Im f D
Lo
al con
rly de
is noti
indire
basal
neocy
pilos
n of t
nt to
on th
ysiolo
sevier
D
mu
me
bee
bo
pe
pre
ing
tro
2 of the major textbooks of dermatology
1,2
promulgate the
notion that diet and acne are unrelated, yet rely only on 2
pri
old
stu
stu
ern
pri
exa
or
ba
(2)
an
ph
oni
acn
P
D
Gi
mi
lish
acne are unrelated and that the 2 articles
11,12
most frequently
cited as definitive evidence against the diet/acne hypothesis
Dep
Add
84mary references
11,12
that are not only more than 30 years
, but which contain key experimental design flaws.
13,14
Because of this virtual absence of well-designed modern
dies, it may be useful to critically examine the 2 dietary
dies that have so strongly influenced the authors of mod-
textbooks of dermatology as well as to briefly examine the
or historical literature. Second, it may be insightful to
mine the current literature showing how, either directly
indirectly, dietary manipulations may influence (1) the
lance of steroid hormones (and hence sebum synthesis),
follicular keratinocyte proliferation and differentiation,
d (3) inflammation. Dysregulation of these 3 fundamental
ysiologic mechanisms along with involvement of Propi-
merely represent capstone studies that confirm previous ob-
servations and conclusions. In actuality, both assumptions
have little factual basis.
Two reviews
7,16
of the diet/acne literature covering 80 ref-
erences and spanning 66 years from 1906 to 1972 clearly
demonstrate the inconclusive and conflicting nature of the
historical literature. Examination of many of these early stud-
ies shows them to be rife with contradictory results and con-
clusions due in part to numerous limitations of study design,
such as the lack of control groups, inadequate sample size, no
statistical treatment of data, the lack of blinding and/or pla-
cebos, and inadequate or no baseline diet data, as well as
imprecise and inconsistent measurement procedures com-
monly seen in early, developing technology. Furthermore,
most early investigators did not have the benefit of research
elucidating the endocrine mechanisms underlying acne’s
pathogenesis, let alone its molecular underpinnings. Accord-
artment of Health and Exercise Science, Colorado State University, Fort
Collins, Colorado.
ress reprint requests to Loren Cordain, Department of Health and Exer-plications for the Role o
ren Cordain, PhD
Within the dermatology community, a gener
lated to the etiology of acne. Except for 2 poo
old, there are few objective data to support th
now exists showing how diet may directly or
causes of acne: (1) increased proliferation of
duct, (2) incomplete separation of ductal cor
apoptosis and subsequent obstruction of the
increases in sebum production, (4) colonizatio
and (5) inflammation both within and adjace
review of the currently available literature
vulgaris as well as a discussion of the ph
association.
Semin Cutan Med Surg 24:84-91 © 2005 El
uring the course of the last 30 to 40 years, a general
consensus has emerged within the dermatology com-
nity that diet has no role in the etiology of acne.
1-8
Com-
nts such as “The association of diet with acne has traditionally
n relegated to the category of myth”
9
are commonplace in
th the past and current literature. This widely accepted
rception is somewhat surprising given that a recent com-
hensive review examining 250 trials of acne therapy dur-
the past 50 years was able to identify only a single con-
lled study in which diet was even mentioned.
10
Moreover,ing
no
cise Science, Colorado State University, Fort Collins, CO 80523. E-mail:
lcordain@cahs.colostate.edu
1085-5629/05/$-see front matter ? 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.sder.2005.04.002iet in Acne
sensus has emerged that diet is unre-
signed studies, now more than 30 years
on. In contrast, a large body of evidence
ctly influence the following 5 proximate
keratinocytes within the pilosebaceous
tes from one another via impairment of
ebaceous duct, (3) androgen-mediated
he comedo by Propionibacterium acnes,
the comedo. This article will provide a
e association between diet and acne
gic principles that may underlie this
Inc. All rights reserved.
bacterium acnes represents the known proximate causes of
e.
15
urported Evidence against
iet as an Etiologic Agent in Acne
ven the dogma in current dermatology textbooks,
1,2
it
ght be assumed that there has been a long and well-estab-
ed literature conclusively demonstrating that diet andly, they commonly did not examine etiologic variables we
w understand to be important. In summary, there is little

sub
siv
acn
hu
in
alo
len
de
fre
the
to
stu
of
Co
foo
mi
68
we
be
tre
sio
pe
da
thi
eta
ica
pre
eac
po
tist
com
exp
Co
me
the
Ac
mi
ing
sub
yo
sw
bu
coc
de
we
con
go
tes
en
tha
stu
the
rie
ch
or
ou
no
var
ch
die
bu
ufa
gen
an
cro
(fr
cor
bu
int
an
Th
sum
ola
cac
con
sug
an
nil
ola
an
an
Figu
rat
Implications for the role of diet in acne 85stantive evidence in the historical literature that conclu-
ely supports or refutes the role of diet in the etiology of
e.
A MEDLINE search revealed that, since 1971, no single
man study has been published examining the role of diet
the etiology of acne. This paucity of recent information
ng with the inconclusive nature of the historical literature
ds little support for the dismissal of the role of diet in the
velopment of acne. Further, careful scrutiny of the 2 most
quently cited studies
11,12
used to refute the role of diet in
development of acne reveals serious design flaws similar
most other historical studies.
In the study conducted by Anderson in 1971,
11
27 medical
dents (no age or gender reported) were asked to identify 1
4 culprit foods (chocolate bars, milk, peanuts, or Coca-
la™). The subjects were then given their single culprit
d (6 servings of either 39-g chocolate bars, 1137 mL of
lk, 113 g of roasted [salted with iodized salt] peanuts, or
2 mL of Coca-Cola™) daily for 7 days. All facial lesions
re counted and graded by size, severity, and depth once
fore the experiment began and then daily during the 7-day
atment period.
Unfortunately, the methods for grading or analyzing le-
ns were not well-described. Further, no pre- or postex-
riment lesion counts were reported, nor were any of the
ta statistically analyzed. Consequently, the reported one-
rd of subjects who developed new lesions during the di-
ry treatment may or may not have been statistically signif-
nt. Additionally, no grouping of data by culprit food was
cisely reported; accordingly, the sample size assigned to
h subgroup is not known and may have lacked statistical
wer to detect a treatment effect, even had appropriate sta-
ics been used. More profoundly, however, these short-
ings pale in comparison to crucial design flaws in the
eriment. The subjects’ baseline diet was not measured.
nsequently, there is no way of knowing whether the treat-
nt diet varied from the subjects’ normal diet. Furthermore,
re was no control group, nor was the experiment blinded.
cordingly, internal validity was substantially compro-
sed in this design, making it very difficult to make mean-
ful interpretations of these data.
In the 1969 experiment by Fulton and co-workers,
12
65
jects (14 adolescent boys, 16 adolescent girls, and 35
ung adult male prisoners) consumed either a 112-g bitter-
eet chocolate bar enriched with chocolate liquor and cocoa
tter or a 112-g control bar without chocolate liquor and
oa butter once a day for 4 weeks in a single blind crossover
sign with a 3-week washout period. Subject lesion counts
re made at the beginning and end of the treatment and
trol legs of the study and organized into 3 ordinal cate-
ries: worse (lesion count increased by 30% at the end of a
t period), improved (lesion count decreased by 30% at the
d of a test period), and unaffected (if lesion count was less
n 30% change). Fig. 1 demonstrates the results of this
dy. Because there were no significant differences betweenchocolate and control bars in the 3 lesion count catego-
s, the authors concluded “ingestion of high amounts of
ten
miocolate did not materially affect the course of acne vulgaris
the output or composition of sebum.”
The Fulton and co-workers study
12
is frequently errone-
sly interpreted to mean that chocolate candy per se does
t influence the development of acne. In fact, the treatment
iable was not chocolate candy or even the bittersweet
ocolate bar used in the study, but rather the specific ingre-
nts within the bittersweet bar, the cacao solids (cocoa
tter and cacao liquor), which were excluded from the man-
cture of the control bar and replaced with partially hydro-
ated vegetable fat (28% by weight). Both the control bar
d the bittersweet bar contained high concentrations of su-
se (53% and 44.3% by weight, respectively). Cacao liquor
equently called cacao paste) is derived from the ground
e (nibs) of cacao seeds (Theobroma cacao), whereas cocoa
tter is the fat extracted from cacao seeds. The only logical
erpretation of this study is that cacao solids (cacao paste
d cocoa butter) may not be involved in the etiology of acne.
e results of the experiment cannot be generalized to as-
e that chocolate candy does not cause acne because choc-
te candy contains many other ingredients in addition to
ao solids. For instance, milk chocolate candy typically
tains 7 ingredients: (1) cacao paste; (2) cocoa butter; (3)
ar, usually sucrose; (4) nonfat milk solids; (5) milk fat; (6)
emulsifier, usually lecithin; and (7) flavoring, usually va-
la. Accordingly, any of the remaining ingredients in choc-
te candy either individually or in combination with one
other cannot be ruled out in the etiology of acne.
As has been previously pointed out,
14
both the treatment
d control bars, because of their similar high sucrose con-
re 1 Effect of chocolate bar and control bar in acne incidence
es. Adapted from Fulton and co-workers.
12ts, would have each elicited high glycemic and insuline-
c responses as a result of their inherent high glycemic