Ten fresh-frozen upper extremities from cadavera with no evidence of
pathological changes at the elbow were used for this study; two were used for
a pilot evaluation and eight, for measurements. One of the latter eight
specimens was excluded because of an iatrogenic lesion of the lateral
collateral ligament during an osteotomy of the lateral epicondyle. Therefore,
a total of seven specimens (two right and five left extremities) from five
donors were available for this experiment. The specimens were from two male
and three female donors with a mean age of eighty-two years (range,
seventy-seven to eighty-eight years) at the time of death. The specimens were
stored at —20°C and thawed overnight before use. The limbs were
amputated through the proximal third of the humerus and disarticulated at the
wrist with preservation of the triangular fibrocartilaginous complex. The
soft-tissue envelope of the arm was left intact as much as possible during the
study. The upper arm was fixed to a specifically designed frame with use of a
large AO external fixator with two bicortical 4.5-mm Steinmann pins placed
through the humeral shaft (Fig.
1). A 4.5-mm-diameter rod was introduced into the medullary canal
of the ulna through its distal end. This rod was secured to the ulnar shaft
with a small AO external fixator. The intramedullary rod of the ulna was laid
onto a horizontally positioned, transverse metal bar allowing unrestricted
varus-valgus and rotational displacement of the forearm. Rotation between the
radius and ulna was also unrestricted throughout the measurements. While the
ulna was kept horizontal, various degrees of elbow flexion were obtained by
tilting the external fixator attached to the frame. Positioning the ulna above
the humerus had the advantage that the weight of the forearm kept the
ulnohumeral joint reduced when no valgus-varus or rotational load was applied,
even if the elbow was unstable, as it was after ligament detachment or radial
head resection.
Valgus laxity was determined by measuring the valgus angular displacement
of the ulna. An arbitrarily chosen valgus torque of 0.825 N-m was applied to
the ulna by attaching a weight of 2.5 N to the intramedullary rod of the ulna
at a distance of 33 cm from the axis of elbow flexion. To eliminate friction
between the horizontal bar and the intramedullary rod, the rod was manually
elevated a few millimeters from the horizontal bar after application of the
valgus torque and then was released. Graduations, in degrees, marked on the
metal bar allowed a direct reading of the angle of valgus angular displacement
of the ulna at the position of the intramedullary rod. The orientation of the
ulna of the intact, unloaded elbow was defined as neutral valgus-varus
angulation and neutral ulnar rotation. Valgus laxity was defined as the
difference between the valgus angular displacement in the unloaded (neutral)
position and that in the loaded position (0.825 N-m).
To test for posterolateral rotatory laxity of the elbow, a constant valgus
torque of 0.825 N-m and an additional supinating torque of 0.75 N-m were
applied to the intramedullary rod of the ulna. Posterolateral rotatory laxity
was defined as the difference between the rotatory displacement of the ulna in
the unloaded (neutral) rotation and that in the loaded rotation measured after
application of both valgus and supinating torques. Application of both valgus
and supinating torques was considered to replicate the clinical situation of
posterolateral rotatory displacement better than application of a supinating
torque alone. The amount of supinating torque was chosen according to the
conclusions of previous investigators, who considered this amount to be
sufficient to demonstrate laxity and not so forceful as to cause permanent
changes in the stabilizing
structures7-9.
The supinating torque was manually applied and was controlled by an instrument
torque wrench10. To
measure the posterolateral rotatory displacement of the ulna, a three-turn
precision potentiometer (type 3507S; Bourns, Riverside, California) calibrated
against a workshop goniometer (Kunkel, Aschaffenburg, Germany) was attached to
the end of the intramedullary rod of the ulna. The potentiometer was kept
vertical in space with a plumb weight of 0.5 N fixed with a cord to its
housing to prevent rotation. The error of the potentiometer was
±0.5°, as determined in a calibration test of twenty repeated
measurements. This was checked with a precision manual goniometer (Kunkel).
The hysteresis due to friction within the goniometer was found to be
<1°. The reproducibility of the rotatory measurements was determined by
repeating ten measurements on one pilot specimen. The mean difference of these
rotatory measurements was found to be 0.7° ± 0.7°.
For the measurements of valgus and posterolateral rotatory displacements
through the range of motion, three different angles of elbow flexion were
arbitrarily chosen: 10°, representing elbow extension; 60°,
representing a midrange position; and 110°, representing a flexed
position. All measurements were performed at these three angles of flexion.
Once the upper arms were fixed to the testing apparatus, all measurements and
surgical procedures were performed without removal of the specimens from the
device. This allowed for repeated-measures analysis to be performed on the
same specimens.
To reduce the ulnohumeral joint and to restore the original neutral
position of the ulna after valgus and rotatory tests, the olecranon was
manually pressed to the trochlea of the humerus. The ulnae were reduced to
within approximately ±1° of their original neutral position as
determined by ten repeated measurements performed on one specimen.
Surgical Approach
A standardized surgical approach was designed to gain access to the
coronoid process and radial head for this experiment. The approach consisted
of two osteotomies—one of the lateral epicondyle and one of the ulnar
insertion of the lateral ulnar collateral ligament—performed with an
oscillating saw (Fig. 2). Prior
to the osteotomy, three 3.5-mm AO screws were placed through the lateral
epicondyle and two 3.5-mm screws were placed through the ulna at the lateral
ligament insertion. This allowed for later reattachment of the lateral
ligament complex in an anatomical and reproducible manner. The length of the
proximal osteotomized fragment was approximately 7 cm, including the lateral
epicondyle and about 5 cm of the supracondylar ridge of the humerus. The
osteotomy was performed through the lateral part of the capitellum to avoid
damage to the ligament. At the level of the lateral epicondyle, the osteotomy
was 0.8 to 1.5 cm thick. The osteotomy at the ulna was marked with two
1-mm-diameter Kirschner wires, which were inserted from the posterior cortex
of the ulna along the inside of the lateral cortex toward the lesser sigmoid
notch. For better visualization, the anconeus muscle was removed. The
osteotomized ulnar fragment was 3.5 cm long and 5 mm thick, and it included a
small lateral part of the lesser sigmoid notch with the annular ligament and
also included the supinator crest with the insertion of the lateral ulnar
collateral ligament. This experimental lateral approach also included
capsulotomy of the anterolateral and posterolateral capsule as well as
transection of the fascial band of the extensor carpi
ulnaris11. These
structures were not repaired for the tests.
Testing Protocol
Starting with the intact specimens, valgus torque was first applied and
valgus angular displacement was measured at the three angles of flexion
(10°, 60°, and 110°). Then the supinating torque was applied in
addition, and the posterolateral rotational displacement was measured. The
same measurements were repeated following the surgical approach and
reattachment of the lateral ligament complex. This condition was termed the
testing model, and all other conditions were compared with this
condition.
The lateral ulnar collateral ligament was next detached by removing the
screws at the osteotomized ulnar bone fragment. The measurements of the
displacements were repeated. All subsequent conditions were tested with the
lateral ulnar collateral ligament both intact and detached.
The displacements were subsequently measured after radial head resection
and then again after resection of 30% of the coronoid process with an
oscillating saw. The plane of coronoid resection was parallel to the
longitudinal axis of the shaft of the ulna and parallel to the axis of
rotation of the elbow joint. To control the plane and height of coronoid
resection, a small external fixator was applied to the ulna and a 1-mm-thick
Kirschner wire was positioned at the level of the resection. The height of the
resection was determined with calipers.
Next, the condition with a radial head defect and resection of 30% of the
coronoid was stabilized, first with a rigid radial head implant and then with
a floating radial head implant. The same measurements were repeated after
resection of 50% and 70% of the coronoid with and without radial head
replacement. The rigid prosthesis was an rHead radial implant system (Avanta
Orthopaedics, San Diego, California) (Fig.
3), which is a metal implant that is available with radial head
diameters of 18, 20, and 22 mm and can be cemented or press-fit without
cementing. The implant that most closely replicated the anatomical diameter of
the resected radial head was used. An implant with a diameter that was larger
than the diameter of the original radial head would probably have overstuffed
the joint. This would have improved stability but also deviated and
lateralized the axis of rotation of the radius, resulting in nonanatomical
kinematics. Similarly, a radial head prosthesis smaller than the intact radial
head would have reduced the tension of the annular ligament that encompasses
the radial head and probably reduced stability. The average diameter (and
standard deviation) of the resected radial heads in this experiment was 22.7
± 1.5 mm. Three 20-mm and four 22-mm rigid radial head implants were
used.
The floating implant was a radial head prosthesis with a floating cup
(Tornier-Zirst, Montbonnot, France) (Fig.
3). This implant consists of a cemented intramedullary stem and a
radial head made of ultra-high molecular weight polyethylene enclosed in a
cobalt-chromium-alloy cup, which articulates with the spherical end of the
stem. The head is available in diameters of 19 mm (used three times in this
experiment) and 22 mm (used four times).
Both the rigid and the floating devices were inserted according to the
instructions of the manufacturers, except for the level of resection for the
rigid prosthesis: 2 mm of the proximal part of the shaft was resected in
addition to the length shown by the resection guide to exactly resect the
amount of bone corresponding to the length of the implant. The level of radial
resection was not the same for both prostheses: it was more proximal for the
rigid device and more distal for the floating device. With use of reamers, it
was possible to insert the collar of the floating prosthesis slightly into the
proximal stump of the radius to obtain a correct radial length without needing
to further resect the proximal radial stump. The movement of the floating head
was not restricted by the protruding cortex of the radial neck. Both the rigid
and the floating prostheses were fixed with a small amount of cement, ensuring
stability and allowing easy removal.
The condition of a radial head defect with 70% of the coronoid resected was
addressed with two different techniques of reconstruction of the coronoid:
first, a coronoid allograft was used (Fig.
4), and second, the tip of the olecranon was transferred to the
deficient coronoid with the technique described by Moritomo et al.
(Fig.
5)12. A
right or left coronoid allograft was used, with the left graft measuring 13 mm
in height and 25 mm in medial-lateral width and the right graft measuring 13
mm in height and 23 mm in medial-lateral width. The grafts were fixed with two
3.5-mm cannulated AO screws. Measurements were again carried out with and
without radial head implants.
After the measurements were completed, the elbows were dissected and the
integrity of the lateral collateral and medial collateral ligaments was
evaluated.
Statistical Analysis
Repeated-measures analysis of variance with the post hoc Bonferroni-Dunn
test was used to test for differences between the laxities of the elbows under
the various test conditions. As a result of the Bonferroni correction factor,
significance of the differences was indicated by a p value of <0.005 or
<0.0083 depending on the groups that were compared.
Laxity of the Normal Elbow and After the Surgical Approach
The posterolateral rotatory laxities were greater than the valgus laxities
because of the rotatory torque applied in addition to the valgus torque
(Table I). Therefore, the main
interest was focused on the posterolateral rotatory laxity, and all subsequent
data refer to this laxity only.
The posterolateral rotatory laxity of the intact elbow averaged 5.4° at
60° of elbow flexion (Fig.
6 and Table I).
Under the various conditions tested, the laxities were usually greatest at
60° of flexion and were rarely greatest at 110° of flexion. Hence, the
results at 60° of elbow flexion are presented.
Release of the lateral ulnar collateral ligament caused significant
posterolateral rotatory laxity (mean, 32.7°; p < 0.0001 compared with
the intact condition), with ulnohumeral dislocation occurring in four of the
seven elbows. Reattachment of the lateral ulnar collateral ligament restored
almost normal stability, with only a slight, insignificant increase in the
mean posterolateral rotatory laxity, to 9.0° (p = 0.24 compared with the
intact condition). This condition with the ligaments reattached was the
"testing model," and all subsequent surgical modifications are
compared with this condition.
Laxity After Resection of the Radial Head and Coronoid Process and
After Stabilization with Radial Head Prostheses
The laxities of the elbows with defects of the radial head and the coronoid
process and after the various stabilizing procedures are listed in the
Appendix. Resection of the radial head significantly increased the
posterolateral rotatory laxity to a mean of 18.6° despite intact
ligaments. Additional removal of 30% of the coronoid fully destabilized the
elbows, always resulting in complete ulnohumeral dislocation at 60° of
elbow flexion. Implantation of a rigid radial head prosthesis restored
stability (mean laxity, 11.4° compared with 9° in the testing model; p
= 0.18). However, a significant mean laxity of 16.9° persisted after the
floating radial head prosthesis was inserted (p < 0.0001). Still, none of
the latter specimens dislocated, indicating that the floating prosthesis also
had a stabilizing effect.
A defect of 50% of the coronoid process combined with a radial head defect
and intact ligaments could not be fully stabilized by replacement alone with
either the rigid or the floating radial head implant. The mean laxity was
16° with the rigid prosthesis (p = 0.0005) and 20.1° with the floating
prosthesis (p = 0.0001). Substantial mean laxities of 22° and 26°
persisted in the elbows with a defect of 70% of the coronoid process and
intact ligaments despite insertion of a radial head prosthesis.
Reconstruction of the Coronoid Process
Reconstruction of the coronoid with an allograft and radial head
replacement restored stability of the elbow. Coronoid reconstruction with the
tip of the olecranon and replacement with a rigid radial head implant resulted
in an increase of laxity of 1.1° compared with the testing model, whereas
use of the floating device resulted in an increase of 4.3°; neither
increase was significant.
Additional Observations
Resection of 30%, 50%, and 70% of the coronoid process resulted in
detachment of the annular ligament at the base of the coronoid of
approximately 50%, 70%, and 90%, respectively. No attempt was made to reinsert
the annular ligament during this experiment.
Manual testing of the medial collateral ligament by applying a valgus
stress did not reveal any opening of the medial joint space. Additional
dissection after the measurements revealed a mainly intact anterior bundle of
the medial collateral ligament despite partial coronoid resection.
In this in vitro study, we investigated the role of the coronoid process
and the radial head as posterolateral rotatory stabilizers and analyzed the
stabilizing effect of two different radial head implants and two different
techniques of coronoid reconstruction.
Isolated removal of the radial head in elbows with intact collateral
ligaments resulted in a significant increase in the posterolateral rotatory
laxity of 9.6° compared with that in the test condition, confirming
similar findings in the study by Jensen et
al.7. Healing of the
periarticular soft tissues such as the capsule may improve stability in such
patients and may explain why they usually do not have chronic symptoms of
instability.
A combined resection of 30% of the coronoid and of the radial head resulted
in a substantially lax condition despite intact collateral ligaments. These in
vitro results reflect the severity of the injury, termed the terrible
triad. Use of a rigid radial head implant restored almost normal
stability in vitro, but bipolar radial head replacement did not restore
stability to the same extent. However, in the current study, it was not
possible to determine the clinical importance of the difference in laxity
between the rigid implant and the testing model (11.4° — 9.0° =
2.4°) or between the floating implant and the testing model (16.9°
— 9.0° = 7.9°). Similar findings were reported in another study,
in which the stabilizing effect of three different radial head prostheses was
analyzed in cadaver elbows with a deficient medial collateral
ligament13. In that
study, the floating implant was associated with a significantly greater valgus
laxity in neutral forearm rotation than were the other two
devices13.
Elbows with a combined resection of the radial head and of 50% or 70% of
the coronoid process could not be fully stabilized in the laboratory by
insertion of a radial head implant alone. Significant laxities of 7° to
17° more than the test condition persisted, with ulnohumeral dislocations
in some elbows, especially those with a defect of 70% of the coronoid process.
Again, it was not possible to determine in this study whether these laxities
are clinically important or to determine the critical size of coronoid defect
that requires reconstruction. Clinical experience suggests that a defect
involving 50% of the coronoid can cause instability and one involving 70%
virtually always causes significant
instability14. In
our experiments, it was possible to restore stability in such elbows by
reconstructing the coronoid and inserting a radial head prosthesis.
This study focused on the role of the radial head and the coronoid as
posterolateral rotatory stabilizers and less on the importance of the
ligaments. The role of the radial collateral ligament has been analyzed
previously1,2,8,9,11,15-18.
The role of the medial collateral ligament combined with coronoid and/or
radial head defects was not addressed in our study. It is very probable that
medial or lateral ligament laxity aggravates elbow instability caused by
radial head and coronoid defects. Additional studies are therefore necessary
to determine the role of the medial collateral ligament in elbows with radial
head and coronoid defects.
Limitations of this study include the small number of specimens tested.
However, the repeated-measures design resulted in a high degree of
reproducibility of the measurements. Although there was some interspecimen
variation, the testing and comparison of the mechanical characteristics were
mainly intraspecimen. Once a humerus was mounted to the testing device, it was
not removed, so its position remained identical for testing of the intact
elbow and for all subsequent testing. Another limitation of the study was that
it was in vitro. The findings are not completely applicable to the clinical
situation, in which soft-tissue healing, particularly of the capsule and the
periarticular structures, occurs. The laxities measured in this experiment may
therefore be greater than those in clinical practice. The current model
included release of the insertion of the lateral ulnar collateral ligament.
This was achieved by osteotomy of an osseous fragment at the crista
supinatoris, including the annular ligament. This might differ from clinical
situations of chronic posterolateral rotatory instability, in which the
annular ligament might be intact. Finally, only one type of rigid metal radial
head prosthesis was tested, even though other rigid models are currently
available19.
Additional investigations are necessary to validate the results of this study
for clinical application.
A table showing the elbow laxities found at the various test conditions is
available with the electronic versions of this article, on our web site at
(go to the article citation and click on "Supplementary Material")
and on our quarterly CD-ROM (call our subscription department, at
781-449-9780, to order the CD-ROM).
Note: The authors acknowledge T.F. Holovacs, MD, for helping to
edit the manuscript, H.R. Sommer for technical support, and B. Seifert, PhD,
Department of Biostatistics, Institute of Social and Preventive Medicine,
University of Zurich, for help with the statistical analysis.