Pe d i a t r i c I m a g i n g • O r i g i n a l R e s e a r c h

Kwong et al.
MRI of Proximal Humerus in Children

Pediatric Imaging
Original Research

Skeletal Development of the

Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved

Proximal Humerus in the Pediatric

Population: MRI Features
Steven Kwong1 OBJECTIVE. The purpose of this study was to provide a guideline of normal MRI devel-
Shefali Kothary 2 opmental anatomy of the proximal humerus in a growing child.
Leonardo Lobo Poncinelli 3 MATERIALS AND METHODS. Retrospective interpretation of 83 consecutive MRI
studies of shoulders in children 2 months to 17 years old was performed in consensus by two
Kwong S, Kothary S, Poncinelli LL radiologists. The following variables were documented: presence, number, and fusion of sec-
ondary ossification centers; appearance and closure of the growth plate; presence of the me-
taphyseal stripe; and proximal metaphyseal marrow signal intensity.
RESULTS. Preossification centers were seen in 2- and 4-month-old patients. Secondary
ossification centers appeared in the medial humeral head and greater tuberosity at 4 and 10
months, respectively, originally depicting red marrow and later converting to yellow marrow.
A separate lesser tuberosity ossification center was not seen. The ossification centers began
fusing by 3 years and gradually conformed to the final shape of the proximal humerus. Ossifi-
cation was completed by 13 years. The multilaminar growth plate initially had a flat and smooth
contour that progressively became irregular and pyramidal, closing at 17 years. The metaphy-
seal stripe was noted at infancy and disappeared by 15.5 years. The metaphyseal marrow signal
intensity was diffusely low in infants but, with conversion to yellow marrow, showed proximal
metaphyseal bright patchy or linear signal-intensity, eventually disappearing by 17 years.
CONCLUSION. The postnatal skeletal development of the proximal humerus in the pe-
diatric population follows distinctive sequential patterns of maturation, which can serve as a
guideline for interpreting MRI studies in children.

Keywords: development, growth plate, MRI, ossification
center, pitfalls, proximal humerus
DOI:10.2214/AJR.13.10711
Received February 2, 2013; accepted after revision
May 15, 2013.
1 growing use of the modality for detecting in-
School of Medicine, NYU Langone Medical Center, 550
1st Ave, New York, NY 10016. Address correspondence to
S. Kwong (steven.kwong@med.nyu.edu).
2
Department of Radiology, Beth Israel Medical Center,
New York, NY.
3
Department of Radiology, NYU Hospital for Joint
Disease, New York, NY.
AJR 2014; 202:418–425
0361–803X/14/2022–418
© American Roentgen Ray Society
O
ver the past few decades, there
has been a tremendous increase
in the use of MRI of the shoulder
in the pediatric patient popula-
tion. MRI of the shoulder in the infant and
young child is particularly useful for assess-
ing brachial plexus injuries, infections, and
tumors. In older children and adolescents, in-
creased athletic activities and increased inci-
dence of sports-related injuries have led to
traarticular shoulder derangements, such as
labral injuries and osseous abnormalities as-
sociated with shoulder dislocation.
The normal postnatal development of the
pediatric shoulder has been elucidated on the
basis of cadaveric and radiographic studies
[1, 2]. MRI, however, has a unique advantage
compared with other modalities because it
enables clear visualization of the nonossified
cartilaginous epiphysis, the secondary ossifi-
cation centers, and the physis. Studies using
MRI have examined the general patterns of
skeletal maturity [3] and the sequential de-
velopment of the elbow and ankle joints [4,
5]. However, aside from a few articles that
briefly describe the developing shoulder
[6–8], there has not been, to the best of our
knowledge, a comprehensive and detailed
MRI study of the sequential skeletal matu-
ration of the proximal humerus in the pe-
diatric age group. Familiarity is paramount
for accurate interpretation of shoulder MRI
studies and for differentiating normal skele-
tal growth from disease. The purpose of our
study was to provide a detailed guideline of
the normal MRI developmental anatomy of
the proximal humerus in the growing child
from infancy through the adolescent years to
skeletal maturity.
Materials and Methods
Our HIPAA-compliant study was approved by
the institutional review board of New York Uni-
versity Langone Medical Center. The requirement
for informed patient consent was waived. A ret-

418 AJR:202, February 2014


rospective search for all MRI examinations of the
shoulder or upper extremity including the shoulder
performed at our institution between January 2003
and May 2012 in patients under 18 years old was
conducted using our digital databank. Exclusion
criteria included inadequate imaging studies, gross-
ly distorted anatomy, or abnormality in the proximal
Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved
humerus signified by abnormal signal intensity.
Two radiologists (with 25 and 7 years of experi-
ence in musculoskeletal radiology) retrospectively
reviewed the MRI studies in consensus. The fol-
lowing variables were reviewed: presence, num-
ber, and fusion of secondary ossification centers;
appearance and closure of the growth plate; pres-
ence of metaphyseal stripe; and presence of proxi-
mal metaphyseal marrow signal intensity. Other
imaging modalities, including radiography and
CT, were also reviewed when available.
MRI was performed with either 1.5 T (Mag-
neton Aera and Avanto, Siemens Healthcare) or
3 T (Magneton Skyra and Magneton Verio, Sie-
mens Healthcare). There was some variability in
the MRI technique depending on patient age and
indication for the study. A head coil was used in
patients 2–6 months old with the shoulders in the
center of the coil, a head and neck coil was used
in patients 12 months to 5 years old, and a shoul-
der surface coil was used in the older patient pop-
ulation. The following sequences were performed
in most patients: coronal proton density-weight-
ed (TR range/TE range, 2100–2300/30–35; FOV,
12–14 cm; section thickness, 2–3 mm; section
gap, 1 mm; matrix, 256–320 × 240) or coronal
proton density-weighted with fat saturation (TR
range/TE, 2000–2100/36; section thickness, 3
mm; matrix, 192 × 115–125), coronal T2-weight-
ed fast spin echo with fat saturation (TR range/
TE range, 2500–4800/54–58; echo train length,
7; FOV, 12–14 cm; section thickness, 3 mm; sec-
tion gap, 1 mm; matrix, 320–512 × 230–392), ax-
ial proton density-weighted (TR range/TE range,
2420–3890/31–39; section thickness, 3 mm; ma-
trix, 256–512 × 187–208) or axial proton density-
weighted with fat saturation (TR range/TE range,
2700–2900/35–41; echo train length, 7; FOV, 12–
14; section thickness, 3 mm; section gap, 1 mm;
matrix, 256 × 256), sagittal T1-weighted (TR
range/TE range, 500–765/8.4–20.0; FOV, 12–14
cm; section thickness, 3 mm; section gap, 1 mm;
matrix, 256–320 × 240–256), and sagittal T2-
weighted with fat saturation (TR range/TE range,
3100–4000/59–80; FOV, 12–14 cm; section thick-
ness, 3 mm; section gap, 1 mm; matrix, 256–336 ×
224–256). In a few patients, coronal and sagittal
STIR imaging was performed instead of fluid-
sensitive sequences (TR range/TE range, 2980–
4820/39–75; slice thickness, 1–3 mm; matrix,
192–320 × 144–321; inversion time, 150–170 ms).
AJR:202, February 2014
MRI of Proximal Humerus in Children
Other sequences, which were occasionally per-
formed, included FLASH (TR/TE, 798/17.9; sec-
tion thickness, 3 mm; matrix, 256 × 144). When
infection or tumors were suspected, the studies
were usually composed of: axial T1-weighted (TR
range/TE range, 400–424;/10–11; section thick-
ness, 3 mm; matrix, 192–256 × 144–192), axial
and sagittal T1-weighted with fat saturation (TR
range/TE range, 423–589/10–11; section thick-
ness, 3 mm; matrix, 192–256 × 134–192) before
and after the administration of IV contrast ma-
terial (gadopentetate dimeglumine, Magnevist,
Bayer HealthCare), axial T2-weighted with fat
saturation (TR range/TE range, 3590–4000/64–
70; section thickness, 3 mm; matrix, 192–256 ×
125–179), and coronal T1-weighted (TR range/TE
range, 400–423/10–11; section thickness, 3 mm;
matrix, 192–256 × 163–192).
Results
A total of 105 MRI studies were reviewed
in 94 patients, ranging in age from 2 months
old to 17 years old. Twenty-two of the 105
studies were excluded for the following rea-
sons: inadequate imaging quality (n = 10) or
significant humeral abnormality, including
fracture (n = 4), tumor (n = 3), osteomyeli-
tis (n = 4), and extensive bony deformity re-
lated to brachial plexus injury (n = 1). The
final patient cohort consisted of 76 patients,
accounting for 83 MRI studies (32 girls and
44 boys). Indications for the MRI studies
included normal findings (n = 9), pain (n =
16), rule out labral tear (n = 12), rule out ro-
tator cuff pathology (n = 9), brachial plex-
us injury (n = 8), soft-tissue tumor (n = 8),
trauma (n =6), dislocation (n = 4), rule out
acromioclavicular-coracoid pathology (n =
4), rule out septic arthritis (n = 2), rule out
bony tumor (n = 2), decreased range of mo-
tion (n = 1), spina bifida (n = 1), and labral
repair (n = 1).
The studies were subdivided into the fol-
lowing categories according to patient age:
0–4 years old, n = 19 (12 girls, seven boys);
5–9 years old, n = 12 (five girls, seven boys);
10–13 years old, n = 17 (five girls, 12 boys);
and 14–17 years old, n = 35 (13 girls, 22 boys).
Humeral Head Ossification
In the 0–4 years age group, one 3-month-
old patient did not show evidence of second-
ary ossification centers; all other patients
had one or more ossification centers. The
nonossified cartilaginous epiphysis was ho-
mogeneously intermediate in signal inten-
sity on T1-weighted images and low in sig-
nal intensity on fluid-sensitive images (Fig.
1). Its shape conformed to the final shape of
the humeral head. Preossification centers,
which manifested as small foci of low sig-
nal intensity measuring a few millimeters in
size on T1-weighted images and bright sig-
nal intensity on fluid-sensitive images, were
seen in three shoulders in two patients. In
a 2-month-old patient with bilateral shoul-
der studies, the preossification centers were
identified at the expected locations of the
medial humeral head secondary ossification
centers. In a 4-month-old patient with an os-
sified medial head secondary ossification
center, the preossification center was found
in the greater tuberosity (Fig. 2).
By 4–10 months (n = 7), a single ossifica-
tion center was apparent in the medial half of
the humeral head in all patients. It was asym-
metrically positioned and had an oval shape
that did not conform to the cartilaginous
shape of the humeral head. Initially, in the
very young age group, the ossification center
was intermediate in signal intensity on both
T1-weighted and fluid-sensitive images, con-
sistent with red marrow signal intensity (Fig.
2). But in the older infants, the ossification
center showed fatty marrow signal intensity
on all pulse sequences (Fig. 3).
By 10 months, a second smaller oval-
shaped ossification center had also formed
in the greater tuberosity (Fig. 3B). This os-
sification center always appeared only after
the formation of the first ossification center
in the medial aspect of the humeral head.
Only one 3-year-old patient showed multi-
ple small ossification centers (Fig. 3C). In all
other cases, only two major secondary ossi-
fication centers were identified. There was a
1-year age gap in the patients sampled from
2 to 3 years old. As a result, we were unable
to examine osseous fusion at that age inter-
val. However, by 3–5 years (n = 4), the two
major secondary ossification centers had al-
ready started the process of osseous fusion
(Fig. 3D), beginning anteriorly and progress-
ing posteriorly.
In the 5- to 9-year-old age group, the two
major secondary ossification centers contin-
ued to expand and further fused together.
As the secondary ossification centers grew,
they lost their oval configuration and became
more hemispherical, with a flat base against
the growth plate and a round superior sur-
face, which progressively conformed to the
articular surface of the humeral head (Fig.
3D). None of the cases showed a separate
lesser tuberosity secondary ossification cen-
ter. Instead, the larger secondary ossification
419
 Kwong et al.

Fig. 1—Cartilaginous tour. The apex of the plate pointed superior-

humeral head with nearly
flat growth plate and
ly and moved from a relatively central posi-
low diffuse proximal tion in the anterior humeral head to a much
humeral marrow signal more lateral position posteriorly. The coni-
intensity in 2-month- cal appearance of the growth plate persist-
old boy. STIR coronal
image (TR/TE, 4820/75) ed throughout skeletal immaturity (Fig.
shows homogeneous 5A). Starting within the 5- to 9-year-old age
Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved

intermediate signal group, the margins of the growth plate were

intensity consistent
with nonossified
cartilage in humeral head
(asterisk). Note slightly
arched growth plate
(arrowheads). There
is also homogeneous
intermediate signal
intensity in proximal
humeral diametaphysis
consistent with red
marrow (stars).
A B
Fig. 2—Preossification center in greater tuberosity and secondary ossification center, depicting red marrow in
medial humeral head in 4-month-old girl.
A and B, Preossification center (white arrow) is low in signal intensity on T1-weighted (TR/TE, 603/11) (A)
and bright in signal intensity on T2-weighted fat-suppressed (TR/TE, 3470/101) (B) axial images, suggesting
fluid collection. Medial head secondary ossification center (black arrow) is low in signal intensity on both
sequences, compatible with red marrow. Lymphatic malformation (asterisk, B) is incidentally seen.
center in the medial humeral head extended insertion of the subscapularis tendon, which
anteriorly into the nonossified lesser tuberos- conformed to the final shape of the lesser tu-
ity (Fig. 4A). berosity (Fig. 4B). In two cases, a 13-year-
The older age groups showed further old patient and a 15-year-old patient, we were
growth of the humeral head secondary ossi- unable to determine whether the lesser tuber-
fication centers. By 13 years old and older, osity ossification was complete.
most cases showed complete ossification of
the proximal humerus (n = 5 in the 10- to Growth Plate
13-year-old age group, n = 26 in the 14- to In the first few months of life, the proxi-
17-year-old age group). In two cases, com- mal humeral growth plate was relatively flat
plete ossification was noted as early as 11 and transversely oriented, with a slight cen-
years old. The lesser tuberosity was the last tral upward arch (Fig. 1). During this period,
portion of the proximal humerus to fully os- the margins of the growth plate were rela-
sify, showing in most studies a persistent in- tively smooth. Beginning at about 4 months,
termediate-signal-intensity, crescentic, sliver the growth plate progressively transformed
of cartilaginous anlage deep in relation to the into a tented or conical (pyramidal) con-
no longer smooth but depicted a more ragged
and irregular appearance (Fig. 5A).
The growth plate showed cartilage MRI
signal-intensity characteristics with interme-
diate signal intensity on T1-weighted images
and increased signal intensity on fluid-sensi-
tive images. A multilaminar appearance was
seen with increased patient age on the fluid-
sensitive images; the normal physeal bright
signal intensity was bordered by a linear
margin of low signal intensity on its epiphy-
seal side, consistent with a bony plate and by
a less-well-defined low-signal-intensity zone
of provisional calcification on its metaphyse-
al side (Fig. 5A). The growth plate measured
a few millimeters in size and was of equal
thickness throughout its course. A ringlike
stripe of bright signal intensity surrounding
the medial head and greater tuberosity ossifi-
cation centers was occasionally seen, compat-
ible with the secondary growth plate (Fig. 5B).
In all but two cases, the growth plate re-
mained completely open until 14 years. The
two exceptions showed partial closure at 9
years. At times, it was difficult to assess the
growth plate closure due to its marked modula-
tion and secondary partial volume averaging.
Leading up to 14 years, the pattern of clo-
sure began with gradual narrowing and de-
creasing signal intensity of the growth plate.
From 14 to 16 years (n = 19), the growth plate
showed partial closure. The posterolateral re-
gion of the growth plate was the last area
to close (Fig. 5C). By 17 years (n = 8), the
growth plate had completely closed in all but
two patients.
Metaphyseal Stripe
For patients between 5 months and 1 year
old, in more than half of the studies (four of
six cases) the proximal humeral metaphysis
showed a circumferential thin stripe of in-
termediate signal intensity on T1-weighted
images and bright signal intensity on fluid-
sensitive images, which paralleled and sep-
arated the humeral cortex from the adja-
cent low-signal-intensity periosteum (Figs.
3D, 5A, and 6). This metaphyseal stripe was
noted in babies a few months old up to the
14-year-old patients, except for two patients,

420 AJR:202, February 2014

 MRI of Proximal Humerus in Children

growth plate closure at about 17 years, the

signal intensity was either imperceptible or
completely absent. In six cases, we were un-
able to adequately assess the proximal me-
taphyseal marrow signal intensity because
the fluid-sensitive sequences were either sub-
optimal or not performed.
Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved

Discussion
Our study assessing the progression of
normal development of the proximal humer-
us on MRI concentrated on the secondary
ossification centers, growth plate, metaphy-
seal stripe, and marrow signal-intensity al-
A B terations in the proximal metaphysis. We
noted a sequential pattern of development
that for the most part was similar to the re-
sults of prior cadaveric and radiographic
studies (Fig. 8).
The secondary ossification centers under-
go a sequential change in MRI signal-inten-
sity characteristics, which correlates with
the histologic and structural changes that
take place within the epiphysis. The progres-
sion of events begins with chondrocyte hy-
pertrophy at the site of future secondary os-
sification centers, followed by vascular and
osteoblast invasion from adjacent cartilage
canals and subsequent lysis of the hypertro-
phied cells. At this time, synthesis of bone
C D and formation of the secondary ossification
Fig. 3—Development of secondary ossification centers. centers occur [1, 9, 10]. The MRI signal in-
A, Proton density–weighted coronal image (TR/TE, 2100/36) in 5-month-old boy shows single ossification tensity of the preossification centers, low on
center (arrow), asymmetrically positioned within medial humeral epiphysis and showing fatty marrow signal T1-weighted and bright on fluid-sensitive im-
intensity. Rest of humeral head and greater tuberosity are intermediate in signal intensity, consistent with
nonossified epiphyseal cartilage (asterisk).
ages, likely correlates with the presence of
B, Proton density–weighted coronal image (TR/TE, 2100/36) in 20-month-old boy shows fatty marrow within vascular invasion and lysis of the chondro-
two secondary ossification centers in medial humeral head and greater tuberosity, respectively (arrows). cytes. Once osteoid has been laid down, the
C, Proton density–weighted coronal image (TR/TE, 2160/35) in 3-year-old girl shows at least three secondary humeral head secondary ossification centers
ossification centers (arrows) within humeral head and greater tuberosity. Rest of nonossified humeral head is
intermediate in signal intensity (asterisk). show low to intermediate signal intensity on
D, T1-weighted coronal image (TR/TE, 400/10) in different 3-year-old girl shows early fusion (white arrow) all pulse sequences, consistent with red mar-
between humeral head and greater tuberosity ossification centers. Centers have become more hemispherical, row [11, 12]. At this stage, the secondary
with flat bases (black arrows) against growth plate and round superior surfaces, which conform to articular
surface of humeral head and to greater tuberosity, respectively (arrowheads). ossification centers can be difficult to dis-
tinguish from the adjacent nonossified epiph-
ages 3 and 9 years. In patients older than 15.5 creasing age, the red marrow signal intensity yseal cartilage on T1-weighted and fluid-sen-
years, the presence of the metaphyseal stripe was progressively more concentrated in the sitive sequences. With red to yellow marrow
began to fade and disappear. In these cases, it proximal metaphysis. Occasionally, in the 0- conversion, the secondary ossification cen-
was either minimal (n = 9) or completely ab- to 4-year-old age group (n = 7), the signal ters develop the easily visualized typical fat-
sent (n = 8). Any residual stripe, before com- intensity appeared on fluid-sensitive images ty marrow signal intensity.
pletely disappearing in these older patients, as a linear stripe of bright signal intensity Ogden et al. [1] reported the histologic pres-
was better visualized posteriorly. on the metaphyseal side of the growth plate, ence of humeral head preossification centers,
contributing to its multilaminar appearance. manifested by chondrocyte hypertrophy, cen-
Proximal Metaphyseal Marrow Signal Intensity In the 10- to 13- (n = 13) and 14- to 17-year- tral vascularization, and early osteoid forma-
During the first few months of life, the old (n = 19) age groups, the residual bright tion. To the best of our knowledge, however,
proximal humerus was diffusely low signal signal intensity was either linear or patchy in the MRI signal-intensity characteristics of the
intensity (brighter than adjacent muscle) on appearance (Fig. 7). By 15 years, when the preossification centers in the proximal humer-
T1-weighted images and bright signal inten- growth plate was partially closed, the signal us have not been documented. Preossification
sity on fluid-sensitive images, consistent with intensity manifested more as a linear band centers on MRI in the trochlea have been de-
red marrow signal intensity (Fig. 1). With in- abutting the epiphysis. Around the time of scribed in several studies [4, 13, 14]; these re-

AJR:202, February 2014 421

 Kwong et al.

graphically evident until 3 years. The time

frame of the study by Ogden et al. was more
comparable to the findings in our study, in
which the earliest medial head and greater tu-
berosity ossification centers appeared on MRI
at 4 months and 10 months, respectively. Our
data also show the process of osseous fusion
Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved

starting at 3 years beginning anteriorly and

progressing posteriorly, again in agreement
with the observations of Ogden et al.
The appearance of a third ossification cen-
ter in the lesser tuberosity continues to be a
topic of debate. Some sources suggest that a
third ossification center may occur in the less-
er tuberosity at 5 years, fusing with the other
A B two ossification centers at 6–7 years [16, 17].
Fig. 4—Maturation of lesser tuberosity. Clarke et al. [8], in an MRI study of the prox-
A, Gradient-echo FLASH axial image (TR/TE, 798/17) in 22-month-old boy shows medial head secondary imal humerus, showed an image with a dis-
ossification center (white arrow) extending anteriorly into nonossified lesser tuberosity (black arrow). Second tinct lesser tuberosity ossification center. On
greater tuberosity ossification center is also seen (arrowhead).
the other hand, Ogden et al. [1], on the basis
B, Fat-suppressed proton density-weighted axial image (TR/TE, 3030/33) in 12-year-old patient shows
intermediate signal intensity in lesser tuberosity (arrow), consistent with nonossified cartilage. of histologic and radiologic studies, did not
mention a third ossification center. Scheuer
port similar MRI signal-intensity characteris- in the humeral head is noted in a small per- and Black [18] questioned the existence of
tics to the preossification centers identified in centage of white newborns. Caffey [2] stat- a separate lesser tuberosity ossification cen-
our shoulder studies, with low signal intensity ed that the humeral head ossification center ter and argued that even if one does exist, it
on T1-weighted images and bright signal in- is radiographically present as early as with- is only for a short time. Our study, in agree-
tensity on fluid-sensitive images. The preossi- in the first month of life and that the greater ment with that of Ogden et al., did not show a
fication centers with their fluid signal-intensity tuberosity ossification center appears by 4–5 separate lesser tuberosity ossification center in
characteristics should not be mistaken for an months. Ogden et al. [1], on the other hand, any of the patients. Instead, the larger ossifica-
intraepiphyseal abscess. Other findings associ- reported that the first ossification center in tion center in the medial humeral epiphysis ex-
ated with infection, such as abnormal soft-tis- the humeral head is evident on radiographs panded anteriorly into the lesser tuberosity in
sue, cartilage, and marrow signal intensity and by 2–3 months and that the greater tuberos- all of our patients. It is conceivable, however,
periosteal reaction should aid in distinguishing ity ossification center appears by 7 months. that a small lesser tuberosity ossification cen-
an abscess from a preossification center. Ogden et al. further stated that although, his- ter was present in some of our patients but had
Kuhns et al. [15] reported, based on chest tologically, osseous fusion beginning anteri- already fused with the medial humeral head
radiography, that the first ossification center orly commences at 2 years, it is not radio- ossification at the time of imaging. Interest-

A B C
Fig. 5—Growth plates in proximal humerus.
A, Proton density-weighted fat-suppressed coronal image (TR/TE, 2840/25) in 14-year-old girl shows pyramidal and jagged, multilaminar growth plate (arrowheads). Bright signal
intensity of growth plate is bordered by low-signal-intensity epiphyseal bone plate (white arrow) and metaphyseal zone of provisional calcification (black arrow).
B, Secondary growth plate, depicted as spherical stripe of bright signal intensity (arrows), surrounding greater tuberosity secondary ossification center (asterisk), is
noted on T2-weighted turbo spin-echo fat-suppressed sagittal image (TR/TE, 3800/64) in 20-month-old boy.
C, Closure of growth plate is seen as loss of normal bright signal intensity in physis (arrowheads) on proton density-weighted fat-suppressed coronal image (TR/TE,
3200/45) in 15-year-old girl. Only posterolateral region remains open (arrow).

422 AJR:202, February 2014


Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved
A B
Fig. 6—Metaphyseal stripe in proximal humerus.
A, T2-weighted axial image (TR/TE, 2100/60) in 20-month-old boy shows circumferential thin bright signal
intensity (white arrow) separating humeral cortex (black arrow) from adjacent low-signal-intensity linear
periosteum (arrowheads).
B, Stripe (arrow), best seen on medial side of humeral metaphysis, is also noted in proton density–weighted
coronal image (TR/TE, 2500/28) in 9-year-old girl.
ingly, our study showed that the lesser tuber-
osity remained incompletely ossified until the
adolescent years, in one patient as late as 15
years. This manifested in the older children
as a sliver of cartilage signal intensity deep
in relation to the insertion of the subscapular-
is tendon. To the best of our knowledge, the
late ossification of the lesser tuberosity has
not been previously reported and may be of
diagnostic importance because cartilaginous
bright signal intensity should not be misinter-
preted as a partial insertional tear of the sub-
scapularis tendon.
AJR:202, February 2014
MRI of Proximal Humerus in Children
Fig. 7—Increased
metaphyseal marrow
signal intensity in 15-year-
old boy. T2-weighted
coronal image (TR/TE,
3500/72) shows patchy
bright signal intensity
in proximal metaphysis
(asterisk). Metaphyseal
stripe is also visualized
(arrow).
The growth plate showed typical multi-
laminar signal-intensity characteristics on
fluid-sensitive sequences. This consisted of
central bright signal intensity of the physeal
cartilage, bordered superiorly by linear low
signal intensity of the epiphyseal bony plate
and by a less well-defined low signal intensi-
ty depicting the zone of provisional calcifica-
tion on the metaphyseal side [3]. The proxi-
mal metaphyseal region occasionally had
linear increased signal intensity approximat-
ing the plate, further contributing to the mul-
tilaminar appearance of the region. The mul-
tilaminar signal-intensity characteristic of
the proximal humeral growth plate is similar
to that of other joints [3]. It is hypothesized
that the central bright signal intensity of the
femoral physeal cartilage, composed of hy-
aline, includes the germinal and prolifera-
tive zones and part of the hypertrophic zone.
This signal intensity is brighter than that of
the surrounding hyaline cartilage because of
increased cellularity and a greater amount of
free water in the physis relative to the rest of
the epiphysis.
Initially, the proximal humeral phy-
sis showed smooth margins and was mild-
ly arched. With growth, the borders of the
physis became more irregular. A gradual
change from a flat to a conical tented ap-
pearance of the plate was also seen, compat-
ible with prior reports [1, 2, 7]. The growth
plate remained open until about 14 years, al-
though partial closure was noted in two pa-
tients, both 9 years old. Gradual thinning of
the plate and eventual closure, manifested
by loss of the bright signal intensity, were
noted by about 17 years, confirming previ-
ous skeletal studies [19]. The posterolateral
aspect of the growth plate was the last re-
gion to disappear. This was compatible with
the prior reports by Ogden et al. [1], who
stated that the closure of the growth plate
begins centrally.
We noted a circumferential metaphyseal
stripe in the proximal humerus in the large
majority of our skeletally immature patients
from a few months of age up to 14 years.
This manifested as linear bright signal in-
tensity on fluid-sensitive images, which par-
alleled and was interposed between the low-
signal-intensity proximal diametaphyseal
cortex and the low-signal-intensity perios-
teum. As the patients reached skeletal ma-
turity around the time of growth plate clo-
sure, the possibility of seeing the stripe
was markedly decreased. The metaphyseal
stripe in the proximal humerus is likely the
equivalent of the stripe noted on MRI along
the posterior surface of the distal femoral
and proximal tibial metaphyses [20]. Sim-
ilar to our results, Laor et al. [20] found a
statistically significant correlation between
the presence of the stripe at the knee and
skeletal immaturity (patency of the growth
plate). Furthermore, histology depicted the
stripe to represent a highly vascularized fi-
brous tissue deep in relation to the loosely
attached periosteum, which was hypothe-
sized to be related to intramembranous bony
growth. Recognizing this stripe as a normal
423

Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved
MRI finding in the proximal humerus as
elsewhere is important to avoid mistaking it
for periosteal elevation related to pathologic
processes, such as tumor, osteomyelitis, or
fracture. The latter disease entities are typi-
cally associated with extensive marrow and
soft-tissue edema. Osseous destruction and
soft-tissue masses in tumors, Brodie abscess
in osteomyelitis, and cortical disruption in
fractures can further aid in distinguishing the
normal metaphyseal stripe from true disease.
Zawin and Jaramillo [6] reported that bone
marrow in the humerus as elsewhere begins
initially as red marrow, which as early as
1–5 years converts to yellow marrow in an
orderly, sequential manner. The conversion
first occurs in the proximal epiphysis, central
diaphysis, and distal metaphysis and, in pa-
tients 15 years old and older, terminates in
the proximal metaphysis [6]. This pattern is
consistent with our findings, in which in in-
fants the marrow signal intensity in the prox-
imal humerus was diffusely low in signal in-
tensity on T1-weighted sequences (higher
than adjacent muscle) and bright in signal
intensity on fluid-sensitive images, suggest-
ing that the entire humerus was composed
of red marrow. With skeletal growth, the red
marrow signal intensity became increasing-
ly concentrated in the proximal metaphysis
and eventually disappeared around 17 years,
comparable to the timeline in the study by
Zawin and Jaramillo. At times, the increased
proximal metaphyseal marrow signal inten-
sity in our patients was quite striking, sug-
424
Kwong et al.
Fig. 8—Drawing shows
developmental stages
of proximal humerus by
age. First humeral head
secondary ossification
center appears in
early infancy in medial
humeral head, followed
by additional secondary
ossification center in
greater tuberosity. Fusion
of ossification centers
begins at 2–3 years
and is complete by late
puberty. Growth plate
metamorphoses from
gentle arc at infancy into
pyramidal appearance in
early years to full closure
by late adolescence.
Metaphyseal stripe
(arrow) disappears at
about time of growth plate
closure. Gray indicates
nonossified epiphyseal
cartilage, and white
circles indicate secondary
ossification centers.
gesting pathologic marrow edema, such as
that which is seen in Little League shoulder.
Comparison of the signal intensity with that
of adjacent muscle can aid in distinguishing
the two entities. In Little League shoulder,
the T1-weighted low signal intensity is typi-
cally lower than that of adjacent muscle. Fur-
thermore, asymmetry and stress-related ir-
regularities in the growth plate will also be
present [21]. In comparison, in healthy chil-
dren, the red marrow T1-weighted signal in-
tensity is higher than adjacent muscle, and
the growth plate is relatively smooth with
equally thick margins.
Aside from the small sample size, the
most significant limitation of our study is
the small number of healthy cases (n = 9) in
our patient cohort. These studies were usu-
ally included in the FOV for comparison,
particularly in the younger patient popula-
tion. The retrospective nature of our study
and lack of histologic and surgical confirma-
tions are other limitations. Many of our MRI
studies lacked gradient-echo images, which
may have been more optimal for assessing
the secondary ossification centers.
In summary, our study provides an MRI
guide to the normal skeletal maturation of
the proximal humerus in the pediatric popu-
lation. This guide, which covers the appear-
ance of the humeral head secondary ossifi-
cation centers, growth plate transformation,
metaphyseal stripe, and proximal metaphy-
seal marrow signal intensity, will hopefully
provide a road map for differentiating nor-
mal developmental changes from disease in
the skeletally immature patient.
Acknowledgment
We sincerely thank Zehava Sadka Rosenberg
for the original idea for this work and her
constant instruction throughout the research.
We are grateful for the countless hours she
spent reading MRI studies, revising this ar-
ticle, and mentoring us throughout the proj-
ect. We all agree that this study would not
have been possible without her supervision
and support.
References
1. Ogden JA, Conlogue GJ, Jensen J. Radiology of
postnatal skeletal development: the proximal
humerus. Skeletal Radiol 1978; 2:153–160
2. Caffey J. Pediatric x-ray diagnosis, 8th ed. Chi-
cago, IL: Year Book Medical Publishers,
1972:434–437
3. Laor T, Jaramillo D. MR Imaging insights into
skeletal maturation: what is normal? Radiology
2009; 250:28–38
4. Jaramillo D, Waters PM. MR Imaging of the
normal developmental anatomy of the elbow.
Magn Reson Imaging Clin N Am 1997; 5:501–
513
5. Chung T, Jaramillo D. Normal maturing distal
tibia and fibula: changes with age at MR imag-
ing. Radiology 1995; 194:227–232
6. Zawin JK, Jaramillo D. Conversion of bone
marrow in the humerus, sternum, and clavicle:
changes with age on MR images. Radiology
1993; 188:159–164
7. Chauvin NA, Jaimes C, Laor T, Jaramillo D.
Magnetic resonance imaging of the pediatric
shoulder. Magn Reson Imaging Clin N Am 2012;
20:327–347
8. Clarke SE, Chafetz RS, Kozin SH. Ossification
of the proximal humerus in children with resid-
ual brachial plexus birth palsy: a magnetic reso-
nance imaging study. J Pediatr Orthop 2010;
30:60–66
9. Gray DJ, Gardner E. The prenatal development
of the human humerus. Am J Anat 1969;
124:431–445
10. Shapiro F, Holtrop ME, Glimcher MJ. Organi-
zation and cellular biology of the perichondrial
ossification groove of Ranvier. J Bone Joint
Surg Am 1977; 59:703–723
11. Jaramillo D, Laor T, Hoffer FA, et al. Epiphy-
seal marrow in infancy: MR imaging. Radiolo-
gy 1991; 180:809–812
12. Taccone A, Oddone M, Dell’Acqua A, Occhi M,
Ciccone MA. MRI “road-map” of normal age-
related bone marrow. Pediatr Radiol 1995;
25:596–606
AJR:202, February 2014
 MRI of Proximal Humerus in Children

13. Jaimes C, Jimenez M, Marin D, Ho-Fung V, Ja-
ramillo D. The trochlear pre-ossification center:
a normal developmental stage and potential pit-
fall on MR images. Pediatr Radiol 2012;
42:1364–1371
14. Chapman VM, Nimkin K, Jaramillo D. The pre-
ossification center: normal CT and MRI find-
Downloaded from ajronline.org by 110.137.72.198 on 01/18/24 from IP address 110.137.72.198. Copyright ARRS. For personal use only; all rights reserved
16. Cocchi U. Zur frage der epiphysenossifikation
des humeruskopfes das tuberculum minus. Ra-
diol Clin 1950; 19:18–23
17. Rockwood CA, Matsen FA. The shoulder, 4th
ed. Philadelphia, PA: Saunders/Elsevier,
2009:12
18. Scheuer L, Black S. The upper limb. In: Scheuer
Epiphyseal union at the upper limb and scapular
girdle in a modern Portuguese skeletal sample.
Am J Phys Anthropol 2008; 137:97–105
20. Laor T, Chun GF, Dardzinski BJ, Bean JA,
Witte DP. Posterior distal femoral and proximal
tibial metaphyseal stripes at MR imaging in
children and young adults. Radiology 2002;

ings in the trochlea. Skeletal Radiol 2004;
33:725–727
15. Kuhns LR, Sherman MP, Poznanski AK, et al.
Humeral-head and coracoid ossification in the
newborn. Radiology 1973; 107:145–149
L, Black S. Developmental juvenile osteology.
San Diego, CA: Academic Press, 2000:278–
285
19. Cardoso HF. Age estimation of adolescent and
young adult male and female skeletons. II.
224:669–674
21. Obembe OO, Gaskin CM, Taffoni MJ, Ander-
son MW. Little leaguer’s shoulder (proximal
humerus epiphysiolysis): MRI findings in four
boys. Pediatr Radiol 2007; 37:885–889

F O R YO U R I N F O R M AT I O N
The AJR has made getting the articles you really want really easy with an online tool, Really Simple
Syndication, available at www.ajronline.org. It’s simple. Click the RSS button located in the
menu on the right side of the page. You’ll be on your way to syndicating your AJR content in no time.

AJR:202, February 2014 425