Visceral fat mass 31501611
Two UKBB subcohorts were constructed: one training dataset with VAT mass measured by DXA (instance 2; n = 5,109), to which the prediction models were calibrated; and one application dataset (instances 0 and 1; n = 502,638), in which VAT mass was predicted using the calibrated prediction models. Adiponectin levels 22479202 Adiponectin levels were measured using ELISA or RIA methods. Leptin levels 26833098 Circulating levels of leptin ESM The power estimation was based on 74124 type 2 diabetes cases and 824006 controls and we assumed type 1 error rate as 0.05. The webtool power calculator: https://shiny.cnsgenomics.com/mRnd/.