R.P.
Smith
et
al.
/
Maturitas
74 (2013) 208–
212
209
patterns
of
photic
input,
sunlight
exposure,
and
activity
levels
impact
circadian
and
hormonal
rhythms
[7]
.
Similarly,
if
con-
Frmed,
predictable
seasonal
variations
in
T
would
require
the
establishment
of
seasonal
reference
standards.
Currently,
studies
on
seasonal
T
variations
have
yielded
signiFcant
heterogeneity
with
no
recognized
consensus.
In
this
review,
the
current
status
of
the
literature
is
summarized
and
recommendations
made
as
to
the
impact
of
seasonal
variations
on
the
current
clinical
standards.
2.
Seasonal
variations
in
testosterone:
the
evidence
In
one
of
the
earliest
manuscripts
on
the
topic,
Reinberg
et
al.
[8]
demonstrated
circannual
variation
in
plasma
T
in
Fve
Parisian
males
with
peak
plasma
levels
and
sexual
activity
occur-
ring
in
the
month
of
October.
Shortly
thereafter,
Smals
et
al.
[9]
examined
a
group
of
15
healthy
male
subjects
and
found
sta-
tistically
signiFcant
seasonal
patterns
in
T
levels.
The
authors
recorded
peaks
in
the
summer
and
early
autumn
and
a
nadir
in
the
winter
and
early
spring
[9]
.
Since
that
time,
seasonal
vari-
ation
has
been
conFrmed
in
several
cross-sectional
studies
of
ethnically,
socially,
and
geographically
varied
populations
of
men
[10–16]
.
Others
however,
have
not
shown
similar
circannual
vari-
ations
[17–20]
.
Inconsistent
variations
have
also
been
identiFed
in
a
number
of
longitudinal
studies
on
seasonal
patterns
of
T
[9,15,21–23]
.
Svartberg
et
al.
[24]
cross-sectionally
examined
the
seasonal
variations
in
total
and
free
testosterone
(TT
and
±T),
luteinizing
hormone
(LH),
and
sex
hormone
binding
globulin
(SHBG)
lev-
els
amongst
1548
men
living
in
northern
Norway.
A
population
selected
based
on
its
exposure
to
a
wide
seasonal
variation
in
both
temperature
and
daylight.
Among
these
men,
TT
showed
bimodal
seasonal
variation
with
peak
values
in
October–November
and
a
nadir
in
June.
Similar
results
were
seen
in
patterns
of
±T
demon-
strating
peaks
in
December
and
a
nadir
in
August.
Notably,
the
lowest
T
levels
occurred
in
months
with
the
highest
tempera-
tures
and
longest
hours
of
daylight.
The
variations
in
hormone
levels
were
large,
with
a
19%
and
31%
difference
between
the
lowest
and
highest
monthly
mean
levels
of
TT
and
±T
respec-
tively.
The
authors
hypothesized
that
these
fluctuations
could
be
explained
by
the
variation
in
daylight
exposure
and
temperature
[24]
.
While
the
study
had
a
signiFcant
confounder
in
that
a
large
length
of
time
(i.e.
0800–1600)
was
deemed
suitable
for
the
lab-
oratory
T
collection,
a
substantial
seasonal
effect
on
T
was
still
highlighted
[24]
.
A
follow-up
cross-sectional
study
[7]
was
con-
ducted
on
men
exposed
to
less
extreme
seasonal
changes
in
sunlight
and
temperature
(San
Diego,
California)
to
determine
whether
seasonal
variations
of
T
persisted
in
this
population.
In
the
915
men
studied,
neither
TT
nor
bioavailable
T
(BT)
varied
by
season
[7]
.
These
results
were
independent
of
age
as
well
as
anthro-
pometric
measurements
and
no
association
with
air
temperature
or
duration
of
sunlight
exposure
was
documented
[7]
.
The
conflicting
results
obtained
in
the
study
set
in
California
and
the
original
Nor-
wegian
study
were
postulated
to
be
due
to
differences
in
climate
and
sleep
patterns
[7,24]
.
In
a
population
with
a
similar
photoperiod
and
climate
(Denmark),
Andersson
et
al.
[25]
,
obtained
monthly
blood
samples
on
27
men
during
a
17-month
period.
Measurements
of
inhibin
B,
follicle
stimulating
hormone
(±SH),
LH,
TT,
and
estradiol
(E2)
levels
were
recorded.
The
authors
found
seasonal
variation
in
LH
and
T
levels,
but
not
in
the
levels
of
other
sex
hormones.
The
sea-
sonal
variation
in
TT
paralleled
the
variation
seen
in
LH
with
peak
levels
in
the
summer
(June–July)
and
nadirs
in
the
winter
and
early
spring
[25]
.
This
data
reported
by
Andersson
et
al.
[25]
is
in
contrast
to
the
Svartberg
study
[24]
,
which
showed
peaks
in
the
fall
and
a
nadir
in
the
summer.
The
lack
of
concordant
seasonal
changes
in
hormone
patterns
in
these
two
geographically
simi-
lar
populations
calls
into
question
the
clinical
signiFcance
of
these
Fndings.
Whereas
the
aforementioned
studies
had
limited
study
pop-
ulations,
Moskovic
et
al.
examined
serum
TT,
E2,
SHBG,
±SH,
LH
and
dehydroepiandrosterone
(DHEA)
in
11,000
men
in
the
southwestern
United
States
[26]
.
Given
the
previous
Fnding
by
Crawford
et
al.
[4]
which
suggested
that
diurnal
variation
dimin-
ished
after
age
60,
men
in
the
Moskovic
study
were
divided
into
cohorts
of
less
than
60
years
of
age
and
those
greater
than
60
years
of
age.
This
division
was
to
control
for
potential
physiologic
differences
in
age-related
hormonal
regulation.
Moskovic
et
al.
[26]
subsequently
found
statistically
signiFcant
differences
in
E2,
testosterone:
estrogen
ratio
(T:E),
±SH,
and
SHBG
between
sea-
sons.
The
magnitude
of
these
differences
was
only
signiFcant
in
the
younger
cohort
of
patients,
although
the
younger
cohort
was
also
larger
(
N
=
9669
versus
1954).
Peak
to
trough
variations
were
16.5%
for
T:
E
ratio,
with
a
peak
in
the
spring
and
nadir
in
the
fall.
The
difference
between
T:E
ratio
in
the
two
cohorts
was
hypothe-
sized
to
be
due
in
part
to
increased
physical
activity
in
the
younger
cohort
and
evidence
to
suggest
that
the
HPG
axis
is
more
sensi-
tive
in
younger
men
[27]
.
Whereas
no
signiFcant
change
in
TT
was
observed,
a
10%
change
was
noted
in
±T
between
peak
(summer)
and
trough
(fall)
values;
however,
this
was
not
clinically
signiF-
cant.
Despite
its
sample
size,
the
Moskovic
study
[26]
still
failed
to
show
statistically
signiFcant
differences
in
TT,
LH,
±T
and
DHEA
in
either
cohort.
This
study
was
also
limited
in
that
samples
were
collected
between
0900
and
1900
and
was
based
on
initial,
sin-
gle
patient
observations
as
opposed
to
longitudinal
data
[26]
.
Along
with
the
prior
Svartberg
study
in
San
Diego
[7]
,
this
repre-
sented
the
second
investigation
in
the
southwestern
United
States
which
failed
to
show
signiFcant
seasonal
TT
variation.
This
sug-
gests
that
the
prior
Scandinavian
studies
were
demonstrating
a
regional
effect.
±urther
research
in
dissimilar
climates
and
popula-
tions
is
required
to
replicate
the
variability
in
these
prior
studies
and
substantiate
their
Fndings.
One
such
study
by
Brambilla
et
al.
[20]
was
based
in
Boston,
Mas-
sachusetts
and
therefore
contrasts
with
the
prior
Moskovic
study
[26]
in
the
potential
regional
effects
of
climate
and
epidemiology.
The
authors
examined
seasonal
fluctuations
in
androgens,
includ-
ing
serum
TT,
±T,
BT,
dihydrotestosterone
(DHT),
SHBG,
LH,
DHEA,
dehydroepiandrosterone
sulfate
(DHEA-S),
E2,
and
cortisol.
One
hundred
and
twenty-one
men,
aged
30–79,
completed
six
morning
blood
draws
at
0,
3,
and
6
months
[20]
.
Time
of
enrollment
was
ran-
dom
in
order
to
capture
data
from
all
twelve
months.
Aside
from
cortisol,
there
was
no
evidence
of
seasonal
variation
in
hormone
levels
[20]
.
Peak
levels
were
within
4%
of
the
mean
level
for
all
hormones
examined.
The
authors
found
that
intra-individual
vari-
ation
was
greater
for
each
hormone
evaluated
when
compared
with
seasonal
variation.
Brambilla
et
al.
[20]
concluded,
therefore,
that
seasonal
effects
are
likely
not
an
important
source
of
signiFcant
variation
clinically.
Whereas
regional
influences
may
account
for
some
of
the
observed
patterns
of
seasonal
variation,
age
within
the
study
popu-
lation
also
has
the
potential
to
confound
these
results.
±or
instance,
Moskovic
et
al.
[26]
found
more
signiFcant
seasonal
variation
in
T:E
ratio
in
younger
men
whereas
Svartberg
[7]
found
neither
TT
nor
BT
varied
by
season
and
these
results
were
independent
of
age.
Simi-
larly,
Tancredi
et
al.
[19]
,
in
a
study
of
5028
men
aged
50
and
over,
showed
that
monthly
variations
in
serum
±T
values
did
not
show
signiFcant
seasonal
variation
(<15%)
[19]
.
Dai
et
al.,
in
an
epidemi-
ologic
study,
demonstrated
that
age
and
obesity
correlated
with
T
levels
in
243
men;
however,
while
diurnal
variation
was
noted,
seasonal
variation
of
T
was
not
[17]
.