Context High nevus density is a risk factor for cutaneous malignant melanoma.
Melanocytic nevi originate in childhood and are largely caused by solar exposure.
Objective To determine whether use of broad-spectrum, high–sun protection
factor (SPF) sunscreen attenuates development of nevi in white children.
Design Randomized trial conducted June 1993 to May 1996.
Setting and Participants A total of 458 Vancouver, British Columbia, schoolchildren in grades
1 and 4 were randomized in 1993. After exclusion of nonwhite children and
those lost to follow-up or with missing data, 309 children remained for analysis.
Each child's nevi were enumerated at the start and end of the study in 1996.
Intervention Parents of children randomly assigned to the treatment group (n=222)
received a supply of SPF 30 broad-spectrum sunscreen with directions to apply
it to exposed sites when the child was expected to be in the sun for 30 minutes
or more. Children randomly assigned to the control group (n=236) received
no sunscreen and were given no advice about sunscreen use.
Main Outcome Measure Number of new nevi acquired during the 3 years of the study, compared
between treatment and control groups.
Results Children in the sunscreen group developed fewer nevi than did children
in the control group (median counts, 24 vs 28; P=.048).
A significant interaction was detected between freckling and study group,
indicating that sunscreen use was much more important for children with freckles
than for children without. Modeling of the data suggests that freckled children
assigned to a broad-spectrum sunscreen intervention would develop 30% to 40%
fewer new nevi than freckled children assigned to the control group.
Conclusions Our data indicate that broad-spectrum sunscreens may attenuate the number
of nevi in white children, especially if they have freckles.
A strong risk factor for the development of cutaneous malignant melanoma
(CMM) in white populations is the presence of acquired melanocytic nevi.1-4 There
is a consistent rise in risk of CMM with increasing number of nevi in virtually
every study that has assessed this relationship.3-6
The presence of remnants of preexisting nevi in about 50% of CMMs7 indicates that acquired nevi are precursor lesions
for many,8,9 although not all,
melanomas.9
Recent work has focused on the origin and etiology of nevi in children,
who are, for the most part, born without nevi. Fewer than 2% of children have
a congenital nevus,10,11 although
acquired nevi begin to become clinically obvious at an early age.12 Etiologic studies have shown that host and pigmentary
characteristics (eg, light skin color, freckling, propensity to burn in the
sun) that raise adult risk of CMM also predispose children to develop high
nevus density.13-16
Genetic factors also influence nevus prevalence, with higher counts of melanocytic
nevi in melanoma-prone families.17,18
The principal environmental risk factor for the development of acquired nevi
is sunlight exposure as measured by sunburn history,13
latitude of residence,19 or reported solar
exposure.14,16
Reducing acquired nevi in children may reduce their risk of CMM as adults.
With this in mind, we have conducted a randomized controlled trial to see
whether broad-spectrum high–sun protection factor (SPF) sunscreen use
might attenuate the number of new nevi that develop in white children.
Study Design and Data Collection
The study was approved by the British Columbia Cancer Agency and University
of British Columbia research ethics committees. Six Vancouver elementary schools
with the largest proportion of white children were selected for the study.
School principals were approached for permission to conduct the study within
their schools. After securing permission from the Vancouver School Board,
the principals released names of all children in grades 1 and 4 (aged 6-7
and 9-10 years, respectively) and their parents to the study. Parents were
sent a letter explaining the study and were asked for written permission to
examine each child and enroll the child and a parent in the 3-year investigation.
At enrollment, each student was examined by either a dermatologist (J.K.R.)
known for his expertise in childhood nevus studies19-21
or by a physician specially trained by him. All nevi, regardless of size,
were counted using techniques outlined in the International Agency for Research
on Cancer counting protocol.21 The scalp, genital
area, and buttocks were not examined, nor was the breast area in girls.
Degree of freckling on the face, shoulders, and arms was estimated using
a chart13 with good observer reproducibility.
Height and weight of each child were taken to allow calculation of body surface
area.22 Skin reflectance on a non–sun-exposed
site (upper inner arm) was measured using a reflectance spectrophotometer
set to 680 nm. Parents of each child completed a detailed questionnaire, assessing
the child's ethnic origin, sun sensitivity, sunburn history, and holiday sunlight
exposure to the time of randomization.
Children were individually randomized by the study statistician (A.J.C.)
to the sunscreen (intervention) or the ambient use (control) group. The statistician
had no contact with the physicians counting nevi or with the study subjects.
Parents of those randomized to the sunscreen group received a bottle of SPF
30 broad-spectrum sunscreen near the end of each school year in June 1993,
1994, and 1995. Parents were instructed to apply the sunscreen in amounts
they usually used to all sun-exposed sites on the enrolled child whenever
he/she was expected to be in the sun for 30 minutes or more. Parents were
specifically asked to use the particular bottle of sunscreen only on the enrolled
child. At the end of July each year, a second bottle of sunscreen was sent.
Parents were then asked to measure and report how much of the original bottle
had been used by marking what remained in the first bottle on an actual-size
diagram of the sunscreen bottle. Parents were instructed to use the second
bottle of sunscreen on the index child for the remainder of the summer and
the next Christmas and spring breaks. Parents whose children were randomized
to the control group were given no advice as to sunscreen use, and no placebo
was provided. Because of the level of general education about sun exposure,
however, use of sunscreen was substantial in the control group.
At the end of each summer vacation, solar exposure during the previous
3 months was determined for children in each study group using an activity-based
questionnaire. Clothing preference and sunscreen use during outdoor activities
were assessed on a semiquantitative basis. Similar instruments were used to
evaluate solar exposure during the Christmas and spring breaks each year.
As Vancouver is a relatively low-sunlight area and records high temperatures
only in the summer, evaluating summer exposure plus the other 2 school holiday
periods each year captures most solar exposure in children.
In May 1996, all children retained in the study were reexamined by physicians,
and their nevi were enumerated once again. Physician-counters did not know
to which study group children had been assigned. To ensure that nevus counts
were concordant among counters, 69 (15%) of the students were counted by 2
of the 3 physicians and 17 (4%) were counted by all 3 physicians. Assuming
the variance among the duplicate and triplicate counts was typical, the proportion
of variance in whole-body nevus counts attributable to the effect of the counter
was less than 5%.
Data were used only if students completed the whole protocol, defined
as the intake and exit nevus counts, the intake questionnaire, and at least
2 of the 3 summer sun update, Christmas break, and spring break questionnaires.
If 1 of the summer sun updates was not completed, mean values from the other
2 such questionnaires were substituted. The same procedure was followed for
missing Christmas and spring break questionnaires.
It is customary in clinical trials to conduct an analysis based on intent-to-treat.
In this study, no intermediate nevus counts were taken between randomization
and conclusion. It is therefore not possible to conduct an intent-to-treat
analysis based on imputed end-point values for subjects who were lost to follow-up
during the course of the study.
Several measures of sun exposure were calculated. Minimal erythemal
dose (MED) information for clear sky conditions by latitude and month of the
year were obtained from Diffey and Elwood.23
Vacation solar exposure in MEDs during the 3 years was assessed using location,
latitude, and month of the vacation, assuming that vacation exposure took
place during peak, daylight UV-B exposure hours. Total UV exposure from vacation
and recreational activities in MEDs, adjusted for clothing worn while outdoors,
was also calculated by anatomic subsite. Because it was not possible to directly
measure whole-body exposure, MED values for each of the 4 anatomic subsites
were simply summed in each subject to get the whole-body score.
Whole-body nevus counts from 1993 were subtracted from 1996 counts for
each child, giving the number of new nevi. All nevi regardless of size were
included in the counts. Comparisons between study groups were based on medians,
and differences were assessed using the Kruskal-Wallis test.
A linear regression model to account for the number of new nevi was
fitted, using the following predictor variables: treatment group, school grade
(equivalent to age), sex, skin reflectance value, facial freckling, hair color,
skin reaction to sunlight, family history of skin cancer, sunburn history
to age 5 years, sunburn history during 1993 through 1996, hours spent outdoors
during 1993 through 1996, vacation sun exposure during 1993 through 1996,
and total sun exposure during 1993 through 1996 adjusted for clothing. The
variables sex, grade, skin reaction to sunlight, treatment group, and hair
color (dark brown, light brown, blond, red) were modeled as categorical variables
and all others as continuous variables.
The baseline model included sex, grade, hair color, and treatment group.
Additional variables were added to the baseline model using a forward-selection
algorithm, with inclusion restricted to factors with a significance level
of P<.10. An inspection of plotted data suggested
potential interactions between treatment group and other predictor variables.
Consequently, in the initial stages of multivariate analysis, variables were
added to the model as a combined main effect and interaction-with-treatment-group
effect. Significance of these variables was assessed according to the P value of the interaction effect rather than the main
effect.
After including variables with significant interaction effects, subsequent
modeling was performed to test for the significance of the remaining independent
variables. Residual plots were used to confirm the independence, normal distribution,
and constant variance of the errors.
A total of 696 children (354 in grade 1 and 342 in grade 4) were ascertained
in the 6 schools. Of these, 458 (66%) were enrolled in the study and randomized
to either the sunscreen or control group. At the completion of the trial 3
years later, 393 (86%) remained. The children in the study were largely white
(323 [82%]); Chinese Canadian and other Asian Canadian students (37 [9%])
made up the second-largest ethnic group. The number of Asian Canadian and
dark-skinned subjects was small, and, as they acquire few new nevi with age20 and are at low risk of eventual cutaneous melanoma,24 they were eliminated from consideration prior to
beginning the analysis. Six grade 1 and 8 grade 4 students with missing nevus-counter
identification were excluded, leaving 309 white children for the final analysis
(Figure 1).
The median nevus counts at intake were 41 for grade 1 students (aged
6-7 years) and 68 for grade 4 children (aged 9-10 years). The distribution
of nevi at intake was skewed positively, with a few children having very high
counts. No child had a count of zero.
Factors such as hair color, skin reaction to sunshine, facial freckling,
and sunburn score in the first 5 years of life demonstrated associations with
nevus counts similar to those seen in previous studies, as shown in Table 1. Skin reflectance value at 680
nm did not demonstrate a significant relationship with nevus frequency.
Analysis of the number of new nevi revealed that children in the sunscreen
group developed significantly fewer new nevi than those in the control group
(median counts, 24.0 vs 28.0; P=.048). A comparison
using mean values showed an even greater difference (28.8 vs 34.6). A few
children had a lower nevus count in 1996 than in 1993.
Table 2 compares measures
of sunlight exposure in the 2 treatment groups. Time spent outdoors from 1993-1996
was very similar among the students in each study group. No difference in
vacation solar exposure from 1993-1996 in MEDs was seen between the sunscreen
and control groups, and no major difference was seen in total sunlight exposure
adjusted for clothing coverage for whole-body or anatomic subsite (Table 2).
Use of sunscreen was assessed by anatomic subsite (Table 3). When sunscreen was used, exposure was defined as protected
and when not used, as unprotected. Median number of episodes of protected
and unprotected exposure during the observation period showed a greater proportion
of unprotected episodes in the control group at each subsite. The majority
of study subjects in both the sunscreen and control groups reported zero episodes
of trunk exposure unprotected by sunscreen, creating artificially low medians.
Mean values provide more credible estimates and show an excess of unprotected
episodes in the control group compared with the sunscreen group (7.7 vs 5.2).
A model of the effects of the independent variables on the whole-body
number of new nevi is presented in Table
4. Total sunlight exposure, adjusted for clothing, school grade
(age), the interaction term for sunscreen group, and degree of facial freckling,
appears to predict nevus counts. The interaction between being randomized
to the broad-spectrum sunscreen group and degree of freckling is statistically
the strongest predictor of new nevi. Figure
2 shows that the importance of being randomized to the sunscreen
group increases with increasing degree of freckling. Removing subjects with
the greatest number of new nevi had little effect on the divergence of the
regression lines for subjects in the 2 study groups.
To further assess the difference in number of new nevi between subjects
with and without facial freckles, models were constructed separately for grade
1 children and grade 4 children, with subjects dichotomized into 2 groups:
those with ≤10% freckling (no freckles), and those with >10% freckling
density (freckles). The model predicted that grade 1 children who had freckles
would have about 40% fewer new nevi after 3 years when randomized to the sunscreen
group rather than the control group. Grade 4 children with freckles randomized
to the sunscreen group would have about 30% fewer new nevi than if they were
randomized to the control group. Children with no freckling in grades 1 and
4 would have little advantage when randomized to the sunscreen group compared
with the control group.
Finally, if sunscreen attenuates the development of new nevi, it might
be expected that, after control for freckling, subjects in the intervention
group who used the most sunscreen would have the fewest new nevi. Figure 3 (grade 4 children presented; grade
1 graph similar) also demonstrates that this is the case, and, although the
differences are not statistically significant, there is an inverse relationship
between sunscreen use and new nevi.
To our knowledge, this is the first randomized trial of the use of sunscreen
as a chemopreventive agent for attenuating nevi in children. Strengths of
the study include individual rather than group randomization and blinding
of the nevus counters to the status of the children. The pattern of association
between phenotypic factors and nevus counts at induction gives reasonable
assurance that the subjects are similar to those recruited for previous studies
of nevi.13,14,16
Subject retention was excellent, and parents and children exhibited a high
degree of compliance in providing data on solar UV exposure during the trial.
Parents in the sunscreen group provided high-quality information on the volume
of sunscreen used.
No information was collected on children who elected not to participate
in the trial, and it is possible that there may be some differences between
participants and nonparticipants. Another potential drawback to the study
is the relatively short period of follow-up; there are no clear data on the
duration of solar exposure needed to initiate nevus formation. Limited data
are available from the study by Harrison et al12
demonstrating the presence of new nevi within 1 year of birth in a cohort
of Australian children. Thus, the initiation period for new nevi was thought
to be short, and the 3-year follow-up was anticipated to be long enough to
see differences develop between the 2 intervention groups.
Another potential problem is the possibility that randomization to the
sunscreen group sensitized parents of these children. If this were the case,
the children's parents might have been attuned to the potential benefits of
sun avoidance and might have restricted solar exposure in the children. However,
reported hours spent outdoors during the 3 years of the study were similar
in the 2 groups. Finally, within the intervention group, there was an inverse
relationship between quantity of sunscreen used and number of new nevi, suggesting
that sunscreen use was the factor of consequence in the study.
A recent study in women has demonstrated a protective effect of chemical
sunscreen against CMM.25 Most previous studies,
however, have suggested either no association26,27
or a positive relationship28-30
between sunscreen use and melanoma risk. Furthermore, results from a recent
cross-sectional study of nevi in European children suggest that sunscreen
is not effective at preventing the appearance of new nevi.31
The authors suggested that this is because use of high-SPF sunscreen promotes
extended duration of sun exposure.32 An increase
in new nevi among children using sunscreen regularly was also seen in a German
study.33 Both of the European studies were
well conducted but assessed sun exposure and sunscreen use retrospectively,
in some cases 3 to 5 years afterward. In addition, it was not clear whether
the sunscreen used in the German study was broad-spectrum and attenuated UV-A
and UV-B or was primarily designed to screen out UV-B radiation. Finally,
as in any retrospective investigation of nevi, incomplete control of host-susceptibility
factors must be considered a potential explanation for these findings.
In our study, the reduced number of new nevi in the broad-spectrum sunscreen
group suggests that these agents may be useful in preventing transformation
of normal melanocytes into nevi, at least in children who freckle. As 1993
and 1996 counts were both conducted prior to the sunny months of the year,
it is unlikely that the effect seen in freckled children was due to differential
misclassification of freckles and nevi between the 1993 and 1996 counts. Both
nevi1-3,34
and freckling35-37
are known to increase risk of CMM, and evidence suggests that nevi and freckling
together have a synergistic effect on risk.3,6
This may indicate that subjects who freckle and develop nevi have an underlying
instability in their melanocytes. If so, these melanocytes might be more likely
to evolve into a clone that ultimately becomes a visible nevus under the influence
of solar UV radiation. Furthermore, formation of a nevus may take place with
a smaller degree of solar UV damage in children who freckle than in those
who do not freckle. If sunscreens protect the melanocytes, the importance
of such protection against the formation of new nevi would be more important
among those who freckle.
An alternative explanation for the greater importance of being in the
sunscreen group for those who freckle may be found in the limited duration
of the intervention. It is possible that the study saw a short-term result
only in those subjects most sensitive to the development of nevi. In a trial
of longer duration, a more clear-cut protective effect might be seen for all
subjects, regardless of freckling status.
In summary, our findings indicate that broad-spectrum sunscreens may
attenuate the development of nevi in children and perhaps ultimately reduce
their risk of developing melanoma.
1.Holman CD, Armstrong BK. Pigmentary traits, ethnic origin, benign nevi, and family history as
risk factors for cutaneous malignant melanoma.
J Natl Cancer Inst.1984;72:257-266.Google Scholar 2.Holly EA, Kelly JW, Shpall SN, Chiu SH. Number of melanocytic nevi as a major risk factor for malignant melanoma.
J Am Acad Dermatol.1987;17:459-468.Google Scholar 3.Marrett LD, King WD, Walter SD, From L. Use of host factors to identify people at high risk for cutaneous malignant
melanoma.
CMAJ.1992;147:445-453.
[published correction appears in CMAJ
1992;147:1764]Google Scholar 4.Grob JJ, Gouvernet J, Aymar D.
et al. Count of benign melanocytic nevi as a major indicator of risk for nonfamilial
nodular and superficial spreading melanoma.
Cancer.1990;66:387-395.Google Scholar 5.Green A, MacLennan R, Siskind V. Common acquired naevi and the risk of malignant melanoma.
Int J Cancer.1985;35:297-300.Google Scholar 6.Osterlind A, Tucker MA, Hou-Jensen K, Stone BJ, Engholm G, Jensen OM. The Danish case-control study of cutaneous malignant melanoma, I: importance
of host factors.
Int J Cancer.1988;42:200-206.Google Scholar 7.Skender-Kalnenas TM, English DR, Heenan PJ. Benign melanocytic lesions: risk markers or precursors of cutaneous
melanoma?
J Am Acad Dermatol.1995;33:1000-1007.Google Scholar 8.Kruger S, Garbe C, Buttner P, Stadler R, Guggenmoos-Holzmann I, Orfanos CE. Epidemiologic evidence for the role of melanocytic nevi as risk markers
and direct precursors of cutaneous malignant melanoma.
J Am Acad Dermatol.1992;26:920-926.Google Scholar 9.Weinstock MA, Colditz GA, Willett WC.
et al. Moles and site-specific risk of nonfamilial cutaneous malignant melanoma
in women.
J Natl Cancer Inst.1989;81:948-952.Google Scholar 10.Rivers JK, Frederiksen PC, Dibdin C. A prevalence survey of dermatoses in the Australian neonate.
J Am Acad Dermatol.1990;23:77-81.Google Scholar 11.Kroon S, Clemmensen OJ, Hastrup N. Incidence of congenital melanocytic nevi in newborn babies in Denmark.
J Am Acad Dermatol.1987;17:422-426.Google Scholar 12.Harrison SL, MacLennan R, Speare R, Wronski I. Sun exposure and melanocytic naevi in young Australian children.
Lancet.1994;344:1529-1532.Google Scholar 13.Gallagher RP, McLean DI, Yang CP.
et al. Suntan, sunburn, and pigmentation factors and the frequency of acquired
melanocytic nevi in children.
Arch Dermatol.1990;126:770-776.Google Scholar 14.Pope DJ, Sorahan T, Marsden JR, Ball PM, Grimley RP, Peck IM. Benign pigmented nevi in children.
Arch Dermatol.1992;128:1201-1206.Google Scholar 15.Coombs BD, Sharples KJ, Cooke KR, Skegg DC, Elwood JM. Variation and covariates of the number of benign nevi in adolescents.
Am J Epidemiol.1992;136:344-355.Google Scholar 16.English DR, Armstrong BK. Melanocytic nevi in children, I: anatomic sites and demographic and
host factors.
Am J Epidemiol.1994;139:390-401.Google Scholar 17.Newton-Bishop JA, Bataille V, Pinney E, Bishop DT. Family studies in melanoma: identification of the atypical mole syndrome
(AMS) phenotype.
Melanoma Res.1994;4:199-206.Google Scholar 18.MacKie RM, McHenry P, Hole D. Accelerated detection with prospective surveillance for cutaneous malignant
melanoma in high-risk groups.
Lancet.1993;341:1618-1620.Google Scholar 19.Kelly JW, Rivers JK, MacLennan R, Harrison S, Lewis AE, Tate BJ. Sunlight: a major factor associated with the development of melanocytic
nevi in Australian schoolchildren.
J Am Acad Dermatol.1994;30:40-48.Google Scholar 20.Gallagher RP, Rivers JK, Yang CP, McLean DI, Coldman AJ, Silver HK. Melanocytic nevus density in Asian, Indo-Pakistani, and white children:
The Vancouver Mole Study.
J Am Acad Dermatol.1991;25:507-512.Google Scholar 21.English DR, MacLennan R, Rivers JK, Kelly J, Armstrong BK. Epidemiological studies of melanocytic naevi: protocol for identifying
and recording naevi. IARC Internal Report No. 90/002. Lyon, France: International Agency
for Research on Cancer; 1990.
22.Mosteller RD. Simplified calculation of body surface area [letter].
N Engl J Med.1987;317:1098.Google Scholar 23.Diffey BL, Elwood JM. Tables of ambient solar ultraviolet radiation for use in epidemiological
studies of malignant melanoma. In: Gallagher RP, Elwood JM, eds. Epidemiological
Aspects of Cutaneous Malignant Melanoma. Boston, Mass: Kluwer Academic
Publishers; 1994:81-105.
24.Parkin DM, Whelan SL, Ferlay J, Raymond L, Young J. Cancer incidence in five continents. Vol 7. IARC Scientific Publication No. 143. Lyon, France: International
Agency for Research on Cancer; 1997.
25.Holly EA, Aston DA, Cress RD, Ahn DK, Kristiansen JJ. Cutaneous melanoma in women, I: exposure to sunlight, ability to tan,
and other risk factors related to ultraviolet light.
Am J Epidemiol.1995;141:923-933.Google Scholar 26.Holman CD, Armstrong BK, Heenan PJ. Relationship of cutaneous malignant melanoma to individual sunlight-exposure
habits.
J Natl Cancer Inst.1986;76:403-414.Google Scholar 27.Elwood JM, Gallagher RP. More about: sunscreen use, wearing clothes, and number of nevi in 6-
to 7-year-old European children [letter].
J Natl Cancer Inst.1999;91:1164-1166.Google Scholar 28.Autier P, Dore JF, Schifflers E.
et al. for the EORTC Melanoma Cooperative Group. Melanoma and use of sunscreens: an EORTC case-control study in Germany,
Belgium, and France.
Int J Cancer.1995;61:749-755.Google Scholar 29.Westerdahl J, Olsson H, Masback A, Ingvar C, Jonsson N. Is the use of sunscreens a risk factor for malignant melanoma?
Melanoma Res.1995;5:59-65.Google Scholar 30.Beitner H, Norell SE, Ringborg U, Wennersten G, Mattson B. Malignant melanoma: aetiological importance of individual pigmentation
and sun exposure.
Br J Dermatol.1990;122:43-51.Google Scholar 31.Autier P, Dore JF, Cattaruzza MS.
et al. for the European Organization for Research and Treatment of Cancer
Melanoma Cooperative Group. Sunscreen use, wearing clothes, and number of nevi in 6- to 7-year-old
European children.
J Natl Cancer Inst.1998;90:1873-1880.Google Scholar 32.Autier P, Dore JF, Negrier S.
et al. Sunscreen use and duration of sun exposure: a double-blind, randomized
trial.
J Natl Cancer Inst.1999;91:1304-1309.Google Scholar 33.Luther H, Altmeyer P, Garbe C.
et al. Increase of melanocytic nevus counts in children during 5 years of
follow-up and analysis of associated factors.
Arch Dermatol.1996;132:1473-1478.Google Scholar 34.White E, Kirkpatrick CS, Lee JA. Case-control study of malignant melanoma in Washington State, I: constitutional
factors and sun exposure.
Am J Epidemiol.1994;139:857-868.Google Scholar 35.Elwood JM, Gallagher RP, Hill GB, Spinelli JJ, Pearson JC, Threlfall W. Pigmentation and skin reaction to sun as risk factors for cutaneous
melanoma: Western Canada Melanoma Study.
BMJ (Clin Res Ed).1984;288:99-102.Google Scholar 36.Holly EA, Aston DA, Cress RD, Ahn DK, Kristiansen JJ. Cutaneous melanoma in women, II: phenotypic characteristics and other
host-related factors.
Am J Epidemiol.1995;141:934-942.Google Scholar 37.Chen YT, Dubrow R, Holford TR.
et al. Malignant melanoma risk factors by anatomic site: a case-control study
and polychotomous logistic regression analysis.
Int J Cancer.1996;67:636-643.Google Scholar