INTRODUCTION: Periprosthetic fracture after total hip arthroplasty (THA) is the number four reason for revision, accounting for 7-10% of all revisions (Swedish Joint Registry 2010, Australian Joint Registry 2014).  In addition to implant loosening, femoral bone quality is a major risk factor of periprosthetic fracture. The overall goal of this study was to establish a model that determines the risk of periprosthetic fracture as a function of periprosthetic BMD and femoral stem material and fixation method. The study incorporated a series of autopsy-retrieved, clinically asymptomatic hip joints with implanted total hip replacements that had been in-vivo for 1 – 29 years, acquired through a willed joint body program at our institution. The specimens varied in stem design, alloy and fixation method, and time in-situ. No patient had been scheduled for a revision at the time of death and all patients were ambulatory. Therefore, the specimens incorporate the loss of BMD or any osteolysis that occurred as a result of in-vivo usage, providing a novel opportunity for assessment of periprosthetic fracture risk. 
 
METHODS:  29 postmortem proximal femurs with THA that had been in clinical service for 1 – 29 years prior to the patient’s death, and known clinical histories, were obtained then disarticulated and radiographed prior to testing. The femoral stems included 5 cemented Ti alloy stems, 6 non-cemented Ti alloy stems, 14 cemented stainless steel or CoCr alloy stems, and 4 non-cemented CoCr stems. For the present study our goal was to simulate in-vivo bodyweight forces on the hip joint. The specimens were flash-frozen within 24 hours of death, and thawed 8 hours prior to experimentation. All specimens (n = 29) were disarticulated, radiographed, and scanned using DEXA to assess bone mineral density prior to testing. Each femur was loaded using a protocol previously established for simulating physiological activities. Specifically, combined axial and torsional loading were applied cyclically, starting from a peak of 100 N and ±1 Nm, and increasing to a peak load of 700 N and ±5 Nm acting on the distal femoral shaft. The femoral head was loaded through a sized cup that allowed free mediolateral translation while transmitting torsional loads to the femoral shaft.    Triaxial rosette strain gauges were applied to the following locations of interest on each cortical shaft: 2 on the lateral surface, one at the stem tip level and one 15 mm distal, and 2 on the anterior surface, again one at the stem tip level and one 15 mm distal.  
 
RESULTS:  Despite the conservatively low maximum load of 1x body weight that was applied, three femurs fractured at the stem tip during loading, so that their data could not be included in the analysis: one non-cemented Ti alloy stem and 2 non-cemented CoCr alloy stems. As this left only 2 viable noncemented CoCr alloy stems, that data is not presented here.   All specimens experienced higher strains on the lateral surface than on the anterior surface, indicating that the bending loads in the frontal plane, rather than axial/torsional loads, had the predominant effect (Figure 1). While average values of principal and shear strains were similar (p = 0.65 and p = 0.86 respectively), maximum strain values were substantially higher in the femurs with cemented stainless-steel or CoCr alloy stems than with cemented Ti alloy stems. At the lateral stem-tip, cemented stainless-steel or CoCr alloy stems (n = 14) experienced maximum principal strains of up to 2640 µ while cemented Ti alloy stems (n = 5) experienced maximum principal strains of up to 1546 µ. Similarly at the anterior stem-tip site, cemented stainless-steel and CoCr alloy stems experienced maximum principal strains of up to 1582 µ while cemented Ti alloy stems only reached maximum values of 962 µ, despite similar average strains (p = 0.83). The same types of differences were observed at the lateral site distal to the stem tip, where cemented stainless-steel or CoCr alloy stems reached maximum principal strains of 2696 µ and cemented Ti alloy stems only reached 1472 µ, with a difference between average values of 102 µ (p = 0.71). Finally, the maximum shear strains in cemented stainless-steel or CoCr alloy stems saw increases of 1000 µ and 1300 µ over the maximum shear values of cemented Ti alloy stems (Table 1).  Similarly, non-cemented Ti alloy stems experienced substantially higher strain than cemented Ti alloy stems. At the lateral stem-tip, non-cemented Ti alloy stems (n = 6) experienced maximum principal strains of up to 2487 µ while cemented Ti alloy stems experienced maximum principal strains of up to 1546 µ. At the lateral site distal to the stem tip non-cemented Ti alloy stems reached maximum principal strains of up to 2323 µ and cemented Ti alloy stems only reached 1472 µ, with a difference between average values of 270 µ (p = 0.48). Finally, the maximum shear strains of non-cemented Ti alloy stems at the lateral sites saw increases of 800 µ over the maximum shear values of cemented Ti alloy stems.   There was a significant correlation between BMD measured by DEXA in Gruen Zone 4 (at the distal stem tip) and principal strains (Pearson’s Correlation Coefficient = -0.51, p = 0.013) (Figure 2), with higher strains corresponding to lower BMD.  
 
DISCUSSION:  The average strains at the lateral and anterior aspects of the femur near the stem tip were not significantly different among the three groups. Nevertheless, the two groups with cemented stainless-steel or CoCr alloy stems and non-cemented Ti alloy stems both included femurs that experienced high strains in excess of 2000 µ while undergoing simulated loads of 1x body weight, indicating that a low-energy trauma event would increase the strain sufficiently to result in Vancouver Type-B fracture in both groups (cortical bone strain failure 4000 – 8000 µ). In contrast, the group with cemented Ti alloy stems never reached the 2000 µ threshold. This was corroborated by the result of dual-energy x-ray absorptiometry which indicated a clear inverse correlation between bone quality and maximum principal strain. The maximum strains correspond to a higher risk of Vancouver Type-B fractures, as these strains of 2000 – 2500 µ resulted only from a conservative body-weight load of 700 N. It is well known that normal activities result in hip joint reaction forces 2-3 times body weight, and sudden low-energy traumatic events such as stumbling can result in even higher loads. Therefore, simple events can easily cause strains in excess of 5000 µ, likely resulting in fracture. 
 
RELEVANCE: Despite the fact that THA in the present study were in successful clinical service until the patient’s death, the results of the present study indicated that some of the femurs with rigid cemented or non-cemented implants were both potentially at high risk for Vancouver Type-B fractures. The results call for prophylactic measures in THA patients, including those with cemented stems, to avoid or preempt even minor traumatic incidents.

1878 - Is a Well-Functioning THA at Risk of Periprosthetic Fracture? A Biomechanical Study of Willed Joints