Injury, Int. J.

Care Injured 47 (2016) 904908

Contents lists available at ScienceDirect

Injury
journal homepage: www.elsevier.com/locate/injury

Simple radiographic assessment of bone quality is associated with loss

of surgical xation in patients with proximal humeral fractures
Ashley W. Newton a,*, Veenesh Selvaratnam b, Satya K. Pydah b, Matthew F. Nixon c
a
Core surgical trainee, Mersey Deanery, UK
b
Specialist registrar in trauma and orthopaedics, Mersey Deanery, UK
c
Consultant upper limb trauma and orthopaedic surgeon, Countess of Chester Hospital, UK

A R T I C L E I N F O A B S T R A C T

Article history: Aim: This study aimed to determine if the ratio of cortical thickness to shaft diameter of the humerus, as
Accepted 25 December 2015 measured on a simple anterior-posterior shoulder radiograph, is associated with surgical xation failure.
Patients and Methods: 64 consecutive fractures in 63 patients (mean age 66.1years, range 3590)
Keywords: operated with surgical xation between March 2011 and July 2014 using PERI-LOC locking plate and
Proximal humerus fracture screws (Smith and Nephew, UK) were identied. Predictors of bone quality were measured from
Fixation failure preoperative radiographs, including ratio of the medial cortex to shaft diameter (medial cortical ratio,
Medial cortical ratio
MCR). Loss of xation (displacement, screw cut out, or change in neck-shaft angle >4 degrees) was
determined on follow-up radiographs.
Results: Loss of xation occurred in 14 patients (21.9%) during the follow up. Patients were older in the
failure group 72.8 vs. 64.2 years (p = 0.007). The MCR was signicantly lower in patients with failed
xation 0.170 vs 0.202, p = 0.019. Loss of xation is three times more likely in patients with a MCR <0.16
(41% vs. 14%, p = 0.015). Increased fracture parts led to increased failure rate (p = 0.0005).
Conclusion: Medial cortex ratio is signicantly associated with loss of surgical xation and may prove to
be a useful adjunct for clinical decision making in patients with proximal humeral fractures.
2016 Elsevier Ltd. All rights reserved.

Introduction Thickness of the humeral cortex has been shown to be a reasonable

Proximal humeral fractures are common, and the incidence is
increasing especially in the elderly [1]. A recent American study
found that surgical management of these fractures had increased
by 40% in the past 20 years [1].
There are a number of treatment options for proximal humeral
fractures including conservative, closed reduction and percutane-
ous wires, open reduction-internal xation (ORIF), intramedullary
nailing, hemiarthroplasty and total shoulder arthroplasty [2].
When weighing up which intervention to choose it is important to
establish the likelihood of treatment success, and a recent
Cochrane systematic review stated that there is insufcient
evidence to determine which treatment method is best [3].
Open reduction-internal xation (ORIF) is a popular choice, and
several studies have explored a number of factors which predict
loss of surgical xation, including age, local bone mineral density
(BMD) in the humerus, as well as generalised osteoporosis [4,5].
* Corresponding author. Tel.: +44 1244365000.
E-mail address: a.w.newton@doctors.org.uk (A.W. Newton).
surrogate for BMD [6]. However no studies have specically
examined whether a simple assessment of the preoperative
radiographs can help predict BMD and then go on to identify
those at risk of premature loss of surgical xation of the proximal
humeral fracture.
The aim of this retrospective study was to determine if the ratio
of medial cortical thickness to shaft diameter of the humerus, as
measured on plain preoperative radiographs is associated with loss
of surgical xation in patients with proximal humerus fractures
treated with humeral locking plate.
Patients and methods
Patients sustaining proximal humerus fractures presenting to
our district general hospital in the North West of England are
discussed at the daily Orthopaedic Consultant led Trauma meeting,
where a treatment decision is made based on multiple factors.
These include age, comorbidities, fracture pattern, and surgeon
preference. We reviewed the radiographs of 64 consecutive
proximal humerus fractures occurring in 63 patients that

http://dx.doi.org/10.1016/j.injury.2015.12.029
00201383/ 2016 Elsevier Ltd. All rights reserved.
 A.W. Newton et al. / Injury, Int. J. Care Injured 47 (2016) 904908 905

underwent surgical xation between March 2011 and July 2014

using PERI-LOC locking plate and screws (Smith and Nephew, UK).
The AO denition of surgical xation was used the application of
a mechanical device to a broken bone to allow healing in a
controlled position [7]. This was achieved using PERI-LOC locking
plate and screws (Smith and Nephew, UK). The mean overall age
was 66.1 years (range 3590 years). Twenty ve patients were
male (39.1%).
Radiographs were reviewed a mean of 24 months following
surgery (range 645). At least one set of follow-up radiographs was
available for each patient.
Firstly, preoperative radiographs were used to determine
number of fracture parts and identify evidence of dislocation.
Then the most recent postoperative radiographs were reviewed to
identify evidence of failed xation by two of the authors (VS/SP).
Krappinger et al [4]. previously dened xation failure with
locking plates for proximal humeral fracture as every kind of Fig. 1. Calculating medial cortical ratio on an AP shoulder radiograph
postoperative relative movement between the implant and the a = proximal level of the humeral diaphysis where the endosteal borders of the
humeral head or shaft. Our denition was broken down into any of lateral and medial cortices become parallel
b = 20 mm below a
the following present on follow up radiographs: fracture displace-
x = diameter of diaphysis at level b
ment, screw cut out, or change in neck-shaft angle greater than 4 y = thickness of medial cortex at level b
degrees. Neck-shaft angle change of 4 degrees or less was not medial cortical ratio (MCR) = yx
dened as failure unless associated with fracture displacement or
screw cut out. This is because small changes in this angle could be xation were signicantly older, 72.8 vs. 64.2 years (mean,
due to variance in radiographs or measurement error. Screw cut unpaired t-test p = 0.008).
out was dened as a screw clearly breaching the medial cortex of
the humeral head on plain radiographs. Radiographs were also
reviewed to establish the presence of avascular necrosis and Medial cortical ratio
determine if further surgery had taken place.
Possible markers of bone mineral density (BMD) were The ratio of the medial cortex thickness to the overall shaft
measured in a standardised fashion from the preoperative diameter (medial cortex ratio, MCR) was signicantly lower in
radiographs by the same author (VS). A number of possible patients with a failed xation (0.170  0.041) compared to those
markers were identied. The medial cortex was chosen (rather with intact xation (0.202  0.044; p = 0.019; Fig. 2). Absolute values
than lateral) as this has previously been demonstrated to relate to for humeral shaft thickness were similar regardless of whether
orthopaedic implant failure at the hip [8]. There is insufcient xation failed (unpaired t-test p = 0.375). The thickness of the medial
evidence to recommend medial or lateral cortex in the humerus so cortex was also similar (p = 0.795) (Table 1).
its selection is arbitrary. To eliminate potential magnication A sample of 10 patients had measurements checked by a second
errors a ratio of the medial cortical thickness to the overall shaft author, showing a signicant correlation in calculation of MCR
diameter at a set point was used, termed the medial cortical ratio, (correlation coefcient 0.80, p = 0.0090).
or MCR. Using a cut off of 0.160, we found that patients with a MCR
The set point was the most proximal level of the humeral <0.160 had a 41% failure rate (7 of 17), compared to a failure rate of
diaphysis where the endosteal borders of the lateral and medial only 14% (7 of 47) in those with MCR  0.160 (Chi square p = 0.015,
cortices were parallel to each other, as described by Tingartet et al Table 2, Fig. 3).
[6]. All measurements were taken 20 mm distal to this (to avoid Avascular necrosis (AVN) of the humeral head occurred in 5 of
potential interference from fracture extension) on a standard the 64 fractures during the follow up period. Screw cut-out
anteriorposterior shoulder radiograph using the digital calliper occurred in 7 patients. Of these, 5 patients underwent reoperation
function within the Picture Archiving Communications System within the follow up period for AVN (n = 1), screw penetration
(PACS). The diameter of the diaphysis and the thickness of the
medial cortex were taken at this point and used to determine the 0.25
medial cortical ratio, MCR (Fig. 1).
A sample of radiographs was assessed by a second author
(AWN) and the inter-observer variability was calculated. Statistical 0.2
and data analyses were conducted within Microsoft Excel 2010
Medial cortex ratio

(Microsoft Corporation, USA). Students t-test, Chi-square test, and

0.15
ANOVA were used as appropriate. A p-value  0.05 was considered
to be statistically signicant.
0.1 0.202 p=0.019
0.170
Results
0.05
From analysis of the operative database 63 patients were
identied (one with bilateral fractures occurring one month apart).
The fractures were considered in two groups, depending on 0
N Y
whether or not the xation failed on radiographs. During the
Fixation failure
follow up period 50 (78.1%) fractures had intact surgical xation,
and 14 (21.9%) had loss of surgical xation. Patients with failed Fig. 2. Chart showing xation failure related to medial cortex ratio
906 A.W. Newton et al. / Injury, Int. J. Care Injured 47 (2016) 904908

Table 1
Comparison of whether xation failed related to cortical measurements.

Fixation maintained (n = 54) Fixation failed (n = 14) p

Medial cortex ratioa 0.200  0.044 0.170  0.043 0.019b
Overall shaft diameter (mm) c 24.21 (18.33 38.26) 24.62 (20.05 30.23) 0.375b
c
Medial cortex thickness (mm) 4.53 (2.26 7.47) 4.51 (2.79 5.86) 0.795b
a
mean  standard deviation
b
unpaired t test
c
mean and range

High quality systematic reviews have been unable to recom-

Table 2 mend the superiority of one treatment modality over another for
Fixation failure compared to Medial Cortex Ratio (MCR). these fractures [3]. Given the wide range of potential options for
MCR < 0.16 7 of 17 (41%) the management of proximal humeral fractures it is important to
MCR  0.16 7 of 47 (14%) consider predictors of success before embarking on surgical
xation as a treatment strategy. Considering our ndings, in
patients who have a low MCR it would be prudent to thoroughly
(n = 2) or a combination of both (n = 2). Three of the ve who were weigh the potential benets of xation against the increased risk of
re-operated were revised to a reverse polarity shoulder arthro- failure, especially in the elderly with complex fractures; arthro-
plasty. plasty or conservative management may be considered more
appropriate for such patients. Alternatively the xation could be
Fracture factors augmented with bone graft to improve the chances of success with
surgical xation [9,10].
Fixation failure was signicantly correlated with fracture Experimental models in cadaveric specimens have demonstrat-
complexity. In fractures with intact xation only 20% (10 of 50) ed that local BMD predicts failure of proximal humerus xation in
had a four-part fracture conguration, rising to 71% (10 of 14) in controlled conditions [1113]. In these models bone with lower
patients with failed xation (Fig. 4, Table 3; Chi squared test BMD predictably fails after fewer stress cycles compared to bone
p = 0.0005). Patients presenting with a fracture-dislocation had a with normal BMD.
21% rate of xation failure compared to 8% in those without Jung et al. retrospectively reviewed 252 patients treated with
dislocation (Chi squared test p = 0.199), however the overall locking plates for proximal humerus fractures. Using logistic
number in the dislocation group is small (n = 7). Medial cortex ratio regression they established that older age (p = 0.023) and
did not signicantly correlate with number of fracture parts, Fig. 5 osteoporosis (p = 0.001) were two of the factors that had signicant
(ANOVA p = 0.461). correlations with reduction loss [5]. They noted that BMD in the
reduction loss group ( 2.8) was signicantly lower than in the
Discussion group with intact reduction ( 1.8, p = 0.001). In their study BMD
was measured in the spine and femur using dual-energy x-ray
This study shows that patients with a lower MCR are absorptiometry (DEXA) the BMD of the proximal humerus was
signicantly more likely to develop loss of xation. Loss of xation not directly measured. Krappinger et al. measured the local BMD of
is three times more likely in patients with a MCR <0.16 (41% vs. the humerus, and also found this to be correlated with an increase
14%). Given that 50% of patients with loss of xation had screw cut in xation failure, especially in elderly patients with multi-
out or penetration, it is not surprising that a lower MCR predicts fragmentary fracture patterns [4]. They also champion the
failure. We postulate that this is because it acts as a surrogate consideration of arthroplasty or xation augmentation in this
marker for bone quality/bone mineral density. Loss of xation was group.
also signicantly more likely with increasing age and more fracture Virtama and Telkka demonstrated over 50 years ago that the
parts. ratio of cortical to shaft width has a highly signicant correlation
with the mineral content of the humerus (p < 0.001) [14].
100 Additionally, several contemporary authors have described meth-
ods of estimating BMD from measurements taken from plain
90
radiographs. Tingart et al. compared the cortical thickness of the
80
70
Fixation failure
60 Y
p=0.015
N
%

50
40
30
20
10
0
<0.160 0.160
MCR

Fig. 3. Chart showing decreased failure rate with MCR  0 160 Fig. 4. Proportion of failed xations depending on number of fracture parts
 A.W. Newton et al. / Injury, Int. J. Care Injured 47 (2016) 904908
Table 3
Comparison of whether xation failed related to fracture factors.
Fixation maintained (n = 50)
Fracture parts
2 part 26 (52%)
3 part 14 (28%)
4 part 10 (20%)
Associated dislocation 4 (8.0%)
a
Chi Squared test.
diaphysis to the bone mineral density of the humeral head, surgical
neck, greater tuberosity, and lesser tuberosity [6]. They found a
signicant correlation between BMD in each region and cortical
thickness (p < 0.01). For the surgical neck the correlation was
r = 0.84 (95% CI 0.62 to 0.94). In their study the measurements
were made in-vitro on cadaveric specimens, allowing cortical
thickness to be measured directly. Other authors have found that
when measurements are taken from plain radiographs in-vivo
there is a strong correlation with femoral BMD measured by DEXA
(r = 0.64, P < 0.0001) [15].
Giannotti et al. found the ratio of cortical thickness to overall
shaft diameter correlated with distant BMD as measured on DEXA
in the spine and femur. Using a ratio of cortical thickness has
advantages over absolute size measurements when the measure-
ments are made from radiographs, as it eliminates potential
magnication errors [16], hence why we used MCR in our study.
This retrospective single centre study is not without limitations.
It included a relatively small number of patients; however it still
managed to demonstrate a statistically signicant difference
between the groups. This suggests that MCR may have a sizeable
clinical difference since it can be statistically demonstrated in a
study of this size.
Due to the retrospective nature of the study the data available
for clinical correlation of outcomes was incomplete. A prospective
study would be able to correlate radiological signs of failure with
patients symptoms and functional outcome. Whilst evidence
supports that the increased failures with a low MCR is related to a
lower BMD, this study did not directly correlate the two variables
with outcomes.
There is excellent scope for future work to further delimitate
the association between MCR and implant failure. The study
should comprise a prospectively designed trial which correlates
0.25
0.2
Medial cortex ratio
0.15
0.1 0.201 0.191
0.185
0.05
0 four-part proximal humerus fractures treated with a locking compression
2 3 4 plate and an autologous iliac bone impaction graft. Injury 2012;43:172431.
Fracture parts
Fig. 5. Chart relating medial cortex ratio with number of fracture parts
907
Fixation failed (n = 14) p
1 (7%) 0.0005a
3 (21%)
10 (71%)
3 (21%) 0.199a
MCR with a formal measure of bone mineral density, and then
relates this to radiological xation failure, pain and function using
validated patient reported outcome measure. Follow up should be
according to a set protocol which continues until fracture union
has been achieved. Multicentre collaboration would increase
patient numbers and improve external validity.
MCR is associated with surgical xation failure of proximal
humeral fractures. Whilst previous papers have shown that
radiographs can predict BMD, and BMD is related to xation
failure, no other author has demonstrated that radiographic
markers of bone quality/BMD are associated with xation failure.
MCR has major advantages over formal BMD measurement in this
setting, since it does not require patients to undergo any additional
testing prior to surgery. Further prospective studies which
measure MCR and correlate this with local and systemic BMD
as well as long term clinical and radiological outcomes will help
to cement MCR as an essential tool for preoperative decision
making in patients with proximal humeral fractures.
Conict of Interest
No authors have any conicts of interest which could inuence
the integrity of the above paper. No funding was received with
respect to the paper.
References
[1] Khatib O, Onyekwelu I, Zuckerman JD. The incidence of proximal humeral
fractures in new york state from 1990 through 2010 with an emphasis on
operative management in patients aged 65 years or older. J Shoulder Elbow
Surg 2014;23:135662.
[2] Ma R, Khan W, Ma P, Hindocha S. Orthopaedic approaches to proximal
humeral fractures following trauma. Open Orthop J 2014;8:43741.
[3] Handoll HH, Ollivere BJ, Rollins KE. Interventions for treating proximal hu-
meral fractures in adults. Cochrane Database Syst Rev 2012;12:CD000434.
[4] Krappinger D, Bizzotto N, Riedmann S, Kammerlander C, Hengg C, Kralinger FS.
Predicting failure after surgical xation of proximal humerus fractures. Injury
2011;42:12838.
[5] Jung SW, Shim SB, Kim HM, Lee JH, Lim HS. Factors that inuence reduction
loss in proximal humerus fracture surgery. J Orthop Trauma 2014.
[6] Tingart M, Apreleva M, Stechow D, Zurakowski D, Warner J. The cortical
thickness of the proximal humeral diaphysis predicts bone mineral density
of the proximal humerus. J Bone Joint Surg [Br] 2003;85:6117.
[7] AO Foundation. Ao surgery reference, glossary, accessed 01/05/15, https://
www2.Aofoundation.Org/wps/portal/!Ut/p/a1/04_sj9cpykssy0xplmnmz0v-
mafgjzokn_a0m3d2ddbz9_ummdrydxq3dw9wmdawctyekiveocdqntr8bdu-
boqeh_qw5okabaevup/dl5/d5/l2djqsevuut3qs80sml1o2xzjpmdbhsvmws09pv-
dewqvnfmudwrjawmeu3/?Solostate=true&popupstyle=diagnosis&showall=sho-
w+all+terms.
[8] Nixon M, Taylor G, Sheldon P, Iqbal SJ, Harper W. Does bone quality predict
loosening of cemented total hip replacements? J Bone Joint Surg Br
2007;89:13038.
[9] Zhu L, Liu Y, Yang Z, Li H, Wang J, Zhao C, et al. Locking plate xation combined
with iliac crest bone autologous graft for proximal humerus comminuted
fracture. Chin Med J (Engl) 2014;127:16726.
[10] Kim SH, Lee YH, Chung SW, Shin SH, Jang WY, Gong HS, et al. Outcomes for
[11] Roderer G, Brianza S, Schiuma D, Schwyn R, Scola A, Gueorguiev B, Gebhard F,
Tami A. Mechanical assessment of local bone quality to predict failure of
locked plating in a proximal humerus fracture model. Orthopedics
2013;36:e1134-L1140.
908 A.W. Newton et al. / Injury, Int. J. Care Injured 47 (2016) 904908

[12] Roderer G, Scola A, Schmolz W, Gebhard F, Windolf M, Hofmann-Fliri L.
Biomechanical in vitro assessment of screw augmentation in locked plating
of proximal humerus fractures. Injury 2013;44:132732.
[13] Hymes RA, Levine MJ, Schulman JE, Westbrook RD, Li J. Mechanisms of failure
of locked-plate xation of the proximal humerus: Acoustic emissions as a
novel assessment modality. J Orthop Trauma 2013;27:3928.
[14] Virtama P, Telkka A. Cortical thickness as an estimate of mineral content of
human humerus and femur. Br J Radiol 1962;35:6323.
[15] Mather J, MacDermid JC, Faber KJ, Athwal GS. Proximal humerus cortical bone
thickness correlates with bone mineral density and can clinically rule out
osteoporosis. J Shoulder Elbow Surg 2013;22:7328.
[16] Giannotti S, Bottai V, Dellosso G, Donati D, Bugelli G, De Paola G, et al. Indices
of risk assessment of fracture of the proximal humerus. Clin Cases Miner Bone
Metab 2012;9:379.