STUDY
Predictive Value of Café au Lait Macules
at Initial Consultation in the Diagnosis
of Neurofibromatosis Type 1
Kara S. Nunley, MD; Feng Gao, MD, PhD; Anne C. Albers, RN, MSN; Susan J. Bayliss, MD; David H. Gutmann, MD, PhD
Objective: To evaluate the predictive utility of the num-
Results: Thirty-four of the children met diagnostic cri-
ber and morphologic appearance of isolated café au lait
macules (CALMs) in establishing the diagnosis of neurofibromatosis type 1 (NF1) in a cohort of children referred to an NF1 subspecialty clinic.
teria for NF1 during the study period. Thirty-two children met criteria prior to age 72 months, and 2 children
met criteria after 72 months. The mean number of CALMs
at presentation in children eventually diagnosed as having NF1 (11.8 CALMs) was significantly higher than the
mean number of CALMs in children not diagnosed as having NF1 (4.6 CALMs). Of the 44 children who had 6 or
more typical CALMs at presentation, 34 children met criteria for NF1. Sixty-eight patients had CALMs described as “typical,” while 42 patients had “atypical”
CALMs. Only 2 patients with atypical CALMs met criteria for NF1.
Design: Retrospective study of patients seen between the
years 2004 and 2007.
Setting: Tertiary care neurofibromatosis referral clinic
at St Louis Children’s Hospital.
Patients: The study population comprised 110 pa-
tients who presented with CALMs and no other diagnostic features of NF1. The median number of CALMs
at initial presentation was 6, while the median age of the
patients was 33 months. The median age at the last follow-up examination was 76.5 months.
Main Outcome Measures: Number and morphologic appearance of CALMs and diagnosis of NF1.
N
Conclusion: The majority of patients with 6 or more
CALMs will eventually meet diagnostic criteria for NF1,
typically by age 6 years, and this likelihood increases with
increasing number and typical morphologic appearance
of CALMs.
Arch Dermatol. 2009;145(8):883-887
EUROFIBROMATOSIS TYPE 1
(NF1) is a common autosomal disorder that affects numerous organ
systems, including the
skin, eyes, bones, blood vessels, and central and peripheral nervous systems.1 The
For editorial comment
see page 929
Author Affiliations:
Departments of Dermatology
(Drs Nunley and Bayliss),
Biostatistics (Dr Gao),
Neurology (Ms Albers and
Dr Gutmann), and Pediatrics
(Dr Bayliss), Washington
University School of Medicine,
St Louis, Missouri.
diagnosis of NF1 is established when
individuals present with 2 or more of the
following features: 6 or more café au lait
macules (CALMs); 2 or more neurofibromas of any type or 1 plexiform neurofibroma; freckling in the axillary or inguinal region; optic glioma; 2 or more Lisch
nodules; a distinctive osseous lesion; and
a first-degree relative with NF1.2 However, infants and young children most frequently exhibit CALMs only1,3-5 and cannot be given the diagnosis of NF1 until a
second feature develops. Because the second disease-defining manifestation of NF1,
(REPRINTED) ARCH DERMATOL/ VOL 145 (NO. 8), AUG 2009
883
such as axillary freckling and neurofibromas, may not develop until later in childhood, it is not uncommon for children with
multiple CALMs and no family history of
NF1 to be followed up for several years before the diagnosis of NF1 is confirmed or
excluded. This presents a particular challenge for both the physician and the family and may limit adequate screening for
serious problems seen in this young age
group (eg, optic pathway glioma).
There is relatively little information available regarding the natural history and diagnostic outcomes of children presenting
only with multiple CALMs. Previous studies have shown that solitary CALMs are
common and may occur in as many as 3%
of infants and 25% of healthy children.6,7
However, the presence of more than 3
CALMs is detected in only 0.2% to 0.3% of
children with no known evidence of an underlying disorder.6,8 Neurofibromatosis 1 is
the most common condition associated with
multiple CALMs. Multiple CALMs can also
be seen in McCune-Albright syndrome, ring
chromosome syndromes, Watson syn-
WWW.ARCHDERMATOL.COM
©2009 American Medical Association. All rights reserved.
Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/derm/5218/ on 06/18/2017
A
B
Figure 1. Examples of a “typical” café au lait macule (CALM) with homogeneous pigmentation and regular borders (A) and an “atypical” CALM with irregular
borders (B).
drome, and Bloom syndrome, but these conditions are far
less prevalent than NF1.9 A syndrome with phenotypic features similar to NF1, but caused by mutations in SPRED1,
as well as an inherited form of multiple CALMs without
other features of NF1, has also been described in the literature.10-13 To our knowledge, the largest study to date examined 41 children with 6 or more CALMs.14 Of these children, 80% were eventually diagnosed as having NF1 by the
end of the study period. While this study provided important insights, it did not include children who had fewer than
6 CALMs at initial presentation and did not focus on the
diagnostic utility of morphologic appearance or number
of CALMs. Moreover, this study did not address the likelihood of rendering the diagnosis of NF1 as either a function of age or number of CALMs. In an effort to address
these important clinical issues, we conducted a retrospective analysis of a large group of children who initially visited a tertiary referral neurofibromatosis subspecialty clinic
between the years 2004 and 2007 with only CALMs and
no other signs of NF1.
METHODS
PATIENTS
This study was performed using an approved Human Studies
Committee Protocol at Washington University School of Medicine, St Louis, Missouri. We reviewed the medical records of
191 patients with an International Classification of Diseases, Ninth
Revision (ICD-9) code of 709.09 (nonspecific dyschromia) who
were seen by a single physician (D.H.G.) in the neurofibromatosis clinic at a large tertiary care pediatric hospital (St Louis
Children’s Hospital) between the years 2004 and 2007. This
ICD-9 code was the code regularly used to designate patients
exhibiting CALMs who did not carry a known diagnosis of NF1.
The diagnosis of NF1 was established using criteria outlined
by the National Institutes of Health Consensus Development
Conference.2 All patients had a complete evaluation including
history and physical and neurologic examination performed by
a single physician (D.H.G.) at the time of initial presentation
to the NF1 clinic. In general, patients were followed up on a
yearly basis. Patients were referred for an annual ophthalmo-
logic evaluation with a slit lamp examination as a part of routine care, beginning at 1 year of age. Patients were excluded
from this study if they met diagnostic criteria for NF1 at their
first visit to the clinic, had a first-degree relative with NF1, had
physical examination findings consistent with segmental NF1,
were incorrectly coded as having CALMs, or had incomplete
or unavailable medical records. Data were collected from the
time of initial presentation to our clinic through the time of
last follow-up examination. Demographic information and clinical characteristics including description of CALMs, date of diagnosis of NF1 (if applicable), and the chronology of developing additional features of NF1 were recorded for each patient.
Similar to previous studies, “typical” CALMs were defined
as lesions with uniform pigmentation and distinct, regular borders, while “atypical” or “irregular” CALMs were defined as lesions with irregular, smudgy borders or nonhomogeneous pigmentation (Figure 1).15 Patients were classified as having
atypical CALMs if their lesions were described as atypical or
irregular at the first clinic visit.
STATISTICAL ANALYSES
Distributions of the demographic and clinical characteristics
across the 3 groups of patients (patients diagnosed as having
NF1 during the period of the study, patients with typical CALMs
who did not meet diagnostic criteria for NF1 at the conclusion
of the study, and patients with atypical CALMs who did not
meet diagnostic criteria for NF1 at the conclusion of the study)
were summarized using means, medians, or frequencies, as appropriate. For categorical features such as sex or race, the differences among groups were compared using the Fisher exact
test. For continuous features such as number of CALMs or age
at presentation, the comparison was made using KruskallWallis rank-sum test. P⬍.05 was considered statistically significant, and all statistical tests were 2-sided.
RESULTS
A total of 191 patients were seen with an ICD-9 code of
709.09 in the NF clinic over the period of this study. Of
these patients, 110 met the inclusion criteria for our study.
There were 63 male and 47 female children in our patient
population (Table). Twenty-four of the children were Afri-
(REPRINTED) ARCH DERMATOL/ VOL 145 (NO. 8), AUG 2009
884
WWW.ARCHDERMATOL.COM
©2009 American Medical Association. All rights reserved.
Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/derm/5218/ on 06/18/2017
Table. Demographic Features of Patient Groups
Demographic
Sex, M/F
Race, No.
African American
Asian
White
Unknown
Age at presentation, mean (range), mo
Age at diagnosis, mean (range), mo
Age at last follow-up examination, mean (range), mo
No. of CALMs at presentation, mean (range)
Age at CALM onset, mean (range), mo
Diagnosed
as Having NF1
(n = 34)
Undiagnosed,
With Regular CALMs
(n = 36)
Undiagnosed,
With Irregular CALMs
(n = 40)
19/15
23/13
21/19
7
1
26
0
13.5 (1 to 50)
33.5 (12 to 96)
71.7 (19 to 144)
11.8 (6 to ⱖ20)
4.2 (0 to 48) a
10
0
26
0
73.8 (4 to 192)
NA
82.4 (4 to 192)
4.5 (1 to 8)
6.8 (0 to 120) b
7
0
31
2
74.0 (7 to 206)
NA
87.3 (16 to 206)
4.7 (1 to 12)
15.3 (0 to 144) c
Abbreviations: CALMs, café au lait macules; NA, not applicable; NF1, neurofibromatosis type 1.
a Data available for 32 patients.
b Data available for 23 patients.
c Data available for 19 patients.
can American, 63 were white, 1 was Asian, and 2 did not
have racial information available in the medical record. The
median age at initial presentation to our clinic was 33
months (range, 1-206 months). The median age at last follow-up was 76.5 months (range, 4-206 months). No children were diagnosed as having syndromes other than NF1.
No undiagnosed children with typical CALMs had a family history of CALMs. One undiagnosed child with irregular CALMs had a parent with CALMs.
The median number of CALMs at presentation of all
patients in the study was 6 (range, 1 to ⱖ20). Thirtyfour (31%) of the children met criteria for diagnosis of
NF1 during the study period. The mean number of
CALMs at initial presentation in children eventually diagnosed as having NF1 (11.8 CALMs; range, 6 to ⱖ20)
was significantly higher than the mean number of CALMs
in children who were not diagnosed as having NF1 (4.6
CALMs; range, 1 to 12) during the study (P ⬍.001).
The percentage of children diagnosed as having NF1
increased with increasing numbers of CALMs at initial
presentation (Figure 2A). Of the 26 children who had
1 to 5 typical CALMs at initial presentation, none met
criteria for a diagnosis of NF1. Of the 44 children who
had 6 or more typical CALMs at presentation, 34 (77%)
met criteria for a diagnosis of NF1 by the end of the study
period.
Sixty-eight patients had CALMs described as typical,
while 42 patients had atypical CALMs. Of the 42 patients with atypical CALMs, 27 had 1 to 5 CALMs and
15 had 6 or more CALMs. Thirty-two patients with typical CALMs met criteria for a diagnosis of NF1 during the
study period. Two patients with atypical CALMs, both
of whom had more than 6 CALMs at initial presentation, met criteria for a diagnosis of NF1 during the study
period.
The mean age at diagnosis of NF1 was 33.5 months
(range, 12-96 months). Twenty-six (76%) met criteria for
diagnosis prior to age 48 months, 32 (94%) met criteria for
diagnosis prior to age 72 months, and 2 (6%) met criteria
for diagnosis after age 72 months (Figure 2B). Of the children with typical CALMs who remained undiagnosed at
the conclusion of the study period, 62 (89%) were older
than 48 months at the last follow-up examination and 57
(81%) were older than 72 months at their last follow-up
examination. Of children with atypical CALMs who remained undiagnosed at the conclusion of the study period, 33 (79%) were older than 48 months at their last follow-up examination and 25 (60%) were older than 72
months at their last follow-up examination.
There were no significant differences in sex (P=.61),
race (P=.54), or mean age at the last follow-up examination (P=.31), but age at presentation for children in
the diagnosed group was younger than for those in the
undiagnosed group (P⬍ .001).
In children eventually diagnosed as having NF1, the
most common second feature used to establish the diagnosis of NF1 was axillary or inguinal freckling, occurring in 26 (77%) patients (Figure 2C). Other signs to occur as the second (confirmatory) diagnostic feature
included the following: Lisch nodules in 6 children, plexiform neurofibroma in 2 children, and tibial pseudarthrosis in 1 child. Following a diagnosis of NF1, children in
this clinic population developed the following additional features of NF1 during the course of our study:
33 patients developed axillary or inguinal freckling, 15
developed Lisch nodules, 8 developed neurofibromas, 4
developed skeletal anomalies, 3 developed macrocephaly, 2 developed optic gliomas, 4 developed plexiform neurofibromas, 1 developed precocious puberty, and
none developed malignant peripheral nerve sheath tumors or pheochromocytomas; 6 of 8 patients who underwent neuroimaging were noted to have T2 hyperintensities on magnetic resonance imaging.
COMMENT
Because of the association of CALMs with NF1, current
guidelines support evaluation of children with multiple
CALMs for this common genetic disorder.9,16,17 In some
instances, the presence of other associated clinical findings or a positive family history of NF1 will allow a di-
(REPRINTED) ARCH DERMATOL/ VOL 145 (NO. 8), AUG 2009
885
WWW.ARCHDERMATOL.COM
©2009 American Medical Association. All rights reserved.
Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/derm/5218/ on 06/18/2017
A
100
Patients Diagnosed as Having NF1, %
90
80
70
60
50
40
30
20
10
0
1
2
3
4
5
6
7
8
9
≥ 10
No. of CALMs at Initial Presentation
B
100
Patients Diagnosed as Having NF1, %
90
80
70
60
50
40
30
20
10
0
0
6
12 18 24 30 36 42 48 54 60 66 72 78 84 90 96
Age, mo
C
80
70
60
Patients, %
50
40
30
20
10
0
Axillary or Inguinal
Freckling
Lisch
Nodules
Pseudarthrosis
Plexiform
Neurofibroma
Figure 2. Likelihood of being diagnosed as having neurofibromatosis type 1
(NF1) based on the number of café au lait macules (CALMs), patient age, and
secondary features. A, The relationship between the number of CALMs at
initial presentation and the likelihood of establishing a diagnosis of NF1 is
shown. Children with irregular CALMs were excluded. B, The percentage of
children eventually diagnosed as having NF1 during the study period is
shown for each age. C, The frequency of developing a second clinical feature
in children presenting with CALMs is shown.
agnosis of NF1 to be made at the initial evaluation.18 However, a notable number of children will come to medical
attention with only CALMs and no other diagnostic features of NF1. At present, there are scant data to estimate
the likelihood that a child who exhibits only CALMs will
develop NF1. To better characterize the clinical course
of these patients, we performed a retrospective analysis
of children with CALMs and no other diagnostic features of NF1 who were seen in our NF clinic over a period of 3 years.
We identified 110 children who were referred to our
clinic for evaluation of NF1 based on the presence of
CALMs. To our knowledge, this is the largest series to
date of children presenting with CALMs alone. However, this was a retrospective study and was limited to
data collected at the time of the office visits. In this regard, 13 patients with typical CALMs and 17 patients with
atypical CALMs who were younger than 6 years at the
last follow-up examination remained undiagnosed. Some
of these patients may eventually develop other features
and meet diagnostic criteria for NF1 at a later date. Thus,
our data may underestimate the number of patients with
multiple CALMs alone who eventually meet criteria for
NF1.
The median age at which children first visited our NF
clinic was 33 months, with a median number of 6 CALMs.
Several experts recommend referral or further evaluation of children who manifest 3 or more CALMs.9,16 In
our cohort, the majority of referring physicians appeared to follow this guideline, as 6% of children referred had fewer than three CALMs.
Thirty-four patients (31%) were eventually diagnosed as having NF1 during the follow-up period. This
is consistent with a previous large study of all children
evaluated at an NF clinic (including children with fewer
than 6 CALMs), which found that 67% of patients had
only 1 diagnostic feature of NF1, most commonly CALMs,
and did not develop other features.5 In our series, patients with a higher number of CALMs at presentation
were more likely to develop NF1: 77% of patients who
had 6 or more typical CALMs at initial presentation were
eventually diagnosed as having NF1. This is consistent
with the one earlier study, in which 75% of 41 patients
with 6 or more CALMs were diagnosed as having NF1.14
These findings suggest that most children with NF1 who
exhibit 6 or more CALMs at initial presentation will be
diagnosed as having NF1. However, it is important for
pediatricians to recognize that there will be a small number of children who present with only 2 to 3 CALMs at
an early age and warrant monitoring for the development of additional CALMs or other features of NF1 over
time.4,5,9 No children were diagnosed as having syndromes other than NF1. One undiagnosed child with an
irregular CALM had a parent with a CALM. It is possible that this represented a case of multiple familial
CALMs without other features of NF1; however, there
was no information available about other family members and no genetic analysis was performed.
In children eventually diagnosed as having NF1 after
an initial presentation with only CALMs, 76% met NF1
criteria by age 4 years, 94% met NF1 criteria by age 6
years, and all patients met NF1 criteria by age 8 years.
The 2 children in our series who were diagnosed after
age 72 months were both lost to follow-up for a period
prior to diagnosis, and it is possible that a diagnosis would
have been established prior to age 72 months had they
(REPRINTED) ARCH DERMATOL/ VOL 145 (NO. 8), AUG 2009
886
WWW.ARCHDERMATOL.COM
©2009 American Medical Association. All rights reserved.
Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/derm/5218/ on 06/18/2017
been seen earlier. This is similar to that previously reported by Korf,14 who found that 75% of patients diagnosed as having NF1 met criteria prior to age 6 years and
92% met criteria prior to age 10 years. In our series, the
most common second feature of NF1 to appear was axillary or inguinal freckling, similar to that reported by
Korf.14 Collectively, these data suggest that the majority
of children who present with CALMs alone will meet diagnostic criteria by age 6 to 8 years and justifies following up these children as though they have NF1 for most
of the first decade of life.
Active NF referral centers are often asked to evaluate
children with unusually shaped CALMs. In our series,
the morphologic appearance of the CALMs was an important indicator of diagnostic outcome. Typical CALMs
in children with NF1 are circular or ovoid macules with
uniform pigment and well-defined borders.9,16 We found
that patients with atypical CALMs were less likely to be
diagnosed as having NF1. Although 47% of patients with
typical CALMs were eventually diagnosed as having NF1,
only 2 patients (5%) with atypical CALMs developed NF1.
While several experts have stated that the morphologic
appearance of CALMs is not predictive of diagnostic outcome, a small study found that 8 of 14 patients with typical CALMs were diagnosed as having NF1, while only 1
of 5 patients with atypical CALMs were diagnosed as having NF1.15 The use of the morphologic appearance of
CALMs as a prognostic indicator has limitations, notably the difficulty of assessing this parameter with a high
degree of reliability; however, in the absence of definitive histologic or other more objective criteria that differentiate NF1-associated from non–NF1-associated
CALMs, this information may provide helpful guidance
for clinicians caring for this group of patients.
Overall, our results from this large series of patients who
present with CALMs confirm the findings from previous
smaller studies that the majority of children with 6 or more
CALMs will eventually meet criteria for NF1, even if they
have no other signs of the disorder at initial presentation.
Given the high percentage of children with 6 or more
CALMs who will eventually be diagnosed as having NF1,
we advocate referring all such children to specialists experienced in the care of children with NF1. We also recommend that these patients be followed up throughout
childhood as though they carry the diagnosis of NF1, even
in the absence of other signs of the disorder. Due to the
infrequency of 3 or more CALMs in the general population without an associated systemic disorder, we continue
to support the current recommendations that these children be referred to a specialist for further evaluation. Our
data further indicate that children with fewer than 6 CALMs
or CALMs with an atypical appearance rarely become diagnosed as having NF1. With few exceptions, the diagnosis of NF1 is established by age 72 months using the established National Institutes of Health clinical criteria.6 In
addition to number of CALMs, the morphologic appearance of CALMs may also be a useful predictive feature in
the diagnosis of NF1. In some cases, genetic testing for NF1
may be useful in providing a definitive diagnosis; however, this analysis remains cost-prohibitive for many patients.19 In conclusion, our findings regarding the number
and morphologic appearance of CALMs should help cli-
nicians provide more accurate counseling to parents with
children harboring multiple CALMs in the absence of other
features of NF1.
Accepted for Publication: March 11, 2009.
Correspondence: David H. Gutmann, MD, PhD, Department of Neurology, Washington University School of
Medicine, Box 8111, 660 S Euclid Ave, St Louis, MO
63110 ([email protected]).
Author Contributions: All of the authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the
data analysis. Study concept and design: Nunley, Bayliss,
and Gutmann. Acquisition of data: Nunley and Albers.
Analysis and interpretation of data: Nunley, Gao, Bayliss,
and Gutmann. Drafting of the manuscript: Nunley and Gutmann. Critical revision of the manuscript for important intellectual content: Gao, Albers, Bayliss, and Gutmann. Statistical analysis: Gao. Administrative, technical, and material
support: Albers. Study supervision: Bayliss and Gutmann.
Financial Disclosure: None reported.
REFERENCES
1. Ward BA, Gutmann DH. Neurofibromatosis 1: from lab bench to clinic. Pediatr
Neurol. 2005;32(4):221-228.
2. National Institutes of Health Consensus Development Conference Statement: neurofibromatosis: Bethesda, Md, USA, July 13-15, 1987. Neurofibromatosis. 1988;
1(3):172-178.
3. Tonsgard JH. Clinical manifestations and management of neurofibromatosis type
1. Semin Pediatr Neurol. 2006;13(1):2-7.
4. North K. Neurofibromatosis type 1: review of the first 200 patients in an Australian clinic. J Child Neurol. 1993;8(4):395-402.
5. Boulanger JM, Larbrisseau A. Neurofibromatosis type 1 in a pediatric population: Ste-Justine’s experience. Can J Neurol Sci. 2005;32(2):225-231.
6. Whitehouse D. Diagnostic value of the cafe-au-lait spot in children. Arch Dis Child.
1966;41(217):316-319.
7. Alper J, Holmes LB, Mihm MC Jr. Birthmarks with serious medical significance:
nevocullular nevi, sebaceous nevi, and multiple cafe au lait spots. J Pediatr. 1979;
95(5, pt 1):696-700.
8. Burwell RG, James NJ, Johnston DI. Cafe-au-lait spots in schoolchildren. Arch
Dis Child. 1982;57(8):631-632.
9. Landau M, Krafchik BR. The diagnostic value of cafe-au-lait macules. J Am Acad
Dermatol. 1999;40(6, pt 1):877-892.
10. Arnsmeier SL, Riccardi VM, Paller AS. Familial multiple cafe au lait spots. Arch
Dermatol. 1994;130(11):1425-1426.
11. Brems H, Chmara M, Sahbatou M, et al. Germline loss-of-function mutations in
SPRED1 cause a neurofibromatosis 1-like phenotype. Nat Genet. 2007;39(9):
1120-1126.
12. Charrow J, Listernick R, Ward K. Autosomal dominant multiple cafe-au-lait spots
and neurofibromatosis-1: evidence of non-linkage. Am J Med Genet. 1993;
45(5):606-608.
13. Hoo JJ, Shrimpton AE. Familial hyper- and hypopigmentation with age-related
pattern change. Am J Med Genet A. 2005;132A(2):215-218.
14. Korf BR. Diagnostic outcome in children with multiple cafe au lait spots. Pediatrics.
1992;90(6):924-927.
15. Fois A, Calistri L, Balestri P, et al. Relationship between cafe-au-lait spots as the
only symptom and peripheral neurofibromatosis (NF1): a follow-up study. Eur J
Pediatr. 1993;152(6):500-504.
16. Tekin M, Bodurtha JN, Riccardi VM. Cafe au lait spots: the pediatrician’s perspective.
Pediatr Rev. 2001;22(3):82-90.
17. Hersh JH; American Academy of Pediatrics Committee on Genetics. Health supervision for children with neurofibromatosis. Pediatrics. 2008;121(3):633-642.
18. Obringer AC, Meadows AT, Zackai EH. The diagnosis of neurofibromatosis-1 in
the child under the age of 6 years. Am J Dis Child. 1989;143(6):717-719.
19. Messiaen LM, Callens T, Mortier G, et al. Exhaustive mutation analysis of the
NF1 gene allows identification of 95% of mutations and reveals a high frequency of unusual splicing defects. Hum Mutat. 2000;15(6):541-555.
(REPRINTED) ARCH DERMATOL/ VOL 145 (NO. 8), AUG 2009
887
WWW.ARCHDERMATOL.COM
©2009 American Medical Association. All rights reserved.
Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/derm/5218/ on 06/18/2017