Lower digit ratio (2D:4D) in alcohol dependence: Confirmation and exploratory analysis in a population‐based study of young men

Moderately sized, case‐control studies have related alcohol dependence in middle‐aged in‐patients to lower second‐to‐fourth finger length ratio (2D:4D), a proxy for prenatal hyperandrogenization. As primary aim, we here intended to confirm that lower 2D:4D is also associated with Diagnostic and Statistical Manual of Mental Disorders (DSM)–IV/‐5 alcohol dependence and alcohol use disorder in a large population‐based cohort of young males. Exploratory aims included underlying mechanisms. We analyzed self‐reported data on 2D:4D, DSM‐IV/‐5 criteria, anticipated subjective responses to alcohol, and willingness to purchase alcoholic drinks from 4989 Swiss men of the Cohort Study on Substance Use Risk Factors (C‐SURF). The mean of right‐hand 2D:4D and left‐hand 2D:4D was lower in men with DSM‐IV alcohol dependence than in those without (0.975 vs 0.981, P = .035) and lower in men with moderate to severe (0.974) than in those with mild (0.982, P = .001) or no (0.981, P = .003) DSM‐5 alcohol use disorder. Moreover, mean 2D:4D was lower in those reporting recent use of health services due to substance use problems (0.968 vs 0.981, P = .046). Lower mean 2D:4D correlated with a stronger anticipation to feel high following alcohol consumption (total cohort: ρ = −0.033, P = .026) and with a willingness to purchase more higher‐priced alcoholic drinks (DSM‐IV alcohol dependence subgroup: ρmin = −0.162, P = .002). This is the first population‐based study on young males to demonstrate lower 2D:4D in DSM‐IV alcohol dependence, DSM‐5 alcohol use disorder, and the related use of health care services. We also provide novel insight into cognitive‐behavioral mechanisms. These results should help to establish more effective preventive and therapeutic strategies targeting 2D:4D and prenatal androgen exposure.


| INTRODUCTION
Alcohol dependence and alcohol use disorder are among the most prevalent and burdening psychiatric disorders worldwide. They are subject to a distinct gender dimorphism. In 2016, alcohol was responsible for more deaths in men than in women (2.3 vs 0.7 million), and more men than women (237 vs 46 million) were affected by alcohol use disorder worldwide. 1 In line with genetic evidence, 2-7 this suggests that sex hormone activity is involved in alcohol dependence.
Hyperandrogenization during prenatal and early postnatal life programs behavioral masculinization and permanently affects neural architecture. 8 It has been speculated that early androgen effects also contribute to greater male phenotypic variability and increased environmental susceptibility. 9 A growing body of evidence from animal and human studies indicates that increased prenatal androgen exposure predisposes to develop alcohol dependence in adulthood in a sex-specific manner.
In male mice, prenatal androgen receptor inhibition by flutamide decreases alcohol intake during adulthood and, in female mice, prenatal androgen receptor activation by dihydrotestosterone increases adult alcohol intake. 10 Prenatal exposure to excess testosterone increases the number of tyrosine hydroxylase-immunoreactive cells in the ventral tegmental area of adult ewes. 11 Higher amniotic testosterone predicts increased behavioral approach tendencies in 8-to 11-year-old children by influencing brain regions (caudate, putamen, nucleus accumbens) to be more responsive to positive compared with negatively valenced cues. 12 These studies suggest that prenatal androgens program the fetal brain's reward system, which is relevant to addictive behavior later in life.
In humans, it is hardly feasible to directly investigate the role of prenatal androgen exposure in alcohol dependence due to ethical constraints and the long time period between the intrauterine window and adulthood. Biomarkers of prenatal androgen exposure have been established. The most widely studied proxy is the second-to-fourth finger length ratio (2D:4D). Lower 2D:4D is indicative of higher prenatal androgen exposure 13,14 (for a critical review, see previous studies 9,15 ).
To date, three case-control studies have investigated the relationship between 2D:4D and alcohol dependence. These have consistently reported lower 2D:4D in male alcohol-dependent in-patients compared with male control subjects. [16][17][18] The findings indicate that higher prenatal androgen exposure predisposes males to develop a dependence on alcohol in adulthood. The prenatal androgen activity model of alcohol dependence 19 has been supported by observations of weaker transient evoked otoacoustic emissions and a higher incidence of late pubertal onset, both also proxies for increased prenatal androgen exposure, in male alcohol-dependent in-patients relative to control subjects. 18 As for females, the data remain controversial because one study supports the effect 17 and the only other failed to find evidence thereof. 18 Three studies have been published on the relationship between 2D:4D and alcohol dependence in humans, [16][17][18] yet these are subject to several limitations: (a) Their sample sizes are limited to less than 500 participants in each of these investigations. (b) The alcoholdependent in-patients were compared with control subjects from the general population. The use of different target populations to recruit participants likely induced bias. Lower 2D:4D has been related to lower educational attainment, 20 16 (e) The cognitive-behavioral mechanisms underlying the association between 2D:4D and alcohol dependence remain unexplored.

| Aims of the study
We conducted this study to address the above-reported limitations. The primary aim was to confirm that lower 2D:4D is related to DSM-IV alcohol dependence and DSM-5 alcohol use disorder by analyzing the data from a large, population-based cohort of young males. To provide novel and mechanistic insight into the association between 2D:4D and alcohol dependence, the exploratory aims were to investigate associations of 2D:4D with DSM-IV alcohol abuse, single DSM-IV/-5 criteria, use of health services due to substance use problems, anticipated subjective responses to standard alcoholic drinks, and willingness to purchase standard alcoholic drinks across a range of prices.

| Study sample
We analyzed data from the third survey wave of the Cohort Study on Substance Use Risk Factors (C-SURF; www.c-surf.ch). 25  The participants were also asked whether they had been in an emergency department, ambulatory care, or a special clinic due to problems with substance use during the last 12 months (Questionnaire 3 ID: B24).

| Behavioral phenotyping
An adapted version of the drug effects questionnaire 26 was used to quantify the participants' anticipation of how they would feel immediately after consuming five standard alcoholic drinks containing approximately 10 g of alcohol each. They were asked about their expectation to feel any effect from the substance, to feel high, to like the effects, to dislike the effects, and to want more alcohol on an 11-point scale from 0 (not at all), over 5 (moderately), to 10 (extremely) (Questionnaire 3 ID: D13). Moreover, an adapted version of the simulated alcohol purchase task 27 was conducted to quantify the participants' willingness to pay for standard alcoholic drinks across a broad price range. They were asked "How many standard drinks with alcohol would you have if … Drinks are free?/Every drink costs 50 cents?/Every drink costs 1/2/3/4/6/8/ 10/15/20 Swiss francs?" (Questionnaire 3 ID: D12).

| 2D:4D
Similar to the methods described by Reimers 28

| Ethical approval
The study was approved by the Ethics Committee for Clinical Research of Lausanne University Medical School (Protocol 15/07).

| Sociodemographic characteristics
The DSM-IV alcohol dependence subgroup did not significantly differ from the DSM-IV nonalcohol dependence subgroup with regard to age, body height, body weight, body mass index, level of education, or whether the participants were gainfully employed (Table 1). DSM-IV alcohol dependence was significantly associated with more standard alcoholic drinks per drinking day and higher smoking prevalence (OR 2.8). Both subgroups differed significantly in how they covered their living expenses and in their civil status. Individuals in the DSM-IV alcohol dependence subgroup were significantly more likely to be eliminated from the M2D:4D (OR 1.9) and the R2D:4D (OR 2.2) subcohorts but not from the L2D:4D sub-cohort (for the total cohort, see Table 1; for descriptive statistics of 2D:4D and details of M2D:4D, R2D:4D, and L2D:4D sub-cohorts, see Tables S1 and S2).  Table S3). The number of fulfilled DSM-IV alcohol dependence criteria, DSM-5 alcohol use disorder criteria, or DSM-IV alcohol abuse criteria did not significantly correlate with M2D:4D, R2D:4D, L2D:4D, or 2D:4Dr-l (for details, see Table S4). We explored associations between 2D:4D and the single criteria for DSM-IV alcohol dependence, DSM-5 alcohol use disorder, and DSM-IV alcohol abuse. These analyses revealed significantly lower M2D:4D and L2D:4D in participants who stated that, during the past 12 months, their drinking had caused them to miss a class, work, or to fail to look after their family more than once  Table S5).

|
We found lower M2D:4D in study subjects with at least one visit in emergency departments, ambulatory care, or a special clinic due to problems with substance use during the past 12 (Table 2).
2D:4D did not significantly correlate with the anticipated subjective responses in the alcohol dependence subgroup (Table S6).

| Exploratory aims: 2D:4D and simulated alcohol purchase task
For higher-priced standard alcoholic drinks (ie, at least six Swiss francs each), lower 2D:4D correlated significantly with the willingness to purchase more standard alcoholic drinks in the DSM-IV alcohol dependence subgroup (Table 3). 2D:4D did not significantly correlate with the number of standard alcoholic drinks when these cost less than six Swiss francs each. There were also no convincing associations between 2D:4D and willingness to purchase standard alcoholic drinks in the total cohort or the DSM-IV nonalcohol dependence subgroup, despite the larger sample size (Table S7).

| Positive control: 2D:4D, body mass index, body height, body weight, and age
We aimed to confirm previously reported correlations of 2D:4D with body mass index and age 29,31,32 as positive controls. In this study, lower M2D:4D correlated significantly with taller body height and lower R2D:4D with lower body mass index and younger age (Table 4).    33 We found only a few minor associations between single DSM-IV/-5 items and 2D:4D, suggesting that 2D:4D influences alcohol dependence via a complex interaction between multiple behavioral traits rather than a single trait.

| Mechanistic investigation
The present study provides the first evidence that to feel high following alcohol consumption, to dislike the effects of alcohol, and to be  35 In light of these investigations, the present study's

| Additional strengths and limitations
The 7.6% prevalence of DSM-IV alcohol dependence observed in this study cohort of young males is consistent with other population-based estimates (4.6% to 6.1% for males and females combined, aged 21-29 years, Germany 21 ; 6.1% for males, all age groups combined, Europe 1 ). and nonalcohol dependence subgroups were well-balanced in terms of many sociodemographic characteristics. We also replicated previous findings as positive controls. The sensitivity analysis not only confirmed the robustness of the major findings but also showed higher effect sizes (see supporting information).

Moreover
Taken as a whole, the observed effect sizes are small, which might be explained by several underlying factors. Addictive behaviors are caused by a complex interaction between bio-psycho-social factors.
Previous case-control studies investigated alcohol-dependent inpatients, whereas this study classified individuals from a population Note. P < .05 in bold print.
of young men as alcohol-dependent regardless of whether they were in treatment or not. Most likely, this resulted in less affected alcohol dependence and alcohol use disorder groups compared with previous investigations. Moreover, no direct clinical interviews have been conducted that may have reduced the diagnostic precision, although the prevalence estimate of alcohol dependence is in the expected range.
Self-measured 2D:4D is said to reach only 46% of the reliability conferred by expert-measured 2D:4D. 46 Furthermore, we did not assess finger deformation in this project, which has reduced precision. Moreover, we acknowledge that the use of 2D:4D as a proxy for prenatal androgen exposure has been criticized for not being a good indicator for individual differences in prenatal androgen exposure. The experimental evidence used to support the validity of 2D:4D as a biomarker of prenatal androgen exposure has not been replicated consistently. 9,15,47 Beyond the a priori-defined group comparisons, we conducted many exploratory statistical tests. We did not correct for multiple hypothesis testing, which might have entailed false-positive findings. Thus, future research is needed to confirm these novel results.

| Perspective
This study's finding of lower 2D:4D in alcohol-dependent individuals in a population-based cohort informs the generation of more effective preventive strategies targeting the development of 2D:4D. There is some evidence from animal and human studies that maternal stress, smoking, and alcohol consumption during pregnancy are related to lower 2D:4D in the offspring. 18,48,49 Currently, a prospective, controlled, and investigator-blinded study is being conducted that examines how a stress-reducing mindfulness-based intervention carried out by pregnant women affects their children's 2D:4D. 50 Moreover, the novel mechanistic findings reported here might encourage future research using 2D:4D as a biomarker and help to establish effective preventive and therapeutic strategies.

| CONCLUSIONS
To the best of our knowledge, this study is the first to establish in a population-based cohort of young males that lower 2D:4D relates to DSM-IV alcohol dependence, DSM-5 alcohol use disorder, and the use of health care services due to problems with substance use and that the anticipation of feeling high following alcohol consumption and the willingness to purchase higher-priced standard alcoholic drinks represent underlying cognitive-behavioral mechanisms. These observations support the model that prenatal androgen exposure increases the risk of alcohol dependence in adulthood and provide a basis for establishing novel preventive and therapeutic strategies.

ACKNOWLEDGMENTS
We would like to thank the C-SURF participants for their continuing support of the research project. Moreover, we thank the reviewers for their constructive suggestions.

CONFLICT OF INTERESTS
The authors declare no conflict of interest.