There is no eternal total hip arthroplasty (THA). With time since the primary procedure, every implant will eventually fail due to the pathological conditions (e.g., infection, fracture, dislocation) or even normal/expected tribological and biological processes, such as wear and loosening (1). The ultimate aim is to provide normal pain-free function for the rest of the patients’ lives (2). Despite patients’ expectations and reliance on surgeons and implants to provide a timeless solution for their conditions, with varying perceptions of the required self-engagement, the surgery is, in fact, only the beginning. This is particularly true for the cementless THA concept, which relies, not only on early mechanical stability, but also on the host's potential to osseointegrate the implanted metallic components (3, 4).

Based on systematic reviews and meta-analysis, the risk of revision gradually increases in the third decade after the primary implantation, with estimates of three-quarters of THAs lasting at least 20 years, and only around 60% to 25 years (1, 5). With the mean age of patients undergoing THA being under 70 years, the female predominance, who have higher expected life span compared to males, and incrementally increasing overall life-expectancy reaching above 80 years, there are more and more implants at risk for failure (6, 7, 8). Among the four most common reasons for revision are infection, periprosthetic fractures (PPFs), loosening and dislocation. These four cumulatively constitutes for approximately 80% of complications (9, 10).

PPFs represent a complex orthopaedic pathology with significant patients’ morbidity and mortality, and socio-economic implications. In more than 80% of cases, the mechanism of injury is a low-energy trauma, mainly fall from the standing height. Most of the PPFs affect the proximal femur (PPFFs), while acetabulum is involved in less than 10% of THA-related fractures (11, 12).

PPFFs are associated with some already known patient- (age, female sex, osteoporosis/osteopenia, neuromuscular diseases, cognitive disorders, Paget's disease, developmental hip dysplasia, rheumatoid arthritis), surgical- (malposition, extensive broaching), and implant-related (cementless, design/type, loosening, stress shielding) risk factors (12–17). Interestingly, the research and developments have been for decades intensively focused mainly on the artificial implants, their materials, composition, design, and other characteristics, while the local host environmental factors have remained more unaddressed (18–22). However, at the end, it is the bone that fractures, which stimulated our efforts to develop a reliable and clinically applicable method for the evaluation of preserved proximal femoral bone stock volume around the implanted femoral stems (V_PF) on routinely available hip radiographs.

The aim of the present study was to introduce and validate a novel method for the evaluation of V_PF on widely available hip anterior-posterior (AP) radiographs, and to assess its influence on the risk of late PPFFs in conventional primary THA.

The mean value of V_PF in the group of cases was 113.8 ± 21.0 cm³ and significantly lower compared to the mean value of 164.0 ± 38.4 cm³ in the control group (P < 0.01). In only 1 out of 30 age-/sex-/implant size-/surgeon-matched strata, the mean V_PF of both controls was lower than V_PF of the case. As shown in Table 1, the significant differences were observed between both groups for the following radiographic parameters: stress shielding, subsidence, CFI and Dorr type (P < 0.01). In contrast, the basic demographic parameters, some of them were also utilized for matching, were well above the level of significance: age (P = 0.57), height (P = 0.90), weight (P = 0.87), BMI (P = 0.82), sex (P = 1.0). Also, there were no significant differences in the prevalence of osteoporosis (P = 0.70) and the levels of femoral neck resection (P = 0.38).

In the ROC curve analysis, the AUC was 0.90 (95% confidence interval: 0.83–0.96). Based on the Youden index, the optimal cut-off value of V_PF was determined at 128.5 cm³.

The conditional logistic regression subanalysis of covariates potentially affecting the risk of PPFFs identified V_PF as the sole parameter influencing this risk, with a hazard ratio of 0.92 for each additional cubic centimetre of V_PF (95% confidence interval: 0.87–0.97; P < 0.01).

In the multivariable regression model for factors correlating with the V_PF, the level of stress shielding, subsidence and sex turned out as significant. The results of both regression analyses were presented in Table 2.

The ICC values for intra- and inter-rater reliabilities, as well as for the comparison of V_PF measurements in standing and supine positions, along with the interpretation of agreement levels were reported in Table 3.

The present study introduced a novel measure of preserved proximal femoral bone stock around the implanted femoral stems and the method for its evaluation on the routinely available hip radiographs. The main finding was the validated importance of bone stock preservation in reducing the risk of late PPFFs, based on significantly lower V_PF values in fractured proximal femora around ABG II stems compared to unfractured control subjects with identical implants of the same size, as well as the results of the regression subanalysis. The difference in preserved bone stock was not only significant, but also substantial.

Despite the longstanding awareness of the increasing incidence and consequences of PPFFs, and the rationale protective role of the preserved bone stock, no method for its evaluation, with the potential for routine clinical application, has been available. The introduced method and V_PF parameter represent a useful tool for the assessment of future PPFFs risk. One of the main advantages of this newly developed method is in its routine and wide availability based on the possibility to utilize standard hip radiographs, which are part of every routine diagnostic evaluation of patients with the indication for primary THA. Moreover, its simplicity, quick learning curve, and time efficiency (taking only a few minutes after some examples measured) enable the surgeon to plan and control the bone stock preservation to some degree for every patient. Also, the intra- and inter-rater reliability demonstrated good to excellent levels of agreement.

Based on the ROC curve analysis with the AUC value of 0.90, the proposed method also demonstrated good to excellent performance with the lowest end of 95% IC interval being 0.83 and above the 0.80 mark for good level of accuracy (35). The cut-off value of V_PF was determined with Youden index at 128.5 cm³, under which the consideration of different approach is advisable. This may include the use of cemented femoral stem at THA or other classes of cementless femoral stems, beside the anatomical type 6 (Khanuja et al. classification), which potentially allow greater bone preservation (33). So far, the conventional stems were being analysed, while the effect of using short femoral stems with their potential to preserve more bone stock and provide better physiological load transfer will show consequences for PPPFs risk in the future studies (36).

The proposed method was developed on the cohort of patients with PPFFs of ABG II femoral stems, for which clinical results have been extensively reported. Previous studies of this implant encompassed arthroplasty registry data, multiple centres and single hospital reports (31, 32, 34, 37–42). Most of them documented satisfactory results with survival rates until the first revision in the 94%–100% range at approximately 6 years (32, 38), 89%–98% range at around 10 years (31, 34, 39), and even up to 96.1% at 14 years of follow-up (40). However, in many studies the increased risk of periprosthetic fractures was observed with notable differences in the revision profiles between the ABG II and all other conventional cementless primary THAs. In some reports, the PPFFs accounted for nearly or even more than half causes for revision (32, 37, 39, 43).

Many factors have been associated with increased risk of PPFFs (12, 13, 17, 19). Especially with time since the implantation, the interactions between the implant and the local host bone seem to be crucial. This relationship appears to depend on the conditions set at the time of implantation, which impact the long-term interactions. In cementless technique, the femoral morphology, implant selection, their compatibility allowing correct stem orientation, fill, stability and early fixation, as well as the newly proposed V_PF parameter, seem to influence the two main long-term processes of osseointegration and stress shielding associated with the risk of PPFFs (3, 21, 44, 45). Based on the results of the present study, the V_PF appears to be at the centre of interplay of these interconnected factors. Therefore, in establishing causal relationships between various parameters and PPFFs, the preserved bone stock could represent an unrecognized confounder.

In the regression model for determining the factors that significantly impact the risk of PPFFs, the V_PF turned out as the sole statistically significant independent predictor after adjustment for age, sex, implant size and surgeon. With every preserved cubic centimetre of bone stock, the risk of PPFFs reduced for approximately 8%. Since the diagnosis of osteoporosis was also included in the model, it may be even speculated that the volume of preserved bone seems to be more vital than its quality. In addition to the positive correlation of V_PF with the male sex, the significant inverse correlations of late PPFFs risk with the two crucial long-term factors, stress shielding and subsidence, were observed. The exact protective mechanism of preserved bone stock is still a ‘black box’, however it may be proposed as a common biological denominator of these parameters at interplay. It appears to support initial stability and enhance osseointegration acutely, while reducing stress shielding and preventing loosening (ensuring stable osseointegration without subsidence) in the long-term. Moreover, the difference between both groups in the indicators of proximal femoral canal morphology (CFI, Dorr type) was significant. That is in line with the established increased risk of late PPFFs in the Dorr type C femurs (45, 46). It may be even hypothesized that lower CFI, which is a numerical counterpart to the Dorr type C, reduces the chance of implantation with sufficient V_PF preservation. Therefore, the bone stock preservation must be considered, starting from the preoperative planning onwards.

Several limitations of the present study should be acknowledged. It employed a retrospective design and included only one type of femoral stem. Further multi-centre studies including different stem types and their comparisons are needed prior to broad generalisation of this new concept. Key parameters (age, sex, implant size, surgeon) were controlled for, and a broad set of established risk factors was included in the analysis; however, residual confounding cannot be completely excluded in a real clinical setting. Some parameters were semiquantitative (stress shielding) or indirectly assessed (osteoporosis). Several measurements were performed on supine radiographs, and the rotational as well as parallax effects were not specially accounted for. The novel method was not yet validated by comparison with other methods for assessment of the risk for periprosthetic fractures. However, based on the reliability testing, not only intra- and inter-observer but also between standing and supine radiographs, the agreements were in the good to excellent range. Moreover, the ROC analysis demonstrated adequate performance of the novel method, and the overall results emphasized the importance of bone stock preservation with the introduced V_PF parameter as its measure.

We introduced a simple and widely accessible method to assess the risk for periprosthetic femoral fractures in THA. The method was tested on two populations: hips that sustained periprosthetic fractures and unfractured hips, demonstrating a considerable and statistically significant difference between the two groups. Reliability testing, including intra- and inter-observer assessments and agreement between supine and standing views was good to excellent. Further validation is needed to develop a protocol that can be used in clinical practice, however, the presented results indicate that bone stock preservation (with V_PF as a quantitative measure) is crucial for the prevention of late PPFFs.