also shown that boys have a greater perineal length
than girls.
Investigators have also used AGD
to show that agents which disrupt androgen signal-
ing in animal models can lead to abnormal genital
length and even altered testicular function as mea-
sured by testosterone and sperm production.
In humans 2 recent studies have correlated AGD
in men to sperm production. A study of healthy male
volunteers demonstrated a positive relationship be-
tween anogenital distance and semen concentration,
motility and morphology.
Another study showed
that fertile men had greater anogenital length com-
pared to infertile men.
In addition, a similar pos-
itive association between anogenital length and
sperm count was identified. Assuming that AGD is
determined in utero, such studies suggest in utero
influences may impact genital development and
adult testicular function. To date, to our knowledge
no correlation of genital measures to hormone pro-
duction exists. As testicular and penile development
and function are related, we determined if human
androgen production is related to anogenital length.
Study Population
The methods of cohort assembly have been previously
After obtaining institutional review board ap-
proval from Baylor College of Medicine, eligible patients
were recruited from a urology clinic specializing in repro-
ductive medicine from August 2010 through November
2010. Men with a history of orchiectomy, testicular torsion
or prior malignancy were excluded from study. A total of
116 men had serum hormone and genital measurements
available for analysis, including 89 evaluated for primary
infertility, 16 for secondary infertility, 8 for sexual dys-
function/hypogonadism and 3 for vasectomy. Mean age
SD was 36.1
8.0 years. Of the cohort 58.6% was white,
13.8% Hispanic and 13.8% black. All men provided writ-
ten consent for participation.
Genital Measurements
The methods of genital measurement have been described
In the supine, frog-leg position with the legs
abducted, allowing the soles of the feet to meet, the dis-
tance from the posterior aspect of the scrotum to the anal
verge was measured using a digital caliper (Neiko USA,
Model No. 01407A) (
fig. 1
). The stretched penile length was
measured from the base of the dorsal surface of the penis to
the tip of the glans. When comparing measurements among
investigators the within subject standard deviation was 4.1
mm for anogenital distance and 5.4 mm for stretched PL.
The correlation coefficient was 0.91 for AGD and PL mea-
surements. It is important to note that other investigators
have defined anogenital distance from the anus to the ante-
rior base of the penis and the distance from the posterior
scrotum to the anus (as was measured in this study) as the
anoscrotal distance.
Given the age of the patients mea-
sured, the posterior scrotum was measured as the anterior
border as it was considered a more comfortable, reliable and
reproducible measure. Testicular volume was estimated
manually during the physical examination by 1 investigator
(LIL) at approximately 25 to 27C.
Hormone Analysis
All hormone assays were processed by a single, experi-
enced laboratory (Laboratory for Male Reproductive Re-
search and Testing, Baylor College of Medicine, Houston,
Texas). Testosterone (normal range 200 to 1,000 ng/dl), LH
(normal range 6 to 19 mIU/ml), FSH (normal range 4 to 10
mIU/ml) and estradiol (0.5 to 5 ng/dl) values were assessed
using an automated, 1-step competitive binding assay with
the Beckman Coulter Access® II Immunoassay system. The
assays were recalibrated daily with controls that spanned
the normal range for all hormones.
Statistical Analysis
ANOVA was used to compare means between groups. In
addition, the Wilcoxon rank sum test was also used given
the nonparametric distribution of the data with no differ-
ence in the overall interpretation or conclusions. Linear
regression models and correlation coefficients were used
to determine the relationship between genital measures
and hormone values. Given the nonparametric distribu-
tion of the genital measures (ie AGD and PL), linear
regression models were also run with log
variables with no differences in the overall conclusions.
Linear regression coefficients between genital measures,
hormone values and anthropomorphic variables were de-
termined, and relationships with p
0.2 were included in the
multivariable models. All p values were 2-sided and anal-
yses were performed using Stata® 10.
Anthropomorphic, hormonal and genital measure-
ments are listed in
table 1
. When stratifying by race
Figure 1.
Anogenital distance as measured with men in supine,
frog-leg position.