Classifications in Brief: Bigliani Classification of Acromial Morphology

History The diagnostic spectrum from rotator cuff tendinopathy through cuff tear arthropathy causes pain and disability that can result in surgical intervention [15]. In 1931, Codman and Akerman [5] proposed that degenerative changes of the tendons initiate rotator cuff tears. Subsequently, in 1949, Armstrong [1] proposed that mechanical impingement of the rotator cuff tendons under the acromion causes supraspinatus syndrome. Neer [21] also postulated that mechanical impingement was responsible for up to 95% of rotator cuff tears and reported successful treatment with anterior acromioplasty [20]. Regardless of the etiology, acromioplasty is commonly used as a surgical adjunct for the treatment of rotator cuff tendinopathy and tears. In 1986, Bigliani and colleagues [3] were the first to classify acromial morphology and correlate it with rotator cuff tears. They described three types of acromion: Type I (flat), Type II (curved), and Type III (hooked) (Fig. 1). A Type IV acromion was later described as a convex acromion [28]. Bigliani et al. [3] and other authors [2, 6–8, 10, 13] have reported that Type III acromia have been associated with rotator cuff tears. Conversely, other authors have found no such correlation [23, 24]. Since Bigliani et al.’s [3] original description, there have been other attempts to quantify acromial morphology. Bigliani et al. [3] and Kitay [12] described the acromial slope, a measure of anterior acromial slope and the relationship between increased acromial slope and rotator cuff tears. In the same paper, Kitay [12] also described acromial tilt, the angle between a line drawn connecting the most posterior point of the inferior acromion to the most anterior point of the inferior acromion and a line drawn connecting the same most posterior point of the inferior acromion to the inferior tip of the coracoid process. Decreased acromial tilt was associated with rotator cuff tears [12]. Nyffeler et al. [22] described the acromial index, a ratio of the distance from the glenoid plane to the acromion divided by the distance from the glenoid plane to the lateral aspect of the humeral head. The authors observed that the acromion of patients with a rotator cuff tear appeared to have a more lateral extension, thus a greater acromial index, than that of patients with an intact cuff [22]. Most recently, the critical shoulder angle (CSA) was described, which not only includes acromial morphology but also considers glenoid morphology [16]. The authors observed patients with rotator cuff tears were more likely to have a CSA > 35° than a control group of patients without rotator cuff tears. Although there have been numerous attempts to find an association between rotator cuff tears and scapular morphology [2, 3, 12, 17, 18], there is no consensus as to the role of scapular morphology in the etiology of rotator cuff tears.Fig. 1: This diagram shows the Bigliani classification of acromial morphology: (A), Type I (flat); (B), Type II (curved); and (C), Type III (hooked). Reprinted with permission from Bright AS, Torpey B, Magid D, Codd T, McFarland EG. Reliability of radiographic evaluation for acromial morphology. Skeletal Radiol. 1997;26:718–721.Purpose The original purpose of the Bigliani classification [3] was to describe acromial morphology and determine whether it was associated with rotator cuff tears. To our knowledge, the Bigliani classification of acromial morphology was the first classification system to try to define acromial morphology. One goal was to facilitate communication among clinicians and to support further research on the relationship between acromial morphology and rotator cuff tendinopathy. However, as we have already noted, this relationship has been questioned (particularly with respect to the concept of impingement syndrome) [2, 23, 24]. In addition, problems with reliability of the Bigliani classification [2, 4, 9, 31] have been reported. Taken together, these concerns should cause both clinicians and researchers to use the Bigliani classification with caution, if at all. In the same publication, Bigliani et al. [3] aimed to determine if there was any correlation between acromial morphology and the presence of rotator cuff tears. Bigliani et al. [3] found that 69.8% of full-thickness rotator cuff tears had Type III acromia, 24.2% had Type II acromia , and 3% had Type I acromia. Further studies have reported conflicting results; some found correlations between acromial morphology and rotator cuff tears [3, 7, 13], and others did not [23, 24]. Later attempts to identify whether acromial morphology types were associated with impingement syndrome were made [2, 6], although this was not originally a purpose of the classification system. Two studies [2, 6] observed that there was no correlation between impingement syndrome and acromial morphology. Because of the proposed correlation between rotator cuff tears and Type III acromia, it was theorized that patients with Type III acromia may be at risk of having impingement syndrome [19]. Classification In 1986, Bigliani et al. [3] described three distinct types of acromial morphology after taking lateral radiographs of 140 shoulders from 71 cadavers. They described Type I as flat, Type II as curved, and Type III as hooked (Fig. 1). No further attempt was made to further define the three different acromial morphology types. Bigliani et al. [3] also observed that acromial morphology and rotator cuff tears were associated with two additional factors. There was an association between increasing acromial slope, which increased from Type I to Type III acromia, and rotator cuff tears, as well as between anterior acromial spurs and rotator cuff tears. Both observations have been verified by further research from several authors [2, 3, 9, 12]. Validation As we suggested earlier, the main problem with the Bigliani classification is that although its intraobserver reliability has varied–sometimes (though not always [4]) achieving kappa values in the good-to-excellent range [11, 25, 27]–its interobserver reliability has been consistently fair-to-poor [4, 9, 11, 23, 25, 27, 31]. This means that it is not suitable as a clinical communication tool among surgeons or clinician-scientists. In one study, in which six fellowship-trained shoulder surgeons reviewed 126 scapular outlet view radiographs (a scapular Y view with 10° of caudal tilt) of the shoulder observed good-to-excellent intraobserver reliability (κ = 0.888) and poor-to-fair interobserver reliability (κ = 0.516) [11]. Several further small studies observed that interobserver reliability was poor (κ = 0.25-0.41) [9, 25, 27, 31]. In another study, which again found only fair interobserver (κ = 0.35) reliability, both interobserver and intraobserver reliability were found not to vary according to surgeon experience level (from PGY-2 resident to fellowship-trained shoulder surgeon), or whether the observers were surgeons or radiologists [4]. Taken together, the interobserver reliability of the Bigliani classification of acromial morphology as reported by multiple studies is fair at best, regardless of level of expertise and training, and as such it should not be used to facilitate communication by orthopaedic surgeons, radiologists, or researchers. The use of MRI to identify acromial morphology has also been studied. One small study compared a single oblique midsagittal MRI image in the plane of acromion and scapular outlet view radiographs of 32 shoulders in patients who had undergone impingement syndrome treatment [29]. Three shoulder surgeons independently reviewed MRI and radiographs. They graded acromial morphology almost identically when radiographs and MRI were compared, with 31 of 32 (97%) having the same acromial type. The authors concluded that MRI was as accurate as radiographs at determining acromial morphology [29]. A similar study compared the value of different MRI planes independently and in combination for assessment of acromial morphology compared with a scapular outlet view radiographs [14]. They found that scapular outlet view radiograph had fair agreement (κ = 0.55) but was superior to any single MRI image position that did not achieve higher than poor agreement (κ = -0.1 to 0.44). However, a combination of two MRI images showed good agreement (κ = 0.66)[14]. This conflicts with the previous study [29], which found that single MRI images were not different from scapular outlet view radiographs. This difference in agreement between the two studies may be due to the already well-documented poor-to-fair interobserver reliability, and as such, to an inability to use the Bigliani classification effectively for research purposes. Although more MRI images showed improved interobserver agreement, employing MRI to identify acromial morphology should be used with caution, if at all. There is conflicting evidence regarding Bigliani’s original observation that rotator cuff tears were associated with Type III acromia [2, 3, 6–8, 10, 13, 17, 23, 24]. Many authors have found a correlation between Type III acromia and rotator cuff tears [2, 3, 6–8, 10, 13]. In one meta-analysis, patients with Type III acromia were more likely to present with rotator cuff tears [18]. By contrast, several other studies found that Bigliani Type III acromia were no more likely to have a rotator cuff tear than shoulders with a Type II or I acromia [17, 23, 24]. On balance, we consider the relationship between acromial morphology as defined by the Bigliani classification and rotator cuff pathology to be weak, if it is present at all. It is not possible to know whether this is because of a flawed anatomic premise (that a more-hooked acromion is more likely to damage the rotator cuff) or because of the considerable problems reported with the reliability of this classification system [4, 9, 11, 25]. Limitations Different observers must be able to agree on the same classification when reviewing the same data (interobserver reliability) before a classification system should be used for clinical decision-making, communication between surgeons, prognosis, or research. Because the interobserver reliability of the Bigliani classification system [3] is only poor-to-fair (κ = 0.28-0.516 [4, 9, 11, 25]), it therefore should not be used for any of these purposes. Some of its problems may derive from the system’s subjective descriptions. The original Bigliani classification does not give objective definitions of flat, curved, or hooked acromia, and thus it entirely depends upon each surgeon’s subjective interpretation. Two studies found that by using standardized criteria to define Type I, Type II, or Type III acromia, the interobserver reliability improved from poor (κ = 0.25 -0.28) to good-to-excellent (κ = 0.62- 0.77) [25, 27]. Whether their finding that standardization of this sort improves the system’s interobserver reliability must be verified by others. Some skepticism may be necessary in this regard, given how sensitive the system appears to be to the quality of the imaging. One study found that the interpretation of acromial morphology is influenced by the angle of the beam in the sagittal, coronal, and axial planes [26]. There is a dearth of research evaluating the effectiveness of the treatment of patients with rotator cuff pathology according to differences in acromial morphology. A single, small study observed that patients with Type III acromia were more likely to undergo surgery after undertaking nonoperative treatment of shoulder impingement than patients with either Types I or II acromia [30]. The utility of this study is limited because it was a small, single-surgeon case series using a questionable shoulder impingement diagnosis [30]. And until or unless surgeons and clinician-scientists find ways to improve the interobserver reliability of the Bigliani classification system, we believe it should not be used to estimate patients’ prognoses. Conclusions The Bigliani classification of acromial morphology is of some historical importance as the first system to classify acromial morphology. However, although it is widely used, we believe it should not be. Its interobserver reliability has generally been in the fair-to-poor range [4, 9], which means that two observers are unlikely to classify the same acromion similarly; consequently, this system simply has no place in clinical communication, determination of prognosis, or research (other than research to try to improve upon the ways we classify the acromion). Attempts to improve reliability have been made using objective criteria to define the distinct types of acromial morphology and were somewhat successful [25, 27], but these findings must be reproduced by others before we should consider them convincing.