Chronic rotator cuff tears are a frequent cause of morbidity in the adult
population. Surgical repair of chronic tears is indicated when conservative
treatment has failed to decrease the
symptoms1. Recent
studies have noted that large chronic cuff tears failed to heal after repair
in twenty-four of 100
shoulders2,
twenty-seven of fifty
shoulders3, and ten
of twenty-seven
shoulders4. Several
factors have been suggested to be responsible for this high failure rate.
These include the size of the
tear5,6,
time from the injury to the
repair7, tendon
quality8, muscle
quality9, biologic
healing
response10,11,
and surgical
technique12,13.
Clinically, chronic large rotator cuff tears are observed to have a
qualitatively shorter and stiffer muscle-tendon unit than normal. Changes to
the muscle, including
atrophy14,15,
fat
accumulation14-17,
and an increase in fibrous
tissue15,18,
have been reported following rotator cuff detachment in animal models. These
data suggest that the difficulty in obtaining an intact rotator cuff repair,
on the basis of imaging criteria, is related to a change in the quality and
quantity of the involved muscle.
We developed a chronic tear of the rotator cuff tendon in a canine model to
investigate and quantify the time-related changes in the passive mechanics,
volume, and fat of the infraspinatus muscle. Specifically, we hypothesized
that infraspinatus muscle stiffness would increase, volume would decrease, and
fat content would increase at twelve weeks following tendon detachment in the
canine model.
All procedures were performed in accordance with the standards of the
American Association for the Accreditation of Laboratory Animal Care. Eight
adult (one to three years old) mixed-breed dogs, including six females and two
males, weighing an average (and standard deviation) of 28.2 ± 2.8 kg
were used. To simulate a rotator cuff tendon tear, we surgically detached the
right infraspinatus tendon. To prevent early spontaneous reattachment of the
tendon and to facilitate tendon identification, the detached tendon was
wrapped with a polymer membrane. Postoperatively, the dogs were allowed free
cage activity. The uninvolved left shoulder served as a control. At twelve
weeks, the dogs underwent a second surgery and were immediately killed.
Changes in the passive mechanics, volume, and fat content of the detached and
control muscles were quantified.
First Surgery
The dogs initially were anesthetized with an intravenous dose of thiopental
(20 mg/kg) to effect. Benzathine penicillin and procaine (20,000 IU/kg) were
administered intramuscularly at induction. The dogs were intubated
orotracheally and maintained on isoflurane in oxygen, titrated to effect (0.5%
to 5%). In a sterile surgical field, a 7-cm-long incision was made between the
acromion medially and the greater tuberosity laterally through skin and
subcutaneous tissue. The interval between the two heads of the deltoid muscle
was developed, exposing the infraspinatus insertion into the proximal part of
the humerus. The tendon was then sharply detached from its insertion. The
tendon and the lateral portion of the muscle were freed from the surrounding
tissues. After release of the tendon, its edge retracted approximately 1 cm
(an average of 11 ± 1 mm) from the greater tuberosity. The free end of
the infraspinatus tendon was then wrapped with polyvinylidene fluoride
(Durapore 7 SVLP; Millipore, Bedford, Massachusetts) with a 125-µm
thickness and a pore size of 5 µm in the first three dogs and with PRECLUDE
membrane (W.L. Gore and Associates, Flagstaff, Arizona) in the last five dogs,
to prevent spontaneous reattachment and to facilitate the identification of
the tendons during the second surgery. We used two different membranes because
of the availability of these materials. The membranes were sutured to the
tendon with a 4-0 monofilament nonabsorbable suture. The wound was irrigated
and closed in two layers with use of 3-0 absorbable suture. A light
compressive bandage was applied. Postoperative analgesia consisted of
buprenorphine (0.3 mg) administered subcutaneously on the day of surgery and
the following day if indicated. Following surgery, the involved shoulders were
not immobilized and the dogs were allowed free cage activity.
Second Surgery
The second surgery was performed twelve weeks after the first surgery. The
dogs were anesthetized, and the shoulder was prepared in a sterile fashion.
The left (control) infraspinatus tendon was approached and detached from its
humeral insertion as described above. The right (tendon-released) shoulder was
approached with use of the previous incision, and the infraspinatus tendon,
wrapped with the membrane, was isolated. On both sides, the tendon and the
lateral portion of the infraspinatus muscle were freed from surrounding
tissues. The passive mechanical properties of the left and right infraspinatus
tendon-muscle units were then measured as described in detail below. The dogs
were administered intravenous pancuronium bromide (0.1 mg/kg) five minutes
prior to the mechanical tests to ensure complete muscle relaxation. The dogs
were then killed with an intravenous lethal injection of pentobarbital sodium
with phenytoin (Beuthansia-D; Schering-Plough Animal Health, Union, New
Jersey). The infraspinatus muscles of both shoulders were dissected for
analysis of their physical properties (volume, length, and weight) and fat
content.
Passive Mechanics
A custom-designed tissue-tension device was used intraoperatively to
stretch the infraspinatus muscle-tendon unit
(Fig. 1). The device consisted
of a large, threaded bore that translated linearly within a mating cylinder by
means of a manual crank and gear mechanism. The device was secured to the
humerus by a separate cleat attached to the bone with two 2.5-mm screws. A
ball-type joint in the device allowed its orientation to be further adjusted
along the line of action of the infraspinatus muscle during testing. The
device components were made of type-6061-T6 aluminum or type-316 stainless
steel and could be autoclaved. Linear digital calipers (Mitutoyo America,
Aurora, Illinois) mounted on a stage that resided outside the sterile field
were used to monitor displacement. The calipers interfaced with the
translating inner bore of the device by means of a steel cable. Load was
monitored by a 250-N (50-lb) load cell in series with the inner bore and
residing entirely within the device housing (Honeywell Sensotec, Columbus,
Ohio).
For tissue tension measurements, a number-1 Dacron suture was passed
through the infraspinatus tendon stump with use of a Krackow
stitch19. The free
ends of the sutures were then attached to the tension device. The device was
secured to the humerus and was manually cranked at a displacement rate of
approximately 0.4 mm/sec. Load and displacement were continuously monitored
throughout the test by a personal computer. Each infraspinatus muscle-tendon
unit was passively stretched laterally along its line of action, 15 to 20 mm
from the point at which it began resisting load. Tests were repeated three to
four times for each muscle.
Load and displacement data were processed with use of an algorithm to
iteratively fit the data from 2-mm displacement to the data point at which the
R2 value of a linear least-squares fit was maximized. The remaining
portion of the data was fit with a cubic equation, maintaining point and slope
continuity with the linear fit. The point bounding the linear and cubic fit
regions was designated as the inflection point, and the inflection load and
inflection displacement were defined at this point. The slope of the linear
fit was defined as stiffness-1, and the slope of the cubic fit at 15 mm was
defined as stiffness-2. The 15-mm point was used because we had recorded data
at 15 mm for all data sets. Load-displacement data were normalized by the
initial muscle physiological cross-sectional area and the initial muscle
length, respectively, to yield engineering stress-strain data. Stress-strain
data were analyzed in a similar manner to that used for the load-displacement
data, providing an inflection stress, inflection strain, modulus-1, and
modulus-2.
Volume Measures
The tendon-released shoulders were scanned by magnetic resonance imaging
just prior to the detachment surgery and every two weeks thereafter. A final
scan was performed just prior to the second surgery. The control shoulder was
scanned only prior to the detachment and the second surgery. The shoulders
were scanned with a 1.5-T magnetic resonance scanner with use of a T1 sequence
with 4-mm wide oblique sagittal views to facilitate the differentiation of the
infraspinatus muscle from the adjacent muscles. The dogs were positioned in a
lateral decubitus position with the scanned shoulder facing down in order to
minimize breathing artifacts. The entire infraspinatus muscle was scanned. The
magnetic resonance scans were analyzed with use of custom image-analysis
software. The perimeter of the infraspinatus muscle was traced in each image
slice by two independent observers, and the muscle cross-section area was
calculated. For each magnetic resonance scan sequence, a total muscle volume
was estimated by summing the slice cross-sectional areas and multiplying by
the slice width (4
mm)20. Total muscle
volume was calculated by averaging the muscle volume obtained by each
observer.
Fluid-Displacement Volume Measures
At the time that the animals were killed, the intact infraspinatus muscles
were dissected and submersed in a graduated cylinder containing saline
solution. Care was taken not to submerse the tendon stump. The volume change
was recorded as the muscle volume. Direct muscle volume measures were compared
with the volume results on magnetic resonance imaging and were used to
validate the magnetic resonance imaging method.
Muscle Fat Content
At the time that the animals were killed, 2-mm-thick transverse sections of
tissue were cut from the medial, central, and lateral regions of the harvested
infraspinatus muscles. Tissue sections were fixed in 10% neutral buffered
formalin for five days. The samples were then post-fixed with a solution of 2%
osmium tetroxide and 5% potassium dichromate for two
weeks14. They then
underwent copious water washes and were dehydrated through ascending series of
alcohols, embedded in paraffin, and cut into 5-µm sections. The sections
were viewed under a light microscope (Olympus BH-2; Olympus America, Melville,
New York) and digitized at four times magnification with use of a camera
(CoolSNAP-Pro 24-bit; Media Cybernetics, Silver Spring, Maryland) and frame
grabber (Prior ProScan; Prior Scientific, Rockland, Massachusetts). The
infraspinatus cross section was traced to identify a region of interest for
image analysis. With use of custom software, the cross-sectional area was
calculated and the area of fat, stained black, was segmented. The ratio of fat
to infraspinatus area constituted the percentage of intramuscular fat for each
muscle section.
Statistical Methods
For tissue tension data, differences between sides were assessed with use
of repeated-measures analysis of variance techniques to account for the
within-subject correlation. For all other parameters, where a single
measurement per subject was available for analysis, paired t tests or Wilcoxon
signed-rank tests were used, as appropriate, for comparisons of the paired
data. Results were considered significant if the p value was =0.05.
Physical Properties of the Infraspinatus Muscle
The physical properties of the control and detached infraspinatus muscles
are shown in Table I. Volume
decreased by a mean of 33% (p < 0.0001); weight, by 34% (p < 0.0001);
and muscle length, by 16% (p < 0.0001) in the detached muscle during the
twelve-week period. The average gap distance (and standard deviation) between
the detached tendon and the greater tuberosity was 34 ± 4 mm.
Passive Mechanics
Three distinct regions were noted in the passive load-displacement response
for both control and detached infraspinatus muscle-tendons: an initial linear
portion21; a second
linear portion of lesser slope than the first; and a nonlinear portion of
increasing slope21.
Because the initial linear portion of the data was somewhat variable among the
samples and was confined to the very early phase of muscle-loading (typically
contained within the first 1 to 2 mm of stretch), we chose not to include it
in our analysis. As described in the Materials and Methods section, only the
data after the 2-mm displacement point were fit with our algorithm.
Accordingly, the passive mechanical properties of the control and detached
muscles are displayed in Table
II, and representative load-displacement curves for both groups
are shown in Figure 2. The
inflection load was not different between the control and detached muscles (p
= 0.74); however, the inflection displacement significantly decreased from a
mean (and standard error) of 10.2 ± 0.33 mm in the control muscle to
6.8 ± 0.33 mm in the detached muscle (p < 0.0001). The chronically
detached infraspinatus was significantly stiffer than the control, with a
twofold increase in stiffness-1 (p < 0.0001) and a fourfold increase in
stiffness-2 (p < 0.0001). The results were similar for the normalized
stress-strain data sets. The inflection stress was not significantly different
between the control and detached muscles (p = 0.13); however, inflection
strain was significantly less in the detached infraspinatus (p = 0.0039). The
chronically detached muscles had significantly higher moduli than the
controls, with a twofold increase in modulus-1 (p < 0.0001) and a fourfold
increase in modulus-2 (p < 0.0001).
Volume Measures
The volume change as measured by magnetic resonance imaging for a
chronically detached infraspinatus muscle is shown in
Figure 3. Detached muscle
volume decreased by an average (and standard deviation) of 32.4% ± 4.6%
at six weeks (p < 0.0001) and by 31.3% ± 5.7% at twelve weeks (p
< 0.0001) relative to the initial muscle volume. No significant volume
change occurred in the detached infraspinatus between six and twelve weeks
after detachment (p = 0.43). No change in volume occurred in the control
muscle during the twelve-week period (p = 0.74). Agreement in the muscle
volume data from the two independent image analysts was high (correlation
coefficient = 0.84), with no significant difference between the data sets (p =
0.49). Furthermore, the magnetic resonance imaging data on the muscle volume
at twelve weeks were highly correlated to the data on the fluid-displacement
volume (Spearman correlation coefficient = 0.93).
Muscle Fat Content
The intramuscular fat content in the control and detached muscles is shown
in Figures 4 and
5. Intramuscular fat increased
significantly in the detached infraspinatus compared with that in the control
with respect to each muscle region as well as the average of all regions (p
< 0.05). The average fat content was 0.8% ± 0.2% in the control and
6.8% ± 1.5% in the detached muscles. There were also significant
differences in the fat content among the different regions of the detached
muscles (p < 0.02). The fat content was an average of 9.1% ± 1.9% in
the lateral regions, 7.0% ± 1.3% in the central regions, and 4.3%
± 2.4% in the medial regions. No regional differences in fat content
were found in the control infraspinatus muscles.
Our results demonstrated that the chronically detached infraspinatus muscle
changed significantly during the detachment period, becoming stiffer, smaller,
and infiltrated with fat. In addition, we showed that the shape of the passive
load-displacement curves of chronically detached muscles was qualitatively
similar to that of the controls and other normal
muscles22. However,
the curves were shifted "up and to the left," indicating a
decrease in the low stiffness region and an overall increase in stiffness of
the detached muscles. These results are similar to the increased passive
elastic stiffness that has been reported for muscles immobilized in shortened
positions23,24.
The fundamental consequence of this shift is that the detached muscle
generates higher loads than normal muscle does for the same amount of
stretch.
The chronically detached muscle is not merely a smaller version of the
original muscle but, rather, a different muscle. After the load-displacement
mechanical data were normalized by the muscle dimensions, the chronically
detached muscles were shown to have decreased inflection strain and increased
moduli compared with control (normal) infraspinatus muscles. In other words,
the material properties or quality of the detached muscles had changed. It is
important to note that canine infraspinatus tendon and number-1 Dacron suture
were mechanically tested in pilot studies and were shown to be two to three
orders of magnitude stiffer than the infraspinatus muscle. Hence, the passive
mechanical properties measured in the present study essentially reflect the
properties of the infraspinatus muscle. This approximation has been made in
previous studies of passive muscle
mechanics25.
Changes in the passive mechanics of lower limb muscles have been attributed
to two major factors: the connective tissue component within and surrounding
the muscle, which increases in volume during muscle
atrophy18,26,
and the noncontractile proteins of the sarcomeric
cytoskeleton25.
More specifically, titin, a giant protein associated with the myosin and actin
filaments in the sarcomere, has been implicated in playing a major role in
passive muscle
mechanics25,27.
In our animal model, whether the change in the muscle material properties
arises from a change in the connective tissue component or a change in other
muscle components, such as titin, is a subject for further research.
Furthermore, whether this muscle material change is reversible with simple
muscle reattachment should be investigated and has important clinical
implications.
Our study is the first, as far as we know, to report on changes in
longitudinal muscle volume over a chronic detachment period. Comparable
reductions in muscle volume have been demonstrated in lower-limb unloading
models, such as lower-limb suspension, muscle immobilization, and Achilles
tenotomy28-30.
Volume reduction was more pronounced in the antigravity muscles than in their
antagonists and varied in intensity between
species30. An
initial decrease in sarcomere length and subsequently in sarcomere number is
believed to be responsible for the observed reduction in muscle
volume18,30,31.
It is assumed that a reduction in sarcomere number is a remodeling response to
optimize the contractile function of the remaining sarcomeres in the shortened
muscles30.
The detached muscle reached a new steady-state volume at six weeks in our
animal model. However, it is important to note that at the time that the
animals were killed, we observed that the detached infraspinatus tendon
(together with the polymeric membrane) had scarred to the joint capsule.
Therefore, we cannot conclude that the muscle was fully unloaded during the
entire detachment period. Reformation of the distal muscle attachment to
surrounding connective tissues and subsequent reapplication of tension has
been shown to decrease muscle
atrophy30. Hence,
it is difficult to speculate whether muscle volume would have decreased
further in our model had the muscle been more fully unloaded throughout the
twelve-week period.
We found a significant increase in the fat content of chronically detached
muscles at twelve weeks. In a chronic supraspinatus model in rabbits,
intramuscular fat content increased from 1.3% in normal muscle to 5.4% in
detached muscle after twenty-four
weeks14; however,
no increase in fat was found at twelve
weeks16. In a
chronic infraspinatus model in sheep, fat content increased from 0.47% in
normal muscle to 5.9% at six weeks and 6.2% at eighteen weeks in the detached
muscle17. All of
these studies evaluated fat content at one location in the involved muscle.
Our study demonstrated that muscle fat changed nonuniformly in the detached
muscle, with the greatest changes occurring in the lateral portion.
Clinically, this means that muscle fat should be used cautiously as an
indicator of muscle change because the lateral muscle fat content does not
represent the entire muscle. Furthermore, subtle changes in the biopsy site
could yield significantly different fat percentages within the same
muscle.
We suggest that the repairability of a chronic rotator cuff tendon tear and
the likelihood of achieving a successful clinical outcome depend on the
passive loads generated during repair. Passive load is influenced not only by
the muscle stiffness but also by the distance required to reattach the tendon
to bone. We represented the loads required to repair the control and detached
infraspinatus muscle-tendons in our animal model by plotting the passive
muscle load as a function of the gap closure distance
(Fig. 6). We approximated 100%
of the gap distance as corresponding to a retraction distance of 10 mm in the
control group and 35 mm in the chronically detached group. These are distances
between the end of the tendon and the greater tuberosity and represent the
distance the tendon would need to be pulled, in the acute and chronic tear
conditions, to reach its attachment site at the time of rotator cuff repair.
One can see that the stretch required to repair the chronically detached
muscle would load the muscle into the high-stiffness portion of the loading
curve, even beyond the region that we achieved with our device. By
extrapolation, the passive loads required to repair the detached muscle-tendon
could reach at least 70 N. Together with any additional loads generated
postoperatively from passive shoulder motion and active muscle contraction,
these excessive loads could precipitate a failed repair.
In summary, we developed a canine model of a chronic rotator cuff tear to
investigate and quantify the changes in the passive mechanics, volume, and fat
of the detached infraspinatus muscle. We found that the infraspinatus muscle
becomes significantly stiffer, smaller, and infiltrated with fat within twelve
weeks after detachment in the canine model. We demonstrated that the
chronically detached muscle is not merely a smaller version of the original
muscle but, rather, a different muscle. The passive loads required to repair
the detached muscle can become quite large and may be important in determining
the ability to repair the tendon to its original site of attachment or for
that tendon to heal after repair.?
Note: The authors acknowledge the helpful contributions of David
Hicks, MD, Hakan Ilaslan, MD, and Jean Tkach, PhD.