Vol.
4. 825-
830.
!)ecember
Sunburn
/995
Cancer
Associated
as Lighter
Terence
Menties
Dwyer,2
Centre
I 7 Liverpool
for
Street.
Leigh
Population
Hobart.
with
Skinned
Blizzard,
Health
Tasmania
Adolescents
and
Research.
Increased
Rosie
University
Ashbobt
of Tasmania,
7(XX). Australia
Abstract
The associations
of sun exposure,
sunburn,
skin color,
and other constitutional
characteristics
with the density
of nevi (2 mm or more in diameter)
were assessed
in a
study of 410 secondary
school
children
ages 14-15
years
in Tasmania,
Australia.
Skin color was estimated
by
using a chromameter
that measures
across
the visible
light spectrum
(400-700
nm). Skin color and lifetime
history
of sunburn
were significant
predictors
of nevus
density
on the arms and legs of girls and boys and on the
shoulders
and backs of boys. The nevus density
ratios
between
the highest
and lowest
exposure
groups
were
2.85 for the arms and legs of boys (P < 0.01), 2.19 for
the arms and legs of girls (P < 0.01), and 1.72 (P = 0.03)
for the shoulders
and backs
of boys. The increase
in
nevus density
appeared
to occur
at lower bevels of lifetime
sunburn
in children
with light or medium
skin than it did
in children
with darker
skin. Darker-skinned
children
with a history
of many sunburns
(11
lifetime
sunburns)
had a similar
number
of nevi compared
with their
lighter-skinned
peers.
Introduction
The investigation
of the relationship
between
sun exposure
and
melanoma
has been hampered
by the time gap between
the
apparent
key exposure
period
in childhood
and disease
onset
( 1 ). In an attempt
to circumvent
this barrier,
investigators
have
substituted
childhood
nevi as a proxy
for melanoma,
nevi
having
been suggested
as an appropriate
marker
for the future
development
of melanoma.
The use of nevi in this research
setting
has been justified
on the grounds
that nevi appear
to
share a common
causal
pathway
with melanoma,
involving
an
interplay
between
constitutional
factors
and sun exposure.
Supporting
this is evidence
that nevi are more prevalent
at lower
latitudes
(2) and among
fairer-skinned
people
(3), as are melanoma,
and evidence
that nevus
density
is the strongest
risk
factor
in case-control
studies
of melanoma
in adults (4).
The attempts
to use the nevus model to answer
key questions
about
sun exposure
that have arisen
in the study
of
melanoma
have thus far yielded
fewer answers
than might have
been hoped for. Several
studies
have shown that constitutional
Received
4/I 9/95: revised
7/6/95:
accepted
7/7/95.
I This
study
is supported
in part by The Medical
Limited.
a registered
health
benefits
organization.
2 To whom
requests
for reprints
should
be addressed.
6 I-02-3577()4.
Benefits
Fund of Australia
which
funded
this research.
Phone:
61-02-357702:
Fax:
Number
of Nevi
of Northern
Epidemiology,
Biomarkers
in Darker
European
& Prevention
825
as Well
Descent’
factors,
such as hair (5, 6), eye color
(6, 7), and skin color
measured
by reflectometer
(2) and by interviewers
using skin
tone panels (8), are associated
with risk. The findings
on history
of sun exposure
and sunburn
have been inconsistent.
In Vancouver
(British
Columbia,
Canada),
Gallagher
et a!. (8) found
that history
of sunburn
was related
positively
to the presence
of
nevi in children,
but Pope et a!. (5) in the West Midlands
of the
United
Kingdom
and Coombs
et a!. (9) in New Zealand
were
unable
to find an association.
Of the sun exposure
variables
measured
by Pope et a!. (5), a history
of the number
of days on
the beach
and of summer
holidays
in a hot location
were
predictive.
In Australia,
Green
et a!. (6) found nonsignificant
associations
for sun exposure
variables.
Even where
associations have been present
for a measure
of sun exposure,
they
have been relatively
weak. As a consequence,
the understanding of some of the key questions
in melanoma
research
has been
advanced
little by this field of investigation
to date. Is cumulative exposure
as important
as exposure
that causes burning?
Is
skin color important
because
it affects the risk of burning’?
Does
the effect
of sun exposure
or burning
increase
steadily
or
diminish
at some point?
On the basis of the hypothesis
that stronger
associations
might be masked
by the inadequately
controlled
confounding
or
modifying
effects
of skin color,
a study
was mounted
that
included
not only many
of the measurement
protocols
used
previously,
but which also used an objective
measure
of skin
color that had not been used previously
in this field. The device
chosen,
a Minolta
Chromameter
II ( 10), measures
across
the
visible
light spectrum
by using the Commission
International
d’Eclairage
color parameters
(L* a* b*).3
Below,
we report the findings
from a study of 14-IS
year
olds in Southern
Tasmania
(Australia;
latitude
43#{176}S).These
subjects
were drawn
from a predominantly
Anglo-Celtic
population,
approximately
90% of which traces its ancestry
to the
British
Isles.
Methods
and
Materials
Study Measurements.
A survey of nevi in grade nine secondary school
students
was conducted
during
1992. The subjects
were
selected
by using
a two-stage
sampling
method.
Ten
schools
were chosen
by using
systematic
sampling,
with a
probability
proportional
to the enrollment
of students
I 5 years
of age. The second
stage of sampling
consisted
of the random
selection
of approximately
60 students
from each school.
A
total of 472 students
(response
rate, 71 .8%) agreed
to participate in the survey.
The students,
wearing
medical
gowns
with back openings,
were examined
at school by two registered
nurses who had been
trained
by a dermatologist
in the recognition
of mebanocytic
3 The
abbreviations
yellowness
of the
used are: color
skin. respectively:
parameters
L* a* h5. lightness.
Cl. confidence
interval.
redness.
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
and
826
Sunburn,
Nevi,
and
Skin
Color
in Adolescents
nevi. Nevi were identified
by using the IARC
protocol
( I I).
Countable
lesions
were defined
as brown
to black-pigmented
macubes
or papubes that were reasonably
well defined
and were
darker
in color
than the surrounding
skin. These
countable
lesions
did not have the features
of freckles,
solar lentigines,
seborrheic
keratoses,
caf#{233}
au bait spots,
or nonmelanocytic
lesions.
Nevi were counted
in three size categories
(by diameter):
<2 mm, 2 mm but <5 mm, and 5
mm. To discriminate
size, black open circles on clear plastic with internal
diameters
of 2 mm or 5 mm were used. Lesions
were considered
to be at
beast 2 mm, or at least 5 mm, if the lesion completely
filled the
applicable
circle.
Separate
counts
of palpable
lesions
were
conducted
in each of the size categories.
Nevi were counted
on
the left arm, right beg, shoulders,
and back.
Freckling
and acne on the face were assessed
by using
charts similar
to those used by Dubin et a!. ( I 2). Hair color was
assessed
by using
hair color
samples
to define
blonde/light
brown,
brown,
black,
and red. In addition,
the nurses
subjectiveby categorized
skin color at the inner part of the left upper
arm as fair. fair/medium,
medium/olive,
olive, or dark. Puberty
was indicated
by the presence
of underarm
hair in boys and by
the onset of menarche
in girls. Height
and weight
were measured to calculate
body surface
area in square meters ( 13). The
proportion
of body surface
area at the body sites examined
was
taken to be 7% (one arm), 15.5% (one leg), and 13% (the back
and both shoulders)
(14). An objective
measure
of the visual
perception
of skin color
was obtained
by using
the Minolta
Chromameter
II, which measures
skin color in the 1976 Cornmission
International
d’Eclairage
L* a* b* color space parameters ( 15). These measurements
were made on the inner part of
the left upper
arm. Reflectometers
that measure
in the 650700-nm
spectrum
have been preferred
in previous
research
on
skin color in this field of investigation
on the grounds
that this
restricted
band
of measurement
more
specifically
measures
melanin
by excluding
the pigmentary
contribution
of carotene
and hemoglobin
( 16). Despite
this reasoning,
reflectometerassessed
skin color was not strongly
associated
with the presence of nevi in a recent Australian
study (2). On the other hand,
interviewer-assessed
skin color was quite strongly
related
to
nevi in Canadian
children,
although
this measure
has been
shown
to involve
input from the entire visible
light spectrum
(400-700
nm; Ref. 17). On the basis of this information
the
chromameter
was chosen
because,
like interviewer
assessment,
it integrates
readings
from across
the entire visual light spectrum. In addition,
the device
is free of interviewer
bias and its
use avoids the unconscious
inclusion
of eye color, hair color, or
freckling
in the assessment
of skin color.
There
was a strong
relationship
between
the L*, a*, b*
readings
and the nurses’
subjective
assessments
of skin color.
The chromameter
readings
predicted
40%
(girls)
and 34%
(boys)
of the variance
in the nurses’
assessment.
Eye and hair
color were also strongly
associated
with the nurses’
assessment
of skin color. The L*, a*, and b* readings
were used throughout
the study as the measure
of skin color because
they were so
strongly
related
to visual assessment
of skin color and had the
advantage
of being objective.
Students
were given
a questionnaire
to take home
and
complete.
This allowed
them
to consult
with their parents,
particularly
on questions
of ethnicity
and family
melanoma
history.
The questionnaire
focused
on the subjects’
ethnic
and
residential
history,
their skin’s response
to sunlight,
history
of
painful
sunburn
basting 2 days or more, and their sun exposure.
The measures
of sun exposure
included
estimates
of the number
of hours spent outside
during
the holidays
and weekends,
the
frequency
of sunbathing,
the proportion
of time spent outdoors,
and memorable,
specific
activities,
such as beach
attendance
and the number
of holidays
taken
overseas
or at a sunny
location
further
north in Australia.
For data analysis,
all subjects
with one or more grandparents born in Asia or in Southern
Europe
were excluded
from the
data set, as were those who were not of age 14 or I 5 years. Also
excluded
was one subject
with 253 nevi of diameter
at least 2
mm and 164 nevi of diameter
<2 mm. The final data set
included
4 10 students
(2 1 2 females
and I 98 males)
Data Analysis.
Alb nevi, both flat and raised, 2 mm or more in
diameter,
were included
in this data analysis.
The distribution
of the nevus
counts
for each sex at each site was positively
skewed,
with a small number
of subjects
having
many nevi.
Median
counts
are reported.
To permit
comparison
of our
results with those of other studies,
the nevus density
at each site
was calculated
by dividing
the nevus count by an estimate
of
the subject’s
body surface
area at that site. Categorical
variables
were ordered
in terms
of increasing
risk based
on a priori
grounds.
The degree of ordered
association
between
categorical
variables
was assessed
with the gamma
statistic.
The association
of nevi with skin color and sun exposure
or sunburning
was examined
by using multiple
linear regression methods.
In each regression
model,
the predictors
included
a single measurement
of skin type and a single measurement
of
sun exposure
or sunburning.
The regression
residuals
were
normally
distributed
when the dependent
variable,
nevus density (the nevus count/square
meter of body
surface
area at the
six sites examined),
was log transformed.
Potential
confounders
of this association
were age, sex, socioeconomic
status,
and
constitutional
characteristics
other than measures
of skin type
(acne on the face and pubertal
status).
A putative
causal
factor
was considered
to statistically
confound
the association
between
bog nevus
density
and a predictor
(skin type or sun
exposure
or sunburning)
if it was associated
with log nevus
density
and with the predictor
in univariate
analysis
and if its
inclusion
in the multivariate
model
changed
the parameter
estimate
for the predictor
by > 10% ( 18). Age ( I 5 versus
14
years)
and pubertal
status
(yes versus
no) were
entered
as
dichotomous
(1-0)
dummy
predictors.
Socioeconomic
status
was assessed
by using
an index
of relative
socioeconomic
disadvantage
based
on the postal
code,
with the weightings
determined
by the Australian
Bureau
of Statistics;
the indices
were categorized
into tertiles.
The L* readings
of skin color
were categorized
into tertibes. A group of dichotomous
dummy
predictors,
one for each category
other than the least-risk
category,
was entered
into the regression
model for each categorical variable.
To compare
the nevus
densities
of groups
of
subjects
cross-classified
by skin color and the number
of their
lifetime
sunburns,
ratios of the geometric
mean of the nevus
densities
in each group to that of the baseline
comparison
group
(never
sunburned,
darkest
tertile of skin color) were estimated
by using dichotomous
dummy
predictors.
The dependent
vanable was the logarithm
of nevus
density,
and the parameter
estimates
were exponentiated
to obtain the estimate
of the ratio.
The data analysis
was undertaken
by using the SAS statistical
package
(19). Estimated
probability
values for the tests
of the null hypotheses
are quoted,
and 95% CIs were estimated
for nevus density
ratios.
Results
The data relating
subjects
and their
presented
in Table
to the constitutional
characteristics
of
history
of sunburn
and sun exposure
1 together
with the median
nevus counts
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
the
are
by
Cancer
:
8
GIRLS
.
U
g0
.<
Boys
o
2.
Ch
o,
.O
.-.e
5.
4.
.2
6.
7.
g
;C
,
g
n’
3.
5
S .s
E
‘a
g=
-‘
tu
<
t.
.
g
a
‘3
8.
‘g
E.0
1:2
9.
10
Table I
sunbum
90
‘3
x
Frequency
distribution
and sun exposure,
and
Acne
0.03
.t7
-.08
#{149}
0.13
0.07
O.2t
#{149} 0.19
0.20
“
0.09
0.08
0.10
color
.07
.14
-.Ot
.13
.19
.09
0.13
0.19
0.04
0.10
.03
0.03
‘3.0!
‘3.03
.03
Lifetime
*
9.05
characteristics
and
nevus counts
‘).2t
association.
P < 0.05:
‘).2e
‘).tC
0.48
S
5*
0.14
#{149} *5*55*
#{149} 0.53
.09
O.l9
0.22
0.29
-.3t
.05
0.28
9.2
‘1.37
0.17
0.05
0.10
0.1 I
(‘t
198)
-
-
52
48
-.02
Age
5
0.41
0.16
0.13
-.22
.09
‘3.77
0.33
12)’
.30
.21
0.64
0.65
.05
.08
‘
14 yr
58
53
15 yr
42
47
Acne
.17
.0t
0.21
0.15
0.29
0.68
.2t
.ffl
0.Ot
12’
‘).22
0.21
‘3.54
..00
.
0.05
0.05
0.24
0.t I
0.M
0.88
.J36
-.32
.t6
.06
‘
0.81’ ‘).2#{128}
10?
‘).0
0.04
.08
‘3W
33
52
49
9
17
-.20
Eye
“
0.30
0.15
‘
lOt
39
Mild
.19
0.t9
‘
#{182}2:.t
None
Moderate/heavy
S
Ordered
associations
and sun exposure.
‘.
Boys
21 2)
=
Girls
.t0
.05
..09
‘
0.2
color
Brown
18
15
Green/hazel
29
25
Blue/gray
52
60
Hair
Fig. 1.
sunburns,
median
Boys
Time in sun on
holidays
Hotidaysinthe
un overseas
history
of
body site
Sex
‘S
burns
Burnsta.st
summer
and
%
Skin reaction
to hour in sun
characteristics,
counts
by sex
Girls
“
Skincotor
(tertitesofL)
Kindoftanat
nd ofsummer
& Prevention
*
0.37
Eye cotor
Hair
).0f
of constitutional
median
nevus
It.
Subject
‘).tf
Blomarkers
;
ie’
aE
mg
Age
U
Epidemiology,
summarize
P < 0.01:
between
Estimates
constitutional
in each cell
the evidence
against
P < 0.001.
the null
characteristics,
are ‘y measures
hypothesis
history
of
of ordinal
that
y
=
0.
‘,
***,
color
Black
Dark
brown
Blonde/light
Skin
color
First
body site. Nearly
all of the girls and most of the boys had
reached
puberty.
Nevi on the shoulders
and back were more
common
in boys than in girls (P < 0.01), but the number
was
similar
on the arm and beg.
Fig. I is the correlation
matrix
for ordered
categorical
variables
constructed
from questionnaire
responses
or measurements
made by the nurses.
Many
of the associations
in this
correlation
matrix
are as expected.
Those
subjects
with blue
eyes were more likely to have fair hair than were those with
brown eyes. Those who described
themselves
as tanning
easily
were less likely to sunburn
and were more likely to spend time
in the sun. The complexity
of the associations
summarized
in
Fig. I highlight
the complexity
of the interrelationships
in this
research
setting.
An apparently
important
observation
is that
girls appeared
more likely than boys to report that they sunburn
rather than tan if they were assessed
as having
light skin by the
chromameter.
Possibly
associated
with this is the stronger
inverse relationship
for girls than for boys between
having
light
skin and spending
time in the sun.
Table 2 provides
information
on the univariate
association
between
nevus
density
and factors
that might be expected
to
influence
the development
of nevi. The associations
are presented
as ratios of the geometric
means
of nevus density
for
groups
of subjects
falling
in the extreme
categories
of each
factor.
The associations
were generally
strongest
for the arm
and leg for both sexes
and weakest
for nevi on the back for
girls. The highest
nevus density
ratios were found on the arm
and beg when comparing
categories
of skin color,
skin type
(kind of tan at the end of summer
and skin reaction
to the sun),
and lifetime
sunburn.
These
ratios
were considerably
higher
than were those for hair or eye color or those for sun exposure
measurements.
Only the L* score associations
are presented
because
they proved
to be much stronger
than the a* or b*
scores.
The associations
of nevi with skin color and sunburn
were
examined
in a multiple
regression.
The self-assessments
by the
brown
Red
(tertiles
(medium)
Third
Kind
5
35
54
56
4
4
32
33
33
31
35
36
of L5 value)
(dark)
Second
2
39
(light)
of tan at end
of summer
Dark
13
14
Medium
44
53
33
29
9
4
Light
Practically
Skin
none
reaction
Tan
to first
hour
in sun
only
Sunbum,
tan
Sunbum,
then
Lifetime
peel
painful
24
21
43
49
33
30
sunbums
Never
14
12
57
54
times
16
23
I 1 + times
13
II
1-5
times
6-10
Painful
sunbums
last
summer
Never
50
49
1-5
times
47
46
6+
times
3
5
Hours
of time
in sun,
weekends,
and
holidays
<2
h
23
12
>2
h
43
44
Just
about
34
44
Holidays
all day
in the
sun,
interstate,
and
overseas
Never
1-5
times
5 times
Nevi
2 mm.
Shoulders
53
23
16
Median
count
Median
count
12.0
16.5
12.0
12.0
Leg
9.0
8.0
21 .0
20.0
33.0
41.0
All sites
and
leg
examined
back
site
31
Arm
Arm
and
by body
30
47
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
827
828
Sunburn,
Nevi,
and
Skin
Color
in Adolescents
Table
2
d ensity
Nevus
ratios”
for c )nstitu
tional
characteristics,
history
of bum s, and
sun exposure
Boys
Girls
Categories
compared
Arm
Ratio
Age.
15 yr s’s. 14 yr
Acne.
moderate/heavy
Eyes.
blue/gray
Hair.
red/blond/light
Skin
color
Kind
oftan,
Skin
reaction,
Lifetime
bums.
Burns
Time
last
summer.
in sun.
Holidays
all day
overseas.
Univariate
ratios
measurements
and
“
I, p <
1+
is.
6+
nevus
95% Cls
(if
(95%
CI)
and
Ratio
leg
(95%
Shoulders
CI)
and
Ratio
back
(95%
CI)
(0.87-1.29)
1.01
(0.86-1.18)
0.85
(0.62-1.17)
0.91
(0.65-1.28)
1.35
(1.02-1.78)”
1.12
(0.89-1.40)
1.18
(t).95-l.45)
0.96
(0.75-1.22)
1.29
(0.99-1.68)
1.12
(0.90-1.39)
t).89
(0.75-1.06)
0.87
(0.72-1.03)
1.23
(1.02-1.50)”
1.12
(0.96-1.32)
1.38
(1.1 ll.73)”
0.87
(0.70-1.08)
1.62
(1.29-2.03)”
1.15
(0.95-1.39)
1.69
(1.27-2.24)”
0.93
(0.69-1.26)
1.56
(1.14-2.13)”
0.87
(0.68-1.12)
1.54
(1.23-1.92)”
0.91
(0.71-1.16)
1.86
(1.40-2.47)”
1.14
(0.90-1.44)
none
1.52
(1.10-2.10?’
0.99
(0.70-1.41)
1.84
(l.26-2.7O)
1.49
(l.06-2.l0)
is.
none
1.17
(0.99-1.39)
0.91
(0.76-1.08)
1.20
(1.00-1.46)”
1.08
(0.92-1.26)
2 h or less
0.94
(0.75-1.19)
1.01
(0.78-1.29)
1.05
(0.79-1.41)
1.09
(0.84-1.41)
is.
i’s. tan
is.
Ratio
Arm
1.06
1st
is. dark
11+
CI)
back
(0.74-1.06)
3rd
sunbum
(95%
and
0.88
vs. brown/black
none/light
Shoulders
(0.704.98)
vs. none
tertiles).
leg
t).83
vs. brown
(L5
and
trips
only
vs. none
(0.77-1.19)
t).96
densities
at the highest
for the ratio estimates.
and
lowest
of ordered
1.35
(1.07-1.72?’
categories
of constitutional,
1.31
skin
color,
(0.96-1.79)
skin
type.
1.17
history
of sunbums.
(0.92-1.50)
and
sun
exposure
0.05.
children
ofdepth
oftan
and skin reaction
were highly correlated
with the L* reading
from the chromameter
and with each other.
On a priori
grounds,
we believe
that all three measurements
assessed
a similar
property
of the skin. We chose to include
the
L* reading
in preference
to tanning
or skin reaction
as an
estimate
of skin color because
we believed
it was most likely to
be independent
of the sunburning
and sun exposure
measures.
It is objective
and, therefore,
unlike the skin reaction
questions,
could not influence
directly
or be influenced
directly
by the sun
exposure/burning
questions.
The subjects’
lifetime
history
of sunburn
and the L* measure of skin color from the chromameter
were independently
associated
with nevi on the arms and legs of boys and girls and
on the shoulders
and backs of boys. Adjusting
for age, socioeconomic
status,
acne on the face, or pubertal
status did not
alter the parameter
estimates
by > 10% and nor did adjusting
for eye color or hair color. Our conclusions
were qualitatively
unchanged
when the other measures
of skin type (tanning
and
skin reaction)
were allowed
to enter the model.
The estimate
of
the number
of holidays
in the sun in warmer
climates
(the sun
exposure
variable
most strongly
associated
with nevi in univariate analysis)
was a significant
predictor
of nevi density
only
on the shoulders
and backs
of girls in multivariate
analysis.
Thus,
the final model
included
only skin color and lifetime
sunburn.
Of the variance
in nevus density
on the arm and leg,
I I % was explained
by the model
for girls and 15% was explained
for boys.
The results
from this model
are depicted
in Fig. 2. The
pattern
of nevus densities
on the arm and leg are similar
for
girls and boys, with significant
independent
effects for lifetime
sunburn
and skin color. The highest
nevus density
ratios occurred for subjects
with many sunburns
and the lowest were for
darker-skinned
children
with few sunburns.
The nevus density
in lighter-skinned
boys who had been sunburned
1 1 or more
times was 2.85 times that of never sunburned,
darker-skinned
boys (P < 0.01); for girls, this ratio was 2.19 (P < 0.01). The
increase
in nevus density
appeared
to occur at lower levels of
lifetime
sunburn
in children
with lighter or medium
skin than in
children
with darker skin. There was no evidence
of linear trend
in skin color effect for children
with 1 1 + sunburns
(P = 0.76
for girls; P = 0.27 for boys),
suggesting
that darker-skinned
children
who are frequently
sunburned
develop
as many nevi as
lighter-skinned
children.
There was no association
of nevi on
the shoulders
and backs of girls with either skin color or history
of sunburning.
For boys, the pattern
of nevus densities
on the
shoulders
and back appeared
to be similar
to, but less pronounced
than, that on their arm and leg. The nevus density
ratio
comparing
lighter-skinned
boys who had been sunburned
often
with darker-skinned,
never-sunburned
boys
was 1 .72 (P
0.03).
Discussion
The findings
from this study highlight
the complexity
of the
interrelationship
among
constitutional
factors,
sun exposure,
and nevi in children.
In addition,
they emphasize
the need to
examine
effects
separately
for boys and girls and by body site
and to include
appropriate
measures
of skin color. They show
that there are likely to be some differences
in the relationships
within
categories
of skin type or color,
both for biological
reasons
and because
sun-seeking
behavior
differs
by skin color
and gender.
The principal
finding
of this study is that lifetime
history
of sunburn
predicts
presence
of nevi in a population
of adolescents of primarily
Anglo-Celtic
origin
after taking
skin color
into account.
The results
reported
here suggest
that this is not
solely because
those who are likely to sunburn
have skin which
is more susceptible
to the development
of nevi but instead
that
sunburn
plays a causal role. Nonetheless,
the effect of sunburn
varies with the relative
darkness
of the individual’s
skin.
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
Cancer
Girls
Epidemiology,
Biomarkers
& Prevention
Boys
Fig. 2.
Nevus
density
ratios
for history
of lifetime
sunbums
and
tertiles
of skin color readings.
The reference
groups
were subjects
with no sunhurns
and with the darkest
skin color. There
were significant linear effects
in nevus density
for skin color (P < 0.01 ). sunbum
on the arms and legs ) P
0.07 for girls and P
0.02 for boys). and
sunburn
on the shoulders
and backs
of boys
(P
0.t).
Med.
medium.
These results suggest
that only a few sunburns
can produce
nevi in the light skinned
and, indeed,
that something
less than
a painful
sunburn
might achieve
this outcome.
It is unlikely
that
the nevi that have arisen in the group of light-skinned
children
who report
no previous
sunburn
are entirely
unrelated
to sun
exposure
because
this group on average
have 34 nevi at the sites
examined.
Their
English
counterparts,
who presumably
have
similar skin but live in a lower total sun exposure
location,
have
fewer than five nevi on their entire bodies
(5, 20).
To explain
the sun exposure-nevus
link, Nicholls
(21)
proposed
that UV radiation
causes
cellular
damage
resulting
in
mutation
of some melanocytes,
which in turn may form nevi. If
one accepts
this explanation,
then the data here suggest
that
lighter-skinned
individuals
either
experience
more
cellular
damage
with a given sunburn
or that their melanocytes
are more
susceptible
to mutation
when irradiated.
The
finding
that lifetime
history
of sunburn
is more
strongly
associated
with nevi than any measure
of sun exposure
suggests
that sunburning
is more important
than cumulative
sun
exposure.
We are reluctant
to confidently
conclude
this. Although
the questions
concerned
with sun exposure
that we
chose had been used previously
by others, a comparison
of sun
exposure
and sunburning
responses
between
this survey
and a
second
conducted
in 1993 showed
that misclassification
on
those questions
was greater
for sun exposure
than sunburning
history.
It is unlikely
that we or others
have yet measured
sun
exposure
with a sufficient
degree
of precision
to estimate
its
effects
well. It is possible
that sunburning
is simply
a surrogate
for sun exposure
but one that happens
to be recalled
with less
misclassification
than responses
to any of the sun-exposure
questions.
However,
in our data the associations
of sunburning
with other variables
exhibited
a sufficiently
different
pattern
to
those for sun exposure,
“lifetime
burns”
being
more strongly
associated
with “skin
reaction”
than with “kind
of tan” and
“time in sun” being more strongly
associated
with “kind of tan”
than with “skin
reaction,”
for it to be appropriate
to treat
sunburning
as an independent
factor.
This has been the approach
also of other investigators
(2, 8).
When the association
of nevi with sunburn
is examined,
there is no evidence
of the inverted
U-shaped
relationship
that
some have found in the adult data on history
of sun exposure
and melanoma
(22). A linear or bog-linear
relationship
beyond
a particular
threshold
is more consistent
with our data. However, these data on nevi and sunburning
do not preclude
the
possibility
that the relationship
of nevi with
sun exposure
follows
an inverted
U-shaped
curve.
If the association
of sun
exposure
with sunburning
had this shape,
then nevi and sun
exposure
would
be related
in just this way. Although
this is
plausible,
the absence
of an inverted
U-shaped
relationship
between
sun exposure
and sunburning
after controlling
for skin
color in these data suggests
that such a relationship
in studies of
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
829
830
Sunburn,
Nevi,
and
Skin
Color
in Adolescents
melanoma
(22) and nevi (23) might be produced
by the confounding
effects
of skin color.
The chromameter
readings
for L* predicted
nevi more
strongly
than similar
objective
measurements
of skin color
obtained
from refiectometers
that have been used previously,
which measure
in the near infrared
region of the spectrum
(685
nm; Ref. 2). It is not concluded
from this that chromameters
are
the ideal way to measure
skin color, but that instruments
that
estimate
skin color across
the visible
light spectrum
appear
to
have a place in this field of research
and that their additional
development
needs to be pursued.
It is possible
that instruments
such as spectrophotometers,
which
have the capacity
to measure more accurately
at specific
wavelengths
in the range 400700 nm, may estimate
even more precisely
the properties
of the
skin that researchers
are seeking
to measure.
The data presented
here suggest
that there is much that can
be learned
about the sun’s effect on melanocytes
in children
by
studying
nevi and that the inferences
will help in understanding
the etiology
of melanoma.
The fact that in these subjects
an
interplay
of skin color and sunburning
was seen to be most
important
for those site- and sex-specific
locations
where melanoma
later occurs
and least important
for the site where
melanoma
is uncommon
(e.g.
, the
back for females),
supports
the inference
that insights
gained
in studying
nevi in children
have relevance
to understanding
the development
of melanoma
in adults.
7. English.
tomic sites
D. R.. and Armstrong.
B. K. Melanocytic
nevi in children.
I. Annand demographic
and host factors.
Am. J. Epidemiol..
139: 390-401,
I 994.
8. Gallagher,
R. P., McLean,
D. I., Yang. C. P., Coldman,
A. J.. Silver,
H. K. B.,
Spinelli,
J. J., and Beagrie,
M. Suntan,
sunbum,
and pigmentation
factors
and the
frequency
of acquired
melanocytic
nevi in children.
Similarities
to melanoma:
The Vancouver
Mole Study.
Arch. Dermatol..
/26: 770-776,
1990.
9. Coombs,
B. D., Sharples,
J. M. Variation
and covariates
J. Epidemiol..
/36:
344-355,
10. Minolta
Camera
Minolta
Camera
Co.,
Co.,
Ltd..
Cancer
Registry.
Tasmania.
Australia)
Colour
Communication.
Osaka,
Japan:
12. Dubin,
N., Moseson,
M., and Pastemak,
B. S. Epidemiology
of malignant
melanoma:
pigmentary
traits, ultraviolet
radiation
and the identification
of high
risk populations.
In: R. Gallagher
ted.).
Recent
Results
in Cancer
Research:
Epidemiology
of Malignant
Melanoma.
Vol. 102, pp. 56-75.
New York: Springer-Verlag.
1986.
13.
Mosteller,
J. Med.,
14.
Lund,
Gynecol.
15.
R. D. Simplified
3/7:
1098,
C. C.,
Obstet..
Weatherall,
colour
CIELAB
absorption
Dace Shugg
(Registrar
of the Tasmanian
provided
technical
assistance.
1992.
Ltd. Precise
1993.
11. English,
D. R., MacLennan,
R., Rivers,
J., Kelly.
J.. and Armstrong,
B. K.
(eds.).
Epidemiological
Studies
of Melanocytic
Naevi:
Protocol
for Identifying
and Recording
Naevi.
IARC
Intemal
Report
No. 90/Ot)2.
Lyon.
France:
IARC,
1990.
and
calculation
ofbody
surface
area
(Letter).
N. EngI.
1987.
and
Browder,
79: 352-358,
N. C. The
estimation
of areas
of burns.
scattering
Surg.
1944.
I. L., and Coombs,
B. D. Skin colour
measurements
space values.
J. Invest.
Dermatol.,
99: 468-473.
16. Dawson.
J. B., Barker,
Fisher,
G. W., and Feather,
Acknowledgments
D. J., Cooke,
K. R., Skegg,
D. C. G., and Elwood,
of the number
of benign
nevi in adolescents.
Am.
D. K.. Ellis.
D. H., Grassman,
E.,
J. W. A theoretical
and experimental
by in visa
skin.
Phys.
Med.
Biol.
25:
in terms
of
1992.
Cotterill,
study
695-709,
17. Hertzman,
C.. Walter,
S. D., From. L., and Alison,
A. Observer
skin colour
in a study ofmalignant
melanoma.
Am. J. Epidemiol.,
J. A.,
of light
1980.
perception
of
/26: 901-910,
1987.
18. Greenland.
Am. J. Public
References
I . Armstrong.
accumulated
B. K. Epidemiology
exposure
to the sun?
of malignant
melanoma:
J. Dermatol.
Surg. Oncol..
intermittent
/4: 835-849.
2. Kelly.
J. W.. Rivers.
J. K., MacLennan.
R.. Harrison,
S.. Lewis.
Tate, B. J. Sunlight:
a major factor associated
with the development
cytic
nevi in Australian
schoolchildren.
J. Am. Acad.
Dermatol..
1994.
3. Rampen.
F. H. J. Frequency
Acad.
Dermatol..
/5: l2(X)-l203.
of moles
1986.
4. MacKie,
R. M.. Freudenberger.
chart for cutaneous
melanoma.
T..
Lancet.
as a key to melanoma
and Aitchison,
2: 487-49)).
5. Pope. D., Sorahan.
T.. Marsden,
J. R.. Ball.
Benign
pigmented
nevi in children.
Prevalence
Midlands.
United
Kingdom
Mole Study.
Arch.
6. Green.
A.. Siskind,
nevi in schoolchildren
1989.
A. E., and
of melano30: 40-48,
incidence.
T. C. Personal
1989.
J. Am.
risk-factor
P. M., Grimley,
P.. and Peck,
and associated
factors:
The
Dermatol.,
128: 1201-1206,
V.. Hansen,
M-E.. Hanson,
L.. and Leech,
in Queensland.
J. Am. Acad.
Dermatol..
or total
1988.
I. M.
West
1992.
P. Melanocytic
20: 1054-1060.
S. Modeling
and variable
selection
Health,
79: 340-349,
1989.
19. SAS Institute.
Inc. SAS/STAT
User’s
Cary, NC: SAS Institute.
Inc.. 1989.
Guide,
in epidemiological
Version
20. Sorahan,
T., Ball, P. M., Grimley,
R. P., and Pope,
in children
from
Kidderminster,
England:
prevalence
Acad.
Dermatol.,
22: 747-750.
1990.
21
.
Nicholls,
anatomical
1973.
E. M. Development
distribution
of malignant
and
elimination
melanoma.
analysis.
6, Ed. 4, Vol.
1 and 2.
D. Benign
pigmented
nevi
and associated
factors.
of pigmented
Cancer
(Phila.).
moles,
and
32:
191-195,
the
22. Elwood,
J. M., Gallagher,
R. P., Hill, G. B.. and Pearson,
J. C. G. Cutaneous
melanoma
in relation
to intermittent
and constant
sun exposure:
the Western
Canada
Melanoma
Study.
Int. J. Cancer,
42: 319-324,
1985.
23. Armstrong,
B. K., de Klerk. N. H., and Holman,
acquired
melanocytic
nevi: constitutional
variables,
NatI. Cancer
Inst., 77: 329-335,
1986.
C. D. J. Etiology
sun exposure.
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.
of common
and diet. J.
Sunburn associated with increased number of nevi in darker as
well as lighter skinned adolescents of northern European
descent.
T Dwyer, L Blizzard and R Ashbolt
Cancer Epidemiol Biomarkers Prev 1995;4:825-830.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cebp.aacrjournals.org/content/4/8/825
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cebp.aacrjournals.org on June 18, 2017. © 1995 American Association for Cancer Research.