Metformin Reduces the Incidence of Shoulder Stiffness After Arthroscopic RC Repair: Letter to the Editor

Dear Editor:

We read with great interest the article by Feng et al ² titled “Metformin Reduces the Incidence of Shoulder Stiffness After Arthroscopic Rotator Cuff Repair: A Randomized, Double-Blinded, Placebo-Controlled Trial”. Their finding of a significantly lower 3-month postoperative stiffness incidence with metformin (11% vs 27%) is promising. We applaud the authors’ efforts to translate an antifibrotic strategy into clinical practice. We appreciate the authors’ contribution, but we would like to raise several points for discussion.

First, it is unclear whether the analysis fully adhered to an intention-to-treat (ITT) principle. The trial enrolled 146 patients, yet outcomes were only reported for 126 completers (approximately 14% loss to follow-up). The authors did not detail how missing data were handled or whether any imputation or sensitivity analyses were performed for the 20 patients lost. ITT analysis dictates that all randomized patients be analyzed in their assigned groups, regardless of protocol deviations or dropout, to avoid attrition bias. ¹ In contrast, excluding patients with no outcome data (a per-protocol approach) can introduce bias if those patients differ systematically from completers. We encourage the authors to clarify whether and how the 20 missing outcomes were accounted for. Clear accounting for losses to follow-up and use of appropriate methods would strengthen confidence that the reported treatment effect is not overstated by attrition.

Second, we wonder whether certain perioperative variables might have influenced stiffness outcomes and should be reported or controlled for. The authors describe a standardized surgical technique for rotator cuff repair, but it is not stated whether adjunct procedures (eg, biceps tenodesis or acromioplasty) were performed in some patients. Such procedures may affect postoperative shoulder motion.^4,6 We acknowledge that randomization tends to balance known and unknown factors between groups. However, explicit reporting and adjustment for key perioperative variables would further strengthen the interpretation of the observed treatment effect.

Third, the study’s use of plasma fibrotic biomarkers to explore the mechanism provides an interesting context, but we urge caution in interpreting these as direct evidence of local shoulder effects. While circulating transforming growth factor-beta and inflammatory cytokines are indeed implicated in fibrosis, their systemic levels may not reflect the microenvironment of the glenohumeral joint capsule. Histopathologic studies of idiopathic adhesive capsulitis (frozen shoulder) demonstrate an inflammatory/fibrotic cascade localized within the joint capsule, with high levels of cytokines such as interleukin-1 beta and tumor necrosis factor alpha driving synovitis and collagen deposition in capsular tissue. ³ In a recent study, specific circulating cytokines associated with shoulder contracture were difficult to pinpoint, underscoring the complex and compartmentalized nature of this pathology. ⁵ Therefore, we thought that the plasma biomarker differences, while intriguing, should be viewed as hypothesis-generating rather than confirmatory proof of mechanism. Furthermore, multiple biomarkers were assayed, and it is unclear whether adjustments for multiple comparisons were applied. Testing numerous plasma proteins raises the risk of false-positive results if each is evaluated at P < .05 without controlling the family-wise error rate.

Again, we sincerely thank the authors for their contributions.