Open Access Research

Short-term abstinence from alcohol and

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
changes in cardiovascular risk factors,
liver function tests and cancer-related
growth factors: a prospective
observational study
Gautam Mehta,1 Stewart Macdonald,1 Alexandra Cronberg,2 Matteo Rosselli,1
Tanya Khera-Butler,2 Colin Sumpter,2 Safa Al-Khatib,1 Anjly Jain,3 James Maurice,1
Christos Charalambous,1 Amir Gander,4 Cynthia Ju,5 Talay Hakan,6
Roy Sherwood,7 Devaki Nair,8 Rajiv Jalan,1 Kevin P Moore1

To cite: Mehta G, Macdonald S, Abstract

Cronberg A, et al. Short-term Strengths and limitations of this study
Objective To assess changes in metabolic risk factors and
abstinence from alcohol and cancer-related growth factors associated with short-term
changes in cardiovascular ►► Prospective study design.
abstinence from alcohol.
risk factors, liver function ►► Recruitment of a control group.
Design Prospective, observational study.
tests and cancer-related ►► Thorough characterisation of the biological and life-
growth factors: a prospective Setting Single tertiary centre. style confounders.
observational study. BMJ Open Participants Healthy subjects were recruited based ►► Lack of randomisation to groups.
2018;8:e020673. doi:10.1136/ on intention to: (1) abstain from alcohol for 1 month ►► Study cohort all from university teaching hospital or
bmjopen-2017-020673 (abstinence group), or (2) continue to drink alcohol science magazine.
►► Prepublication history and
(control group). Inclusion criteria were baseline alcohol
additional material for this consumption >64 g/week (men) or >48 g/week (women).
paper are available online. To Exclusion criteria were known liver disease or alcohol risk of metabolic diseases such as type 2 diabetes and
view these files, please visit dependence. fatty liver disease.
the journal online (http://​dx.​doi.​ Primary and secondary outcome measures The
org/​10.​1136/​bmjopen-​2017-​ primary outcome was change in insulin resistance
020673). (homeostatic model assessment (HOMA) score). Secondary Introduction
outcomes were changes in weight, blood pressure (BP), Alcohol is a major cause of disability and
Received 16 November 2017
vascular endothelial growth factor (VEGF), epidermal
Revised 2 March 2018 preventable death. Globally, alcohol is the
Accepted 5 April 2018 growth factor (EGF) and liver function tests. Primary and
seventh leading risk factor overall in terms of
secondary outcomes were adjusted for changes in diet,
exercise and cigarette smoking. disability-adjusted life years (DALYs), and is
Results The abstinence group comprised 94 participants the leading risk factor globally in working age
(mean age 45.5 years, SD ±1.2) and the control group individuals (ages 15–59). Moreover, alcohol
47 participants (mean age 48.7 years, SD ±1.8). Baseline use-attributable DALYs have increased by over
alcohol consumption in the abstinence group was 25% in the last 25 years.1 European countries
258.2 g/week, SD ±9.4, and in the control group 233.8 g, have among the highest alcohol consump-
SD ±19.0. Significant reductions from baseline in the tion. Eastern Europe has the highest per
abstinence group (all p<0.001) were found in: HOMA score capita consumption worldwide,2 and in the
(−25.9%, IQR −48.6% to +0.3%), systolic BP (−6.6%, UK over 25% of the adult population drink
IQR −11.8% to 0.0%), diastolic BP (−6.3%, IQR −14.1% in excess of recommended guidelines.3
to +1.3%), weight (−1.5%, IQR −2.9% to −0.4%), VEGF Aside from liver disease, which is the third
(−41.8%, IQR −64.9% to −17.9%) and EGF (−73.9%,
most common cause of preventable death
IQR −86.1% to −36.4%). None of these changes were
in the UK, there is also a significant burden
associated with changes in diet, exercise or cigarette
smoking. No significant changes from baseline in primary
from alcohol-related cancer and meta-
or secondary outcomes were noted in the control group. bolic syndrome.3 Alcohol has been classi-
For numbered affiliations see Conclusion These findings demonstrate that abstinence fied by the WHO as a class I carcinogen for
end of article. some decades, and a report from the World
from alcohol in moderate–heavy drinkers improves insulin
Correspondence to resistance, weight, BP and cancer-related growth factors. Cancer Research Fund/American Insti-
Dr Gautam Mehta; These data support an independent association of alcohol tute for Cancer Research states that there is
​gautam.​mehta@​ucl.​ac.u​ k consumption with cancer risk, and suggest an increased convincing evidence that alcohol is causally

Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673 1

Open Access

related to cancers of the oral cavity, pharynx, larynx, AUDIT questionnaire (modified to capture data for the

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
oesophagus, breast and colorectum.4 preceding 6–8 months).
Moreover, it has long been recognised that there is Sample size calculation for the control group was
an important interaction between alcohol misuse and performed, based on pre/post data acquired from the
fatty liver disease.5 One of the main factors driving the abstinence group (table 1). Specifically, based on these
development of fatty liver disease and steatohepatitis is data, a power calculation determined that the following
insulin resistance. Thus, any action that improves insulin sample sizes were required to detect statistically signifi-
resistance will have a major impact on the development cant differences of the same magnitude (80% power,
and severity of fatty liver disease. However, there remains alpha 5%, two-sided test): HOMA score n=47, weight
debate as to the impact of alcohol consumption on fatty n=21, VEGF n=31, EGF n=30.
liver disease driven predominantly by insulin resistance BP was measured seated, following a 2 min rest period,
and metabolic factors.6 7 and the mean of three measurements was recorded.
In this climate of increased awareness of alcohol-re- Fasting blood was taken, between 08:00 and midday, for
lated morbidity, the UK Chief Medical Officers have measurement of glucose, insulin, liver function tests,
revised downwards their weekly guidance limits.8 Addi- lipids, carbohydrate deficient transferrin (abstinence
tionally, public health campaigns, where non-dependent group only) and VEGF (isoforms 165, 145 and 121) and
drinkers are encouraged to commit to short-term absti- EGF (Randox Investigator, Randox, Belfast, UK). The
nence from alcohol, are increasingly common. The aim HOMA score was calculated according to the methods
of this study was to assess changes in insulin resistance, of Matthews et al.11 Participants with diabetes requiring
metabolic risk factors and cancer-related growth factors treatment were excluded from HOMA measurements.
with short-term abstinence from alcohol in moderate
drinkers. Statistical analysis
Baseline and 1-month differences were analysed by
paired t-test for normally distributed differences in
Methods continuous variables, by Wilcoxon signed-rank test for
Study design non-normally distributed differences in continuous
This was a single-centre, prospective, observational study variables, and differences in categorical variables by Χ2
conducted at the Royal Free London NHS Foundation test. Differences between abstinence and control groups
Trust. Written informed consent was obtained from all were analysed by unpaired t-test for normally distributed
participants. Study recruitment was initiated through email variables, and Mann-Whitney test for variables that were
advertising within University College London, Queen Mary not normally distributed. Lifestyle factors were categor-
University of London and New Scientist Magazine. The ically graded (better/same/worse), and delta change in
entry criteria were baseline alcohol consumption of >64 g/ biological variables between lifestyle groups was assessed
week (eight units) for men or >48 g/week (six units) for by Kruskal-Wallis test. Multivariable logistic regression
women. Exclusion criteria were >3 days abstinence from analysis was also used to test the effect of abstinence on
alcohol prior to commencement of the study, the presence improvement in HOMA, weight, BP, VEGF and EGF once
of known liver disease or alcohol dependence. Participants other lifestyle factors (diet and exercise) were taken into
were not randomised to group, but were allocated based account. Correlation between biological variables was
on intention to maintain abstinence for 1 month (absti- assessed by Spearman’s correlation. All analyses were
nence group) or to continue alcohol consumption (control performed using STATA V.13.1 and SPSS Statistics V.21.0.
group). SD is reported for means and IQR for medians where
Participants were assessed at baseline, and after 1 month. applicable. All p values are two sided; p<0.01 was consid-
The primary outcome was change in insulin resistance ered significant to account for multiple comparisons.
(homeostatic model assessment (HOMA) score) at base-
Patient and public involvement
line and 1 month. Secondary outcomes were changes in
The research question was developed following public
weight, blood pressure (BP), vascular endothelial growth
feedback to a pilot project, conducted in collaboration
factor (VEGF), epidermal growth factor (EGF) and liver
with, and published by, New Scientist magazine (New
function tests. Information on diet, exercise and smoking
Scientist, 31 December 2013). Additionally, the research
history were obtained by self-reporting using components
question was informed by focus groups, funded through
of the Simple Lifestyle Indicator Questionnaire (SLIQ).9
the National Institute for Health Research Enabling
Self-reported alcohol intake was assessed at baseline
Involvement Fund. No specific patient advisers were
using the full Alcohol Use Disorders Identification Test
involved in the design or conduct of the study. Results of
(AUDIT) questionnaire, and a direct interview by a single
the study will be disseminated to all participants by email.
interviewer (KM) was also conducted to assess alcohol
intake over the preceding 2 months, using the timeline
follow back method.10 Additionally, a follow-up telephone Results
interview was conducted at 6–8 months to determine Ninety-seven participants were recruited to the absti-
drinking habits following the study period, using the full nence group, and forty-eight participants to the control

2 Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673

 Table 1 Baseline and 1 month variables for abstinence and control groups
Abstinence group Control group
Number P values Number P values
of paired (pre vs Effect of paired (pre vs Effect
Variable (units) values* Baseline 1 month post) size values* Baseline 1 month post) size
HOMA score, median (IQR) 86 1.4 (1.0–2.1) 1.0 (0.7–1.4) <0.001 −0.40 47 1.2 (0.9–1.5) 1.1 (0.9–1.5) 0.42 −0.08
Systolic blood pressure 93 136.0 (121.0–147.5) 125.0 (115.0–141.0) <0.001 −0.57 47 121.0 (118.0–134.0) 122.0 (116.0–131.0) 0.17 −0.14
(mm Hg), median (IQR)
Diastolic blood pressure 93 89.0 (80.0–95.5) 82.0 (77.0–89.0) <0.001 −0.34 47 71.0 (68.0–80.0) 75.0 (68.0–80.0) 0.77 −0.03
(mm Hg), mean (SD)
Weight (kg), median (IQR) 89 81.1 (67.0–89.7) 79.5 (66.7–87.2) <0.001 −0.49 47 73.8 (65.2–85.0) 72.6 (64.9–84.6) 0.09 −0.18
VEGF (ng/L), median (IQR) 81 7.6 (5.5–16.0) 4.0 (3.1–5.8) <0.001 −0.55 41 13.4 (9.0–17.9) 12.9 (8.3–18.5) 0.34 −0.10

Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673

EGF (ng/L), median (IQR) 81 7.2 (3.6–13.5) 1.5 (1.0–2.5) <0.001 −0.56 41 3.8 (1.0–10.3) 4.9 (1.0–9.5) 0.55 −0.07
Total cholesterol (mg/dL), 88 228.7 (198.4–258.4) 202.1 (176.3–227.4) <0.001 −0.85 47 208.8 (177.9–232.0) 205.0 (177.9–239.8) 0.25 −0.12
median (IQR)
LDL (mg/dL), median (IQR) 88 131.5 (112.3–163.1) 127.0 (97.4–153.8) <0.001 −0.42 47 112.1 (88.9–139.2) 112.1 (85.1–143.1) 0.16 −0.15
HDL (mg/dL), median (IQR) 88 74.4 (±22.3) 60.9 (±17.8) <0.001 −0.57 47 73.5 (58.0–88.9) 73.5 (58.0–92.8) 0.95 −0.02
Triglycerides (mg/dL), 88 79.3 (57.6–115.8) 76.6 (60.4–103.2) 0.05 −0.15 47 79.7 (62.0–124.0) 79.7 (62.0–106.3) 0.70 −0.05
median (IQR)
Gamma GT (IU/L), median 92 22.5 (14.0–34.8) 15.0 (10.3–24.0) <0.001 −0.53 46 39.8±8.3 44.8±11.0 0.26 −0.12
(IQR)
ALT (IU/L), median (IQR) 94 25.0 (18.0–37.0) 20.0 (16.0–29.3) <0.001 −0.29 47 24.0 (18.0–31.0) 25.0 (20.0–33.0) 0.06 −0.21
AST (IU/L), median (IQR) 94 23.0 (19.0–28.0) 21.0 (18.0–27.0) 0.03 −0.16 47 20.0 (17.0–24.0) 22.0 (19.0–27.0) <0.01 −0.27
Carbohydrate-deficient 81 0.7 (0.5–1.1) 0.6 (0.4–0.8) <0.001 −0.32 n/a n/a n/a n/a n/a
transferrin (%), median (IQR)

Effect size for normally distributed variables is calculated as [mean change in variable]/SD. Effect size for non-normally distributed variables is calculated as Wilcoxon signed-rank [test
statistic]/√ [ number of observations].
*Where the number of paired observations is less than 94 (abstinence group) or 47 (control group), this is due to missing data points.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; EGF, epidermal growth factor; HDL, high-density lipoprotein; HOMA, homeostatic model assessment; LDL, low-density
lipoprotein; VEGF, vascular endothelial growth factor.
Open Access

3
BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
Open Access

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
Figure 1 Flow chart of study participants.

group. Three subjects in the abstinence group and one type 2 diabetes. Levels of VEGF and EGF also markedly
subject in the control group did not attend for follow-up. reduced in the abstinence group, at −41.8% (IQR −64.9%
Thus, the final abstinence group comprised 94 partici- to −17.9%) and −73.9% (IQR −86.1% to −36.4%) respec-
pants (43 men, 51 women) mean age 45.5 years, SD ±1.2, tively (figure 2). Serum lipids (pre vs post) also improved
and the control group comprised 47 participants (22 in the abstinence group: fasting total serum cholesterol
men, 25 women) mean age 48.7 years, SD ±1.8. Mean (−13.4%, IQR −18.9% to −2.7%), low-density lipopro-
baseline alcohol intake for the abstinence group was tein (LDL) cholesterol (−9.4%, IQR −20.1% to +4.8%),
258.2 g/week, SD ±9.4 (men 275.9, SD ±25.5; women high-density lipoprotein (HDL) cholesterol (−16.7%,
243.1, SD ±12.8). All subjects in this group, except one IQR −25.0% to 0.0%). All the above variables were signifi-
individual, remained abstinent for the study period—this cantly reduced from baseline, p<0.001. By contrast, the
participant was included in all analyses. Mean weekly control group did not show significant changes from
baseline alcohol intake for the control group was 233.8 g, baseline in any of the above variables. Changes from base-
SD ±19.0 (men 270.2, SD ±26.6; women 200.2, SD ±25.8), line in HOMA score, VEGF, EGF, weight and systolic and
and was not significantly different at 1 month 260.1 g, diastolic BP are shown in figure 3.
SD ±20.8 (men 286.4, SD ±26.6; women 235.8, SD ±31.1), Liver function tests also improved in the abstinence
p=0.11. A flowchart of participants and observations is group; thus, there was a significant reduction in serum
shown in figure 1. alanine aminotransferase (ALT) (−14.5%, IQR −28.9
Baseline and 1-month variables for the abstinence and to +6.7%, p<0.001) and gamma GT (−28.6%, IQR −43.5
control groups are listed in table 1. There were no signif- to −14.4%, p<0.001), and a trend towards reduction
icant differences in baseline characteristics between absti- in serum aspartate aminotransferase (AST) (−5.4%,
nence and control groups, aside from baseline BP which IQR −​16.​two to +9.5%, p=0.03). No significant change
was significantly lower in the control group (systolic BP: in these variables was seen in the control group, aside
135.8 SD ±1.9 mm Hg vs 125.7 SD ±2.0 mm Hg, p<0.01; from a small rise in AST (+4.5%, IQR −5.6 to +23.1%,
diastolic BP: 87.7 SD ±1.2 mm Hg vs 74.3.7 SD ±1.5 mm p<0.01).
Hg, p<0.01). Antihypertensives were used in one partic- Lifestyle factors did not account for changes in the
ipant in the abstinence group, and one participant in abstinence group. No changes were seen in exercise
the control group. Lipid-lowering agents were used in score (10.9 SD ±4.7 vs 10.7 SD ±4.6, p=0.82) or cigarette
two participants in the abstinence group, one subject smoking (1.3 SD ±0.7 vs 1.4 SD ±0.7, p=0.17). A small
in the control group. These participants were excluded change in diet score was noted (from 8.2 SD ±3.3 to 8.8
from analyses for BP and lipids, respectively. Significant SD ±3.0, p=0.03). The pre/post differences in HOMA
reductions from baseline (pre vs post) in the abstinence score, weight, VEGF, EGF, triglycerides and HDL were
group were observed in: HOMA score (−25.9%, IQR distributed with a left (negative) skew. Therefore,
−48.6 to +0.3%), systolic BP (−6.6%, IQR −11.8% to non-parametric approaches were adopted to account
0.0%), diastolic BP (−6.3%, IQR −14.1% to +1.3%) and for lifestyle variables. Changes in HOMA score, BP and
weight (−1.5%, IQR −2.9% to −0.4%). HOMA score was weight in the abstinence group were not associated with
not performed due to type 1 diabetes in one participant changes in any lifestyle score (see online supplemen-
in the abstinence group. By chance, no participants had tary table 1). There was also no association between

4 Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673

 Open Access

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.

Figure 2 Baseline and 1-month data for the abstinence group presented as pre/post scatter plot (left) and bar chart chart
(right). Bar chart data are presented as median (IQR). Panels (clockwise from top right): homeostatic model assessment (HOMA)
score, weight, diastolic blood pressure (bp), epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), systolic
bp. Baseline and 1-month values were compared with Wilcoxon signed-rank test, p<0.01 taken as level of significance.

Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673 5

Open Access

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
Figure 3 Percentage change from baseline in homeostatic model assessment (HOMA) score, vascular endothelial growth
factor (VEGF), epidermal growth factor (EGF), systolic blood pressure (bp), diastolic bp and weight in abstinence (dark bar) and
control (light bar) groups. Data are presented as median (IQR). Changes from baseline in abstinence and control groups were
compared with Mann-Whitney test, p<0.01 taken as level of significance.

changes in HOMA score and weight (r=0.04, p=0.73). diet score was not associated with improvement in any of
However, changes in total cholesterol and LDL choles- these variables (table 2).
terol attained borderline significance between groups A further important result relates to follow-up ques-
when compared by change in diet in the abstinence tionnaire data, obtained in 77 individuals (81.9%) in
group (see online supplementary table 1; p=0.01 and the abstinence group and 40 (83.3%) in the control
p=0.02, respectively). group, at 6–8 months following the study period. In
Additionally, multivariable logistic regression anal- the abstinence group, a significant reduction in alcohol
ysis was used across the whole cohort, combining the consumption was maintained from their prestudy
abstinence and control groups, to determine predictors assessment. Thus, there was a significant reduction in
of: HOMA score reduction ≥20%, systolic BP reduc- overall AUDIT score from 10.0 (IQR 7.0 to 15.0) to 7.0
tion ≥5%, weight reduction ≥2%, VEGF reduction ≥20%, (IQR 5.0 to 9.0), p<0.001, and in the proportion of indi-
EGF reduction ≥20%. The model used covariates of absti- viduals with harmful use of alcohol (AUDIT score >8)
nence (yes/no) or change in exercise and diet SLIQ (61.0% vs 28.5%, p<0.001) at 6–8 months compared
score (better/same/worse). Abstinence was a highly with baseline. By contrast, in the control group, there
significant predictor of improvement in these biological was a non-significant trend to reduction in overall
variables (all p<0.01). By contrast, change in exercise and AUDIT score from 8.5 (IQR 6.3 to 12.0) to 8.0 (IQR

6 Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673

 Open Access

factors contributing to metabolic risk with good retest reli-

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
Table 2 Independent predictors of improvement in HOMA
score, systolic BP, weight, VEGF and EGF ability.9 As such, changes in HOMA score, weight and BP
were independent of changes in lifestyle as measured by
Biological variable (target
reduction) All study participants (n=141) the SLIQ score. Nevertheless, it remains possible that
the questionnaire scoring for diet, exercise and cigarette
Covariate OR 95% CI P values
smoking has inadequate sensitivity for all lifestyle changes
HOMA score (reduction ≥20%) within this cohort.
 Abstinence 3.48 1.60 to 7.53 0.002 The primary endpoint of insulin resistance, measured
 Exercise+diet SLIQ score 0.87 0.43 to 1.77 0.694 by HOMA score, showed a marked decrease (~25%)
Systolic BP (reduction ≥5%) following the cessation of alcohol consumption. Some
 Abstinence 6.47 2.71 to 15.46 <0.001 previous epidemiological data have supported a protec-
 Exercise+diet SLIQ score 1.47 0.70 to 3.08 0.310
tive effect of low-dose alcohol use on the risk of type 2
diabetes,12 although more recent work suggests this may
Weight (reduction ≥2%)
be due to incomplete adjustment for ‘sick quitters’,13 and
 Abstinence 15.63 3.54 to 68.95 <0.001 prospective alcohol intervention studies have provided
 Exercise+diet SLIQ score 0.74 0.323 to 1.69 0.475 mixed results.14 15 Our data support a positive associa-
VEGF (reduction ≥20%) tion of moderate–heavy alcohol use with an increased
 Abstinence 4.35 1.93 to 9.81 <0.001 risk of type 2 diabetes. Moreover, the observed effects of
 Exercise+diet SLIQ score 2.17 0.97 to 4.86 0.59 abstinence on HOMA score noted in this study are too
EGF (reduction≥20%)
dramatic to be accounted for by weight loss alone, and
no specific association was found between change in
 Abstinence 48.81 15.26 to 156.06 <0.001
HOMA score and weight. To our knowledge, this is the
 Exercise+diet SLIQ score 2.52 0.80 to 7.95 0.115
first paper to prospectively demonstrate a link between
Results are presented as adjusted odds ratios and 95% excess alcohol consumption and insulin resistance.
confidence intervals using multivariable logistic regression A major novel finding of this study is the rapid decrease
analysis.
in serum VEGF and EGF with short-term abstinence from
SLIQ, Simple Lifestyle Indicator Questionnaire; EGF, epidermal
growth factor; HOMA, homeostatic model assessment; VEGF, alcohol, which was seen in 90% of subjects in the absti-
vascular endothelial growth factor. nence group. Importantly, these changes were not seen in
the control group with continued alcohol consumption.
6.0 to 10.8), p=0.06), and no significant change in the Alcohol is causally related to the development of several
proportion with harmful use (50% vs 40%, p=0.37). cancers, including the digestive tract, nasopharynx, liver
and breast, and is classified as a class I carcinogen.4 16 The
increased risk caused by alcohol persists even at low levels
Discussion of consumption. The mechanism of mutagenesis is
This study is the first to comprehensively assess the effects thought to relate to direct effects of the alcohol metab-
of short-term abstinence from alcohol in a population olite, acetaldehyde.4 However, in this study, we chose to
of ‘healthy’ individuals, who are representative of the study VEGF and EGF, since they are key molecules in the
25% of the wider population who drink alcohol above multistep progression of cancer, are both highly expressed
national guidelines. The key findings of this study are in the solid tumours listed above and are common thera-
improvements in insulin resistance, BP, body weight and peutic targets for these tumours.17 VEGF plays a key role
a decrease in circulating concentrations of cancer-related in tumour progression through angiogenic pathways, and
growth factors following a month of abstinence from VEGF expression is driven by oncogene expression (eg,
alcohol. Ras, src, HER2, EGFR).18 EGF signalling contributes to
The strengths of this study are the prospective study oncogenesis by directly promoting cell proliferation,19
design, the recruitment of a control group and the thor- and expression levels are correlated with progressive
ough characterisation of the participant’s biological tumour growth and metastasis.20–22
and lifestyle data. A weakness is the lack of randomisa- Mechanistically, rodent models have demonstrated
tion of groups, although for ethical reasons the alloca- that alcohol exposure directly promotes the progression
tion of individuals to a predefined alcohol consumption of several cancers, including breast cancer. Lu et al have
regimen was inappropriate. A further weakness relates shown, in a mouse model of breast cancer, that alcohol
to the study cohort, who were recruited through staff at directly induces tumour angiogenesis and accelerated
university teaching hospitals and a science magazine, and tumour growth through a VEGF-dependent mechanism.23
thus probably had higher educational attainment and Similar evidence for an alcohol-VEGF pathway exists in
health-related motivation than the average population. A mouse models of colon cancer and melanoma.24 25 The
further confounder is the possibility of lifestyle change in EGF pathway has also been implicated in alcohol-related
the abstinence group, alongside abstinence from alcohol. breast cancer.26–28 The baseline levels of VEGF and EGF
We have tried to minimise the impact of these using the reported in this study are lower than reported in other
SLIQ questionnaire, a self-reported measure of lifestyle studies exploring associations of circulating VEGF/EGF

Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673 7

Open Access

levels with the occurrence of solid tumours.29–31 These to cancer risk, and the synergistic relationship between

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
differences are explained by the method of sample collec- alcohol and metabolic syndrome. Further attention
tion. The collection of blood into EDTA tubes, as in this should be directed to determining the durability of these
study, leads to reduced contribution of platelet-derived biological effects of abstinence, and conveying these
VEGF and EGF, and thus lower plasma concentrations.32 33 complex public health messages to the public.
Here, we demonstrate for the first time in humans an
association of abstinence from alcohol with a marked Author affiliations
1
Division of Medicine, UCL Institute for Liver and Digestive Health, London, UK
reduction in circulating concentrations of VEGF and 2
Camden and Islington Public Health, London, UK
EGF, which suggests that alcohol consumption per se 3
Department of Clinical Biochemistry, Royal Free London NHS Foundation Trust,
increases the concentrations of these growth factors. London, London, UK
There is strong evidence that these growth factors play 4
UCL Tissue Access for Patient Benefit, London, UK
5
an important role in oncogenesis. However, it would be Department of Anesthesiology, McGovern Medical School, University of Texas
Health Science Center at Houston, Houston, Texas, USA
wrong to speculate further on this observation without 6
Department of Liver Transplantation, HPB and Hepatology, Royal Free London NHS
longitudinal study in subjects who continue moderate Foundation Trust, London, UK
alcohol consumption. 7
Department of Clinical Biochemistry, King’s College Hospital NHS Foundation Trust,
These data also show the dynamic effect of regular London, UK
8
alcohol consumption on BP, an effect that is maintained Department of Clinical Biochemistry, Royal Free London NHS Foundation Trust,
London, UK
in healthy individuals with no history of hypertension
requiring medication. An effect of alcohol on BP has
Acknowledgements Administrative support for the study was provided by Patricia
long been recognised, with consumption greater than Langley.
two daily doses considered to be one of the most common Contributors GM contributed to the study design, participated in data collection,
reversible causes of hypertension.34 wrote the analytical plan and drafted and revised the paper. He is the guarantor.
Collectively, the above findings have implications for the SM participated in the study design, participated in data collection and drafted
risk of synergistic liver injury among individuals with risk and revised the paper. AC and TK-B analysed the data, and drafted and revised the
paper. CS participated in study design and revision of the paper. MR, SA-K, AJ, CC,
factors for alcohol-related liver disease (ALD) and fatty JM, AG and TH participated in data collection and revision of the paper. CJ, RS, DN
liver disease. Previous studies have emphasised an associ- and RJ contributed to the study design and revision of the paper. KM supervised the
ation between these pathways of liver injury, since serum study, contributed to the study design, participated in data collection and drafted
ALT among moderate drinkers is elevated to a greater and revised the paper.
extent in those with higher body mass index (BMI), Funding This work was funded by the Royal Free Charity, Camden and Islington
and ALD and fatty liver are pathologically similar. Two Public Health and the Royal Free London NHS Foundation Trust.
prospective cohort studies from Scotland have demon- Competing interests None declared.
strated an increased risk of liver disease with alcohol use Patient consent Not required.
and elevated BMI.35 More recently, a large prospective Ethics approval NRES Committee North West (14/NW/1510).
study of over 100 000 women in the UK confirmed a syner- Provenance and peer review Not commissioned; externally peer reviewed.
gistic association between alcohol and high BMI and risk
Data sharing statement All raw data are available on request from the
of chronic liver disease.36 Since alcohol use and insulin corresponding author.
resistance are both directly implicated in the develop- Open Access This is an Open Access article distributed in accordance with the
ment of steatohepatitis, the results of this study provide Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which
further support for this common causal pathway. Further, permits others to distribute, remix, adapt, build upon this work non-commercially,
changes in the gut microbiome have also been implicated and license their derivative works on different terms, provided the original work is
properly cited and the use is non-commercial. See: http://​creativecommons.​org/​
in the pathogenesis of steatohepatitis and obesity,37 and licenses/​by-​nc/​4.​0/
therefore changes in gut microbe populations following
© Article author(s) (or their employer(s) unless otherwise stated in the text of the
abstinence from alcohol are a further possible explana- article) 2018. All rights reserved. No commercial use is permitted unless otherwise
tion for the biological changes observed in this study. expressly granted.
These hypotheses merit further attention in subsequent
mechanistic studies.
A frequent criticism of public health strategies of short-
term abstinence (eg, Dry January) has been the lack of References
evidence of health benefits, or even negative effects on 1. GBD 2016 Risk Factors Collaborators. Global, regional, and national
comparative risk assessment of 84 behavioural, environmental and
longer-term alcohol consumption. occupational, and metabolic risks or clusters of risks, 1990-2016: a
Although this study has demonstrated health benefits systematic analysis for the Global Burden of Disease Study 2016.
Lancet 2017;390:1345–422.
from short-term abstinence, a possible misrepresenta- 2. Rehm J, Samokhvalov AV, Shield KD. Global burden of alcoholic liver
tion of these results is the concept that a 'detox' period is diseases. J Hepatol 2013;59:160–8.
all that is required to 'refresh' the liver or achieve other 3. Williams R, Aspinall R, Bellis M, et al. Addressing liver disease in the
UK: a blueprint for attaining excellence in health care and reducing
health gains. This is clearly untrue, since the durability of premature mortality from lifestyle issues of excess consumption of
the observed biological effects remains to be established. alcohol, obesity, and viral hepatitis. Lancet 2014;384:1953–97.
4. LoConte NK, Brewster AM, Kaur JS, et al. Alcohol and cancer: a
The data presented here provide supportive mechanistic statement of the american society of clinical oncology. J Clin Oncol
evidence for the recent changes in alcohol guidance due 2018;36:83–93.

8 Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673

 Open Access

5. Becker U, Deis A, Sørensen TI, et al. Prediction of risk of liver disease 21. Lo HW, Hung MC. Nuclear EGFR signalling network in cancers:

BMJ Open: first published as 10.1136/bmjopen-2017-020673 on 5 May 2018. Downloaded from http://bmjopen.bmj.com/ on October 7, 2024 by guest. Protected by copyright.
by alcohol intake, sex, and age: a prospective population study. linking EGFR pathway to cell cycle progression, nitric oxide pathway
Hepatology 1996;23:1025–9. and patient survival. Br J Cancer 2006;94:184–8.
6. Dunn W, Sanyal AJ, Brunt EM, et al. Modest alcohol consumption 22. Bracher A, Cardona AS, Tauber S, et al. Epidermal growth factor
is associated with decreased prevalence of steatohepatitis in facilitates melanoma lymph node metastasis by influencing tumor
patients with non-alcoholic fatty liver disease (NAFLD). J Hepatol lymphangiogenesis. J Invest Dermatol 2013;133:230–8.
2012;57:384–91. 23. Lu Y, Ni F, Xu M, et al. Alcohol promotes mammary tumor growth
7. Ekstedt M, Franzén LE, Holmqvist M, et al. Alcohol consumption is through activation of VEGF-dependent tumor angiogenesis. Oncol
associated with progression of hepatic fibrosis in non-alcoholic fatty Lett 2014;8:673–8.
liver disease. Scand J Gastroenterol 2009;44:366–74. 24. Tan W, Bailey AP, Shparago M, et al. Chronic alcohol consumption
8. Department of Health. UK chief medical officers’ alcohol guidelines stimulates VEGF expression, tumor angiogenesis and progression of
review: summary of the proposed new guidelines. https://www.​ melanoma in mice. Cancer Biol Ther 2007;6:1222–8.
gov.​uk/​government/​uploads/​system/​uploads/​attachment_​data/​file/​ 25. Gu JW, Bailey AP, Sartin A, et al. Ethanol stimulates tumor
489795/​summary progression and expression of vascular endothelial growth factor in
9. Godwin M, Streight S, Dyachuk E, et al. Testing the simple lifestyle chick embryos. Cancer 2005;103:422–31.
indicator questionnaire: initial psychometric study. Can Fam 26. Ciarloni L, Mallepell S, Brisken C. Amphiregulin is an essential
Physician 2008;54:76–7. mediator of estrogen receptor alpha function in mammary gland
10. Sobell LCS MB. Timeline follow-back: a technique for assessing development. Proc Natl Acad Sci U S A 2007;104:5455–60.
self-reported ethanol consumption. Measuring alcohol consumption: 27. Willmarth NE, Baillo A, Dziubinski ML, et al. Altered EGFR localization
psychosocial and biological methods. Totowa: Humana Press, 1992. and degradation in human breast cancer cells with an amphiregulin/
11. Matthews DR, Hosker JP, Rudenski AS, et al. Homeostasis model EGFR autocrine loop. Cell Signal 2009;21:212–9.
assessment: insulin resistance and beta-cell function from fasting 28. Meng Q, Gao B, Goldberg ID, et al. Stimulation of cell invasion and
plasma glucose and insulin concentrations in man. Diabetologia migration by alcohol in breast cancer cells. Biochem Biophys Res
1985;28:412–9. Commun 2000;273:448–53.
12. Baliunas DO, Taylor BJ, Irving H, et al. Alcohol as a risk factor for
29. Burstein HJ, Chen YH, Parker LM, et al. VEGF as a marker for
type 2 diabetes: a systematic review and meta-analysis. Diabetes
outcome among advanced breast cancer patients receiving anti-
Care 2009;32:2123–32.
VEGF therapy with bevacizumab and vinorelbine chemotherapy. Clin
13. Knott C, Bell S, Britton A. Alcohol Consumption and the risk of type
Cancer Res 2008;14:7871–7.
2 diabetes: a systematic review and dose-response meta-analysis
30. Meggiato T, Plebani M, Basso D, et al. Serum growth factors in
of more than 1.9 million individuals from 38 observational studies.
patients with pancreatic cancer. Tumour Biol 1999;20:65–71.
Diabetes Care 2015;38:1804–12.
14. Schrieks IC, Heil AL, Hendriks HF, et al. The effect of alcohol 31. Yoon SS, Kim SH, Gonen M, et al. Profile of plasma angiogenic
consumption on insulin sensitivity and glycemic status: a systematic factors before and after hepatectomy for colorectal cancer liver
review and meta-analysis of intervention studies. Diabetes Care metastases. Ann Surg Oncol 2006;13:353–62.
2015;38:723–32. 32. Webb NJ, Bottomley MJ, Watson CJ, et al. Vascular endothelial
15. Gepner Y, Golan R, Harman-Boehm I, et al. Effects of initiating growth factor (VEGF) is released from platelets during blood clotting:
moderate alcohol intake on cardiometabolic risk in adults with type implications for measurement of circulating VEGF levels in clinical
2 diabetes: a 2-year randomized, controlled trial. Ann Intern Med disease. Clin Sci 1998;94:395–404.
2015;163:569–79. 33. Yucel A, Karakus R, Cemalettin A. Effect of blood collection tube
16. Committee on Carcinogenicity of Chemicals in Food. Statement on types on the measurement of human epidermal growth factor. J
consumption of alcoholic beverages and risk of cancer. 2015 https:// Immunoassay Immunochem 2007;28:47–60.
wwwgovuk/government/groups/committee-on-carcinogenicity-of- 34. Beilin LJ, Puddey IB. Alcohol and hypertension: an update.
chemicals- in-food-consumer-products-and-the-environment-coc Hypertension 2006;47:1035–8.
17. Tabernero J. The role of VEGF and EGFR inhibition: implications 35. Hart CL, Morrison DS, Batty GD, et al. Effect of body mass index
for combining anti-VEGF and anti-EGFR agents. Mol Cancer Res and alcohol consumption on liver disease: analysis of data from two
2007;5:203–20. prospective cohort studies. BMJ 2010;340:c1240.
18. Ferrara N, Gerber HP, LeCouter J. The biology of VEGF and its 36. Trembling PM, Apostolidou S, Gentry-Maharaj A, et al. Risk of
receptors. Nat Med 2003;9:669–76. chronic liver disease in post-menopausal women due to body
19. Arteaga CL. The epidermal growth factor receptor: from mutant mass index, alcohol and their interaction: a prospective nested
oncogene in nonhuman cancers to therapeutic target in human cohort study within the United Kingdom Collaborative Trial of
neoplasia. J Clin Oncol 2001;19(Suppl):32S–40. Ovarian Cancer Screening (UKCTOCS). BMC Public Health
20. De Jong KP, Stellema R, Karrenbeld A, et al. Clinical relevance of 2017;17:603.
transforming growth factor alpha, epidermal growth factor receptor, 37. Henao-Mejia J, Elinav E, Jin C, et al. Inflammasome-mediated
p53, and Ki67 in colorectal liver metastases and corresponding dysbiosis regulates progression of NAFLD and obesity. Nature
primary tumors. Hepatology 1998;28:971–9. 2012;482:179–85.

Mehta G, et al. BMJ Open 2018;8:e020673. doi:10.1136/bmjopen-2017-020673 9