Content uploaded by Robyn N. Larsen
Author content
All content in this area was uploaded by Robyn N. Larsen on Aug 26, 2019
Content may be subject to copyright.
The effect of a high-protein, low glycemiceload
diet versus a conventional, high glycemiceload
diet on biochemical parameters associated
with acne vulgaris: A randomized,
investigator-masked, controlled trial
Robyn N. Smith, BAppSc (Hons),
a
NeilJ.Mann,BSc(Hons),BAppSc,PhD,
a
Anna Braue, MBBS, MMed,
b
Henna Ma¨kela¨inen, BAppSc,
d
and George A. Varigos, MBBS, FACD, PhD
b,c
Melbourne and Parkville, Australia; and Turku, Finland
Background: No previous study has sought to examine the influence of dietary composition on acne
vulgaris.
Objective: We sought to compare the effect of an experimental low glycemiceload diet with a conven-
tional high glycemiceload diet on clinical and endocrine aspects of acne vulgaris.
Methods: A total of 43 male patients with acne completed a 12-week, parallel, dietary intervention study
with investigator-masked dermatology assessments. Primary outcomes measures were changes in lesion
counts, sex hormone binding globulin, free androgen index, insulin-like growth factor-I, and insulin-like
growth factor binding proteins.
Results: At 12 weeks, total lesion counts had decreased more in the experimental group (21.9 [95%
confidence interval, 26.8 to 19.0]) compared with the control group (13.8 [19.1 to 8.5], P= .01). The
experimental diet also reduced weight (P= .001), reduced the free androgen index (P= .04), and increased
insulin-like growth factor binding protein-1 (P= .001) when compared with a high glycemiceload diet.
Limitations: We could not preclude the role of weight loss in the overall treatment effect.
Conclusion: This suggests nutrition-related lifestyle factors play a role in acne pathogenesis.
However, these preliminary findings should be confirmed by similar studies. ( J Am Acad Dermatol
2007;57:247-56.)
The pathogenesis of acne is complex, with
strong evidence supporting the involvement
of sebaceous hyperplasia, follicular hyper-
keratinization, bacterial and yeast intrafollicular
colonization, and inflammation.
1
Although andro-
gens play an essential role in the development of
acne, few studies have demonstrated a direct corre-
lation between acne severity and plasma androgen
From the School of Applied Sciences, RMIT University, Melbourne
a
;
Department of Dermatology, Royal Melbourne Hospital, Park-
ville
b
; Department of Dermatology, Royal Children’s Hospital,
Parkville
c
; and Department of Biochemistry and Food Chemis-
try, Turku University.
d
The study was supported by a research grant from Meat and
Livestock Australia.
Disclosure: This study was the responsibility of the investigators.
Meat and Livestock Australia had no role in data collection,
data analysis, data interpretation, or submission for publication.
Ms Smith receives a postgraduate scholarship from MINTRAC
(National Meat Industry Training Council of Australia). Authors
Mann, Braue, Ma
¨kela
¨inen, and Varigos had no conflicts of
interest to disclose.
Presented at the 15th Congress of the European Academy of
Dermatology and Venereology Meeting in Rhodes, October 5,
2006, and at the Nutrition Society of Australia Meeting in
Melbourne, December 2, 2005.
Accepted for publication January 9, 2007.
Reprint requests: Robyn Smith, BAppSc (Hons), School of Applied
Sciences, RMIT University, GPO Box 2476V, Melbourne, Victoria,
Australia, 3001.
Published online April 20, 2007.
0190-9622/$32.00
ª2007 by the American Academy of Dermatology, Inc.
doi:10.1016/j.jaad.2007.01.046
247
levels.
2
Variations in the clinical response to andro-
gens suggests that the endocrine control of acne is
complex.
3
Some studies suggest that acne severity
correlates better with sex hormone binding globulin
(SHBG) than circulating testosterone levels.
4-6
Elevated dehydroepiandrosterone sulfate (DHEAS),
the major adrenal androgen, has also been shown to
correlate with acne severity in adolescent girls and
with lesion counts in adults.
7,8
Other biologic factors,
such as insulin and insulin-like growth factor (IGF)-I,
may also augment sebum production, one of the 4
proximate causes of acne. Clinically, IGF-I has been
shown to correlate with acne lesion counts in adult
women and significantly higher IGF-I levels have
been described in women with acne compared with
control subjects.
8-10
Acne is also a common feature of women with
polycystic ovary syndrome (PCOS), a condition
characterized by hyperandrogenism and hyperinsu-
linemia.
11
Clinical and experimental evidence sug-
gests that insulin resistance and its compensatory
hyperinsulinemia are the underlying disturbance in
PCOS, as insulin resistance generally precedes and
gives rise to hyperandrogenism.
12
Insulin has been
shown to stimulate ovarian androgen production
through effects on steroidogenic enzymes and by
amplifying gonadotrophin-releasing hormone secre-
tion.
13
Insulin and IGF-I stimulate adrenal androgen
synthesis
14
and inhibit hepatic SHBG production,
15
allowing for an increase in androgen bioavailability.
Furthermore, insulin has been shown to decrease
IGF-binding protein (IGFBP)-1, allowing free IGF-I
concentrations to act on target tissues.
16
Treatments
aimed at reducing insulin secretion and/or increase
insulin sensitivity, such as metformin or acarbose,
have been shown to improve clinical symptoms of
acne in patients with PCOS.
17-19
Expression of acne during adolescence may also
be affected by endocrine changes, which are closely
related to changes in insulin sensitivity. During
normal puberty and adolescence, there is a transient
decline in insulin sensitivity,
20,21
which is accompa-
nied by a reciprocal decrease in levels of SHBG and
IGFBP-1.
22
According to cross-sectional observa-
tions, acne begins about the same time as the
gradual increase in plasma insulin,
20
the preadoles-
cent increase in body mass index (BMI),
23
and the
increase in IGF-I concentrations.
20,21
Acne incidence
more closely corresponds to the changing course of
insulin and IGF-I levels than to changes in plasma
androgens. This is because insulin and IGF-I levels
peak during late puberty and gradually decline until
the third decade.
20
Acne generally resolves by this
time despite circulating androgens remaining
unchanged.
Hyperinsulinemia may provide an important
link between nutrition-related lifestyle factors and
the incidence of acne. Accumulating evidence sug-
gests that low glycemiceload (LGL) diets may play
a dual role in the prevention of hyperinsulinemia
by lowering the postprandial insulin demand and
improving insulin sensitivity.
24-27
Dietary glycemic
load may be interpreted as a measure of the blood
glucose and insulin-increasing potential, as it repre-
sents the rate of carbohydrate absorption (indicated
by the glycemic index [GI]) and the quantity of
carbohydrate consumed.
28
It has recently been pos-
tulated that high intakes of refined, high-GI carbo-
hydrates may be a significant contributor to the high
incidence of acne in Western countries.
29
However,
the impact of such a dietary change on acne and
hormone levels has not been previously investi-
gated. Therefore, the aim of this study was to
investigate the effect of a LGL diet, compared with
a typical high glycemiceload diet, on acne severity
and metabolic and endocrine variables associated
with insulin resistance.
METHODS
Study population
Male patients with acne were recruited using
flyers posted at a university and newspaper adver-
tisements. Informed consent was obtained from
participants and guardians (if age \18 years) and
the study was conducted with the approval of our
human ethics committee. This study included only
male participants, age 15 to 25 years, with mild-
moderate facial acne. Eligible participants were re-
quired to have acne for more than 6 months and a
severity grade of greater than 0.25 but less than 2.0
as defined by the Leeds acne grading technique.
30
Volunteers were excluded if they were taking med-
ications known to affect acne or glucose metabolism.
A washout period of 6 months was required for oral
Abbreviations used:
BMI: body mass index
CV: coefficient of variation
DHEAS: dehydroepiandrosterone sulfate
FAI: free androgen index
GI: glycemic index
HOMA-IR: homeostasis model assessment of
insulin resistance
IGF: insulin-like growth factor
IGFBP: insulin-like growth factor binding
protein
LGL: low glycemic load
PCOS: polycystic ovary syndrome
SHBG: sex hormone binding globulin
JAMACAD DERMATOL
AUGUST 2007
248 Smith et al
retinoids or 2 months for oral antibiotics or topical
agents.
Study design
It was calculated that 20 participants per group
would provide 80% power (at the 2-sided 5% level)
to detect a difference of 20% in acne lesion counts,
2.7 U/mL in fasting insulin, 5.5 nmol/L in SHBG, and
8.3 ng/mL in IGFBP-1. To compensate for expected
patient withdrawal, 54 patients were enrolled.
Eligible participants were recruited between
June 2003 and June 2004. Approximately 2 to 3
weeks after recruitment, participants attended their
baseline appointment and were randomly assigned
(1:1) to either the LGL or control group (Fig 1).
Randomization was carried out by computer-gen-
erated random numbers and allocation to groups
was performed by a third party.
This study was designed as a parallel dietary
intervention study with investigator-masked derma-
tology assessments. Topical therapy, in the form of
a noncomedogenic cleanser, was standardized for
both groups and facial acne was scored at monthly
visits (weeks 0, 4, 8, and 12) at the academic research
clinic. On all visits, height, weight, percentage body
fat, and hip and waist circumferences were mea-
sured. All patients were weighed in light clothes and
BMI was calculated as: weight (kg)/height (m)
2
.
Fig 1. Recruitment to completion of participants in trial.
JAMACAD DERMATOL
VOLUME 57, NUMBER 2
Smith et al 249
Percentage body fat was measured using a bioelectric
impedance analyzer (TBF-521, Tanita Corp, Arlington
Heights, Ill). At baseline and 12 weeks, a venous
blood sample was taken after an overnight fast.
Dietary intervention
Participants were informed that the study was
comparing the carbohydrate to protein ratio in the
diet and were not informed of the study’s hypothesis.
The LGL diet was low in glycemic load, achieved
through modifications to the amount and type of
carbohydrate. The LGL group members were edu-
cated on how to substitute high-GI foods with foods
higher in protein (eg, lean meat, poultry, or fish) and
lower in GI (eg, whole grain bread, pasta, and fruits).
Some staple foods were supplied and participants
were urged to consume these or similar foods on
a daily basis. Each participant’s dietary directions
were isocalorically matched with their baseline diet
as determined from 7-day weighed/measured food
records. The recommended LGL diet consisted of
25% energy from protein, 45% from low GI carbo-
hydrates, and 30% energy from fats. In contrast, the
control group received carbohydrate-dense staples
and were instructed to eat these or similar foods
daily. The foods provided had moderate to high GI
values and were typical of their normal diet as
evidenced from 7-day weighed/measured food rec-
ords. The control group were not informed about the
GI, but were urged to include carbohydrates as a
regular part of their diet. All participants received
initial education on how to use foods scales and
keep food records. During the study period, nutrient
intakes were calculated from 3-day weighed/
measured food records each month using Australia-
specific dietary analysis software (Foodworks, Xyris
Software, Highgate Hill, Australia). Dietary compli-
ance was monitored by regular telephone inter-
views, assessments of glycemic load per day and
24-hour urine samples (weeks 0 and 12) for an
assessment of urea excretion relative to urinary
creatinine.
Standardized topical cleanser
All patients were provided with a topical cleanser
(Cetaphil gentle skin cleanser, Galderma, Forrests
Hill, Australia) and were advised to use it in place
of their normal wash, soap, or cleanser. The cleanser
provided contained no active agents for acne and its
formulation is identical to that which is currently
available overseas. Participants began using the
topical cleanser 2 weeks before baseline and were
asked to maintain a standard level of use during the
trial. Compliance was determined from self-report
at each visit.
Dermatology assessment
Scaling of the acne was performed by a derma-
tology registrar who was masked to the group
assignment of the participants. The registrar assessed
facial acne occurrence and severity using a modified
lesion count technique (Leeds) form Burke and
Cunliffe.
30
To ensure all acne lesions were counted,
located, and graded by size and severity, lesions
were mapped by placing a transparent plastic film
with a laser-printed grid gently against the skin.
Facial anatomic landmarks, such as the ear, chin,
and tip of nose, were used to ensure consistency
between assessments. Each side of the face was
assessed separately. Where necessary, the registrar
palpated the skin to determine the lesion type. All
assessments were performed under fluorescent
background lighting and with a halogen lamp, which
could be easily moved to illuminate both sides of
the patient’s face. To maintain reproducibility of
the above method, one physician performed all the
dermatology assessments. A small group of volun-
teers (n = 4) was counted 1 week apart to evaluate
the reproducibility by the same physician (9.5%
coefficient of variation [CV]).
Biochemical measurements
Code-labelled samples were stored at 808C for
analysis poststudy by an independent laboratory.
Baseline and 12-week samples for each participant
were included in the same assay run to avoid
interassay variability. Serum insulin was measured
using a commercially available microparticle
Table I. Baseline characteristics of the participants
by dietary group for the per protocol population
Variable
LGL
(n = 23)
Control
(n = 20) P
Age, y 18.2 60.5 18.5 60.5 .66
Body mass index,
kg/m
2
22.9 60.6 22.5 60.7 .34
Weight, kg 73.5 62.5 73.3 63.3 .90
Waist
circumference, cm
79.2 61.7 79.0 62.2 .81
Total count average 40.6 65.0 34.9 64.3 .40
Inflammatory count
average
31.9 63.9 28.4 63.6 .72
Noninflammatory
count average
8.8 62.2 6.5 61.6 .62
Fasting glucose,
mmol/L
4.76 60.07 4.47 60.09 .02
Fasting insulin, U/mL 7.37 60.66 7.35 60.74 .99
Testosterone, nmol/L 20.97 61.19 20.36 61.48 .75
SHBG, nmol/L 25.39 61.27 24.75 61.92 .78
LGL, Low glycemic load; SHBG, sex hormone binding globulin.
JAMACAD DERMATOL
AUGUST 2007
250 Smith et al
enzyme immunoassay (Abbott Laboratories, Tokyo,
Japan) (intra-assay CV 4.0%). Capillary blood glu-
cose was measured using a glucose analyzer
(2011, HemoCue, A
¨ngelholm, Sweden) (intra-assay
CV 1.6%). The homeostasis model assessment of
insulin resistance (HOMA-IR) was used as a surro-
gate measure of insulin sensitivity, calculated as:
fasting glucose (mmol/L) 3insulin (U/mL)/22.5.
31
SHBG concentrations were assayed with a commer-
cially available radioimmunoassay (Orion Diagnos-
tica, Espoo, Finland) (intra-assay CV 2.5%). Total
testosterone was measured using solid-phase radio-
immunoassay (Diagnostic Products, Los Angeles,
Calif) (intra-assay CV 2.7%). The free androgen
index (FAI) was calculated as: testosterone con-
centration (nmol/L) 3100/SHBG concentration
(nmol/L). IGF-I (intra-assay CV 2.9%) and DHEAS
(intra-assay CV 8.1%) were measured using semi-
automated technology (Immulite, Diagnostic Pro-
ducts). IGFBP-1 and IGFBP-3 were assayed with a
noncommercial radioimmunoassay as previously
described.
32,33
An automated analyzer (Olympus,
Melville, NY) was used to measure total cholesterol,
high density lipoprotein cholesterol, and triglyceride
levels (intra-assay CV were 1%-2% for all tests). Low-
density lipoprotein cholesterol was calculated by the
formula of Friedewald et al.
34
Statistical analysis
All statistical analyses were performed for the per
protocol population using software (SPSS 11.0 for
Windows, SPSS Inc, Chicago, Ill) and significance
was set at Pless than .05. Baseline variables were
analyzed for between group significance using
Mann-Whitney U test or 1-way analysis of variance,
depending on whether the data was normally dis-
tributed. General linear regression models were
used to test for overall treatment differences, with
adjustments made for potential cofounders, includ-
ing age, ethnicity, and baseline data. Secondary
analyses of acne end points and hormone variables
were performed adjusting for changes in BMI. All
covariates were individually entered into the models
to determine whether there was an interaction
between the covariate and the treatment. Bivariate
linear regression analysis was also conducted, pool-
ing data from both groups, to explore relationships
between endocrine variables and the change in
lesion counts.
RESULTS
Patients
In all, 43 patients completed the study per proto-
col (Fig 1). Seven participants did not complete the
study (5 in control and two in LGL groups) and 4
were removed from data set (two were noncompli-
ant; two began medications known to affect acne,
an exclusion criterion). Baseline characteristics are
shown in Table I.
Diet composition
Table II lists the diet composition of the LGL and
control groups before randomization and during the
trial. No significant group differences were observed
Table II. Dietary composition of the low glycemic load and control diets before randomization and during
the trial period
Trial period
Nutrient
Before randomization
(n = 43)
LGL
(n = 23)
Control
(n = 20) P
Energy, kJ/d 10,585 9320*10,620 .06
Dietary glycemic index
y
57 43*56 \.001
Dietary glycemic load
z
178 101*174 \.001
Carbohydrate, % total kJ 49 44*50 \.001
Protein, % total kJ 17 23*17 \.001
Total fat, % total kJ 32 31 32 \.74
Fat subgroups, % total kJ
Saturated 13.2 9.0*12.6 \.001
Polyunsaturated 4.5 6.8*4.8 \.001
Monounsaturated 11.7 12.4 11.9 \.51
Cholesterol, mg/d 296 309 304 \.93
Dietary fiber, g/d 25.3 36.9*26.2 \.001
Pvalue corresponds with 1-way analysis of variance or Mann-Whitney.
LGL, Low glycemic load.
*Significantly different from baseline (P\.05).
y
Calculated as the weighted average glycemic index of all carbohydrate-containing foods in the diet.
z
Calculated as the food’s carbohydrate amount (grams) 3the respective glycemic index value and divided by 100, then totalled for all foods
each day.
JAMACAD DERMATOL
VOLUME 57, NUMBER 2
Smith et al 251
in any of the dietary variables at baseline. During the
trial period, dietary glycemic load was significantly
lower in the LGL group compared with the control
group and this was achieved by a reduction in
carbohydrate intake and by means of low-GI foods
(as indicated by a reduction in the calculated dietary
GI) (Table II). Protein intake increased in the LGL
compared with the control group (P\.001), indi-
cating that some carbohydrates were replaced with
foods higher in protein. This was substantiated by
a 15.4% increase in urinary urea/creatinine ratio at
12 weeks for the LGL group compared with a 12.3%
decrease for the control group (P= .009), indicating
good dietary compliance. Although the LGL group
received isocaloric dietary advice, energy intake
decreased relative to their baseline diet (P= .02).
Compliance with topical nonacne cleanser
There was no discontinuation of the topical non-
acne therapy among the study completers.
Study outcomes
Table III shows the mean change in lesion counts
and anthropometric measures at 12 weeks according
to dietary group. After adjusting for age, ethnicity,
and baseline counts, the reduction in total lesion
counts was significantly greater in the LGL group
compared with the control group (P= .01). Similar
results were observed for the mean decrease in
inflammatory counts (P= .02). Figure 3 shows
examples of acne improvement in the LGL group.
The LGL group also showed significantly greater
reductions in weight (P\.001), BMI (P= .001), body
Table IV. Mean change in hormone concentrations and plasma lipids for the per protocol population
at 12 weeks according to dietary group
Adjusted means*(95% CI)
LGL (n = 23) Control (n = 20) P
Fasting insulin, U/mL
y
0.90 (2.59, 0.79) 1.96 (0.15, 3.77) .03
Log(HOMA-IR) 0.05 (0.13, 0.03) 0.09 (0.01, 0.18) .02
Testosterone, nmol/L 1.32 (2.75, 0.12) 1.20 (2.74, 0.34) .91
SHBG, nmol/L 0.27 (1.56, 2.09) 2.71 (4.67, 0.74)
z
.03
FAI, nmol/L 8.73 (17.34, 0.12)
z
4.51 (4.74, 13.75) .04
DHEAS, mol/L 0.68 (1.30, 0.06)
z
0.12 (0.79, 0.55) .22
IGF-I, nmol/L 2.93 (6.21, 0.35) 2.79 (6.31, 0.73) .95
Log(IGFBP-1), ng/mL
§
0.14 (0.05, 0.22)
z
0.09 (0.18, 0.03) .001
IGFBP-3, mg/mL 0.11 (0.10, 0.32) 0.16 (0.06, 0.38) .75
Total cholesterol, mmol/L 0.42 (0.63, 0.21)
z
0.05 (0.27, 0.17) .02
LDL cholesterol, mmol/L 0.32 (0.51, 0.13)
z
0.04 (0.25, 0.16) .06
HDL cholesterol, mmol/L 0.04 (0.10, 0.10) 0.05 (0.11, 0.01) .92
Log(triglycerides), mmol/L 0.06 (0.13, 0.08) 0.03 (0.05, 0.10) .10
CI, Confidence interval; DHEAS, dehydroepiandrosterone sulfate; FAI, free androgen index; HDL, high-density lipoprotein; HOMA-IR,
homeostasis model assessment of insulin resistance; IGF, insulin-like growth factor; IGFBP, IGF binding protein; LDL, low-density lipoprotein;
LGL, low glycemic load; SHBG, sex hormone binding globulin.
*Means are adjusted for differences in baseline values, age, and ethnicity.
y
n = 42 As a result of an outlier of more than 3 SD from the mean.
z
Significant difference from baseline to 12 weeks (P\.05).
§
Analysis was performed on log-transformed variable as the data was log-normally distributed.
Table III. Mean change in outcome variables for the per protocol population at 12 weeks according
to dietary group
Adjusted means (95% CI)*
Variable LGL group (n = 23) Control group (n = 20) P
Total lesion counts 21.9 (26.8, 19.0) 13.8 (19.1, 8.5) .01
Inflammatory lesion counts 16.0 (20.3, 11.8) 8.4 (13.0, 3.8) .02
Weight, kg 2.9 (4.0, 1.7) 0.4 (0.8, 1.7) \.001
BMI, kg/m
2
0.89 (1.25, 0.54) 0.02 (0.41, 0.36) .002
Percentage body fat (%) 2.2 (3.0, 1.4) 0.5 (1.3, 0.4) .006
Waist circumference, cm 2.2 (3.6, 0.9) 0.2 (1.6, 1.2) .04
BMI, Body mass index; CI, confidence interval; LGL, low glycemic load.
*Means are adjusted for differences in baseline values, age and ethnicity.
JAMACAD DERMATOL
AUGUST 2007
252 Smith et al
fat percentage (P= .006), and waist circumference
(P= .04) when compared with the control group.
Statistical adjustment of the mean change in acne
scores for changes in BMI altered the outcome for
total lesion counts (21.6 [95% confidence interval
28.9 to 14.4] vs 14.1 [21.9 to 6.2], P= .07), but
not inflammatory counts (16.0 [95% confidence
interval 20.6 to 11.4] vs 8.5 [13.4 to 3.5], P=
.04). However, we found no significant interaction
effect of dietary treatment and the change in BMI on
acne end points.
Table IV shows the mean change in hormonal
variables and plasma lipids at 12 weeks according
to dietary group. The mean change in fasting insulin
levels (P= .03) and log-transformed HOMA-IR (P=
.02) was significantly different between groups with
the LGL group showing a trend for improved insulin
sensitivity and the control group showing a trend for
increasing insulin resistance. SHBG levels decreased
in the control group compared with the LGL group
(P= .031). The effect of dietary treatment on FAI
(P= .041) was marginally significant, with the LGL
group showing a decrease in testosterone bioavail-
ability compared with the control group. IGFBP-
1 increased significantly in the LGL group relative
to baseline (P= .001) and this was significantly
different from the mean change in the control group.
Statistical adjustment for changes in BMI was found
to affect the results for HOMA-IR (0.04 for the LGL
group vs 0.08 for the control group, P= .08), SHBG
(0.04 for the LGL group vs 2.35 for the control
group, P= .13), and FAI (8.8 for the LGL group vs
4.6 for the control group, P= .07), but not IGFBP-
1 (0.14 for the control group vs 0.09 for the control
group, P= .003).
Hormonal variables as predictors
of acne improvement
Fig 2 depicts the results from bivariate linear
regression analysis. A positive relationship was ob-
served between the change in total lesion counts and
the change in insulin sensitivity as determined by
HOMA-IR (r= 0.38, P= .01). A change in SHBG levels
was also shown to correlate negatively with a change
in lesion counts (r=0.38, P= .01). In contrast,
a change in FAI was not significantly associated with
the change in lesion counts (r= 0.10, P= .50).
DISCUSSION
This pilot study investigated the independent
effects of an experimental LGL diet versus a conven-
tional high glycemiceload diet, combined with a
standardized noncomedogenic cleanser. Although
both groups showed improvements in acne, the
LGL group showed significantly greater reductions
in the clinical and endocrine assessments of acne.
In addition, participants in the LGL group showed
reductions in weight and measures of adiposity,
despite receiving dietary advice that was isocalori-
cally matched with diets at baseline. In contrast,
participants on the conventional, high glycemic
loadediet showed no change in weight or body
composition.
This study found a positive effect of a LGL diet on
insulin sensitivity when compared with a conven-
tional high glycemic loadediet. The improvement in
insulin sensitivity may be attributable not only to the
decrease in glycemic load,
35
but also to the decrease
in total energy intake and subsequent body weight
loss.
36
It has been hypothesized that a decrease in
insulin may mediate a reduction in underlying path-
ological aspects of acne.
29
In accordance with this
hypothesis, we observed a moderate relationship
Fig 2. Relationships between acne improvement and
hormone variables homeostasis model assessment of insu-
lin resistance (A) and sex hormone binding globulin (B).
JAMACAD DERMATOL
VOLUME 57, NUMBER 2
Smith et al 253
between the change in insulin sensitivity and the
change in acne lesion counts. This suggests that the
therapeutic effect may be a factor of the change in
insulin sensitivity, or simply that improved insulin
sensitivity is another manifestation of a LGL diet.
An association between acne and mild peripheral
insulin resistance has been previously described in
healthy eumenorrheic women.
37
The authors found
that patients with acne exhibited significant hyper-
insulinemia during an oral glucose tolerance test
compared with age-matched control subjects. This
relationship was also found to be independent of
obesity, as BMI was similar in both groups.
Our results suggest that the improvement in
insulin sensitivity may be linked to reductions in
androgenicity. We observed a reduction in testoster-
one bioavailability and DHEAS concentrations in the
LGL group and this may help to explain the lessening
of acne severity. The observed reduction in free
testosterone was probably related to the dual effect
of insulin on androgen production in testicular
tissues
38
and the hepatic production of SHBG.
15
Plasma concentrations of DHEAS, the major adrenal
androgen, also decreased in the LGL group, possibly
owing to insulin’s effect on the expression of adrenal
steroidogenic enzymes.
14
In contrast, the control
group showed a decline in insulin sensitivity and
SHBG concentrations. Why acne improved in the
control group despite no significant change in an-
drogen levels remains unanswered, but the possible
direct effect of the topical cleansing agent should
be considered.
39
As SHBG correlates inversely with
insulin, it was not surprising to find that acne also
correlated with the change in SHBG. Our results also
corroborate previous evidence that SHBG may be
a marker of acne.
5,6,40
However, we observed no
relationship between acne severity and free testos-
terone levels, an association that has been demon-
strated in some
7,41
but not all studies.
3,42
Normal sebaceous gland growth is also influ-
enced by factors other than androgens, such as
IGF-I.
9
Therefore, increased expression of IGF-I or
Fig 3. Photographs of acne improvement in low glycemic load group. Patient A at baseline
(A) and 12 weeks (B). Patient B at baseline (C) and 12 weeks (D).
JAMACAD DERMATOL
AUGUST 2007
254 Smith et al
a reduction in the level of its carrier proteins could
influence acne. In the current study, IGFBP-1 levels
increased significantly in the LGL group compared
with the control group and we speculate that this is
a compensatory adaptation to the improvement in
insulin sensitivity and the reduction in basal insulin.
Insulin is the principle determinant of plasma
IGFBP-1 levels and low basal IGFBP-1 levels have
been observed in insulin-resistant individuals, pos-
sibly as a result of increased portal insulin over-
night.
43
Therefore, it is possible that the LGL diet
may also induce changes to the IGF system that may
be clinically relevant to events involved in acne
pathogenesis.
There are some limitations regarding the study
design and intervention that should be addressed.
Firstly, as participants in the LGL group lost weight,
we cannot preclude the change in BMI to the overall
treatment effect. When we statistically adjusted the
data for changes in BMI, the effect of the LGL diet on
several clinical and endocrine parameters was lost.
However, this does not necessarily imply that acne is
influenced by weight loss per se. Weight loss trials,
involving women with PCOS, have consistently
shown increased SHBG levels
44,45
and decreased
FAI.
44,45
However, low-fat dietary interventions in
nonobese women have shown no change in SHBG
levels after weight loss.
46,47
Currently, there is a
paucity of evidence to indicate that acne is more
prevalent or severe in overweight adolescents. One
study revealed a relationship between weight and
acne in men age 20 to 40 years, but this was not true
for young males aged 15 to 19 years.
48
Although we
cannot determine an aspecific effect of weight loss
on acne, one could speculate that a reduction in
hyperinsulinemia, either through weight loss or
dietary composition, may reduce precipitating fac-
tors involved in acne. Another limitation of this study
was the use of a fasting index to quantitatively
estimate hyperinsulinemia and insulin resistance.
Although this index correlates with the euglycemic
clamp technique and has proven to be useful in large
studies, its applicability to small intervention trials
remains uncertain.
49
Furthermore, this test reflects
insulin action in a basal state, whereas in life much of
insulin action is postprandial. Therefore, it is possi-
ble that this index may provide an underestimation
or overestimation of the relationship between acne
and the extent of hyperinsulinemia.
To our knowledge, this is the first study to
demonstrate a therapeutic effect of dietary interven-
tion on acne. The results of this study open the
prospect that nutrition-related lifestyle factors may
affect the pathogenesis of acne. After 12 weeks, a
LGL diet was shown to reduce weight, acne severity,
and hormonal aspects of acne (eg, testosterone
bioavailability, IGFBP-1, and HOMA-IR) when com-
pared with a conventional high glycemiceload diet.
Although we could not determine an aspecific effect
of diet from that of weight loss, the finding that
insulin sensitivity correlated with acne suggests that
both may be involved. Therefore, these results
should be considered preliminary and larger-scale
studies are needed to confirm the effect of dietary
intervention on acne.
REFERENCES
1. Deplewski D, Rosenfield R. Role of hormones in pilosebaceous
unit development. Endocr Rev 2000;21:363-92.
2. Shaw J. Acne: the effect of hormones on pathogenesis and
management. Am J Clin Dermatol 2002;3:571-8.
3. Levell M, Cawood M, Burke B, Cunliffe W. Acne is not
associated with abnormal plasma androgens. Br J Dermatol
1989;120:649-54.
4. Slayden S, Moran C, Sams W, Boots L, Azziz R. Hyperandrogene-
mia in patients presentingwith acne. Fertil Steril 2001;75:889-92.
5. Lawrence D. Elevated free testosterone concentration in men
and women with acne vulgaris. Clin Exp Dermatol 1986;11:
263-73.
6. Odlind V, Carlstro
¨m K, Michae
¨lsson G, Vahlqvist A, Victor A,
Mellbin T. Plasma androgenic activity in women with acne
vulgaris and in healthy girls before, during and after puberty.
Clin Endocrinol 1982;16:243-9.
7. Lucky A, McGuire J, Rosenfield R, Lucky P, Rich B. Plasma
androgens in women with acne vulgaris. J Invest Dermatol
1983;81:70-4.
8. Cappel M, Mauger D, Thiboutot D. Correlation between serum
levels of insulin-like growth factor 1, dehydroepidandroster-
one sulfate, and dihydrotestosterone and acne lesion counts
in adult women. Arch Dermatol 2005;141:333-8.
9. Deplewski D, Rosenfield R. Growth hormone and insulin-like
growth factors have different effects on sebaceous cell growth
and differentiation. Endocrinology 1999;140:4089-94.
10. Aizawa H, Niimura M. Elevated serum insulin-like growth
factor-I (IGF-I) levels in women with postadolescent acne.
J Dermatol 1995;22:249-52.
11. Franks S. Polycystic ovary syndrome. N Engl J Med 2003;13:
853-61.
12. Dunaif A, Segal K, Shelley D, Green G, Dorbrjansky A. Profound
peripheral insulin resistance, independent of obesity in poly-
cystic ovary syndrome. Diabetes 1989;38:1165-74.
13. Willis D, Mason H, Gilling-Smith C, Franks S. Modulation by
insulin of follicle stimulating and luteinizing hormone action
in human granulosa cells of normal and polycystic ovaries.
J Clin Endocrinol Metab 1996;81:302-9.
14. Kristiansen S, Endoh A, Casson P, Buster J, Hornsby P.
Induction of steroidogenic enzyme genes by insulin and
IGF-I in cultured adult human adrenocortical cells. Steroids
1997;62:258-65.
15. Singh A, Hamilton-Fairley D, Koistinen R, Seppala M, James V,
Franks S, et al. Effect of insulin-like growth factor-type I (IGF-I)
and insulin on the secretion of sex hormone binding globulin
and IGF-I binding protein (IBP-I) by human hepatoma cells.
J Endocrinol 1990;124:R1-3.
16. Powell D, Suwanichkul A, Cubbage M, DePaolis L, Snuggs M,
Lee P. Insulin inhibits transcription of the human gene for
insulin-like growth factor-binding protein-1. J Biol Chem 1991;
266:18868-76.
JAMACAD DERMATOL
VOLUME 57, NUMBER 2
Smith et al 255
17. Ciotta L, Calogero AE, Farina M, De Leo V, La Marca A, Cianci A.
Clinical, endocrine and metabolic effects of acarbose, an
a-glucosidase inhibitor; in PCOS patients with increased
insulin response and normal glucose tolerance. Hum Reprod
2001;16:2066-72.
18. Kazerooni T, Dehghan-Kooshkghazi M. Effect of metformin
therapy on hyperandrogenism in women with polycystic
ovary syndrome. Gynecol Endocrinol 2003;17:51-6.
19. Kolodziejczyk B, Duleba AJ, Spaczynski RZ, Pawelczyk L.
Metformin therapy decreases hyperandrogenism and hyper-
insulinemia in women with polycystic ovary syndrome. Fertil
Steril 2000;73:1149-54.
20. Smith C, Dunger D, AJK W, Taylor A, Perry L, Gale E, et al.
Relationship between insulin, insulin-like growth factor I, and
dehydroepiandrosterone sulfate concentrations during child-
hood, puberty and adult life. J Clin Endocrinol Metab 1989;68:
932-7.
21. Caprio S, Plewe G, Diamond M, Simonson D, Boulware S,
Sherwin R, et al. Increased insulin secretion in puberty:
a compensatory response to reductions in insulin sensitivity.
J Pediatr 1989;114:963-7.
22. Holly J, Smith C, Dunger D, Howell R, Chard T, Perry L, et al.
Relationship between the pubertal fall in sex hormone bind-
ing globulin and insulin-like growth factor binding protein-I:
a synchronized approach to pubertal development? Clin
Endocrinol 1989;31:277-84.
23. Rolland-Cachera M. Body composition during adolescence:
methods, limitations and determinants. Horm Res 1993;
39(Suppl):25-40.
24. Willett W, Manson J, Liu S. Glycemic index, glycemic load and
risk of type 2 diabetes. Am J Clin Nutr 2002;76(Suppl):274S-80S.
25. Jenkins D, Kendall C, Augustin L, Franceschi S, Hamidi M,
Marchie A, et al. Glycemic index: overview of implications
in healthy and disease. Am J Clin Nutr 2002;76(Suppl):266S-73S.
26. Carnethon MR, Loria CM, Hill JO, Sidney S, Savage PJ, Liu K.
Risk factors for the metabolic syndrome: the coronary artery
risk development in young adults (CARDIA) study, 1985-2001.
Diabetes Care 2004;27:2707-15.
27. McKeown N, Meigs J, Liu S, Saltzman E, Wilson P, Jacques P.
Carbohydrate nutrition, insulin resistance and the prevalence
of the metabolic syndrome in the Framingham offspring
cohort. Diabetes Care 2004;27:538-46.
28. Brand-Miller J, Thomas M, Swan V, Ahmad Z, Petocz P,
Colagiuri S. Physiological validation of the concept of glycemic
load in lean young adults. J Nutr 2003;133:2695-6.
29. Cordain L, Lindeberg S, Hurtado M, Hill K, Eaton B, Brand-Miller
B. Acne vulgarisea disease of Western civilization. Arch
Dermatol 2002;138:1584-90.
30. Burke B, Cunliffe W. The assessment of acne vulgarisethe
Leeds technique. Br J Dermatol 1984;111:83-92.
31. Matthews D, Hosker J, Rudenski A, Naylor B, Treacher D,
Turner R. Homeostasis model assessment: insulin resistance
and beta-cell function from fasting plasma glucose and insulin
concentrations in man. Diabetologica 1985;28:412-9.
32. Baxter R, Martin J. Radioimmunoassay of growth hormone-
dependent insulin-like growth factor binding protein in
human plasma. J Clin Invest 1986;78:1504-12.
33. Baxter R, Martin J, Wood M. Two immunoreactive binding
proteins for insulin-like growth factors in human amniotic
fluid: relationship to fetal maturity. J Clin Endocrinol Metab
1987;65:423-31.
34. Friedewald W, Levy R, Fredrickson D. Estimation of the
concentration of low-density lipoprotein cholesterol in
plasma, without use of the preparative ultracentrifuge. Clin
Chem 1972;18:499-502.
35. Frost G, Keogh B, Smith D, Akinsanya K, Leeds A. The effect of
low-glycemic carbohydrate on insulin and glucose response
in vivo and in vitro in patients with coronary heat disease.
Metabolism 1996;45:669-72.
36. Hannah J, Howard B. Dietary fats, insulin resistance and
diabetes. J Cardiovasc Risk 1994;1:31-7.
37. Aizawa H, Niimura M. Mild insulin resistance during oral
glucose tolerance test (OGTT) in women with acne. J Dermatol
1996;23:526-9.
38. Bebakar W, Honour J, Foster D, Liu Y, Jacobs H. Regulation of
testicular function by insulin and transforming growth factor-
beta. Steroids 1990;55:266-70.
39. Draelos Z. The effect of a daily facial cleanser for normal to oily
skin on the skin barrier of subjects with acne. Cutis 2006;
78(Suppl):34-40.
40. Marynick S, Chakmajian Z, McCaffree D, Herdon J. Androgen
excess in cystic acne. N Engl J Med 1983;308:981-6.
41. Schiavone F, Rietschel R, Sgoutas D, Harris R. Elevated free
testosterone levels in women with acne. J Invest Dermatol
1983;81:70-4.
42. Cibula D, Hill M, Vohradnikova O, Kuzel D, Fanta M, Zivny J.
The role of androgens in determining acne severity in adult
women. Br J Dermatol 2000;143:399-404.
43. Attia N, Tamborlane W, Heptulla R, Maggs D, Grozman A,
Sherwin R, et al. The metabolic syndrome and insulin-like
growth factor I regulation in adolescent obesity. J Clin
Endocrinol Metab 1998;83:1467-71.
44. Kiddy D, Hamilton-Fairley D, Seppala M, Koistinen R, James
V, Reed M, et al. Diet-induced changes in sex hormone
binding globulin and free testosterone in women with
normal or polycystic ovaries: correlation with serum insulin
and insulin-like growth factor-I. Clin Endocrinol 1989;31:
757-63.
45. Guzick D, Wing R, Smith D, Berga S, Winters S. Endocrine
consequences of weight loss in obese, hyperandrogenic,
anovulatory women. Fertil Steril 1994;61:598-604.
46. Prentice R, Thompson D, Clifford C, Gorbach S, Goldin B, Byar
D. Dietary fat reduction and plasma estradiol concentration
in healthy postmenopausal women: the women’s health trial
study group. J Natl Cancer Inst 1990;82:129-34.
47. Rose D, Chlebowski R, Connolly J, Hones L, Wynder E.
Effects of tamoxifen adjuvant therapy and a low-fat diet on
serum binding proteins and estradiol bioavailability in
postmenopausal breast cancer patients. Cancer Res 1992;52:
5386-90.
48. Bourne S, Jacobs A. Observations on acne, seborrhoea, and
obesity. Br Med J 1956;1:1268-70.
49. Daly M. Sugars, insulin sensitivity, and the postprandial state.
Am J Clin Nutr 2003;78(Suppl):865S-72S.
JAMACAD DERMATOL
AUGUST 2007
256 Smith et al
