The coracoacromial ligament: The morphology and relation torotator cuff pathology

Hayrettin Kesmezacar, MD,a Isik Akgun, MD,a Tahir Ogut, MD,a Selim Gokay, MD,a and Ibrahim Uzun, MD,b

Istanbul, Turkey

We dissected 80 shoulders from 44 fresh cadavers todefine variants of the coracoacromial ligament and theirrelationship to rotator cuff degeneration. The shapesand the geometric data of the ligaments wereinvestigated, and the rotator cuffs of the cadavers wereevaluated macroscopically. Five main types ofcoracoacromial ligaments were found: Y-shaped, broadband, quadrangular, V-shaped, and multiple-banded.The Y-shaped ligament was the most frequent type, witha frequency of 41.3%, and the V-shaped ligament(11.2%) has not been previously reported. Of thecadavers that were dissected bilaterally, 64% showedthe same type of ligament. There was no statisticalsignificance between rotator cuff degeneration and thetype or geometric measurement of the ligament.However, the coracoacromial ligaments with more than1 bundle showed significant association with rotatorcuff degeneration with a longer lateral border andlarger coracoid insertion. (J Shoulder Elbow Surg2008;17:182-188.)

The relation between the anatomy of the coracoacro-mial arch and the impingement syndrome was firstpopularized by Neer8 in 1972, and since that time,investigations have pointed out the role of the coracoa-cromial arch in rotator cuff pathology. The coracoa-cromial arch is composed of the acromion, thecoracoacromial ligament (CAL), and the coracoid. An-atomic variants of the acromion and the association ofthe curved and hook-shaped acromions with cuff de-generation were first described by Bigliani et al,1

and since then, numerous studies have documentedan increased incidence of rotator cuff tears in patientswith a type III acromion.2,9,15

From the aCerrahpasa Faculty of Medicine, Department of Ortho-paedics and Traumatology, Istanbul University, and bInstitute ofForensic Medicine.

Reprint requests: Hayrettin Kesmezacar, MD, Abide-i Hurriyet cadRuyam Palas, No. 144/16, 80260, Sisli, Istanbul, Turkey(E-mail: hkesmezacar@yahoo.com).

Copyright ª 2008 by Journal of Shoulder and Elbow SurgeryBoard of Trustees.

1058-2746/2008/$34.00doi:10.1016/j.jse.2007.05.015

182

Regarding the morphology of the CAL, 3 maintypes—quadrangular, Y-shaped, and broad band—were identified by Holt and Allibone.6 Other studieshave focused on the histologic and biomechanicalproperties of this ligament.3,9,14,17 An increase incross-sectional area, accompanied by a decrease inmaterial properties of the ligament, has been foundin patients with subacromial impingement.14 Fre-merey et al3 observed that the geometric propertiesof the CAL in asymptomatic shoulders differed fromthose in shoulders with rotator cuff degeneration,having a shorter lateral and medial band.

The length and cross-sectional area of the CAL withrelation to rotator cuff degeneration have been dis-cussed,3,14,17 but no study has yet evaluated thetype of ligament in relation to the impingement syn-drome. The purpose of this study was to determinethe anatomic features of the CAL and to investigatewhether any relationship exists between its morphol-ogy and rotator cuff lesions.

MATERIALS AND METHODS

We dissected 80 shoulders from autopsy specimens of 44persons, who were of various ages at death. The fresh ca-davers were all obtained from the Institute of Forensic Med-icine (Istanbul, Turkey), and the period between dissectionand death did not exceed 24 hours. None had an upperlimb deformity or a surgical scar on the shoulder. The ageat death ranged from 25 to 90 years (mean, 51 years).There were 39 male and 5 female cadavers; the meanheight and weight were 169 cm (range, 152-184 cm) and66.50 kg (range, 39-93 kg), respectively. Dissectionswere performed bilaterally in 36 cadavers and unilaterallyin the remaining 8, because of the forensic procedures. Un-fortunately, the dominant side, previous occupation, andmedical history regarding shoulder symptoms could not beconsidered because of insufficient records.

All shoulders were dissected through an anterosuperiorincision followed by total detachment of the deltoid musclefrom the clavicle and scapula. The CAL was identified withcareful blunt dissection to prevent overlooking any thinnerbands. The shapes of the ligaments, the lengths of the lateralborders, and the widths of the attachments to the acromionand coracoid were recorded.

The rotator cuff tendons were examined macroscopicallyfrom the bursal side for the presence of partial- or full-thick-ness tears. Delamination or fraying at the tendon surface

J Shoulder Elbow Surg Kesmezacar et al 183Volume 17, Number 1

Figure 1 Five main types of CAL. Type I, Y-shaped; Type II, broad band; Type III, quadrangular; Type IV, V-shaped;Type V, multiple-banded.

was accepted as a partial tear. Examination of the gleno-humeral and acromioclavicular joints was performed. Allacromion processes were evaluated to determine theirshapes according to the classification of Bigliani et al.1

The thickness of the acromion at the midpoint of the lateralborder was also measured. The shoulders were dissectedby the same surgeon, and all were photographed.

The CALs were determined from a superior-to-inferiorview and were classified by their types with the data col-lected. Statistical analyses of the results were performed,and the following parameters were considered: age, CALtype, geometric measurement of the CAL morphology, pres-ence of degeneration of the bursal side of the rotator cuff,acromion type, and acromion thickness. A combination ofc2 and t tests was used for variable parameters. Nonpara-metric analyses were also performed with the Mann-Whit-ney U and Kruskal Wallis H tests. P < .05 was consideredstatistically significant.

RESULTS

The morphologic results revealed 5 main typesof CAL: Y-shaped, broad band, quadrangular,V-shaped,and multiple-banded (Figure 1).

Y-shaped ligament

The Y-shaped ligament started as a unique bandand truncated into 2 separate bands before its in-sertion onto the lateral and medial border of thecoracoid. A thin membranous tissue, which formeda diaphanous region, was detected lying betweenthe separate bands (Figure 2). In all cases, lateral

bands showed larger and thicker structures com-pared with medial bands. In some cases, the medialbranches of the ligaments were so thin that they couldbe missed if dissection was not carefully performed.

Broad band ligament

The broad band ligament was formed by a uniqueband, which had a difference of less than 2 mm inwidth between the insertions at both ends, as was for-merly described in Holt and Allibone.6 This type of lig-ament started from the acromion and extended to themost lateral border of the coracoid without change inits width (Figure 3).

Figure 2 Y-shaped CAL.

184 Kesmezacar et al J Shoulder Elbow SurgJanuary/February 2008

Quadrangular ligament

In the quadrangular ligament CAL type, ligamen-tous tissue replaced the membranous layer of theY-shaped type and attached to the coracoid withoutany division. It had a unique band as in the broadband type, but the widths of the attachment to thebones were different to form their quadrangular shape(Figure 4). The lateral and medial borders of the liga-ment were curved slightly rather than straight.

V-shaped ligament

The V-shaped ligament consisted of 2 completelyseparate bands from the acromion to the coracoid. Itappeared similar to the Y-shaped, having 2 branches;however, the ligament started from the acromion asa unique band in the Y-shaped type rather than having2 completely separate bands (Figure 5). The acromialattachments of the bundles were close to each other.The one that was situated medially was thinner andstarted just inferior to the main lateral band.

Figure 3 Broad band CAL.

Figure 4 Quadrangular CAL.

Multiple-banded ligament

The multiple-banded ligament did not have homog-enous morphologic characteristics. A third thinnerbranch was localized medially and accompaniedthe 2 main branches in some cases (Figure 6),whereas some ligaments looked like the Y-shapedtype, with a medial band dividing into 2 or morebranches before attaching to the coracoid (Figure 7).When there was a completely separated bundle, itextended inferomedially to the coracoid, as wasobserved in the V-shaped ligament.

Morphologic findings

The Y-shaped ligament, the most frequent type, wasobserved in 33 shoulders (41.3%), and the broadband type existed in 22.5%. No significant differ-ences were found in the distribution of the other 3types (Table I). The cases with Y-shaped ligamentswere younger than the other types. The mean age,the length of the lateral border, the width of the acro-mial attachment, the width of the coracoid attachmentof the lateral band, and the thickness of the mid acro-mion based on CAL type are presented in Table II. Thestandard measurements of the length of the medialborder and the width of the coracoid attachmentwere not possible because of the morphologic differ-ences of the types; Y-shaped, V-shaped, and multiple-banded ligaments had more than 1 insertion into thecoracoid, and each branch also had different medialborders. No statistical difference was detected be-tween the parameters of each type of CAL, exceptfor the significantly larger lateral band attachment ofthe quadrangular pattern.

There were 22 type I (27.5%), 55 type II (68.7%),and 3 type III (3.8%) acromions based on the system ofBigliani et al.1 Regarding the status of the rotator cuff,53 shoulders (66.3%) had macroscopically normal ro-tator cuffs, whereas cuff degeneration or complete

Figure 5 V-shaped CAL with 2 completely separated bundles.

J Shoulder Elbow Surg Kesmezacar et al 185Volume 17, Number 1

tear (degeneration in 23 and complete tears in 4) wasdetected in 27 shoulders (33.7%). The distribution ofthe acromion types and the pathology of the rotatorcuff according to the CAL types are given in TablesIII and IV. No statistically significant correlation wasfound between the rotator cuff degeneration and theligament type. However, all shoulders with quadran-gular CALs had a normal rotator cuff, and the percent-age of cuff degeneration in the shoulders that hada broad band ligament was lower than that in theother 3 types. These 2 patterns, the quadrangularand broad band types, had unique bundles. This find-ing led us to divide all ligaments into 2 groups: uniqueband and more than 1 band. There were 29 ligamentswith 1 bundle (quadrangular and broad band type)and 51 with multiple bands. Ligaments with multiplebands showed a significant association with rotatorcuff degeneration (Table V), with a longer lateral bor-der and a larger coracoid insertion (Table VI).

Of the 36 cadavers dissected bilaterally, 23 (64%)showed the same type in both shoulders. In addition,

Figure 6 Variation of multiple-banded CAL with 2 separatedbundles situated at medial site of lateral band.

Figure 7 Another variation of multiple-banded CAL with division of1 bundle into multiple parts.

the geometric measurements of these ligaments werealmost similar. Despite the fact that the same type ofCAL was observed on both sides, tendon degenera-tion or a tear was detected in 1 shoulder whereasthe rotator cuff was normal on the other side in 4cadavers. The remaining 19 cadavers had a similarstatus on both sides.

DISCUSSION

Previous reports about the CAL described a varietyof types, from triangular shape to 2 main bands. Salteret al13 observed a Y-shaped type of ligament in all 20dissections. Prescher11 reported that the ligament wascomposed of medial and lateral tracts and that thearea between these tracts might be either weak orcompletely absent. Gagey et al4 also found all theCALs to be Y-shaped in 15 dissections. The anatomicvariants and their incidences were not clear until thecadaveric study of Holt and Allibone,6 who found 3main patterns: quadrangular, Y-shaped, and broadband. They also described a multiple-banded liga-ment that had been seen in only 1 shoulder. Our dis-sections of 80 shoulders showed different structuresof the CAL from what was described in the previousreports. In our study, 5 main forms were observed:Y-shaped, broad band, quadrangular, V-shaped,and multiple-banded. Most of the ligaments were Y-shaped (41.3%) or broad band (22.5%). These find-ings were not in agreement with those of the previousstudy,6 where the most frequent type was quadrangu-lar, occurring in 48% of all ligaments. However, themean age of their cadaveric group was 81 years. Dif-ferences in the distribution pattern may exist accord-ing to age or race. Age-related alteration of themorphology of the ligament was reported by Wasmeret al,16 where 80% of the CALs consisted of 2 parts inolder persons, with a mean age of 75 years, and thesame rate was 100% in younger persons. Regardingthe mean age for each type of CAL, the cadavers ofthe Y-shaped group were younger than the others inour study. However, the other 2 types (V-shaped andmultiple-banded) were the oldest groups. No statisticalsignificance was found when the mean ages of the

Table I Distribution of CAL types

n %

TypeY-shaped 33 41.3Broad band 18 22.5Quadrangular 11 13.8V-shaped 9 11.2Multiple-banded 9 11.2

Total 80 100

186 Kesmezacar et al J Shoulder Elbow SurgJanuary/February 2008

Table II Mean age and geometric properties of ligament types

TypeMeanage (y)

Length of lateralborder (mm)

Width of acromionattachment (mm)

Width of coracoidattachment

(lateral band) (mm)Thickness of midacromion (mm)

Y-shaped 45.21 6 19.6 34.57 6 3.1 15.60 6 2.7 12.93 6 2.9 9.03 6 1.3Broad band 52.44 6 9.4 35.11 6 5.8 13.66 6 3.5 13.83 6 3.2 9.50 6 1.1Quadrangular 50.00 6 11.6 38.36 6 8.9 14.18 6 4.7 22.09 6 1.2* 8.18 6 1.6V-shaped 62.00 6 16.7 34.00 6 4.7 11.88 6 2.4 13.88 6 4.9 9.22 6 1.3Multiple-banded 60.88 6 23.4 32.00 6 2.7 15.44 6 2.8 10.66 6 4.1 8.77 6 1.0

*P < .05.

types were compared. In addition, almost two thirdsof the cadavers had the same ligament patterns bilat-erally. In our opinion, age is not a factor that affectsthe type of CAL, as in acromial morphology. Our inves-tigation supports the hypothesis of Kopuz et al,7

who suggested that the types of CAL were alreadypresented at birth and that the final shapes weredetermined by developmental factors, rather thandegenerative changes. Three patterns of ligaments(quadrangular, broad band, and U-shaped) wereseen in neonatal cadavers in their study.

In our study, each shoulder showed a ligament thathad a different shape and different dimension, even if

Table III Relation between morphologic type of acromion andligament type

Type Iacromion

Type IIacromion

Type IIIacromion

TypeY-shaped (n ¼ 33) 13 18 2Broad band (n ¼ 18) 3 15 0Quadrangular (n ¼ 11) 3 7 1V-shaped (n ¼ 9) 0 9 0Multiple-banded (n ¼ 9) 3 6 0

Total 22 (27.5%) 55 (68.7%) 3 (3.8%)

Table IV Distribution of normal and degenerated rotator cuffsregarding types of CAL

Normal cuffDegenerated cuff or

cuff tear

TypeY-shaped (n ¼ 33) 20 (61%) 13 (39%)Broad band (n ¼ 18) 13 (72%) 5 (28%)Quadrangular (n ¼ 11) 11 (100%) 0 (0%)V-shaped (n ¼ 9) 5 (56%) 4 (44%)Multiple-banded (n ¼ 9) 4 (44%) 5 (56%)

Total 53 (66.3%) 27 (33.7%)

it was within one determined group. Division of the lig-ament into its branch was observed at different pointsfrom one ligament to another. A multiple-banded liga-ment, which was seen only in 1 shoulder in Holt andAllibone’s dissections,6 was not rare in the cadavericgroup of another study, where Pieper et al10 reporteda 14.5% incidence of 3-part CAL. In agreement withPieper et al, the multiple-banded ligament type wasobserved in 11.2% in our study. The V-shaped liga-ment was not a subject of any of the previous studies.This type of ligament consisted of two completelyseparate bands, and the one situated medially hadan acromial attachment inferior to the lateral band.In some cases, it was not easily distinguished fromsome of the Y-shaped types; however, the medialband of the V-shaped type was attached to the acro-mion separately from the thicker lateral band. This pat-tern may be the developing form of the U-shapedligament that was described by Kopuz et al.7

The lack of uniformity of the ligament was empha-sized by the absence of statistical correlation between

Table V Incidence of normal rotator cuff and rotator cuff degenerationor tear in unique-bundle group and group more than 1 bundle

Normal cuffDegenerated cuff

or cuff tear

Unique bundle 24 (83%) 5 (17%)*More than 1 bundle 29 (57%) 22 (43%)*

*P < .05.

Table VI Coracoid width and lateral length of lateral band in unique-bundle group and group with more than 1 bundle

Unique bundleMore than1 bundle

Mean length of lateral border (mm) 36.34 34.02*Mean width of lateral band

attachment to coracoid (mm)16.97 12.71*

*P < .05.

J Shoulder Elbow Surg Kesmezacar et al 187Volume 17, Number 1

the type of the ligament and sex, age, width of theacromial attachment, length of the lateral border, orthickness of the mid acromion. There was no correla-tion between the types and acromial morphology orbetween the types and rotator cuff degeneration. Al-though the type of CAL did not affect rotator cuffdegeneration, statistical analyses showed that the lig-aments that were composed of more than 1 bundlewere associated more frequently with rotator cuff de-generation or tear. The number of bands changedonly the dimension of the coracoid attachment of thelateral branch. Because the most lateral band of theCAL is the region that would most likely impinge onthe rotator cuff, any change in the width of the cora-coid attachment of this band would be biomechani-cally important. Our measurements suggested thatthe mean width of the quadrangular and broadband types at the insertion to the coracoid was foundto be higher compared with the mean width of the lat-eral band of the other types. In addition, the mean lat-eral border length of the unique-band ligaments waslonger than the CAL composed of more than 1 bundle.These data support the findings of Zuckerman et al.17

They reported a shorter distance between the acro-mion and the coracoid in shoulders with cuff tearsthan in those with intact rotator cuffs. The same resultswere observed by Soslowsky et al14 and Fremereyet al.3 Soslowsky et al also stated that the cross-sec-tional area of the lateral band was greater in rotatorcuff tear specimens than in normal specimens. How-ever, they reported that all the specimens had 2 pre-dominant bands with a space between them filledwith extremely thin tissue and that the cross-sectionalarea of the lateral band was greater in rotator cuffspecimens. They also noted that no difference was ob-served between the length of the medial bands of thetear and no-tear groups. In contrast to the findings ofSoslowsky et al, Pieper et al10 reported similar patho-logic changes in all parts of the degenerated ligament.Fremerey et al supported this observation witha shorter medial band in specimens with rotator cuffdegeneration compared with normal shoulders. Be-cause the medial band was not located in the impinge-ment region, their hypothesis was that the anatomicvariants of the CAL damaged the rotator cuff tendonrather than changes in the tissue of the ligament as aresultof the load alteration after cuff tear.

The CAL, which interestingly lies between 2 projec-tions of the same bone, acts as a tension band for cor-acoid and acromion. According to the biomechanicalstudy of Putz et al,12 the acromion had a 10 timeshigher bending force after division of the ligament,and this rate was 3 times higher for the coracoid.The CAL, supporting the acromion and the coracoid,has a significant role of transmitting forces from the sur-rounding musculature. Any muscular imbalancearound the shoulder affects the tension of the ligament

and probably subsequently alters material properties.On the other hand, Gallino et al5 observed that allCALs in their cases showed a different structure ex-tending to the undersurface of the acromion. They sug-gested that the potential variation in origin andthickness of the ligament should be taken into consid-eration when performing surgery for impingementsyndrome. At this point, geometric parameters of thelateral band that act primarily against superior transla-tion of the humeral head may be an important factor inthe resistance against the tensile forces. As reportedpreviously,14,17 when a cuff tear was present, the lat-eral bundle was longer and broader. The CAL, whencomposed of a unique band, possesses all of its fibersin 1 branch without any division in the impingementregion. Thus, the tension effected is concentrated ina unique area and may play a role in dampening ten-sion forces. Conversely, if it has multiple components,the lateral part of the ligament has a shorter coracoidattachment compared with an undivided type andprobably concentrates the tensile forces mostly onthe lateral band. Our study showed that the quadran-gular type and the broad band type had longer andbroader lateral bundles compared with the other typesand were infrequently associated with cuff tears. Al-though the morphology of the lateral band is importantfor force transmission, the type of CAL and the numberof bundles affect this morphology.

A limitation of our study was the lack of informationregarding the previous medical history and occupa-tion of the cadavers and the lack of histologic evalua-tion of the specimens, which could not be performedfor forensic reasons. The results of our dissectionshave shown that 5 main types of CAL existed, withthe addition of the V-shaped and multiple-bandedtypes. However, the geometric measurements of theligaments varied. Sixty-four percent of the cadavershad the same type of CAL bilaterally. No statistical cor-relation was found between the type of ligament androtator cuff degeneration. The etiology of the impinge-ment syndrome is multifactorial. Regarding the role ofthe morphology of the CAL in the impingement, it is un-clear whether it causes impingement or is affected byimpingement. However, in shoulders that have morethan 1 band, the association of cuff lesions was ob-served accompanied by a shorter lateral border andcoracoid attachment compared with the unique-bandligament.

REFERENCES

1. Bigliani LU, Morrison DS, April EW. The morphology of the acro-mion and its relationship to rotator cuff tears. Orthop Trans 1986;10:228.

2. Epstein RE, Schweitzer ME, Frieman BG, Fenlin JM, Mitchell DG.Hooked acromion: prevalence on MR images of painful shoulders.Radiology 1993;187:479-81.

188 Kesmezacar et al J Shoulder Elbow SurgJanuary/February 2008

3. Fremerey R, Bastian L, Siebert WE. The coracoacromial ligament:anatomical and biomechanical properties with respect to age androtator cuff disease. Knee Surg Sports Traumatol Arthrosc 2000;8:309-13.

4. Gagey N, Ravaud E, Lassau JP. Anatomy of the acromial arch: cor-relation of anatomy and magnetic resonance imaging. Surg RadiolAnat 1993;15:63-70.

5. Gallino M, Battiston B, Annaratone G, Terragnoli F. Coracoacro-mial ligament: a comparative arthroscopic and anatomic study.Arthroscopy 1995;11:564-7.

6. Holt EM, Allibone RO. Anatomic variants of the coracoacromialligament. J Shoulder Elbow Surg 1995;4:370-5.

7. Kopuz C, Baris S, Yildirim M, Gulman B. Anatomic variations ofthe coracoacromial ligament in neonatal cadavers: a neonatalcadaver study. J Pediatr Orthop B 2002;11:350-4.

8. Neer CS II. Anterior acromioplasty for the chronic impingementsyndrome in the shoulder: a preliminary report. J Bone Joint SurgAm 1972;54:41-50.

9. Panni AS, Milano G, Lucania L, Fabbriciani C, Logroscino CA.Histological analysis of the coracoacromial arch: correlation be-tween age-related changes and rotator cuff tears. Arthroscopy1996;12:531-40.

10. Pieper HG, Radas CB, Krahl H, Blank M. Anatomic variation of thecoracoacromial ligament: a macroscopic and microscopic cadav-eric study. J Shoulder Elbow Surg 1997;6:291-6.

11. Prescher A. Anatomical basics, variations, and degenerativechanges of the shoulder joint and shoulder girdle. European J Ra-diol 2000;35:88-102.

12. Putz R, Lieberman J, Reichelt A. Funktion des ligamentum coracoa-cromiale. Acta Anat 1988;131:140-5.

13. Salter EG, Nasca RJ, Shelley BS. Anatomic observations on theacromioclavicular joint and supporting ligaments. Am J SportsMed 1987;15:199-206.

14. Soslowsky LJ, An CH, Johnston SP, Carpenter JE. Geometric andmechanical properties of the coracoacromial ligament and theirrelationship to rotator cuff disease. Clin Orthop Relat Res 1994;314:10-7.

15. Toivonen DA, Tuite MJ, Orwin JF. Acromial structure and tears ofthe rotator cuff. J Shoulder Elbow Surg 1995;4:376-83.

16. Wasmer G, Hagena FW, Bergman M, Mittelmeier T. Anatomi-sche und Biomechanische Untersuchunges des Ligamentum cora-coacromiale am Menshen. In: Refior HI, Plitz W, Jaeger M,Hackenbroch MH, editors. Biomechanik der Gesunden unKranken Schulter. Stuttgart: Thieme; 1985. p. 61-5.

17. Zuckerman JD, Kummer FJ, Cuomo F, Simon J, Rosenblum S. Theinfluence of coracoacromial arch anatomy on rotator cuff tears.J Shoulder Elbow Surg 1992;1:4-14.