We released the infraspinatus tendons of six sheep, allowed retraction of the musculotendinous unit over a period of 40 weeks and then performed a repair. We studied retraction of the musculotendinous unit 35 weeks later using CT, MRI and macroscopic dissection.

The tendon was retracted by a mean of 4.7 cm (3.8 to 5.1) 40 weeks after release and remained at a mean of 4.2 cm (3.3 to 4.7) 35 weeks after the repair. Retraction of the muscle was only a mean of 2.7 cm (2.0 to 3.3) and 1.7 cm (1.1 to 2.2) respectively at these two points. Thus, the musculotendinous junction had shifted distally by a mean of 2.5 cm (2.0 to 2.8) relative to the tendon. Sheep muscle showed an ability to compensate for approximately 60% of the tendon retraction in a hitherto unknown fashion. Such retraction may not be a quantitatively reliable indicator of retraction of the muscle and may overestimate the need for elongation of the musculotendinous unit during repair.

We assessed the predictive value of the macroscopic and detailed microscopic appearance of the coracoacromial ligament, subacromial bursa and rotator-cuff tendon in 20 patients undergoing subacromial decompression for impingement in the absence of full-thickness tears of the rotator cuff. Histologically, all specimens had features of degenerative change and oedema in the extracellular matrix. Inflammatory cells were seen, but there was no evidence of chronic inflammation. However, the outcome was not related to cell counts.

At three months the mean Oxford shoulder score had improved from 29.2 (20 to 40) to 39.4 (28 to 48) (p < 0.0001) and at six months to 45.5 (36 to 48) (p < 0.0001). At six months, although all patients had improved, the seven patients with a hooked acromion had done so to a less extent than those with a flat or curved acromion judged by their mean Oxford shoulder scores of 43.5 and 46.5 respectively (p = 0.046). All five patients with partial-thickness tears were within this group and demonstrated less improvement than the patients with no tear (mean Oxford shoulder scores 43.2 and 46.4, respectively, p = 0.04). These findings imply that in the presence of a partial-thickness tear subacromial decompression may require additional specific treatment to the rotator cuff if the outcome is to be improved further.

The establishment of a suitable animal model of repair of the rotator cuff is difficult since the presence of a true rotator cuff anatomically appears to be restricted almost exclusively to advanced primates. Our observational study describes the healing process after repair of the cuff in a primate model. Lesions were prepared and repaired in eight ‘middle-aged’ baboons. Two each were killed at four, eight, 12 and 15 weeks post-operatively. The bone-tendon repair zones were assessed macroscopically and histologically.

Healing of the baboon supraspinatus involved a sequence of stages resulting in the reestablishment of the bone-tendon junction. It was not uniform and occurred more rapidly at the sites of suture fixation than between them. Four weeks after repair the bone-tendon healing was immature. Whereas macroscopically the repair appeared to be healed at eight weeks, the Sharpey fibres holding the repair together did not appear in any considerable number before 12 weeks. By 15 weeks, the bone-tendon junction was almost, but not quite mature.

Our results support the use of a post-operative rehabilitation programme in man which protects the surgical repair for at least 12 to 15 weeks in order to allow maturation of tendon-to-bone healing.