Quantitative Morphology of Full ThicknessRotator Cuff Tears
JASON D. WENING, RONALD F. HOLLIS, RICHARD E. HUGHES,* AND JOHN E. KUHN
University of Michigan, Orthopaedic Research Lab, Ann Arbor, Michigan
The occurrence of full thickness rotator cuff tears (RCTs) at time of death, the size anddistribution of those tears among the tendons of the rotator cuff, and the anterior to posteriorand medial to lateral dimensions of the RCTs were determined. A subset of 57 shoulders froma group of 414 were found to have full thickness rotator cuff tears. The three-dimensionalcoordinates of specific bony landmarks and points defining the circumference of the RCTwere digitized using a Flock of Birds DC electromagnetic tracking device. Bony landmarkswere used to determine the separation between adjacent tendons of the rotator cuff. Pointsdescribing the circumference of the RCT were used to calculate the area of the tear. Themajority of tears occurred in the supraspinatus tendon alone (25), or in both the supraspinatusand infraspinatus tendons (22). The areas of the RCTs ranged from 0.07 cm2 to 19.17 cm2with an average of 4.43 cm2. The mean anterior to posterior length was 1.95 cm. The meanmedial to lateral length was 1.98 cm. Clin. Anat. 15:1822, 2002. 2002 Wiley-Liss, Inc.
Key words: rotator cuff tendons; prevalence; supraspinatus; infraspinatus; sub-scapularis
Several studies have reported the prevalence of fullthickness rotator cuff tears (RCTs) in the generalpopulation at the time of death (Bigliani et al., 1986;Hijioka et al., 1993; Keyes, 1933; Lehman et al., 1995;Ogata and Uhthoff, 1990; Wilson and Duff, 1943;Yamanaka et al., 1983; Zuckerman et al., 1992). A fewauthors have reported qualitative assessments of thedistribution of full thickness tears among the rotatorcuff tendons (Itoi et al., 1995). Others have reportedqualitative assessments of RCT size using the Bate-man scale (Tempelhof et al., 1999). Sher et al. (1995)reported quantitative measures of RCTs in asymp-tomatic individuals based on MR images. Althoughgrading tears according to small, medium, large, andmassive sizes is clinically useful, new advances incomputational modeling of the rotator cuff (Chang etal., 2000; Hughes and An, 1997; Hughes et al., 1999)create a need for more accurate quantitative measuresof tear size and percentage of each tendon involved ina tear. Shoulder strength in normal subjects has beenpredicted from geometric and physiological data(Hughes et al., 1999), but prediction of strength def-icits in a population of shoulders with RCTs requiresinformation about the statistical distribution of cuff
tear morphology. Obtaining the most accurate mea-surements requires digital collection of data from care-fully dissected specimens and computational estima-tion of the area of each tear.
The purpose of this study was to accurately deter-mine the area of full thickness RCTs and to deter-mine the portion of the tear included in each of therotator cuff tendons. Descriptive statistics for theprevalence of full thickness RCTs at the time of deathbased on age, sex, and side of body are presented. Thedimensions of tears in the supraspinatus tendon weremeasured in the anterior-to-posterior and medial-to-lateral directions.
MATERIALS AND METHODS
Between May, 1994, and March, 1999, a convenientsample of 414 cadaver shoulders were collected from apopulation of men and women who donated theirbodies to the University of Michigan Anatomical Do-
*Correspondence to: Richard Hughes, Ph.D., University of Mich-igan, Orthopaedic Research Lab, 400 North Ingalls BuildingG-161, Ann Arbor, MI 48109-0486.
Received 8 May 2000; Revised 6 October 2000
Clinical Anatomy 15:1822 (2002)
2002 Wiley-Liss, Inc.DOI 10.1002/ca.1086
nations Program. Since the program provides tissue tomultiple institutions, it was impossible to collect thespecimens in any patterned or randomized fashion.Institutional Review Board approval was obtained forthis study. Of the shoulders collected, 357 specimens,ages 5386 (mean 70), had no signs of full thicknessRCTs. No information regarding the subjects symp-toms was available. Data were collected from theremaining 57 fresh frozen cadaver shoulders, ages6194 (mean 78). Non-rotator cuff soft tissue wasdissected from each specimen. A blunt dissection ofthe supraspinatus muscle was preformed so that thedigitizing probe could be inserted under the musclebelly to contact the base of the spine of the scapula inthe supraspinatus fossa. The medial border of thescapula and the mid-shaft of the humerus weremounted to a testing frame such that the humerus wassuspended in a neutral position parallel to the medialborder of the scapula in the plane of the scapula. Oncethe bony landmarks had been digitized, the acromionwas removed with a Stryker Bone Saw (Stryker,Kalamazoo, MI) for better access to the RCT.
A Flock of Birds system (Ascension TechnologiesCorp., Burlington, VT) was used to digitize the three-dimensional coordinates of specic bony and soft tis-sue landmarks on each specimen. Prior to mountingthe specimen, two Flock of Birds sensors were rigidlyattached to each specimen using nylon screws. Therst sensor was bolted to the medial region of theinfraspinatus fossa. The second sensor was bolted tothe distal shaft of the humerus. A stylus similar to thatdescribed by Meskers et al. (1999) was mounted to athird Flock of Birds sensor to digitize the landmarksand the most internal border of the rotator cuff tear(Fig. 1). Data points circumscribing the RCT weretaken at approximately 2 mm intervals. Prior to be-ginning each session of digitizing, the workspace wascalibrated to account for any variation in the environ-ment, as suggested by Meskers et al. (1999). Thesame person (R.F.H.) digitized every point on eachspecimen. The interoperator digitizing error was in-distinguishable from the error inherent to the Flock ofBirds. Interoperator error was determined by execut-ing the digitization protocol for one specimen four
Fig. 1. A point on the border of the RCT being digitized with a probe attached to the base of a Flockof Birds sensor. The medial border of the scapula is affixed to the testing frame. A Flock of Birds sensoris rigidly mounted to the medial rim of the infraspinatus fossa.
Quantitative Morphology of Rotator Cuff Tears 19
times. The variance was then computed for each ofthe digitized landmarks.
The collected data were converted to a coordinatesystem referenced to the scapular sensor using soft-ware coded in Matlab (MathWorks, Natick, MA). Thistransformation eliminated the effects of any motion ofthe specimen on the testing frame during digitizing.The following measurements were computed usingsoftware coded in Matlab: the area of the RCT, thetendons included in each tear, the area of the tear ineach of the included tendons, the anterior-to-posteriorlength of the tear at the widest point, and the medial-to-lateral length of the tear at the widest point.
The area of the RCT was dened as the area insidea two-dimensional polygon described by the perpen-dicular projection of each of the RCT data points ontoa plane. The plane of the tear was determined bytaking the planar least-squares t (Van der Helm et al.,1992) of the data points circumscribing the RCT. Twotendon separation planes were then used to locate theborder of the included tendons: one plane representedthe border between the infraspinatus and supraspina-tus and the second represented the border betweenthe supraspinatus and the subscapularis. The infraspi-natus-supraspinatus separation plane was dened bypoints digitized at three scapular landmarks: medialborder at the intersection of the scapular spine, pos-terolateral acromion, and a point at the base of thespine of the scapula in the supraspinatus fossa. Thesupraspinatussubscapularis separation plane was de-termined in a like manner, using points digitized atthe superior angle, scapular notch, and superior ante-rior corner of the coracoid. The RCT data points werethen segregated into their respective tendons suchthat any point posterior to the infraspinatussupraspi-natus plane belonged to the infraspinatus, any pointbetween the two planes belonged to the supraspina-tus, and any point anterior to the supraspinatussub-scapularis plane belonged to the subscapularis. Thearea of the tear in a given tendon was computed in asimilar fashion to that used for the entire tear. Thepoints where the tear intersected a separation planewere included in the segregated datasets to ensurethat none of the area of the tear was lost in the processof distributing the data points to their respective ten-dons.
For tears including the supraspinatus tendon, thedimension of the tear in the medial-to-lateral directionand anterior-to-posterior direction were computed atthe widest point. A line in the anteriorposterior di-rection was determined by the following algorithm:the lines describing the intersection of the plane ofthe supraspinatus tear and the planes dening thetendon borders were established; a line in the plane of
the tear in the anteriorposterior direction was chosensuch that it intersected both of these lines at equalangles (Fig. 2). The length of the tear in the anteriorposterior direction was calculated along this line at thewidest portion of the tear. The line in the mediallateral direction was chosen as the line in the plane ofthe tear perpendicular to the anteriorposterior line.The length of the tear in the mediallateral directionwas calculated along this line at the widest portion ofthe tear.
Of the 414 shoulders collected during the course ofthis study, 57 (16%) had full thickness RCTs. Of thenormal shoulders, 184 were female and 173 weremale. Of the shoulders with RCTs, 37 were female, 20male, 25 were left arms, and 32 were right arms. Theshoulders came from 38 cadavers, 23 female and 15male. Nineteen of the cadavers had bilateral RCTswhile the remaining 19 possessed unilateral RCTs.No correlation between area of RCT and age wasfound (P 0.564).
Of the 57 specimens collected with full thicknessRCTs, six had tears only in the infraspinatus, 25 onlyin the supraspinatus, one only in the subscapularis, 22in the infraspinatus and supraspinatus, one in thesupraspinatus and subscapularis, and two had a tearincluding all three of the major tendons (Table 1).
The mean, minimum, maximum, and median cal-culated areas of the RCT are listed in Table 2 by theincluded tendons. The overall largest tear includedthe supraspinatus and the infraspinatus with a size of19.17 cm2. The smallest tear was contained within the
Fig. 2. A superior view of the shoulder with the acromion re-moved to show an RCT in the lateral region of the humeral head. Aline intersecting the infraspinatussupraspinatus separation line andthe supraspinatussubscapularis separation line is used to measure thetear in the anteriorposterior direction. A line perpendicular to theanteriorposterior line is used to measure the tear in the mediallateraldirection. Measurements are taken at the widest point of the tear.
20 Wening et al.
supraspinatus and was 0.07 cm2. The overall meantear size was 4.40 cm2. The anteriorposterior lengthmeasurement ranged from 0.206.00 cm with a meanof 1.95 cm and a median of 1.60 cm. The mediallateral measurement ranged from 0.305.31 cm with amean of 1.98 cm and a median of 1.38 cm.
In this study the prevalence of full thickness RCTsat the time of death was found to be 16%. A quanti-tative description of the area of those full thicknesstears, and the distribution of those tears among thetendons of the rotator cuff, was also determined. Forany tear including the supraspinatus tendon, anteriorposterior and mediallateral dimensions of the tearwere calculated.
The percentage of shoulders with full thicknessRCTs in this population was found to be similar to thepercentages currently reported in the literature (Table3). Of these studies only one categorized the tearbased on the individual tendons of the rotator cuff(Itoi et al., 1995). Itoi et al. (1995) studied differencesbetween the extramuscular tendon length and thefunctional tendon length to determine the size of theRCT but did not specically report tear sizes. Sher etal. (1995) noted tear sizes ranging from 0.514 cm2
(mean 5.0 4.3 cm2) based on MR images of asymp-tomatic individuals. A description of the measurementtechnique is not provided. It should be noted that themean and standard deviation for this study and theSher et al. (1995) study are very similar. A cadaverstudy comparing the sizes of RCTs found using MRimages to those found via dissection and digitizing
with the Flock of Birds system may be an effectivemethod for validating the accuracy of MR images formeasuring RCT areas.
The correlation between increased age and preva-lence of RCTs is well documented (Keyes, 1933;Ogata and Uhthoff, 1990; Sher et al., 1995; Tempelhofet al., 1999; Wilson and Duff, 1943; Zuckerman et al.,1992). In this study it was also noted that the preva-lence of RCTs increased as age increased. A correla-tion between the area of the RCT and the age of thesubject was not found (P 0.564). This is not entirelysurprising, since the date of onset and cause (traumat-ic, nontraumatic) of the lesion was unknown.
Due to the nature of the data collection and pro-cessing, it is possible that the areas reported for theRCTs in this study are slightly smaller than the truearea. The digitizer was consistently placed such thatthe probe tip was inside the tear and lightly contactingthe tear. The area reported is the planar projection ofthe data points dening the border of the RCT, andtherefore does not account for the cylindrical nature ofthe capsule. However, it should be noted that theerrors in measurement produced by this planar pro-jection are negligibly small for small tears and only ofminor concern for the largest tears reported in thisstudy.
In this study the supraspinatus was clearly the mostcommonly affected tendon, while RCTs affectingboth the supraspinatus and infraspinatus followedclosely. Involvement of the subscapularis was rare.The rotator interval between the supraspinatus andthe subscapularis contains the biceps tendon. Al-though this region of the capsule is relatively thin, thesheath surrounding the biceps is reinforced by bers
TABLE 2. Smallest, Largest, Mean, and Median Tear Size for Each Combination of Tendons Torn
Tendons included in tearInfcm2
Number with tear 6 25 1 22 1 2 57Smallest tear 0.16 0.07 0.25 0.22 7.09 9.85 0.07Largest tear 3.97 12.16 0.25 19.17 7.09 18.60 19.17Mean tear 1.41 2.38 0.25 6.77 7.09 14.23 4.43Median tear 0.69 0.65 0.25 4.20 7.09 14.23 2.40
Infraspinatus (Inf), Supraspinatus (Sup), Subscapularis (Sub).
TABLE 1. Tendons Studied
Tendons included in tear Inf Sup Sub Inf Sup Sup Sub All tendons All specimens
Percent of all shoulders 1.68 7 0.28 6.16 0.28 0.56 15.96Number with tear 6 25 1 22 1 2 57Percent of shoulders with...