Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Nov 1.
Published in final edited form as: Skeletal Radiol. 2021 May 6;50(11):2233–2243. doi: 10.1007/s00256-021-03805-9

Reliability of supraspinatus intramuscular fatty infiltration estimates on T1-weighted MRI in potential candidates for rotator cuff repair surgery: full-thickness tear versus high-grade partial-thickness tear

Derik L Davis 1, Mohit N Gilotra 2, Rodolfo Calderon 3, Andrew Roberts 3, S Ashfaq Hasan 2
PMCID: PMC8565455  NIHMSID: NIHMS1742661  PMID: 33959799

Abstract

Objective

Patients with supraspinatus high-grade partial-thickness tear or full-thickness tear are potential candidates for rotator cuff repair surgery. We sought (1) to compare supraspinatus intramuscular fatty infiltration between these groups by Goutallier grade, fuzzy C-means and an orthopaedic surgeon visible percentage estimate, (2) and to determine the reliability of each method.

Materials and methods

We performed a retrospective cross-sectional study of supraspinatus intramuscular fatty infiltration on T1-weighted MR images for 93 shoulders with either supraspinatus partial-thickness tear > 50% tendon thickness or full-thickness tear by Goutallier grade, fuzzy C-means and an orthopaedic surgeon visible percentage estimate, by two observers for each method. Descriptive statistics were performed to compare groups. Inter- and intra-observer reliability was determined. Correlative analysis among the three methods was performed.

Results

Significant differences of mean supraspinatus intramuscular fatty infiltration were present when comparing supraspinatus high-grade partial-thickness tear versus full-thickness tears by Goutallier grade (p = 0.004), fuzzy C-means (p = 0.002) and orthopaedic surgeon visible percentage estimate (p = 0.001). There was no significant difference for age (55.0 ± 11.1 years versus 56.1 ± 9.6 years) or sex (35.4% male versus 47.8% male) for supraspinatus high-grade partial-thickness tear and full-thickness tear, respectively. A significant difference existed among the subgroup of full-thickness tears stratified by tear size by all three methods (p < 0.020). Inter- and intra-observer reliability was Goutallier grade 0.590 and 0.624, fuzzy C-means 0.768 and 0.925 and orthopaedic surgeon visible percentage estimate 0.858 and 0.686, respectively. For shoulders with mean Goutallier grade ≥ 2.0, inter-observer reliability was 0.878 and 0.802 for fuzzy C-means and orthopaedic surgeon visible percentage estimate, respectively. A strong correlation was present among the three methods of supraspinatus FI analysis (rho ≥ 0.72).

Conclusion

Supraspinatus full-thickness tears have higher amounts of intramuscular fatty infiltration compared to high-grade partial-thickness tear. Quantitative fuzzy C-means shows excellent inter-observer reliability for estimating supraspinatus intramuscular fat. Experienced orthopaedic surgeons’ semi-quantitative estimation of supraspinatus visible intramuscular fat may offer improved reliability as compared to semi-quantitative Goutallier grade.

Keywords: Shoulder, Rotator cuff, Intramuscular fatty infiltration, Goutallier grade, Orthopaedic surgeon, MRI, Fuzzy C-means

Introduction

Rotator cuff (RC) tear is a leading cause of shoulder dysfunction in middle-aged and older adults, with a rate of more than 4 million patient visits every year in the USA to healthcare providers for evaluation of RC disease [1, 2]. Most cases of RC tear involve the supraspinatus tendon (SST) [3, 4]. In the USA, orthopaedic surgeons perform over 270,000 rotator cuff repair (RCR) surgeries every year [5, 6]. The primary indication for RCR surgery is symptomatic full-thickness RC tear or partial-thickness RC tear > 50% tendon thickness, which is also known as high-grade partial-thickness tear. For full-thickness SST tear, orthopaedic surgeons anchor the torn tendon edge to the greater tuberosity of the humerus. Operative management strategies for high-grade partial-thickness SST tear include either (1) conversion of the tear into a full-thickness SST tear by completely transecting the small band of remaining tendon attached to the humeral greater tuberosity intraoperatively during RCR surgery or alternatively (2) a transtendinous repair for partial-thickness articular-surface tears while leaving the bursal-surface tissue intact [7–9].

Currently, consensus among orthopaedic surgeons for one universal standard to select patients for operative management of RC tear remains eluvise. Several patient-specific factors inform orthopaedic surgeons’ clinical decision-making for selecting conservative versus operative management of symptomatic full-thickness RC tear and partial-thickness RC tear > 50% tendon thickness. Factors include patient age, comorbidity, shoulder dysfunction and degree of RC intramuscular fatty infiltration (FI), among others. There is increasing evidence in animal and human studies that the degree of pre-operative FI is predictive of post-operative healing, shoulder function and tendon re-tear rate following RCR surgery, with higher levels of FI showing inferior post-operative outcomes as compared to lower levels of FI. There is a growing consensus among orthopaedic surgeons to perform early RCR surgery for symptomatic RC tear in patients who lack significant FI [10–15].

Shoulder imaging is a key component for orthopaedic surgeons’ pre-operative evaluation of RC tear. Magnetic resonance imaging (MRI) is the primary modality used in the USA to quantify FI and to identify a full-thickness RC tear or partial-thickness RC tear > 50% tendon thickness [7, 16]. The semi-quantitative Goutallier classification system traditionally has been the leading method to classify FI on T1-weighted MR images in clinical medicine, with a Goutallier grade ≥ 3 considered a relative contraindication to RCR surgery [11, 17–19]. Quantitative fuzzy C-means segmentation is an alternative method for the assessment of rotator cuff FI on T1-weighted MR images that may provide superior reliability as compared to semi-quantitative Goutallier grade [20]. Fuzzy C-means is a mathematical clustering algorithm that facilitates evaluation of FI by pattern recognition, through a process of iterative optimization [21, 22].

Although RCR surgery is commonly performed for symptomatic shoulders, there is a gap in the literature explicitly describing differences for supraspinatus muscle FI in shoulders with full-thickness SST tear as compared to shoulders with high-grade partial-thickness SST tear. Therefore, our first objective is to determine differences in the degree of FI in a cohort of shoulders with full-thickness SST tear as compared to a cohort of shoulders with high-grade partial-thickness SST tear on T1-weighted MR images by three different methods, including semi-quantitative Goutallier grade [18, 19, 23], quantitative fuzzy C-means segmentation [20] and a proposed method of semi-quantitative visible FI percentage estimation by subspecialized orthopaedic surgeons with fellowship training in shoulder surgery. Our second objective is to determine inter-rater reliability for each method of FI evaluation. Our first hypothesis is that shoulders with full-thickness SST tear will have a higher mean amount of FI as compared to shoulders with high-grade partial-thickness SST tear. Our second hypothesis is that the quantitative fuzzy C-means technique will have superior reliability as compared to the semi-quantitative Goutallier grade and orthopaedic surgeon visible FI percentage estimate.

Methods

Study population

The study was approved by an institutional review board and complied with Health Insurance Portability and Accountability Act guidelines. The requirement for patient informed consent was waived for this retrospective study. A retrospective search was conducted using a database local to the authors’ institution. The database is inclusive of all clinical shoulder MRI examinations performed from November 2013 to December 2014. The authors searched the shoulder MRI database for all diagnostic radiology final reports which contained a diagnosis of either full-thickness tear or partial-thickness tear of the supraspinatus tendon. Men and women ≥ 18 years were included in the study population. A total of 195 shoulder MRIs were identified. Subsequently, a musculoskeletal radiologist prospectively evaluated the images of each shoulder MRI to verify the tear status of each supraspinatus tendon as either full thickness, partial thickness > 50% tendon thickness or partial thickness ≤ 50% tendon thickness. Inclusion criteria for the study are (1) full-thickness SST tear or (2) partial-thickness SST tear > 50% tendon thickness. Exclusion criteria for the study are (1) absent sagittal T1-weighted MRI sequence; (2) poor quality sagittal T1-weighted MRI sequence; (3) supraspinatus muscle disease process not related to rotator cuff disease; (4) prior joint replacement surgery or (5) prior RCR surgery. A total of 16 shoulder MRIs were excluded from the study for the following reasons: absent sagittal T1-weighted MRI sequence (N = 2); poor quality sagittal T1-weighted MRI sequence (N = 5); metastatic tumor involving the supraspinatus muscle (N = 1); glenohumeral joint replacement (N = 1) and prior RCR surgery (N = 7). Eighty-six shoulders demonstrated a partial-thickness SST tear ≤ 50% tendon thickness. Therefore, a total of 93 shoulders were included in the analysis.

MR imaging

All shoulder MRI examinations were performed at 3.0 T (Magnetom Trio, Siemens, Erlangen, Germany) or 1.5 T (Magnetom Aera/Avanto/Espree, Siemens, Erlangen, Germany) with a dedicated shoulder coil and in general included a two-dimensional (2D) oblique sagittal spin-echo (SE) T1-weighted MR sequence with slice thickness ranging from 3 to 4 mm; repetition time (TR) ranging from 400 to 778 ms; echo time (TE) ranging from 10 to 27 ms and FOV 160 mm2. All shoulder MRI examinations also contained either 2D short-tau inversion recovery (STIR) or turbo spin-echo (TSE) T2-weighted fat saturation sequence in the oblique coronal and oblique sagittal planes and also 2D TSE proton density-weighted sequences in the axial and oblique coronal planes.

MRI evaluation of the supraspinatus tendon

A board-certified musculoskeletal radiologist with 10 years of experience determined the status of the supraspinatus tendon for each shoulder MRI. Each supraspinatus tendon included in the analysis was defined as either (1) full-thickness tear or (2) partial-thickness tear > 50% tendon thickness (Fig. 1). Full-thickness tears were stratified by size according to the Cofield classification system: small (< 1 cm), medium (1 cm to < 3 cm), large (3 cm to < 5 cm) and massive (≥ 5 cm) [16].

Fig. 1.

Fig. 1

Open in a new tab

Example shoulders in each cohort. a Oblique coronal short-tau inversion recovery (STIR) MR image demonstrating a full-thickness tear of the supraspinatus tendon. The arrow shows the edge of the medially retracted tendon. b Oblique coronal STIR MR image shows a high-grade partial-thickness articular-surface tear of the supraspinatus tendon involving > 50% thickness of the tendon. The arrow shows a thin strand of supraspinatus tendon remaining attached to the greater tuberosity footprint of the humerus

Quantitative MRI evaluation of supraspinatus intramuscular fatty infiltration

DICOM image modules were created containing a single oblique sagittal T1-weighted image corresponding to the Y-shaped view for each shoulder (n = 93). Two blinded medical student research assistants performed a quantitative assessment of supraspinatus FI with Medical Image Processing, Analysis and Visualization software (MIPAV, version 7, National Institutes of Health, Bethesda, Maryland, USA). The National Institutes of Health provides MIPAV software which is freely available for download at https://mipav.cit.nih.gov/. The MIPAV provides an automated version of fuzzy C-means segmentation, following the placement of a region of interest (ROI) by human input, which allows for the feasible application of quantitative intramuscular fat percentage estimation in skeletal muscles on T1-weighted MRI images in biomedical clinical research studies or during the clinical care of patients, with the rationale that fuzzy C-means segmentation stratifies MR images into two dominant classes in a similar manner that a human observer would distinguish muscle from fat when assigning a Goutallier grade [20].

In this study, two medical student research assistants independently quantified the visible (1) intramuscular fat and (2) muscle of the supraspinatus on oblique sagittal T1-weighted images with fuzzy C-means by manual segmentation (Fig. 2) [20]. The medical student research assistants used the automated process provided by the MIPAV, following the manual placement of the ROI and input of these parameters: algorithms ➔ segmentation ➔ fuzzy C-means ➔ single channel – number of desired classes, 2; desired exponent value, 2; end tolerance, 0.01; maximum number of iteration, 200; signal threshold, 0.0; region, VOI region; segmentation, hard and fuzzy both. The medical student research assistants placed the ROI at the border edge of the supraspinatus muscle during FI quantification, excluding the extramuscular fat that is also present in the supraspinatus fossa. The percentage of visible FI for each supraspinatus muscle was determined as the cross-sectional area (CSA) of visible intramuscular fat divided by the sum of the CSA for muscle and visible intramuscular fat. To allow for the determination of intra-observer reliability, one medical student research assistant repeated the analysis for 50 shoulders after a period of > 2 weeks.

Fig. 2.

Fig. 2

Open in a new tab

Method of quantitative supraspinatus visible intramuscular fatty infiltration (FI) percentage estimation. a and b Oblique sagittal T1-weighted MR images demonstrating two examples of supraspinatus muscles with mean fuzzy C-means visible FI percentage estimates of 6.0% and 34.7%, respectively. The fuzzy C-means segmentation algorithm estimates the amount of visible FI within a region of interest (gold outline) manually placed at the periphery of the supraspinatus muscle

Semi-quantitative MRI evaluation of supraspinatus intramuscular fatty infiltration

Two blinded board-certified orthopaedic surgeons performed a semi-quantitative assessment of supraspinatus FI. One orthopaedic surgeon with > 15 years of experience and fellowship training both in shoulder and elbow surgery and in sports medicine, and another orthopaedic surgeon with 7 years of experience and fellowship training in shoulder and elbow surgery, independently used a DICOM image viewer module (Aquarius iNtuition Edition, version 4.4; TeraRecon Inc., Foster City, California, USA) to assign a Goutallier grade for each supraspinatus muscle on a single oblique sagittal T1-weighted image corresponding to the Y-shaped view for each shoulder (n = 93) using the 5-point Goutallier classification system: grade 0, no fat; grade 1, streaks of fat; grade 2, muscle > fat; grade 3, muscle = fat; grade 4, muscle < fat. In addition to assigning a Goutallier grade, each orthopaedic surgeon independently estimated the percentage of supraspinatus visible intramuscular fat by visual inspection, on a scale from 0 to 100% (Fig. 3). The orthopaedic surgeons assigned the supraspinatus Goutallier grade and the visible FI percentage estimate based on the amount of intramuscular fat perceived within the supraspinatus muscle, excluding the extramuscular fat that is also present in the supraspinatus fossa from the assessment. To allow for the determination of intra-observer reliability, one orthopaedic surgeon repeated the analysis for 50 shoulders after a period of > 2 weeks.

Fig. 3.

Fig. 3

Open in a new tab

Method of semi-quantitative supraspinatus visible intramuscular fatty infiltration (FI) percentage estimation. a and b oblique sagittal T1-weighted MR images demonstrating two examples of supraspinatus muscles with varying amounts of visible FI. First, a Goutallier grade was assigned for the supraspinatus muscle using the 5-point Goutallier classification system—mean Goutallier grades of 1.5 and 3.0 respectively. Then, the percentage of visible FI was estimated based on visual inspection of the supraspinatus muscle—mean orthopaedic surgeon visible FI percentage estimates of 8.5% and 46.0%, respectively

Statistical analysis

Statistical analysis was performed using SAS statistical software version 9.4 (SAS, Cary, North Carolina). Mean cohort characteristics between the full-thickness SST cohort and partial-thickness SST tear > 50% tendon thickness cohort were compared with the Mann–Whitney U test and the chi-square test. Comparison of full-thickness SST tear sub-groups stratified by tear size was performed with the Kruskal–Wallis test. The Spearman rank correlation (Rho) test was used to determine the correlation among the three methods of FI evaluation. Rho correlation was interpreted as 0.00 to 0.35, weak; 0.36 to 0.67, moderate and 0.68 to 1.00, strong [24]. Inter-observer and intra-observer reliability for the fuzzy C-means and the orthopaedic surgeon estimate of supraspinatus visible FI percentage were assessed by the intraclass correlation coefficient. Goutallier grade was assessed by weighted kappa statistic. Inter-observer and intra-observer reliability for was interpreted as 0.00 to 0.39, poor; 0.40 to 0.59, fair; 0.60 to 0.74, good and 0.75 to 1.00, excellent [25]. A p value < 0.05 was considered a statistically significant difference.

Results

The study population had a mean age of 55.7 ± 10.0 years and was 47.8% male. The characteristics of shoulders by cohort are shown in Table 1. There was no significant difference between cohorts for age or sex. Significantly greater amounts of supraspinatus intramuscular fat were present in shoulders with full-thickness SST tear versus those with high-grade partial-thickness SST tear for all measures: mean fuzzy C-means visible FI percentage estimate, 16.4% ± 9.3% versus 10.7% ± 7.3%, p = 0.002; mean Goutallier grade 1.6 ± 0.8 versus 1.1 ± 0.6, p = 0.004 and mean orthopaedic surgeon visible FI percentage estimate 15.4% ± 14.8% versus 6.2% ± 4.8%, p = 0.001. The distribution of the study population by Goutallier grade is as follows: mean grade < 1 (n = 18), 1 ≤ mean grade < 2 (n = 41), 2 ≤ mean grade < 3 (n = 30), 3 ≤ mean grade < 4 (n = 3) and mean grade = 4 (n = 1).

Table 1.

Characteristics of the study population

Supraspinatus full-thickness tear cohort (n = 62) Supraspinatus partial-thickness tear (> 50%) cohort (n = 31) P-value
Age, years 56.1 ± 9.6 55.0 ± 11. 1 0.925
Male, % 47.8 35.4 0.141
Fuzzy C-means FIA estimate, % 16.4 ± 9.3 10.7 ± 7.3 0.002
Goutallier gradeB 1.6 ± 0.8 1.1 ± 0.6 0.004
Orthopaedic surgeon FIA estimate, % 15.4 ± 14.8 6.2 ± 4.8 0.001
Open in a new tab

Values are mean ± standard deviation unless otherwise noted

A

Supraspinatus visible intramuscular fatty infiltration

B

Supraspinatus muscle

For the sub-population of shoulders with full-thickness SST tear stratified by tear size, the frequency was 37.1% small (n = 23), 41.9% medium (n = 26), 14.5% large (n = 9) and 6.5% massive (n = 4). There was no significant difference in mean age among the sub-populations of full-thickness SST tear. Significant differences existed among the sub-populations of full-thickness SST tear for fuzzy C-means visible FI percentage estimate (p = 0.001), Goutallier grade (p = 0.014) and orthopaedic surgeon visible FI percentage estimate (p = 0.004) (Fig. 4). Mean and standard deviation of fuzzy C-means visible FI percentage estimate were stratified by size: small, 11.2% ± 6.6%; medium, 17.7% ± 8.2%; large, 20.5% ± 11.5; massive, 28.3% ± 8.7%. Mean and standard deviation of Goutallier grade were stratified by size: small, 1.2 ± 0.7; medium, 1.7 ± 0.6; large, 1.9 ± 0.6; massive, 2.4 ± 1.3. The maximum degree of discordance for Goutallier grade was 1 grade. There were no differences in Goutallier grade by 2 or more grades between the blinded orthopaedic surgeons for the study population. Mean and standard deviation of orthopaedic surgeon visible FI percentage estimate were stratified by size: small, 8.5% ± 9.2%; medium, 16.9% ± 12.5%; large, 17.8% ± 13.3%; massive, 39.0 ± 29.9%.

Fig. 4.

Fig. 4

Open in a new tab

Boxplots showing the distribution of supraspinatus intramuscular fatty infiltration (FI) for shoulders with full-thickness tear stratified by tear size for a mean fuzzy C-means visible FI percentage estimate, b mean Goutallier grade and c mean orthopaedic surgeon visible FI estimate. For each method, the order of mean (diamond) FI is small < medium < large < massive when stratified by tear size

There was excellent inter-observer reliability for the fuzzy C-means supraspinatus visible FI percentage estimate (ICC, 0.768) and also for the inter-observer orthopaedic surgeon supraspinatus visible FI percentage estimate (ICC, 0.858) (Table 2). By contrast, there was fair inter-observer reliability for Goutallier grade (weighted kappa, 0.590). There was excellent intra-observer reliability for the fuzzy C-means supraspinatus visible FI estimate (ICC, 0.925). Good intra-observer reliability was present for orthopaedic surgeon supraspinatus visible FI estimate (ICC, 0.686) and Goutallier grade (weighted kappa, 0.624). For the subset of supraspinatus muscles (n = 34) with clearly more than streaks of fat showing a mean Goutallier grade ≥ 2.0, the inter-observer ICC for fuzzy C-means supraspinatus visible FI percentage estimate and orthopaedic surgeon visible FI percentage estimate was excellent at 0.878 and 0.802, respectively.

Table 2.

Inter-observer and intra-observer reliability for supraspinatus visible intramuscular fatty infiltration (FI) for the study population

Inter-observer reliability Intra-observer reliability
Fuzzy C-means visible FI % estimateA 0.768 0.925
Goutallier gradeB 0.590 0.624
Orthopaedic surgeon visible FI % estimateA 0.858 0.686
Open in a new tab
A

Intraclass correlation coefficient

B

Weighted kappa statistic

A strong correlation was found among the methods of supraspinatus FI on T1-weighted MR images for the study population. The correlation between fuzzy C-means visible FI percentage estimate and orthopaedic surgeon visible FI percentage estimate was strong: rho = 0.74, p < 0.001 (Fig. 5). The fuzzy C-means visible FI percentage estimate and orthopaedic surgeon visible FI percentage estimate were discrepant by an absolute difference of ≥ 20 percentage points in less than 5% (4 out of 93). A strong correlation also was found between fuzzy C-means and Goutallier grade: rho = 0.72, p < 0.001. The correlation between orthopaedic surgeon visible FI percentage estimate and Goutallier grade was strong: rho = 0.95, p < 0.001.

Fig. 5.

Fig. 5

Open in a new tab

Correlation of visible supraspinatus intramuscular fatty infiltration between mean fuzzy C-means percentage estimation and mean orthopaedic surgeon percentage estimation for the study population

Discussion

Our study shows supraspinatus intramuscular fatty infiltration differs significantly in potential candidates for RCR surgery when comparing full-thickness SST tear to high-grade partial-thickness SST tear. All three methods—quantitative fuzzy C-means, semi-quantitative Goutallier grade and semi-quantitative orthopaedic surgeon visible FI percentage estimate—showed a statistically significant higher mean level of FI for the full-thickness SST tear cohort as compared to the high-grade partial-thickness SST tear cohort. Quantitative fuzzy C-means demonstrates excellent inter-observer reliability and intra-observer reliability for supraspinatus visible FI percentage estimate. Our study also suggests that subspecialized orthopaedic surgeon estimation of supraspinatus visible FI percentage may be a step toward improving inter-observer reliability for semi-quantitative evaluation of FI as compared to the Goutallier classification system.

Research studies in humans show an advanced progression of FI shortly after full-thickness RC tear [10, 11, 17]. Animal studies in tenotomy models of full-thickness rotator cuff tear provide experimental evidence that support the observation of abnormal rapid proliferation of FI occurring secondary to muscle disuse shortly following the surgical transection of the tendon [12, 14, 15, 26]. There is also increasing evidence that pre-operative FI is predictive of post-operative shoulder function, healing and tendon re-tear after RCR surgery, and there is growing consensus to perform the early operative intervention in patients who lack significant FI [10–13]. In a rabbit model of full-thickness SST tear, Valencia et al. demonstrated a strong inverse correlation between supraspinatus muscle contractile strength and degree of FI, with the weakest muscles having the greatest amounts of FI [27]. The increase in cytokines, adipokines and chemokines such as IL-1ß58 released by the increasing number of intramuscular adipocytes are posited to induce increasing levels of oxidative stress and inflammation that reduce the contractile capacity of local myofibers [28–30]. Since FI is considered irreversible, successful RCR surgery mitigates further FI deposition and helps preserve muscle function [10, 11, 13–16, 26, 31]. Early RCR surgery increases the likelihood to improve post-operative outcomes as compared to delayed RCR surgery [17, 24, 31].

Pre-operative imaging assessment to quantify FI in potential candidates for RCR surgery is a key component of the clinical decision-making process since traditionally a Goutallier grade of ≥ 3 is a relative contraindication to operative management [11, 17–19, 23]. In the USA, MRI is the primary modality used to both evaluate FI and accurately characterize RC tear pattern and size for patients who are candidates for RCR surgery [16, 32]. MRI enables pre-operative identification of full-thickness RC tear or partial-thickness RC tear > 50% tendon thickness and is of value since arthroscopy has limited intraoperative sensitivity to diagnose partial-thickness RC tear > 50% tendon thickness with confidence [16]. However, best practices and clinical indications for RCR surgery remain controversial. Patients’ age, body mass index, comorbidities, objective shoulder function, perception of shoulder pain, physical limitations, future activity level and rotator cuff FI are all pertinent factors which inform orthopaedic surgeons’ decision to offer RCR surgery [11, 33–36]. Appropriate patient selection improves the likelihood that RCR surgery will decrease shoulder pain and restore shoulder strength and range of motion [10, 11, 34, 37].

Human studies suggest that significantly higher levels of FI can be expected to develop in full-thickness RC tears, although not necessarily in partial-thickness RC tears, as compared to shoulders with no RC tear [38–40]. Thus, patients with symptomatic high-grade partial-thickness tear who are potential candidates for RCR surgery should most likely be expected to exhibit lower levels of FI before any progression to development of a full-thickness RC tear. Our study investigates explicitly differences in supraspinatus FI between full-thickness SST tear and high-grade partial-thickness SST tear since these are the two groups relevant for a potential RCR surgery [16, 41–45]. Historically, high-grade partial-thickness SST tear is converted to a full-thickness tear intraoperatively then receives a RCR surgery similar to a full-thickness SST tear [7–9]. A retrospective study performed by Lee et al. suggested that high-grade partial-thickness SST tear had a lower mean FI by chemical shift MR fat fraction than full-thickness SST tear in a population of clinical patients receiving shoulder MRI [40]. However, these investigators neither evaluated explicitly the differences between the group of patients with full-thickness SST tear and high-grade partial-thickness SST tear nor reported differences in FI for these two groups by fuzzy C-means visible FI percentage estimate, Goutallier grade or orthopaedic surgeon visible FI percentage estimate.

Our study is consistent with prior animal and human studies that suggest an association between the size of full-thickness SST tear and the degree of supraspinatus FI [11, 32, 40, 46]. Shoulders with full-thickness SST tear in our study demonstrated a stepwise increase in mean FI for the sub-populations with small to massive full-thickness SST tear size by all methods—fuzzy C-means, Goutallier grade and the orthopaedic surgeon visible FI percentage estimate.

As expected, the quantitative supraspinatus fuzzy C-means visible FI estimate showed excellent inter-observer and intra-observer reliability in our study [20]. Fuzzy C-means is a mathematical clustering algorithm that facilitates pattern recognition by allocating each voxel within a region of interest (ROI) to a pre-defined category of signal intensity based on a process of iterative optimization [21]. Fuzzy C-means segmentation has been used in medical imaging for clinical scenarios on MRI, such as separating muscle from fat in the extremity or fibroglandular tissue from fat in the breast [20, 22, 47]. When evaluating supraspinatus muscles with clearly more than streaks of fat (Goutallier grade ≥ 2.0), there was excellent inter-observer reliability for the fuzzy C-means visible FI percentage estimate. Quantitative evaluation for these types of muscles with the fuzzy C-means technique can offer increased confidence in clinical decision-making when considering RCR surgery for patients with symptomatic full-thickness SST tear or high-grade partial-thickness SST tear. This is especially true when there is uncertainty as to whether there is more muscle than fat (Goutallier grade 2) versus muscle = fat (Goutallier grade 3) since Goutallier grade 3 is a relative contraindication to RCR surgery [11, 17–19, 23].

The excellent inter-observer reliability of our study’s fellowship-trained subspecialist orthopaedic surgeon semi-quantitative estimate of visible FI percentage was not expected. The fact that the orthopaedic surgeon’s visible FI percentage estimate showed a higher level of agreement as compared to the well-established Goutallier grade was surprising. The Goutallier classification system is currently the most commonly used method to quantify rotator cuff FI in clinical practice. Yet, the Goutallier classification system is well known for having poor to fair inter-observer reliability, in line with our study [40, 48–52]. In the typical workflow of clinical practice, experienced orthopaedic surgeons in the care of patients routinely evaluate rotator cuff FI as part of the clinical decision-making process. Orthopaedic surgeons’ gestalt to estimate supraspinatus visible FI percentage is inherent in their daily workflow since this process directly influences their patients’ eligibility for RCR surgery. In general, however, these gestalts are converted into the language of Goutallier grade for routine communication with other healthcare providers or in research studies. Only Goutallier grade 0 (no fat, 0%) and Goutallier grade 3 (muscle = fat, 50%) theoretically translate into exact percentages of FI. The fair-at-best inter-observer reliability of Goutallier grade is in part related to differences among raters’ opinion for which grade best characterizes their own gestalt of the FI percentage estimate. To improve reproducibility, researchers have used modified 2- or 3-point ordinal scale versions of the Goutallier classification system. Although the modified Goutallier grading scales show improved reliability as compared to the 5-point ordinal scale, potential bias is introduced into the research study design since there is a reduced stratification of study participants and a greater concern for an increased number of misclassification errors [48]. Therefore, our study sought to evaluate the reliability of a proposed subspecialized orthopaedic surgeons’ visual estimation of FI percentage, to test its feasibility as a semi-quantitative alternative to the Goutallier classification system.

Our study is not without limitations. The study was retrospective and cross-sectional in design. We identified pre-existing shoulder MRI examinations at one time point among study participants receiving clinical evaluation at our institution. The local database used for the study contains only information that was available in our institution’s imaging picture archiving and communication system (PACS) and does not include the clinical information present in our institution’s separate electronic health record (EHR). The MR images, along with the age and sex of the study participants included with the images on PACS, were included in the analysis of this imaging study. The study was not designed to compare clinical outcomes among study participants nor was the study designed to compare clinical outcomes between study participants who went on to receive a RCR surgery versus those who only received non-operative treatment. The study also did not account for the chronicity of supraspinatus tendon tears. Despite our study’s excellent inter-observer reliability for experienced subspecialty fellowship-trained orthopaedic surgeons to estimate supraspinatus visible FI percentage, future research is needed to determine the reliability (1) among subspecialty trained orthopaedic surgeons who work at different institutions or practices, (2) among general orthopaedic surgeons with no fellowship training in shoulder surgery, (3) among orthopaedic surgery residents, (4) among radiologists and (5) among radiologistsin-training. Future studies with larger sample sizes and ≥ 3 observers are also warranted to evaluate the reproducibility of results for Goutallier grade and the orthopaedic surgeon visible FI percentage estimate, given potential differences in the statistical power between ordinal and continuous variables. Our study may lack generalizability to populations of adults with RC tear and supraspinatus muscles classified as Goutallier grade ≥ 3 since only 4% of study participants had this presentation. However, prior studies have suggested that shoulders that lack advanced changes of FI are the ideal population for RC tear clinical research studies since symptomatic patients with Goutallier grade ≤ 2 are the favored population by orthopaedic surgeons to be considered for a potential RCR surgery [23]. Distinguishing higher grades of FI is unnecessary as both Goutallier grades 3 and 4 are relative contraindications for repair. We did not include study participants who had a prior history of RCR surgery. Future studies will be needed to compare supraspinatus FI in patients with full-thickness SST re-tear versus patients with high-grade partial-thickness SST tear, following RCR surgery.

Conclusion

Patients with full-thickness SST tear have greater amounts of supraspinatus intramuscular fatty infiltration as compared to those with high-grade partial-thickness SST tear, in the population of individuals who are potential candidates for RCR surgery. Quantitative fuzzy C-means visible FI percentage estimate has excellent inter-observer and intra-observer reliability. Our study also suggests that inter-observer agreement among experienced orthopaedic surgeons with fellowship training in shoulder surgery for the semi-quantitative estimation of the visible percentage of FI in the supraspinatus muscle may offer improved reliability as compared to Goutallier grade, but future research is warranted to vet the reproducibility of this observation.

Acknowledgements

The authors acknowledge Dr. Ranyah Almardawi, Department of Diagnostic Radiology & Nuclear Medicine, University of Maryland School of Medicine, for her contribution to data management for this study.

Conflict of interest

Dr. Derik L. Davis receives partial salary support from the University of Maryland Claude D. Pepper Older Americans Independence Center (NIA 3P30AG028747-13S1) and the National Institute on Aging (NIA 1R03AG067927-01). Dr. Mohit N. Gilotra, MD is a consultant for Tigon.

References

  • 1.McElvany MD, McGoldrick E, Gee AO, Neradilek MB, Matsen FA 3rd. Rotator cuff repair: published evidence on factors associated with repair integrity and clinical outcome. Am J Sports Med. 2015;43:491–500. [DOI] [PubMed] [Google Scholar]
  • 2.Opsha O, Malik A, Baltazar R, et al. MRI of the rotator cuff and internal derangement. Eur J Radiol. 2008;68:36–56. [DOI] [PubMed] [Google Scholar]
  • 3.Yamamoto A, Takagishi K, Osawa T, et al. Prevalence and risk factors of a rotator cuff tear in the general population. J Shoulder Elbow Surg. 2010;19:116–20. [DOI] [PubMed] [Google Scholar]
  • 4.Melis B, Nemoz C, Walch G. Muscle fatty infiltration in rotator cuff tears: descriptive analysis of 1688 cases. Orthop Traumatol Surg Res. 2009;95:319–24. [DOI] [PubMed] [Google Scholar]
  • 5.Jain NB, Higgins LD, Losina E, Collins J, Blazar PE, Katz JN. Epidemiology of musculoskeletal upper extremity ambulatory surgery in the United States. BMC Musculoskelet Disord. 2014;15:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Colvin AC, Egorova N, Harrison AK, Moskowitz A, Flatow EL. National trends in rotator cuff repair. J Bone Joint Surg Am. 2012;94:227–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Franceschi F, Papalia R, Del Buono A, et al. Articular-sided rotator cuff tears: which is the best repair? A three-year prospective randomised controlled trial. Int Orthop. 2013;37:1487–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Shin SJ. A comparison of 2 repair techniques for partial-thickness articular-sided rotator cuff tears. Arthroscopy. 2012;28:25–33. [DOI] [PubMed] [Google Scholar]
  • 9.Shin SJ, Kook SH, Rao N, Seo MJ. Clinical outcomes of modified Mason-Allen single-row repair for bursal-sided partial-thickness totator cuff tears: comparison with the double-row suture-bridge technique. Am J Sports Med. 2015;43:1976–82. [DOI] [PubMed] [Google Scholar]
  • 10.Liem D, Lichtenberg S, Magosch P, Habermeyer P. Magnetic resonance imaging of arthroscopic supraspinatus tendon repair. J Bone Joint Surg Am. 2007;89:1770–6. [DOI] [PubMed] [Google Scholar]
  • 11.Gladstone JN, Bishop JY, Lo IK, Flatow EL. Fatty infiltration and atrophy of the rotator cuff do not improve after rotator cuff repair and correlate with poor functional outcome. Am J Sports Med. 2007;35:719–28. [DOI] [PubMed] [Google Scholar]
  • 12.Gerber C, Meyer DC, Schneeberger AG, Hoppeler H, von Rechenberg B. Effect of tendon release and delayed repair on the structure of the muscles of the rotator cuff: an experimental study in sheep. J Bone Joint Surg Am. 2004;86(9):1973–82. [DOI] [PubMed] [Google Scholar]
  • 13.Goutallier D, Postel JM, Gleyze P, Leguilloux P, Van Driessche S. Influence of cuff muscle fatty degeneration on anatomic and functional outcomes after simple suture of full-thickness tears. J Shoulder Elbow Surg. 2003;12:550–4. [DOI] [PubMed] [Google Scholar]
  • 14.Coleman SH, Fealy S, Ehteshami JR, et al. Chronic rotator cuff injury and repair model in sheep. J Bone Joint Surg Am. 2003;85(12):2391–402. [DOI] [PubMed] [Google Scholar]
  • 15.Uhthoff HK, Coletta E, Trudel G. Effect of timing of surgical SSP tendon repair on muscle alterations. J Orthop Res. 2014;32:1430–5. [DOI] [PubMed] [Google Scholar]
  • 16.Morag Y, Jacobson JA, Miller B, De Maeseneer M, Girish G, Jamadar D. MR imaging of rotator cuff injury: what the clinician needs to know. Radiographics. 2006;26:1045–65. [DOI] [PubMed] [Google Scholar]
  • 17.Deniz G, Kose O, Tugay A, Guler F, Turan A. Fatty degeneration and atrophy of the rotator cuff muscles after arthroscopic repair: does it improve, halt or deteriorate? Arch Orthop Trauma Surg. 2014;134:985–90. [DOI] [PubMed] [Google Scholar]
  • 18.Goutallier D, Postel JM, Bernageau J, Lavau L, Voisin MC. Fatty muscle degeneration in cuff ruptures Pre-and postoperative evaluation by CT scan. Clin Orthop Relat Res. 1994;304:78–83. [PubMed] [Google Scholar]
  • 19.Ashry R, Schweitzer ME, Cunningham P, Cohen J, Babb J, Cantos A. Muscle atrophy as a consequence of rotator cuff tears: should we compare the muscles of the rotator cuff with those of the deltoid? Skeletal Radiol. 2007;36:841–5. [DOI] [PubMed] [Google Scholar]
  • 20.Davis DL, Kesler T, Gilotra MN, et al. Quantification of shoulder muscle intramuscular fatty infiltration on T1-weighted MRI: a viable alternative to the Goutallier classification system. Skeletal Radiol. 2019;48:535–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Feng Y, Guo H, Zhang H, et al. A modified fuzzy C-means method for segmenting MR images using non-local information. Technol Health Care. 2016;24(Suppl 2):S785–93. [DOI] [PubMed] [Google Scholar]
  • 22.Clendenen TV, Zeleniuch-Jacquotte A, Moy L, Pike MC, Rusinek H, Kim S. Comparison of 3-point Dixon imaging and fuzzy C-means clustering methods for breast density measurement. J Magn Reson Imaging. 2013;38:474–81. [DOI] [PubMed] [Google Scholar]
  • 23.Fuchs B, Weishaupt D, Zanetti M, Hodler J, Gerber C. Fatty degeneration of the muscles of the rotator cuff: assessment by computed tomography versus magnetic resonance imaging. J Shoulder Elbow Surg. 1999;8:599–605. [DOI] [PubMed] [Google Scholar]
  • 24.Vidt ME, Santago AC 2nd, Hegedus EJ, et al. Can self-report instruments of shoulder function capture functional differences in older adults with and without a rotator cuff tear? J Electromyogr Kinesiol. 2016;29:90–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess. 1994;6:284–90. [Google Scholar]
  • 26.Uhthoff HK, Matsumoto F, Trudel G, Himori K. Early reattachment does not reverse atrophy and fat accumulation of the supraspinatus–an experimental study in rabbits. J Orthop Res. 2003;21:386–92. [DOI] [PubMed] [Google Scholar]
  • 27.Valencia AP, Lai JK, Iyer SR, et al. Fatty infiltration is a prognostic marker of muscle function after rotator cuff tear. Am J Sports Med. 2018;46:2161–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Buck M, Chojkier M. Muscle wasting and dedifferentiation induced by oxidative stress in a murine model of cachexia is prevented by inhibitors of nitric oxide synthesis and antioxidants. EMBO J. 1996;15:1753–65. [PMC free article] [PubMed] [Google Scholar]
  • 29.Baumann CW, Kwak D, Liu HM, Thompson LV. Age-induced oxidative stress: how does it influence skeletal muscle quantity and quality? J Appl Physiol. 2016;121:1047–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Lamounier-Zepter V, Ehrhart-Bornstein M, Karczewski P, Haase H, Bornstein SR, Morano I. Human adipocytes attenuate cardiomyocyte contraction: characterization of an adipocyte-derived negative inotropic activity. FASEB J. 2006;20:1653–9. [DOI] [PubMed] [Google Scholar]
  • 31.Hantes ME, Karidakis GK, Vlychou M, Varitimidis S, Dailiana Z, Malizos KN. A comparison of early versus delayed repair of traumatic rotator cuff tears. Knee Surg Sports Traumatol Arthrosc. 2011;19:1766–70. [DOI] [PubMed] [Google Scholar]
  • 32.Davis DL, Gilotra MN, Hovis JP, Almardawi R, Hasan SA. Association of rotator cuff tear patterns and intramuscular fatty infiltration on magnetic resonance imaging. J Clin Imaging Sci. 2019;9:38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Dunn WR, Schackman BR, Walsh C, et al. Variation in orthopaedic surgeons’ perceptions about the indications for rotator cuff surgery. J Bone Joint Surg Am. 2005;87:1978–84. [DOI] [PubMed] [Google Scholar]
  • 34.McMahon PJ, Prasad A, Francis KA. What is the prevalence of senior-athlete rotator cuff injuries and are they associated with pain and dysfunction? Clin Orthop Relat Res. 2014;472:2427–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Varkey DT, Patterson BM, Creighton RA, Spang JT, Kamath GV. Initial medical management of rotator cuff tears: a demographic analysis of surgical and nonsurgical treatment in the United States Medicare population. J Shoulder Elbow Surg. 2016;25:e378–85. [DOI] [PubMed] [Google Scholar]
  • 36.Kweon C, Gagnier JJ, Robbins CB, Bedi A, Carpenter JE, Miller BS. Surgical versus nonsurgical management of rotator cuff tears: predictors of treatment allocation. Am J Sports Med. 2015;43:2368–72. [DOI] [PubMed] [Google Scholar]
  • 37.Piper CC, Hughes AJ, Ma Y, Wang H, Neviaser AS. Operative versus nonoperative treatment for the management of full-thickness rotator cuff tears: a systematic review and meta-analysis. J Shoulder Elbow Surg. 2018;27:572–6. [DOI] [PubMed] [Google Scholar]
  • 38.Davis DL, Almardawi R, Henn RF 3rd, et al. Correlation of quantitative versus semiquantitative measures of supraspinatus intramuscular fatty infiltration to shoulder range of motion and strength: a pilot study. Curr Probl Diagn Radiol. 2020. [epub ahead of print]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Kim HM, Dahiya N, Teefey SA, Keener JD, Galatz LM, Yamaguchi K. Relationship of tear size and location to fatty degeneration of the rotator cuff. J Bone Joint Surg Am. 2010;92:829–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Lee S, Lucas RM, Lansdown DA, et al. Magnetic resonance rotator cuff fat fraction and its relationship with tendon tear severity and subject characteristics. J Shoulder Elbow Surg. 2015;24:1442–51. [DOI] [PubMed] [Google Scholar]
  • 41.Deutsch A Arthroscopic repair of partial-thickness tears of the rotator cuff. J Shoulder Elbow Surg. 2007;16:193–201. [DOI] [PubMed] [Google Scholar]
  • 42.Longo UG, Risi Ambrogioni L, Candela V, et al. Conservative versus surgical management for patients with rotator cuff tears: a systematic review and META-analysis. BMC Musculoskelet Disord. 2021;22:50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Liu JN, Garcia GH, Gowd AK, et al. Treatment of partial thickness rotator cuff tears in overhead athletes. Curr Rev Musculoskelet Med. 2018;11:55–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Katthagen JC, Bucci G, Moatshe G, Tahal DS, Millett PJ. Improved outcomes with arthroscopic repair of partial-thickness rotator cuff tears: a systematic review. Knee Surg Sports Traumatol Arthrosc. 2018;26:113–24. [DOI] [PubMed] [Google Scholar]
  • 45.Matthewson G, Beach CJ, Nelson AA, et al. Partial thickness rotator cuff tears: current concepts. Adv Orthop. 2015;2015:458786. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Kim HM, Galatz LM, Lim C, Havlioglu N, Thomopoulos S. The effect of tear size and nerve injury on rotator cuff muscle fatty degeneration in a rodent animal model. J Shoulder Elbow Surg. 2012;21:847–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Orgiu S, Lafortuna CL, Rastelli F, Cadioli M, Falini A, Rizzo G. Automatic muscle and fat segmentation in the thigh from T1-Weighted MRI. J Magn Reson Imaging. 2016;43:601–10. [DOI] [PubMed] [Google Scholar]
  • 48.Davis DL, Zhuo J, Almardawi R, Mulligan ME, Resnik CR, Abdullah SB, Al Khalifah H, Henn RF 3rd, Gilotra MN, Hasan SA, Gullapalli RP. Association of patient self-reported shoulder scores to quantitative and semi-quantitative MRI measures of rotator cuff intramuscular fatty infiltration: a pilot study. AJR Am J Roentgen. 2019;213:1307–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Horiuchi S, Nozaki T, Tasaki A, et al. Reliability of MR quantification of rotator cuff muscle fatty degeneration using a 2-point Dixon technique in comparison with the Goutallier classification: validation study by multiple readers. Acad Radiol. 2017;24:1343–51. [DOI] [PubMed] [Google Scholar]
  • 50.Lippe J, Spang JT, Leger RR, Arciero RA, Mazzocca AD, Shea KP. Inter-rater agreement of the Goutallier, Patte, and Warner classification scores using preoperative magnetic resonance imaging in patients with rotator cuff tears. Arthroscopy. 2012;28:154–9. [DOI] [PubMed] [Google Scholar]
  • 51.Slabaugh MA, Friel NA, Karas V, Romeo AA, Verma NN, Cole BJ. Interobserver and intraobserver reliability of the Goutallier classification using magnetic resonance imaging: proposal of a simplified classification system to increase reliability. Am J Sports Med. 2012;40:1728–34. [DOI] [PubMed] [Google Scholar]
  • 52.Spencer EE Jr, Dunn WR, Wright RW, et al. Interobserver agreement in the classification of rotator cuff tears using magnetic resonance imaging. Am J Sports Med. 2008;36:99–103. [DOI] [PubMed] [Google Scholar]

RESOURCES