Anatomical Total Shoulder Arthroplasty with Keeled glenoids in Patients under 60 years at 10 years minimum: which risk factors of failure are still valid at long term follow-up?

Aim: To assess the long-term (>10 years) outcomes in anatomical total shoulder arthroplasty (aTSA) and implant survival in patient under 60 years and identify risk factors for complications and revision.

Methods: This was a retrospective, multicenter study conducted from 1993 to 2008. Over 104 aTSA, 87 in 82 patients (55 years, 36 to 60yo) were included at a mean follow-up of 14±4 years (10 to 25y). Outcome measures included pain, motion, Constant score and subjective shoulder value at 10 years minimum. On AP radiograph, the radiolucent line (RLL) score of Molé was used to assess loosening around the glenoid. A glenoid was considered "loose" in 3 circumstances: (1) revision for glenoid loosening, (2) radiological migration of the implant or (3) RLL score≥12. Preoperative glenoid morphology according to Walch, glenohumeral mismatch and cementing technic were evaluated. Survivorship free of revision and free of glenoid loosening were calculated at 10 and 15 years. The mean follow-up was 14±4 years (10 to 25y) or until revision.

Results: Revision-free survivorship was 81% at 10 years and 65% at 15 years. Glenoid failure was the main cause of revision: among the 28 (32%) revised shoulders, 19 (22%) were revised for glenoid loosening. Heavy labor was a risk factor for glenoid component loosening (p=0.029). The curettage technique and flat-back glenoids were risk factors for glenoid revision (p= 0.035) but presented a longer follow-up than compaction technique and convex-back glenoids. The type of preoperative glenoid erosion (Walch type) and glenohumeral mismatch did not correlate with a higher glenoid loosening rate.

Conclusion: aTSA is a reliable procedure for primary OA under 60, but survivorship declines after 10 years. Glenoid loosening, often combined with cuff deficiency or infection, is the main cause of failure and revision. Glenoid morphology as classified by the modified Walch classification does not influence revision rate beyond 10 years.