The main finding of this study is that MRI evaluation revealed the complete absence of fatty infiltration in half of the shoulders and only minimal infiltration in the remaining half (Fuchs grade 1), demonstrating the positive and effective biological response of the tissues, and thus confirming the hypothesis that remplissage is a safe procedure which, over the years, does not lead to pathological changes in the involved structures. MRI was chosen to detect the appearance of fatty infiltration in the context of the infraspinatus muscle due to the high-quality scans produced, which allowed for more accurate detection of fatty infiltration and an evaluation of the quality of the scar tissue placed at the tendon–bone interface.
High-grade filling (FISOR grade 3 or 4) of the Hill–Sachs lesion was successfully observed in 12 shoulders (85.71%), with hypointense healthy dehydrated fibrous scar tissue appearing between the infraspinatus tendon and the humeral head, demonstrating the healing of the capsulotendodesis. In the two cases of recurrence, the filling of the Hill–Sachs was < 25% (FISOR grade 1).
These results are in agreement with those reported by Park et al. [20], who noticed a complete absence of consistent fatty infiltration in MRI scans of 11 shoulders subjected to remplissage (with a mean Hill–Sachs volume of 334.3 mm3, significantly shorter than that reported in our study) during a mean follow-up of 18 months (minimum follow-up: 8.8 months). A time period of 18 months is probably too short to detect early signs of degeneration and assess the complete healing of the capsulotenodesis; in fact, Park et al. found the presence of granulation tissue in 72.72% of cases, and this finding seems to demonstrate that the healing process of those shoulders was still ongoing.
This finding is similar to that reported by Boileau et al. [23], who performed a radiological evaluation of 42 patients (CT arthrography in 38 patients and MRI in 4 patients; mean follow-up: 24 months) who underwent the remplissage procedure. Those authors highlighted capsulotenodesis healing in all the patients and noted that the Hill–Sachs lesions were more than 75% filled in 73.8% of the patients. This result, which was based on a larger sample size, confirms that filling the Hill–Sachs lesion is an essential condition to obtain healing of the tissue; however, the use of CT as the primary imaging device could represent an inherent bias of this study, as it is not the best choice to evaluate the quality of scar tissue.
Radiological evaluation of the two recurrences (14.29%) in the present study revealed significantly larger Hill-Sachs volumes (1963.9 mm3 and 2049.32 mm3, respectively) and adipose tissue located at the tendon-bone interface of the capsulotenodesis, suggesting that complete filling could be essential for the healing process.
The use of a single metallic anchor in Hill–Sachs of this size might cause recurrences. Still, it is impossible to know precisely if the larger size of the Hill–Sachs lesion, the tendon detachment, and the appearance of fatty tissue were direct consequences of the new trauma or were previously present.
No complete thickness rotator cuff tears were identified in any of the shoulders upon MRI examination, and the infraspinatus tendon was completely intact, showing no signs of fraying or irritation. Supraspinatus tendon partial tears were found in four shoulders, with the larger ones (A2B2 according to Snyder’s classification) affecting the two recurrences. A possible explanation for these findings could be the uncoupling between static and dynamic stabilizers in these two shoulders following the stabilization failure, with eccentric tensile forces causing significant stress at the supraspinatus tendon and, over time, a consequent tear [24].
The postoperative clinical and functional evaluation showed good-to-excellent outcomes; in particular, the Walch–Duplay Score (95.00 [80.00–100.00]) was excellent in nine shoulders (64.28%), good in four shoulders (28.67%), and medium in one shoulder (7.14%). Similar results were also reported by Merolla et al. [25] based on 61 remplissage procedures; good functional outcomes were maintained, with a mean CMS of 90 points and a mean Walch–Duplay Score of 90.4 after a mean follow-up of 39.5 months.
A relevant finding of this study is the statistically significant loss (median: 10°) of ROM in external rotation, both in ER1 (p = 0.0005) and in ER2 (p = 0.0010), compared to healthy contralateral shoulders. These findings follow those reported by Boileau et al. [23], who referred to mean deficits of 8° (± 7°) in ER1 (p < 0.001) and 9° (± 7°) in ER2 (p < 0.001), and Merolla et al. [25], who found a significantly lower ER1 (68.3° vs. 89.1°, p < 0.001) and ER2 (74.1° vs. 89.4°, p < 0.001) compared to healthy shoulders. On the contrary, Garcia et al. [26] did not detect significant differences in postoperative ER2 (83.95° vs. 89.21°, p = 0.13) compared to the contralateral shoulders, and Franceschi et al. [27] found only a minimal reduction in postoperative ER1 (56.0° vs. 60.6°, p = 0.4) compared to the preoperative assessment. The latter authors also compared the postoperative external rotation reached after Bankart repair and remplissage to that achieved following Bankart repair alone and did not find any statistically significant differences (p = 0.02).
Isometric infraspinatus strength testing revealed superior strength of the contralateral healthy shoulder, and we found that only ER2 showed a statistically significant decrease (p = 0.0342). Merolla et al. [25] were first to assess the infraspinatus strength: the measurement was performed with the arm at the side and with the examiner standing in front of the patient and resisting the external rotation force. They did not find a significant reduction in ER1, but they did not measure ER2. This type of measurement with the examiner resisting the force of the patient could be less precise and more variable than that achieved with the dynamometer stably fixed to a post, so the results obtained could have been influenced by the strength exerted by the examiner himself.
However, none of the patients included in the present study had a loss of ROM or strength that affected their daily and sports activities, as reported by Buza et al. [24] in a systematic review of six clinical studies (167 patients) with a mean follow-up of 26.8 months. No ROM reduction, especially in ER1, was perceived by our patients. In ER2, a feeling of tension when the patient or the examiner brought the arm to the maximum angle of abduction external rotation was sometimes referred to. The patients did not perceive a significant limitation in strength at any degree of abduction (despite the appearance of minimal signs of adipose atrophy in the context of its muscular belly in 50% of the shoulders).
The rate of return to sport was 100%, with 92.3% returning to their previous sport and 76.92% returning to their pre-injury level, which are similar results to those of Boileau et al. [23] and Garcia et al. [26].
The recurrence rate of 14.28% is higher than in some other studies (0% [28], 4.4% [29], 9.1% [26]) but similar to that reported by Park et al. [30], who noted a 15% recurrence rate at a mean follow-up of 30 months. In a recent retrospective study, Pandey et al. [31] compared the recurrence rates of off-track Hill–Sachs lesions treated with and without the use of remplissage, and found a significantly lower recurrence rate of 3.4% in the remplissage group compared with the Bankart repair alone group (30%). These findings support the hypothesis that the remplissage procedure, in association with Bankart repair, is effective and reliable for treating anterior shoulder instability. It is also safe, with no appearance of fatty degeneration of the infraspinatus muscle or other complications, and yields a low recurrence rate. It brings a loss of ROM and a reduction of strength in external rotation that are not a source of discomfort for the patients.
This study's strength is that the same surgeon performed all the procedures and all the patients underwent the same postoperative rehabilitation protocol. The cohort presented in this study, with a mean follow-up of 55.9 months, is the second for which such long-term results are described [23, 26,27,28,29,30,31].
Patients were chosen with a maximum age of 45 years to exclude those who could already have degenerative changes of the rotator cuff at the preoperative level. The Hill–Sachs lesions in older patients are usually more extensive due to poorer bone quality. Additionally, these patients have lower shoulder function and lower functional request than younger patients. All of this could have led to a bias in the analysis of the results.
Limitations of this study include the small sample size due to the infrequent indication for this surgical procedure and the relatively high rate of patients who were unwilling or unable to return to the institution for clinical evaluation.
A measurement of the contralateral healthy shoulder was conducted to compare the results to the affected one. The fact that preoperative ipsilateral measurements were not available could have represented a slight bias in the results. However, the discrepancy between the affected side and the contralateral side could have arisen because patients did not perform the complete extra rotation of the arm preoperatively due to a feeling of apprehension that the shoulder was experiencing subluxation or dislocation, as recently reported by Pandey et al. [31].
Furthermore, the measurement of the infraspinatus strength in this study could have also been influenced by the fact that eight of the shoulders (66.66%) represented the non-dominant limb, which may have distorted the side-to-side comparison, as they would have naturally been less intense than the dominant one.