Does Cement Viscosity Influence 10-year Outcomes in Primary Total Knee Arthroplasty Under Standardized Surgical Technique?

Abstract

Aseptic loosening remained a leading cause of failure following total knee arthroplasty (TKA). While cement viscosity was a potentially modifiable factor influencing fixation, prior clinical and registry studies have reported conflicting results regarding its association with aseptic loosening. This study evaluated whether cement viscosity independently influences the risk of revision for aseptic loosening and the occurrence of radiolucent lines (RLLs). We retrospectively reviewed 3,961 primary TKAs in 2,521 patients performed between 2003 and 2013 using a standardized cementation protocol. Knees were categorized by cement viscosity: low-viscosity (LV; n = 2,118), medium-viscosity (MV; n = 324), and high-viscosity (HV; n = 1,519). The primary endpoint was revision for aseptic loosening. Hazard ratios (HR) were calculated using cause-specific and Firth-penalized Cox regression models. The secondary outcomes included RLL presence analyzed via generalized estimating equations (GEE). The overall revision rate for aseptic loosening was 0.43%. In the penalized Cox model, cement viscosity was not significantly associated with revision for aseptic loosening when comparing LV (HR [hazard ratio], 0.47; 95% confidence interval (CI), 0.12 to 1.50; P = 0.206) and MV (HR, 2.99; 95% CI, 0.72 to 12.27; P = 0.129) groups to the HV group. Radiographically, no significant differences in RLLs were observed between LV and HV groups. However, the MV group demonstrated significantly higher odds of tibial RLLs at one year (OR [odds ratio], 3.14; P < 0.001), two years (OR, 3.12; P < 0.001), and 10 years (OR, 2.41; P = 0.005) compared with the HV group. Cement viscosity was not independently associated with revision for aseptic loosening or radiographic fixation when standardized contemporary techniques were applied. A favorable trend toward lower revision risk was observed with low-viscosity cement, while increased RLLs in the medium-viscosity group likely reflected formulation-specific characteristics rather than viscosity alone.