Fractures of the radial head constitute about 30% of
all fractures around the elbow joint1,2.
Five to 10% of all elbow dislocations are associated with
periarticular fracture, most commonly of the radial head3-6. Previous biomechanical studies
have identified the medial collateral ligament as the primary constraint and
the radial head as a secondary constraint against valgus instability7,8. It is not always possible to treat
a comminuted fracture of the radial head with anatomic reduction
and stable fixation. The treatment of such fractures—that
is, whether resection or replacement should be performed—is
controversial. The complications after radial head resection are
well known9, and prosthetic replacement
is clearly indicated after radial head resection in the presence
of an injury of the medial collateral ligament or the interosseous
membrane2. With the increasing
recognition of "complex instability," the value
of the radial head in providing elbow stability is emerging10.
Unfortunately, no currently available silicon rubber or metal
monoblock radial head prostheses have proven to be entirely satisfactory.
The most common complications include loosening11-13 and
damage to the implant11,13-17.
In addition, implantation of a silicon rubber prosthesis has been
associated with osteoporosis of the capitellum and silicone synovitis11,16,18-21. Cadaveric studies have
also shown that silicon rubber implants are unable to adequately
resist valgus stress applied to the elbow joint12,18,22-24.
Recently, bipolar radial head prostheses have been introduced;
they are commercially available in a few countries and are used
in some medical centers in Europe. The rationale for the design
of these prostheses is that the additional freedom of movement may
reduce stress on the implant and at the implant-bone interface, which
could decrease the risk of implant loosening and capitellar wear25-28. Initial clinical results of
the use of the Judet prosthesis have been promising26,29. It has a long stem (5.5 cm),
a neck-shaft angle of 15°, and an arc of 35° of angular movement
(Fig. 1).
One of us (S.P.) and colleagues introduced a bipolar radial head
prosthesis with a short (2-cm) straight stem and an arc of 30° of
angular movement (Fig. 1)27 and
performed a preliminary evaluation of stress distribution with finite
element analysis28.
As the head of a bipolar implant is mobile, the extent to which
such a prosthesis can contribute to stability of the elbow is not
clear, and we are not aware of any biomechanical studies evaluating
the contribution of bipolar implants to elbow stability. We hypothesized
that a bipolar implant can be as effective as a monoblock radial head
prosthesis in restoring valgus stability of the elbow after injury
to the medial collateral ligament.
Nine fresh-frozen upper extremities from cadavera with no evidence
of pathological changes at the elbow were used for this study. The
specimens included six right limbs and three left limbs. Four were
from female donors, and five were from male donors. The median age
of the donors at the time of death was seventy-nine years (range,
fifty-eight to ninety-one years). The specimens were stored at —20°C
and then were thawed overnight before testing.
The humerus was transected in its midportion, and the wrist was
disarticulated with care taken to preserve the ligaments of the
distal radioulnar joint. Soft tissues were removed from the proximal
10 cm of the humerus to allow the shaft to be cemented into an acrylic
tube. Nylon lines were attached to the biceps, brachialis, and triceps
tendons to allow the application of simulated muscle-loading, and
the acrylic tube was mounted in a testing fixture with the humerus
parallel to the floor. The entire testing fixture could be rotated
to allow the elbow to flex in either a horizontal or a vertical
plane8,30.
The three-dimensional spatial orientation of the ulna relative
to the humerus was measured with use of an electromagnetic tracking
device (3Space Fastrak; Polhemus, Colchester, Vermont) sampling
at a rate of 30 Hz30,31. This
system is accurate to 0.5° for angular rotation8.
Although the accuracy of the electromagnetic tracking device is
affected by certain metals, orthopaedic implants are not ferromagnetic
and thus are relatively transparent to this device32.
To maintain different degrees of forearm supination and pronation,
a custom supination/pronation holder was mounted on two
fiberglass rods, which were fixed with surgical cement into the
medullary canals of the distal parts of the ulna and radius (Fig. 2-A). The device
was adjustable to accommodate different specimen sizes. A ball-and-socket
junction connecting the rod in the radius to a rotating scale allowed
the radius to rotate freely. This additional freedom of movement
was necessary because the rotation of the radius about the ulna
does not follow a perfectly circular path33.
The desired rotation of the forearm was then fixed with use of a
set-screw when the pointer showed 0° or 80° of forearm rotation
in pronation or supination.
Passive motion of the elbow from full flexion to extension was
performed with the forearm in the vertical plane (neutral orientation)
as a reference motion pattern and then in the horizontal plane with
the elbow subjected to valgus and varus stresses, respectively (Fig. 2-B)30,34.
Tests were performed with 20, 20, and 40 N applied to the biceps,
brachialis, and triceps tendons, respectively. These loading conditions
allow maintenance of optimum elbow tracking as dictated by osseous
and soft-tissue constraints30,35.
The total valgus-varus laxity of the elbow joint at any given
flexion angle was calculated as the difference between the valgus
and varus angulations with the application of valgus and varus stresses,
respectively, as was previously described by King et al.23. These data were recorded throughout
the range of elbow flexion. Elbow flexion was repeated three times
at each forearm rotation. The test procedure measured both varus
and valgus laxity, and the difference in the valgus angulation from
the varus position was the measurement parameter. Since the varus
displacement from the intact state varied little (<2° on
average), the full varus position was the starting position and
the data are described as "valgus laxity" as this
is the most meaningful way to express the data for the clinician.
The valgus elbow laxity during elbow flexion was measured under
the following conditions:
1. In the intact elbow.
2. After the surgical approach (lateral epicondylar osteotomy
of the distal part of the humerus). A skin incision was made over
the lateral epicondyle. To allow for access to the radial head,
a lateral epicondylar osteotomy was performed with an oscillating
saw. Next, the lateral epicondyle was reattached with two 4-mm cancellous
screws placed into the previously drilled holes filled with surgical
cement. We performed this approach in order to preserve the integrity
of the lateral collateral ligament and the anterolateral aspect
of the joint capsule.
3. After release of the anterior bundle of the medial collateral
ligament8.
4. After release of the anterior bundle of the medial collateral
ligament and resection of the radial head. We used the previously
described surgical approach to resect the radial head at its junction
with the radial neck.
5. After release of the anterior bundle of the medial collateral
ligament, resection of the radial head, and replacement of the radial
head with one of three different prostheses (Fig. 1) implanted in
varying order into the same cadaver elbow. The prostheses included
a Wright monoblock titanium implant (Wright Medical Technology,
Arlington, Tennessee), a KPS (Kedzior/Pomianowski/Skalski) bipolar
Vitallium (cobalt-chromium)-polyethylene implant (Technika Medyczna,
Warsaw, Poland), and a Judet bipolar Vitallium-polyethylene-Vitallium
implant (Tornier SA, Saint-Ismier, France). Each type
of prosthesis was implanted three times, as a first, second, and third
implant. We used the previously described surgical approach to restore
the continuity of the lateral column of the elbow joint after each
implantation.
The diameter of the prosthesis was matched as closely as possible
to that of the removed radial head. The level of resection of the
radial head was determined according to the height of the proximal
part of the prosthesis, which protrudes above the resected neck
of the radius. In order to maintain the original radial length among the
different implants, spacers made of polyethylene or of diaphyseal
bone (the previously resected neck of the radius), equal to the
change in length with the particular design, were wedged under the
proximal part of the prosthesis when necessary.
All implants were designed for implantation with cement. Cement
was applied only to the proximal part of the stem to make it possible
to remove the implant without any damage. This amount of cement
was sufficient to maintain the proper position of the stem within
the medullary canal. After implantation, an image intensifier was
used to confirm correct alignment and seating of each prosthesis.
The specimens were kept moist with a physiological saline solution
during the experiments. After completion of the kinematic study,
the elbow joint was disarticulated to digitize the osseous landmarks
and the articular surface geometry with use of a sensor with a calibrated probe
attached. The anatomic sites that were digitized included the trochlea,
capitellum, greater sigmoid notch, humeral shaft around its circumference
above the level of the metaphysis, and distal part of the ulna (the distal
point of the rod of the supination/pronation holder). Data
obtained from the electromagnetic tracking device were used to measure
the three-dimensional spatial orientation of the ulna relative to
the humerus and were analyzed with use of the Euler angle description30,31.
Angular deviations of the ulna from the optimum tracking position
were calculated through the arc of the elbow flexion from 130° to
17° in 1° intervals. The range of elbow flexion was truncated at
17° because six specimens had a mild flexion contracture (ranging
from 5° to 16°), although none had evidence of arthritis or other pathological
change.
Data are shown as the mean and one standard deviation in degrees.
In the first analysis, the valgus elbow laxity values were compared
among all elbow conditions, three forearm rotation angles (80° of
supination, neutral rotation, and 80° of pronation), and three flexion
angles (20°, 60°, and 100°) with use of three-factor repeated-measures
analysis of variance. Valgus displacement was measured throughout
the arc of elbow flexion, but the values at 20°, 60°, and 100° are
reported to simplify data interpretation. Significant main effects
were further analyzed with use of the Student-Newman-Keuls multiple-comparisons
procedure. Additional analysis was undertaken to more completely
examine the effects of the rotation angle and the radial head prosthesis.
Specifically, the experimental groups were compared at each of the
three forearm rotation angles with use of one-factor analysis of
variance models with repeated measures. In this way, valgus laxity
values in the presence of deficiency of the medial collateral ligament
were compared among the specimens with an intact radial head, those with
each of the three radial head prostheses, and those without a radial
head. Again, significant main effects were further analyzed with
use of the Student-Newman-Keuls multiple-comparisons procedure.
All statistical tests were two-sided, and the threshold of significance was
set at a = 0.05. All analysis was performed with use of
SAS software (SAS Institute, Cary, North Carolina).
No difference in the mean valgus laxity was identified between
the intact elbows (3.4° 1.6°) and those in which the surgical approach
had been carried out (3.8° 1.8°). As expected, the mean valgus
laxity changed significantly (p < 0.001) as a factor of
constraint alteration. The greatest mean laxity was observed after
the medial collateral ligament release and radial head resection
(11.1° 5.6°). Less laxity was seen following medial collateral
ligament release alone (6.8° 3.4°), and the least was seen in the
intact elbows (3.4° 1.6°). A significant difference in the mean
valgus laxity was identified among the medial collateral ligament-deficient
elbows with an intact radial head, those with an implant, and those
without a radial head across all analyzed forearm rotation and flexion
angles (p < 0.001). All implants provided some stability,
but none restored stability in the medial collateral ligament-deficient elbows
to the same degree as was provided by the native radial head.
Significant differences in valgus laxity among the three forearm
rotations were found in all of the medial collateral ligament-deficient
elbows (p = 0.003). Post hoc multiple-comparisons
testing showed that the mean valgus laxity values in 80° of pronation
were significantly greater than those in neutral rotation or in
80° of supination (Table I) under all of the conditions except
the Judet prosthesis at 100° of flexion. In 80° of supination, the
mean valgus laxity values associated with each of the implants were significantly
greater than those for the elbows with an intact radial head and
significantly less than those for the elbows without a radial head
(p < 0.001). In neutral forearm rotation, the mean valgus
laxity values associated with each of the implants were significantly
greater than those for the elbows with an intact radial head. Also,
the mean valgus laxity values for the elbows without a radial head
were significantly greater than those for the elbows with an intact
radial head, a Wright implant, or a KPS implant (p < 0.001).
However, there was no significant difference in laxity between the elbows
without a radial head and those with a Judet implant. In 80° of
pronation, the mean valgus laxity values associated with each of
the implants and with the elbows without a radial head were significantly
greater than those for the elbows with an intact radial head (p = 0.002).
There was no significant difference in valgus laxity between the
elbows without a radial head and those with the implants.
The valgus laxity values did not differ significantly across
elbow flexion angles (p = 0.26), but there was a tendency
for laxity to increase at lower angles of flexion (Table I).
This study demonstrated the effect of bipolar radial head prostheses
on the valgus stability of the elbow compared with that of a monoblock
design. We confirmed that the experimental surgical approach did
not produce changes in laxity, and we demonstrated a significant
increase in laxity after release of the medial collateral ligament
and subsequent resection of the radial head.
In addition, forearm rotation was observed to have a significant
effect on valgus elbow laxity. Valgus laxity was always
greatest in pronation and least in supination. This finding was
demonstrated in a previous study from our laboratory36. In the pronated position, in which
the ulna is internally rotated, the decreased contact on the medial
aspect of the trochlea permits increased valgus laxity by decreasing
the contribution of the osteoarticular geometry to valgus stability.
Since the anterior bundle of the medial collateral ligament limits
internal rotation of the ulna during forearm pronation, release
of this bundle results in loss of this constraint, thereby allowing
increased internal ulnohumeral rotation and consequent increased valgus
laxity of the elbow. Conversely, external rotation of the ulna tends
to lock the ulnohumeral joint, providing greater stability during
supination.
None of the implants restored valgus stability of the medial
collateral ligament-deficient elbow to the same degree
as that found in the elbows with a native radial head. This was
probably due to the inability to fully replicate the physiological
shape and size of the original radial head.
In 80° of forearm supination, all implants significantly improved
valgus stability but did not provide the same stability as was provided
by the native radial head. In contrast, in 80° of forearm pronation,
no implant significantly improved valgus stability. In neutral forearm rotation,
valgus laxity was greater with the Judet implant than it was with
the Wright or KPS implant, although this difference was not significant.
This rotation-dependence of the Judet prosthesis might be due to the
angled neck design. The fact that the KPS bipolar implant and the
Wright monoblock prosthesis provided equivalent degrees of elbow
stability suggests that a bipolar implant may be as effective as
a solid prosthesis in restoring stability in clinical practice,
although extrapolation from this in vitro study
to the clinical situation must be made with caution. Also, the additional
freedom of movement of a bipolar prosthesis may reduce stress on
the implant and at the implant-bone interface, thereby decreasing
the risk of loosening of the implant as well as decreasing wear
on the capitellum.
In conclusion, all of the implants provided some stability in
the medial collateral ligament-deficient elbows. None of the prostheses
restored stability to the same degree as was provided by the native
radial head, particularly in forearm pronation. There were some
differences in the valgus stability provided by the different implants,
particularly in neutral forearm rotation. The inability of any prosthesis
to function as well as a native radial head suggests that open reduction
and internal fixation to restore radial head anatomy is preferable
to replacement when possible.