It is well recognized that the radial head is an important secondary
stabilizer of the elbow and
forearm1, and the
importance of preserving it after a complex elbow fracture has been well
documented2-6.
Open reduction and internal fixation of comminuted and severely displaced
radial head fractures may be technically very difficult to achieve. Metal
radial head prostheses were developed to restore elbow stability and permit
early joint
mobility7-9,
and the short-term clinical results following use of these implants have
generally been
promising10-17.
There are two different types of metal radial head prostheses, based on
different concepts: a loose monoblock prosthesis and a bipolar prosthesis.
Most monoblock metal radial head prostheses have an axisymmetric head with
a smooth stem. They are not fixed with cement. The designers believe that a
small amount of movement of the stem within the radial neck may compensate for
the lack of head asymmetry and improve the congruency of the articular surface
of the implant with the capitellum and the proximal part of the
ulna15. However,
this intentionally loose intramedullary fixation may have been an important
factor in the development of radiolucencies surrounding the uncemented stem
that was reported by Moro et
al.15, who noted
that the lucencies tend to develop early and remain stable over time.
Bipolar radial head prostheses have a symmetric head that articulates in a
semiconstrained fashion with a cemented
stem11. The
rationale for the design of this prosthesis is that the additional freedom of
movement may reduce stress at the implant-bone interface and increase the
contact area at the radiocapitellar
joint11. However,
generation of wear debris and osteolysis could be a disadvantage. To our
knowledge, this has not yet been reported in the literature, perhaps because
the number of patients available for study has been limited and follow-up has
been relatively
short11,13,14,18.
The purpose of this study was to evaluate the midterm radiographic outcomes
of the use of a bipolar radial head prosthesis in a consecutive group of
patients presenting with a comminuted fracture of the radial head associated
with other destabilizing injuries of the elbow.
Aretrospective study was undertaken to evaluate fifty-five consecutive
patients who had been admitted to our hospital between January 1994 and
December 2001 with an acute radial head fracture associated with other
destabilizing elbow injuries. The associated elbow injurie6s included elbow
dislocation, ligamentous injuries, and fractures. The radial head was excised
because open reduction and internal fixation was not considered to be
feasible, and a bipolar radial head prosthesis (Tornier, Saint Ismier, France)
was implanted. The prosthesis is available with two radial head diameters: 19
and 22 mm. The radial stem is available in two sizes: a diameter of 8 mm with
a length of 60 mm and a diameter of 6.5 mm with a length of 55 mm. These
components are interchangeable and allow four different combinations of the
head and stem (Fig. 1).
Our institutional review board approved this review. The physical and
radiographic examinations, including the assessment protocol, were carried out
after the patients gave informed consent to participate in the study. The
outcomes of fifty-one arthroplasties in fifty-one patients were reviewed. Four
patients were lost to follow-up after two years. The dominant arm of forty-one
patients and the nondominant arm of ten patients were involved. There were
thirty-two men and nineteen women with a mean age of fifty-one years (range,
twenty-two to sixty-nine years). Thirty-two patients were injured in a
ground-level fall on the outstretched hand; six, in a fall on stairs; five, in
a fall from a roof; and eight, in a motor-vehicle accident. All patients had a
closed elbow injury.
Preoperative anteroposterior and lateral radiographs were made of each
elbow to identify the radial head fracture and whether there was an elbow
dislocation. Plain anteroposterior and lateral radiographs of the forearm and
wrist were made as well. Computed tomographic evaluation was performed
routinely for all patients to identify any other osseous injury that was not
identified during the initial radiographic evaluation.
The fractures of the radial head were classified according to Mason's
system19. Nine
fractures involved part of the radial head (Mason type II), and forty-two
involved the entire head (Mason type III). Eight patients had a simple
fracture of the entire neck with the head completely displaced from the shaft,
sixteen had a fracture of the entire head with more than two large displaced
fragments, and eighteen had a fracture with a tilted and impacted articular
fragment and small, comminuted markedly displaced fragments. All fractures
were subclassified with regard to the absence or presence of a dislocation of
the elbow. Eighteen patients had a posterior elbow dislocation with an
isolated radial head fracture, and sixteen other patients had a posterior
elbow dislocation associated with fractures of the coronoid process and radial
head. The fractures of the coronoid process were categorized into three types
according to the system of Regan and
Morrey20. There
were ten type-IB fractures, five type-IIB fractures, and one type-IIIB
fracture. Two patients had an associated fracture of the proximal part of the
ipsilateral humerus. Six patients had both a radial head fracture and a
posterior fracture-dislocation of the elbow (a posterior Monteggia lesion).
The mean time from the injury to surgery was 4.1 days (range, two to seven
days).
The surgical technique has been previously
described13. The
patient was placed in the supine position on the operating table with the arm
on a side arm-table for the entire procedure. A Kocher exposure was used with
the forearm pronated to protect the posterior interosseous branch of the
radial nerve. The interval between the anconeus and the extensor carpi ulnaris
was developed. In patients with an elbow dislocation, the lateral collateral
ligament was already avulsed from its proximal attachment, facilitating
surgical exposure. In some patients, the lateral collateral ligament was
carefully released and then repaired at the end of the operation. The medial
collateral ligament and the interosseous radioulnar
membrane16 were
assessed for stability during the surgery. With the forearm placed in
pronation and the elbow in 30° of flexion, the stability of the medial
collateral ligament was assessed by exerting a valgus stress on the elbow. A
decrease in the distance between the radial neck and the capitellum of >2
mm was considered abnormal. A Kocher clamp was placed on the radial neck and
the radius was axially loaded to assess the competence of the interosseous
membrane. A change in the distance between the radial neck and the capitellum
of >2 mm was considered abnormal. Lateral ligament injuries were assessed
during the surgical approach.
The medullary canal was prepared with small rasps of 6.5 or 8 mm in
diameter sized with trial stems. All of the stems were cemented. The canal was
plugged distally with bone fragments from the fractured radial head. With use
of a syringe, cement was inserted in an antegrade manner, and the
corresponding radial shaft component was inserted manually. The inclination of
the neck was set in the same plane as that of the thumb positioned in
abduction and forward
flexion13. Radial
head templates (19 and 22 mm) were used to assess stability and the head
dimensions and to avoid impingement against the humerus while allowing a
clearance of 1 mm. Particular attention was paid to the surgical repair of the
lesions of the lateral ligament complex to prevent posterolateral rotatory
instability of the elbow. The ulnar attachment of the lateral ulnar collateral
ligament was palpated at the tubercle of the supinator crest and was
dissected. On the humeral side, the supracondylar ridge was exposed
posteriorly and anteriorly. We reattached the lateral collateral ligament with
interrupted transosseous number-1 nonabsorbable braided sutures after
insertion of the prosthesis in thirty-three patients.
The torn medial collateral ligament was repaired in seven patients because
the elbow was unstable following radial head replacement. The ligament was
exposed through a medial incision centered over the medial epicondyle, and the
ulnar nerve was identified. In four patients, a muscle-splitting approach was
made in the flexor-pronator mass extending from the medial humeral epicondyle
to a point distal to the tubercle of the ulna. In three patients, to provide
additional exposure, the common flexor-pronator mass and the pronator teres
were released from the medial epicondyle in such a way that reinsertion was
possible. Mobilization and subcutaneous transposition of the ulnar nerve was
performed in these patients. The medial epicondyle was exposed completely. The
medial collateral ligament was found to be disrupted from its insertion on the
medial epicondyle. The proximal point of isometry was determined, and the
medial collateral ligament was reattached with 5.0-mm FastinRC suture anchors
with number-2 Ethibond (DePuy-Mitek Products, Raynham, Massachusetts). The
suture in the medial collateral ligament was tied after the radial head was
inserted.
The mean duration of follow-up was 8.4 years (range, four to thirteen
years). All patients were examined by us clinically and radiographically two
and six months after the operation and then annually with plain radiographs of
the elbow and wrist. All fifty-one patients were evaluated by one of us (N.P.)
at the time of the last follow-up.
Anteroposterior, lateral, and radiocapitellar radiographs of the elbow were
made with the patient sitting on a stool and the upper extremity resting on a
table. For the anterior radiograph, the shoulder was flexed forward and the
elbow was extended as much as possible with the forearm supinated. The beam
was centered perpendicular to the elbow on the midpart of the antecubital
crease. The lateral radiograph was made with the elbow flexed to 90°, the
forearm in neutral rotation, and the shoulder internally rotated. The
radiocapitellar radiograph was made with the elbow positioned as it was for
the lateral radiograph but with the beam angled 45° anteriorly. Our
radiographic system used phosphorus crystal plates and an analogue-to-digital
converter system (ADC; Agfa, Mortsel, Belgium) for development. This allowed
zooming of the digital image on a screen and provided more accurate
measurements.
Assessment of the patients included measurement of the range of motion of
the elbow and forearm as well as evaluation of pain and grip strength. Flexion
and extension of the elbow were measured with a handheld goniometer held along
the lateral aspect of the arm with the forearm in neutral rotation. Pronation
and supination were measured at the extremes of active motion with the elbow
in 90° of flexion. One arm of the goniometer was held along the patient's
arm and the second was placed parallel to the dorsal or volar aspect of the
wrist. Pain was graded as absent, mild, moderate, or severe, and the need for
pain medication was recorded. The isometric strength in flexion, extension,
supination, and pronation was measured with a tensiometer (Pesola, Paris,
France) at 90° of elbow flexion. All patients were evaluated with use of
the Mayo Elbow Performance
Index21.
Radiographic evaluation consisted of assessment and quantification of
changes in the elbow joint and around the stem of the prosthesis in the
proximal part of the radius. Degenerative changes in the elbow joint were
classified according to the system of Broberg and
Morrey22 and
heterotopic ossification was graded according to the system of Hastings and
Graham23 on the
postoperative and most recent radiographs. The presence or absence of
overstuffing of the radial head prosthesis was determined on follow-up
radiographs by comparing the medial ulnohumeral joint space of the operatively
treated and untreated elbows. If the medial ulnohumeral joint surfaces were
parallel and the joint space was equal to that of the contralateral elbow, it
was deemed that there was no overstuffing.
Changes in the proximal part of the radius, including radiolucent lines,
osteolysis, and proximal resorption of the radial neck, were recorded.
Resorption sites and lengths were assessed on standard anteroposterior
radiographs with use of the method developed by Gruen et
al.24 for the hip.
The proximal part of the radius was divided into seven zones
(Fig. 2). The width of the
radiolucent zones was measured at their widest point with a ruler. Progression
was deemed to have occurred when the resorption had either extended into an
additional zone or had widened by at least 1 mm between consecutive
assessments25.
Complete or incomplete radiolucent lines were defined by the presence of an
adjacent layer of reactive bone sclerosis appearing as a radiopaque line no
more than 1 mm thick. Osteolysis was identified as balloon-shaped radiolucent
zones around the stem or as isolated proximal bone resorption at the radial
neck, without any reactive bone sclerosis.
Migration of the stem of the prosthesis relative to the bone was evaluated
with use of the Imagika image analysis system (GreyStone, Paris, France). An
analysis model (a custom-made program built with the Imagika software) was
created in order to evaluate the reliability of several possible landmarks on
the proximal part of the radius and on the stem and to assess the possibility
of using a correction factor to improve the comparability of successive
radiographs made under nonstandardized conditions. The Imagika method can be
applied to marked reference points on scanned radiographs. On the basis of two
points determined manually on the contour of the prosthetic head, the software
identifies the largest diameter of the head with use of automatic edge
detection. The known diameter of the prosthetic head is used for calibration.
For assessment of migration, a reference point on the prosthesis is determined
at the intersection of the longitudinal axis of the stem and the transverse
axis of the collar of the prosthesis. The reference point on the radius is the
most distal point of the radial tuberosity; from this point, a line is drawn
perpendicular to the longitudinal axis of the stem, and a second reference
point is marked at the intersection. The distance between these two points is
compared on consecutive radiographs, and any reduction in the distance
indicates subsidence of the stem (Fig.
3).
The stability of the implant in the coronal plan was evaluated by measuring
valgus laxity with stress radiography. Stress radiographs were made with the
aid of a Telos GA-II/E stress device (Telos, Weiterstadt, Germany) for all
fifty-one patients. The Telos stress device was used to provide consistent
extremity positioning and a valgus stress. The reliability of the Telos stress
device was tested in a previous study, in which bilateral radiographs of the
elbows of fifty normal individuals were made with use of an identical
procedure26. The
intraclass correlation coefficient was found to be 0.95. The Telos device can
provide medial joint space widths that can be reliably measured on
anteroposterior radiographs when the radiographs are interpreted by one
radiologist. The elbow was flexed 30° with the forearm in neutral
rotation. The elbow flexion position was verified with use of a standard
universal goniometer. The patient was seated on a stool with the Telos device
placed on the radiography table and the shoulder abducted 65° and
externally rotated. The arm was placed within the Telos apparatus with a
proximal stabilizing pad just distal to the axilla and a distal stabilizing
pad at the hand and the wrist. This was accomplished by having the patient
grasp a handle on the Telos device with the forearm in neutral rotation.
Anteroposterior radiographs of both elbows were then made with 0 and 150 N of
applied valgus force. The medial joint space was measured before and after
application of the valgus stress. A dependent t test was used to identify
differences in medial laxity between the operatively treated and contralateral
arms, with significance set at the 0.05 level. All radiographs were made by
one technician with the assistance of the senior author (N.P.), and they were
interpreted by one independent senior radiologist.
Clinical Results
At the time of follow-up, the average Mayo Elbow Performance Index was 83
points (range, 59 to 95 points). Elbow flexion averaged 130°; extension,
14°; supination, 71°; and pronation, 68°. The mean functional arc
of motion was 116° (range, 50° to 135°). A correlation (r = 0.84)
was noted between restriction of elbow mobility and the severity of the
initial injury. Twenty-three patients had complete pain relief; eighteen
patients had mild elbow pain after strenuous daily activities; eight patients
had moderate pain, requiring some form of medication, with daily activities;
and two patients, both with overstuffing of the radial head prosthesis, had
severe pain. The mean grip strength on the injured side (29 kg) was
significantly reduced compared with that of the contralateral extremity (34
kg) (p = 0.017). One patient, in whom the initial injury was a comminuted
radial head fracture associated with elbow dislocation and proximal detachment
of the lateral collateral ligament, had persistent ligamentous laxity. The
lateral ulnar collateral ligament had been inadequately restored. At six
months, the patient reported mild pain. During the pivot shift maneuver, a
clunk as a result of spontaneous radial head reduction was noted. A
subcutaneous dimple was present just proximal to the radial head as the
prosthesis subluxated posteriorly, and it disappeared with reduction of the
radial head. The patient deferred operative intervention. None of the patients
had any symptoms related to the radiocarpal or distal radioulnar joint.
Radiographic Assessment
Twenty-seven patients (53%) had radiographic evidence of periprosthetic
lucency within the medullary canal of the radius. Progressive loss of bone in
the radial neck region, adjacent to the area beneath the collar of the implant
(zones 1 and 7), was observed in sixteen patients (31%)
(Fig. 4). Five patients (10%)
had progressive balloon-shaped osteolysis in the midstem region (zones 2, 3,
5, and 6) with important changes in cortical wall thickness
(Fig. 5). The stem of the
prosthesis had migrated more than 1 mm in these five patients.
Six patients were seen to have a reduction in the thickness of the
articular cartilage of the capitellum on the radiocapitellar radiograph, and
marked flattening of the capitellum was noted in two patients. Compared with
the contralateral elbow, thirty involved elbows had radiographic evidence of
posttraumatic osteoarthritis according to the classification system of Broberg
and Morrey22. The
posttraumatic ulnohumeral osteoarthritis was graded as mild in twenty-one
patients, moderate in seven, and severe in two. An increase in pain after
daily activities could be attributed to degenerative arthritis of the elbow in
these patients. Heterotopic ossification was observed in twenty-one patients
at the time of final follow-up. According to the Hastings-Graham
classification23,
thirteen had class-I ossification, six had class-II (three had class-IIa, two
had class-IIb, and one had class-IIc), and two had class-III. There were no
fractures of any of the prosthetic components.
With no stress applied, the radiographic measurements of the medial joint
space widths did not show a significant difference between the operatively
treated elbows (mean and standard deviation, 3.5 ± 0.5 mm) and those
that had not been operated on (3.4 ± 0.5 mm) (p = 0.36). Application of
the valgus stress (150 N) resulted in a significant increase in the medial
ulnohumeral distance in both elbows as compared with the values under the
unstressed condition (p = 0.018); on the average, the operatively treated
extremities had 0.3 ± 0.1-mm greater increase in the medial joint space
width with the application of 150 N of valgus stress (p = 0.06). There was no
significant difference in the amount of valgus stress opening between the
operatively treated and contralateral elbows.
Complications
There were ten complications. Complex regional pain syndrome developed in
one patient. Four patients sustained a posterior interosseous nerve palsy
related to exposure that resolved spontaneously four months postoperatively.
In these four patients, a more anterior interval through the extensor
digitorum communis had been used and retractors had been placed around the
radial neck. We believe that these factors were responsible for the transient
neurological complications. A progressive ulnar neuropathy developed in one
patient who had had a comminuted radial head fracture associated with a
posterior elbow dislocation and a type-IB coronoid fracture; nerve
decompression was performed at eight months, with a satisfactory result. A
complete proximal radioulnar synostosis developed in another patient. Two
patients had early persistent pain and early degenerative changes, which were
attributed to overstuffing of the radial head replacement. One patient had
posterolateral subluxation of the prosthesis. There were no infections. At the
time of writing, no implant has been permanently removed.
Overall, the short-term results following treatment with the bipolar radial
head prosthesis have been promising, so that this implant appeared to be a
safe and effective
option11,13,14,18,27.
The clinical results in the present study are largely comparable with those in
other
studies10,12,15,17
on metal radial head replacement. They suggest that the bipolar radial head
prosthesis provides an overall satisfactory clinical outcome with respect to
the four most important functional parameters: stability, motion, pain, and
strength. Our observation that the severity of the initial elbow injury
influenced the functional outcomes, especially elbow mobility, is in agreement
with the findings of
others15.
Previous reports on the short-term results of treatment with a bipolar
radial prosthesis did not mention progressive radiographic changes at the
elbow11,13,18—i.e.,
the authors did not note lucencies surrounding the cemented stems, progressive
joint space narrowing, stem loosening, or osteoporosis of the humeral condyle.
To our knowledge, only Smets et
al.14 reported
progression of degenerative changes in a few patients. These excellent
short-term radiographic results were attributed to automatic (self-correcting)
positioning of the prosthetic head against the capitellum and the radial notch
of the ulna, reducing the possibility of aseptic loosening secondary to the
complex movements of the elbow. However, in our midterm study, we observed
three different types of radiographic changes around the prosthetic stem:
complete or incomplete radiolucent lines, balloon-shaped radiolucent zones,
and proximal bone resorption at the radial neck.
Radiolucent lines surrounding the stem of the bipolar radial head
prosthesis were observed in twenty-seven patients. In eleven of these
patients, incomplete radiolucent lines developed early in the first
postoperative year and remained stable over time. It is our impression that
the lucencies in these eleven patients can be attributed to suboptimal initial
fixation as a result of suboptimal cementing technique because of the small
diameter of the medullary canal of the proximal part of the radius.
Progressive radiolucent lines were observed in the other sixteen patients.
They may reflect progressive stem loosening as a result of mechanical factors,
possibly in conjunction with progressive osteolysis.
Sixteen of our fifty-one patients had partial or complete loss of proximal
bone support at the neck of the radius, whereas the distal end of the
prosthesis remained rigidly fixed in the cement. Proximal bone resorption
could result from osteolysis secondary to wear debris from the bipolar device
or from altered load transmission to the osteotomized neck of the radius. A
possible indication for operative intervention is local pain with radiographic
evidence of complete loosening and bone destruction around the stem with a
danger of cortical fracture. If intervention is necessary, removal of the
prosthesis is the only procedure that we would currently recommend,
considering our limited experience with failed radial head prostheses. As we
have not reintervened to date and therefore have no histologic studies, we can
only speculate that the bone destruction may have been related to tissue
reaction to wear particles, a process that has been studied extensively at the
hip joint28.
Whether stress-shielding played any role in the bone loss at the radial neck
is difficult to say, although one would expect that it would result first in
cortical atrophy, which we did not observe. Longer follow-up is needed to
assess whether the proximal resorption is a continuous process. We did observe
progressive ballooning osteolysis with bone destruction in the midstem region
and radiographic evidence of complete loosening in five patients.
A possible explanation for the high prevalence of radiographic changes
reflecting or suggesting osteolysis in our patients is that the bipolar radial
head prosthesis consists of two parts. The radial head is made of high-density
polyethylene enclosed in a cobalt-chromium cup that articulates in a
semiconstrained manner with the spherical end of a cemented intramedullary
stem11.
Biomechanical studies have shown that the radiocapitellar joint carries
considerable
loads29,30
and that the contact area of a metal head is reduced by approximately
two-thirds as compared with the native radiocapitellar
joint31. We
hypothesize that there is likely an excess of contact stresses between the
small spherical metal head and the thin (3-mm-thick) polyethylene insert.
Previous studies on hip arthroplasty have suggested that the yield strength of
ultra-high molecular weight polyethylene inlays of <8 mm in thickness is
exceeded by the joint contact stresses, which may result in catastrophic
wear32. However,
this finding may not be applicable to the elbow joint, where the biomechanical
conditions are different, or to the particular design concept of the bipolar
radial head prosthesis. It is well established that polyethylene wear can lead
to osteolysis and bone loss as a result of a macrophage response to
particulate
debris28.
Osteolysis is often silent, and it is common for patients not to experience
pain until the bone loss is advanced, as was the case in five of our patients.
In our series, stem migration was observed only in the patients with obvious
balloon-shaped osteolysis and not in the patients with radiolucent lines or
radial neck resorption.
The results of this study confirm that, despite midterm satisfactory
clinical results with the bipolar radial head prosthesis, radiographic changes
occur over time, possibly reflecting osteolysis at the bone-cement interface
in a number of patients. It also appears that the fixed cemented stem of a
bipolar radial head prosthesis can loosen over time and that osteolysis
secondary to wear debris from the bipolar interface may be one of the
contributing factors.
Clinicians should be aware that one possible long-term drawback of use of
the bipolar radial head prosthesis is major loss of bone in the area of the
stem, which warrants caution regarding the use of this prosthesis in young and
active individuals. A longer duration of follow-up and studies of larger
series of patients will be required before definitive treatment
recommendations can be made. ?