Radius of proximal femur curvature as a significant predictor for long stem-canal match in total hip arthroplasty: a retrospective cohort study

Abstract

Background

The match between femoral stems and the medullary canal is crucial for successful total hip arthroplasty (THA), particularly for long stems often used in complex primary and revision surgeries. This study aimed to investigate the association between the radius of proximal femur curvature (RPFC) on long stem-canal match in THA.

Methods

In this retrospective study, 65 patients who underwent revision or complex primary THA using a 200 mm long stem between June 2019 and June 2024 were analyzed. Preoperative CT scans were used to measure RPFC and bone volume fraction (BV/TV). Stem-canal match was assessed using postoperative radiographs. Multivariate logistic regression and ROC curve analyses were performed to evaluate the predictive value of RPFC for stem-canal match.

Results

Significant differences in RPFC were observed between the match and mismatch groups (656.0 ± 198.9 mm vs. 401.4 ± 73.3 mm, p < 0.001). Height and gender also differed significantly between groups in univariate analysis. After adjusting for height and gender in a multivariate logistic regression model, RPFC remained an independent protective factor against stem-canal mismatch (OR = 0.98, 95% CI: 0.97–0.99, p < 0.001). ROC analysis revealed an optimal RPFC cut-off value of 496.4 mm for predicting good stem-canal match (sensitivity: 94.4%, specificity: 85.1%, AUC = 0.923). Compared to height (AUC = 0.725) and gender (AUC < 0.7), RPFC demonstrated superior predictive value for stem-canal match.

Conclusion

RPFC is a significant predictor of long stem-canal match in THA. An RPFC < 496.4 mm is associated with a higher risk of mismatch for the implant used in this study. While these findings are promising, this is a single-center study with a limited sample size. Further, larger prospective studies are needed to validate these conclusions before widespread clinical adoption can be recommended. Preoperative RPFC measurement shows potential to improve implant selection and reduce mismatch-related complications in THA.

Introduction

Total hip arthroplasty (THA) is the gold standard treatment for end-stage hip diseases, including avascular necrosis of the femoral head, degenerative arthritis, and developmental dysplasia of the hip. As the longevity of primary THA increases, the demand for revision procedures has risen significantly, often necessitating the use of long stems to address proximal femoral bone loss and structural abnormalities [1, 2].

The successful implantation of long stems presents unique challenges, particularly in achieving optimal match within the femoral canal [3]. Mismatch between the long stem and the medullary cavity can lead to serious complications, including periprosthetic fractures, thigh pain, and early implant failure. For instance, one study on modular stems reported that 15.4% of hips developed unstable fixation, which was linked to poor canal fill and led to aseptic loosening [4]. Similarly, a study on straight intramedullary nails in an Asian population found that the implant abutted the anterior cortex in 34.8% of cases due to femoral bow mismatch, a scenario predisposing to technical challenges and complications [5]. These findings underscore the clinical importance of understanding the anatomical factors that predict mismatch. Recent studies have highlighted the multifactorial nature of stem-canal mismatch, including racial differences, gender, height, proximal femoral anatomy, and surgical technique [6–9].

The anterior bow of the proximal femur has been recognized as a critical anatomical feature. While its importance is well-documented in the context of femoral nailing, where mismatch can lead to anterior cortex perforation [10, 11], its specific role in long-stem THA has been less explored. This highlights a novel aspect of our study, focusing specifically on the arthroplasty context.

The three-dimensional morphology of the proximal femur, particularly the anterior bow, varies significantly among individuals and populations [12]. Studies have shown considerable anatomical differences between Asian and Western populations, and even within different ethnic groups, affecting parameters like femoral neck-shaft angle, offset, and canal curvature [13–16]. This variability poses a challenge in designing and selecting appropriate femoral stems, especially for Asian populations where commercially available implants may not optimally match the native anatomy [12].

Given the potential impact of femoral bow mismatch on surgical outcomes, there is a pressing need to investigate the role of RPFC in predicting the match of long stems. We hypothesize that RPFC is a key factor in determining stem-canal match and may serve as a valuable preoperative planning tool.

The purpose of this study was to evaluate the association between RPFC on the matching of long stems with the medullary cavity in THA. By quantifying this relationship, we aim to provide surgeons with a practical metric for preoperative assessment and implant selection, potentially reducing the risk of mismatch-related complications and improving overall surgical outcomes. Specifically, this study sought to answer the following questions:

1. Is there a significant difference in RPFC between patients with good stem-canal match and those with mismatch?

2. Can RPFC, after accounting for other variables, serve as an independent predictor of stem-canal match in long-stem THA?

3. What is the optimal RPFC threshold for achieving good stem-canal match with the specific implant used in this study?

Materials and methods

Study design and patient selection

This retrospective study was conducted at the Department of Orthopaedics, Nanjing First Hospital, Nanjing Medical University, in accordance with the Declaration of Helsinki. The study protocol was approved by the institutional ethics committee (approval number: KY20240514-09-KS-01), and written informed consent was obtained from all participants.

Inclusion criteria

Patients were included in this retrospective analysis if: (1) they underwent total hip arthroplasty at our institution between June 2019 and June 2024; (2) they were implanted with a specific 200 mm long, cementless femoral stem (type-160, CHUNLI, China); (3) the clinical indication for the long stem was either revision THA or a complex primary THA requiring enhanced distal fixation (e.g., due to proximal femoral bone defects or poor bone quality); (4) they had complete preoperative CT scans and immediate postoperative anteroposterior and lateral radiographs of sufficient quality for analysis.

Exclusion criteria

We excluded patients if they had: (1) bilateral THA, to ensure the independence of data points; (2) incomplete imaging data; (3) CT scans unsuitable for computer modeling; (4) intraoperative periprosthetic femoral fractures; (5) conditions known to significantly affect bone morphology, such as Paget’s disease, acromegaly, or Marfan syndrome.

All surgeries were performed by a single highly experienced senior surgeon (Fig. 1).

Fig. 1.

Flowchart of patient inclusion in the study

Sample size justification

This study was retrospective in nature, and the sample size was determined by the number of eligible patients meeting the criteria within the study period. A formal a priori sample size calculation was not performed. We acknowledge that the relatively small sample size may limit the statistical power of our analyses, and this is addressed in the limitations section.

Stem-canal match classification

Postoperative anteroposterior and lateral radiographs were used to assess stem positioning within the femoral canal. The stem-canal match was classified into five grades based on the position of the stem tip (Table 1), adapting the method described by Chang [17]. A posterior position of the stem tip relative to the canal centerline was recorded as a negative value. Based on the potential for cortical impingement or perforation, grades 0, 1, and − 1 were categorized as the “match group”, while grades 2 (anterior impingement) and − 2 (posterior impingement) were classified as the “mismatch group”.

Table 1.

Position Grade of the Tip of Femoral Prosthesis and Match Classification

Grade		Classification	Description
0	Match		located in the center of medullary cavity
1	Match		Located anteriorly but no contact with the anterolateral cortex, and no trend of anterior perforation of cortex
2	Mismatch		Contacted with the anterolateral cortex, the underlying trend of anterior perforation of cortex
-1	Match		Located posteriorly but no contact with the posterolateral cortex
-2	Mismatch		Contacted with the posterolateral cortex, with an underlying trend of posterior perforation

Measurement of Radius of Proximal Femur Curvature (RPFC) and Bone Volume Fraction (BV/TV)

CT scans were performed using a 64-slice spiral CT scanner (Siemens, Germany) with patients in supine position and lower limbs in neutral rotation. The scanning range included the proximal femur and pelvis, with a slice thickness of 1 mm. Data were stored in DICOM format.

All radiological measurements were performed by two independent orthopedic residents (W.H. and H.Z.) with three years of experience in medical imaging software. Both were trained on the measurement protocol by a senior author (C.T.) to ensure standardization. Discrepancies were resolved by consensus. The reliability of these measurement methods has been previously established in the literature [18–20].

Image processing and 3D modeling were conducted using MIMICS (v20.0, Materialise, Belgium) and 3-matic Medical (v13.0, Materialise, Belgium) software to calculate RPFC and BV/TV (Fig. 2). The process involved the following steps:

Fig. 2.

Femur modeling process. (A) 3D femur model reconstructed from patient CT data; (B) Segmentation of medullary cavity; (C) Isolation of proximal femur medullary canal; (D) Centerline fitting; (E) Spline curve fitting; (F) Extension of fitted curve to a complete circle

1. Femur Segmentation: DICOM images were imported into MIMICS. Using thresholding, region growing, and manual editing tools, the femur was segmented from surrounding tissues.

2. 3D Model Generation: A 3D model of the femur was created using the “Calculate 3D” function in MIMICS.

3. Medullary Canal Isolation: The femoral model was imported into 3-matic Medical. The medullary canal was isolated using Boolean operations to subtract the cortical bone from the total femur volume.

4. Proximal Canal Segmentation: In 3-matic Medical, two parallel planes, 100 mm apart and perpendicular to the femoral shaft axis, were created. The proximal canal was defined as the region between these planes, starting from the lower edge of the lesser trochanter.

5. Centerline Extraction: The “Create Centerline” function in 3-matic Medical was used to generate a centerline through the proximal canal.

6. RPFC Calculation: The centerline was fitted to a circular arc using the “Curve Fitting” tool in 3-matic Medical. The radius of this best-fit circle was defined as the RPFC.

7. BV/TV Calculation: Using the “Volume Measurement” tool in 3-matic Medical, the cortical bone volume and total bone volume of the proximal femur were measured. BV/TV was calculated as the ratio of these two volumes.

Analysis of full femoral bow

To analyze curvature differences along the femur, CT data from 45 patients with normal hips were collected. The femur was modeled from the lesser trochanter to 30 mm above the femoral condyles. The medullary cavity was divided into proximal and distal halves to measure RPFC and radius of distal femur curvature (RDFC), respectively.

Data analyses

Data analyses were performed using SPSS 27.0 (IBM, USA) and GraphPad Prism 10 (GraphPad Software, USA). The distribution of continuous data was assessed using the Shapiro-Wilk test, which confirmed normal distribution, justifying the use of parametric tests. Continuous variables were expressed as mean ± standard deviation and categorical variables as frequencies and percentages. Between-group comparisons of continuous variables were conducted using independent t-tests, whereas RPFC and RDFC were compared using paired t-tests. Chi-square tests were used for categorical variables. Variables showing significant differences between groups in the univariate analysis (defined as p < 0.05) were included in a multivariate logistic regression analysis to identify risk factors for stem-canal mismatch. Receiver operating characteristic (ROC) curves were constructed to assess the diagnostic value of each factor, with area under the curve (AUC), sensitivity, specificity, and optimal cut-off values calculated. AUC values were interpreted as: <0.7 poor, 0.7–0.8 fair, 0.8–0.9 good, and ≥ 0.9 excellent diagnostic value. Intraclass correlation coefficients (ICC) were used to assess inter-observer reliability, with ICC > 0.75 considered high consistency. A p-value < 0.05 was considered statistically significant for all analyses.

Results

Cohort characteristics

Of 906 THA performed, 65 patients (24 men, 41 women) met the inclusion and exclusion criteria. The mean age was 80.2 ± 9.8 years (range: 47–96), average height 161.4 ± 8.9 cm, and weight 62.9 ± 12.1 kg. The cohort included 10 revision and 55 complex primary THA procedures. There were 34 left and 31 right hip procedures.

Reliability of measurements

Inter-observer reliability was excellent for all measurements. The intraclass correlation coefficients (ICC) were 0.95 (95% CI: 0.93–0.97) for RPFC and 0.90 (95% CI: 0.83–0.94) for BV/TV. In the full femoral bow analysis, ICC values were 0.98 (95% CI: 0.95–0.99) for RPFC and 0.93 (95% CI: 0.86–0.96) for RDFC, indicating high consistency and reliability of the measurements.

Stem-canal match classification

Of the 65 patients included in the study, stem-canal match was classified as follows: Grade 0 in 17 cases (26.2%), Grade 1 in 22 cases (33.8%), Grade 2 in 18 cases (27.7%), and Grade − 1 in 8 cases (12.3%). No cases were classified as Grade − 2 (Fig. 3; Table 2). Based on these classifications, 47 patients were assigned to the match group and 18 to the mismatch group.

Fig. 3.

Typical radiographic examples of each stem-canal match classification grade

Table 2.

Distribution of Stem-Canal Match Classification

Grade	0	1	2	-1	-2
%	26.2	33.8	27.7	12.3	0.0
Frequency	17	22	18	8	0

Comparison of RPFC and demographic factors between match and mismatch groups

There were no significant differences between the match and mismatch groups in age, surgical site, weight or BV/TV (p > 0.05). However, significant differences were observed in RPFC (656.0 ± 198.9 mm vs. 401.4 ± 73.3 mm, p < 0.001), height (163.3 ± 9.0 cm vs. 156.4 ± 6.7 cm, p = 0.005), and proportion of women (55.3% vs. 83.3%, p = 0.036) between the match and mismatch groups, respectively (Table 3).

Table 3.

Demographic and Clinical Characteristics of Population Groups

Characteristics	Match group	Mismatch group	p-Value
RPFC (mm)	656.0 ± 198.9	401.4 ± 73.3	< 0.001
Gender, n (%)			0.036
- Men	21 (44.7)	3 (16.7)
- Women	26 (55.3)	15 (83.3)
Age (year)	79.2 ± 10.5	82.7 ± 7.4	0.199
Height (cm)	163.3 ± 9.0	156.4 ± 6.7	0.005
Weight (kg)	63.9 ± 12.8	60.5 ± 9.9	0.321
BV/TV	0.595 ± 0.095	0.591 ± 0.080	0.869
Operation side, n (%)			0.432
- Left hip	26 (55.3)	8 (44.4)
- Right hip	21 (44.7)	10 (55.6)

Data are presented as mean ± SD or n (%)

RPFC Radius of Proximal Femur Curvature, BV/TV Bone Volume/Total Volume

RPFC as an independent predictor of stem-canal match

A multivariate logistic regression analysis (Table 4), using variables that showed significant differences in univariate analysis, identified RPFC as an independent predictor for stem-canal mismatch (OR = 0.980, 95% CI: 0.969–0.992, p < 0.001). This suggests that for every 100 mm increase in RPFC, the odds of achieving good stem-canal match increase by approximately 2-fold.

Table 4.

Multivariate Logistic Regression Analysis for Predictors of Stem-Canal Mismatch

Parameter	B Coefficient	Standard Error	Wald Statistic	Odds Ratio (95% CI)	p-Value
RPFC	0.020	0.006	11.060	0.980 (0.969–0.992)	< 0.001
Gender (Women)	1.198	1.234	0.943	3.313 (0.295–37.173)	0.332
Height	-0.008	0.078	0.010	0.992 (0.852–1.155)	0.919

Gender: Men is the reference category

RPFC Radius of Proximal Femur Curvature

Optimal RPFC threshold for predicting stem-canal match

ROC curve analysis revealed that RPFC had excellent predictive value for stem-canal mismatch (Fig. 4; Table 4, AUC = 0.923, 95% CI: 0.859–0.988, p < 0.001). The optimal cut-off value was 496.4 mm, which yielded a sensitivity of 94.4% and a specificity of 85.1%. Height showed fair predictive value (Fig. 4, AUC = 0.725, 95% CI: 0.593–0.857, p = 0.005), while gender showed poor predictive performance (Fig. 4, AUC < 0.7).

Fig. 4.

Receiver operating characteristic (ROC) curves for radius of proximal femur curvature (RPFC), height, and gender diagnostic models

Analysis of full femoral bow radius

Comparison of femoral curvature at different segments revealed that RPFC (719.29 ± 183.97 mm) was significantly lower than RDFC (1,279.38 ± 377.09 mm, Table 5, p < 0.001). This indicates a non-uniform curvature along the femoral shaft, with the proximal femur exhibiting a greater curvature compared to the straighter distal femur. This analysis is valuable as it quantitatively demonstrates the non-uniform curvature along the femur, providing a rationale for why a single-curvature long stem may not fit the entire canal, particularly the more pronounced proximal bow.

Table 5.

Difference in Anterior Bow of Proximal and Distal Femur

Parameter	Anterior bow (mm)	t-Value	p-Value
Proximal femur	697.59 ± 154.77	-8.959	< 0.001
Distal femur	1,265.36 ± 370.82

Values are presented as mean ± SD

Discussion

Importance of stem-canal match in total hip arthroplasty

In this study, we investigated the relationship between the radius of proximal femur curvature (RPFC) and the fit of a long femoral stem in THA. Our primary finding is that a smaller RPFC—indicating a more pronounced anterior bow—is a strong and independent predictor of stem-canal mismatch. This provides quantitative evidence supporting the clinical intuition that femoral anatomy is a critical determinant of implant fit.

The morphology of the femoral canal varies significantly across different racial, geographical, gender, and age groups. Western populations generally have larger femoral canals compared to Eastern populations, with notable variations even within regions of China [21–23]. Despite the maturity of prosthesis design concepts in Western countries, most hip implants used in China are still sourced from global manufacturers. This often leads to mismatches between femoral stems and medullary canals, particularly with long stems used in revision surgeries, such as Zimmer’s Wagner SL, DePuy’s Solution, Smith & Nephew’s ECHELON, and Stryker’s Restoration stems. These classic long stem designs, many of which are straight, are not universally applicable and can lead to increased postoperative complications, directly impacting clinical outcomes.

Achieving optimal fixation requires excellent stem-canal match, minimizing stress shielding and preventing varus or valgus positioning [24]. Poor matching can lead to periprosthetic fractures, thigh pain, stress shielding-induced bone resorption, and osteolysis, ultimately increasing the risk of long-term implant loosening [25–28]. Our study provides a specific anatomical parameter, RPFC, that can be used preoperatively to anticipate and potentially avoid such issues. The key message from this is that a simple radiological measurement could significantly enhance preoperative planning.

Factors influencing stem-canal match

Multiple factors are associated with stem-canal match, including race, gender, height, proximal femoral bow, implant type and size selection, surgical technique, and proximal femoral bone quality. Our study focused on patient-specific factors and found that gender (p = 0.036), height (p = 0.005), and RPFC (p < 0.001) significantly affect stem-canal match, with RPFC being the most influential. Su et al. reported that the average femoral canal anterior bow radius in Chinese populations is 971.44 ± 211.68 mm, with women (872.81 ± 174.57 mm) having a significantly greater curvature than men (1,015.72 ± 212.19 mm). Maratt et al., in a CT-based reconstruction study of 3,922 femurs, reported mean radii of curvature of 112 ± 26 cm for the medullary canal and 145 ± 55 cm for the anterior cortex. They observed a positive correlation between femoral bow curvature and femoral length, suggesting that taller individuals tend to have larger radii of curvature, indicating straighter femurs [29].

Surgical technique and implant selection are crucial factors [30]. Successful biological fixation of femoral stems requires tight press-fitting, with higher stem-canal match associated with stronger fixation and lower loosening rates [30]. However, most current implant designs, especially long straight stems, struggle to achieve physiological matching [31]. This can lead to bony impingement during implantation, necessitating smaller stem sizes, weakening fixation, and increasing the risk of loosening or cortical perforation [32]. Intraoperative fluoroscopy is essential for immediate assessment of stem-canal match and appropriate implant selection. However, preoperative planning based on objective measurements like RPFC can reduce intraoperative uncertainty and the need for unexpected changes.

Proximal femoral bone quality also impacts stem-canal match. In our cohort, nearly half of the patients with periprosthetic fractures (4/10) initially had cemented stem fixation. Factors such as osteolysis-induced canal enlargement, residual cement, and localized cortical sclerosis can complicate stem implantation in revision surgeries, leading to stress concentration and mismatch [33]. Some studies suggest that for Dorr type C femurs with a “chimney-shaped” proximal canal, often seen in older patients with wide canals and thin cortices, cemented prostheses may be preferable due to concerns about stability and osseointegration. Another study on primary THA found higher complication rates and lower stem survivorship when using cementless tapered wedge stems in Dorr type A femurs compared to type B [34].

Value of RPFC in preoperative planning

Our study identifies RPFC as a valuable indicator for predicting mismatch. We found a threshold of 496.4 mm, below which the risk of mismatch with the studied implant is substantially higher. Special attention should be given to patients with smaller anterior bow radii, particularly women with lower height, when considering long stems. Chang et al. cautioned that even commonly used short, straight implants like PFNA-II may not be suitable for all Asian populations due to potential mismatch with femoral canal anatomy [5].

Evolution has resulted in an anterolateral bow of the femur, with a normal anterior bow angle of 10.6 ± 1.8° and lateral bow generally < 3°. Femoral bow curvature is associated with gender, age, body habitus, and ethnicity, with greater curvatures typically observed in shorter individuals, women, older adults, and Eastern populations [35–38]. Our study further demonstrated significant differences in canal curvature between proximal and distal femoral segments (p < 0.05), highlighting the non-uniform nature of femoral canal curvature. This variability must be considered in implant selection and design, particularly regarding insertion depth and population differences.

This study used the CHUNLI type-160 stem, designed to accommodate Chinese femoral anatomy. The 200 mm anatomical stem features a straight proximal 100 mm section and a curved distal 100 mm section with a 950 mm radius of curvature in the coronal plane. Despite this design, we still observed a significant rate of stem-canal mismatch in our cohort. This finding underscores the need for further research into proximal femoral canal characteristics and improved prosthesis designs to meet the growing demand for revision surgeries.

Limitations

This study has several important limitations that must be acknowledged. First, its single-center, retrospective design is susceptible to selection bias. Second, the sample size is small, which limits the statistical power to detect smaller effects and increases the risk of Type II statistical error for non-significant findings. Third, we used a single type of long-stem implant; therefore, our findings, particularly the specific cut-off value of 496.4 mm, may not be generalizable to other implant designs with different geometries. Fourth, due to the retrospective nature, we could not consistently collect data on ethnicity, specific comorbidities, or bone mineral density, which could be confounding factors. Fifth, we excluded patients with intraoperative fractures to focus on initial fit, but future studies should specifically analyze the relationship between RPFC and fracture risk. Finally, this study lacks long-term clinical follow-up; therefore, our assessment is confined to the initial postoperative match. We cannot correlate our findings with long-term clinical outcomes (such as pain scores or functional recovery), nor can we account for time-dependent changes such as prosthesis subsidence due to factors like bone remodeling or osteoporosis. This represents an important direction for future research.

Conclusion

This study demonstrates that the radius of proximal femur curvature (RPFC) is a strong, independent predictor of long stem-canal match in total hip arthroplasty. A smaller RPFC (< 496.4 mm for the studied implant) is associated with a significantly higher risk of mismatch. While the conclusions are constrained by the study’s limitations, they suggest that incorporating preoperative RPFC measurement into surgical planning could be a valuable tool to improve implant selection, anticipate potential difficulties, and potentially reduce complications, particularly in populations with a higher degree of femoral bow.

Authors’ contributions

C.T and L.W conceived and designed the trial. W.H collected patient data and wrote the original draft. J.Y and H.Z analyzed and interpreted the data. S.Z and C.T reviewed the manuscript. M.W and Z.D are responsible for the integrity of the data. L.W approved the final version of the manuscript. All authors have read and approved the final submitted manuscript.

Funding

This work was financially supported by the National Nature Science Foundation of China (No. 81702205), Nature Science Foundation of Jiangsu Province (No. BK20170141) and Nanjing Medical Science and Technology Development Foundation (No. ZKX22032).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the institutional ethics committee of Nanjing First Hospital (approval number: KY20240514-09-KS-01). Written informed consent was obtained from all participants included in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.