Coronal and rotational femoral anatomy in a Spanish cohort: Anatomical description and surgical implications for total knee arthroplasty

Abstract

Purpose

The primary objective is to analyze the distribution of three femoral anatomical parameters in a Spanish osteoarthritic population using CT-based measurements: (1) anatomical–mechanical axis (AA–MA), (2) posterior condylar–transepicondylar axis (PCA–TEA) and (3) lateral distal femoral angle (LDFA). The secondary objective is to assess the surgical relevance of these findings in TKA.

Methods

It is a multicentric cross-sectional observational study. Preoperative CT-scans of 318 patients from a Spanish cohort were analyzed to determine AA–MA, PCA–TEA and LDFA. Then, femoral implant positioning was simulated as 5° of valgus from the AA and 3° of external rotation from the PCA. For each patient, this simulated positioning was compared to the theoretical mechanical alignment target. Finally, the native femoral JLO (LDFA) of each patient was compared to the mechanical alignment target (90°).

Results

Mean AA–MA relationship was 6.0° ± 0.9°, mean PCA–TEA angle was 3.1° ± 1.8° and mean native LDFA was 87.3° ± 2.9°. Positioning the femoral component systematically at 5° of valgus from the AA and 3° of external rotation from the PCA would result in 74.7% of implants being placed in varus and 46.1% in internal rotation from the TEA. Moreover, mechanically alignment would produce a more varus femoral JLO than the native LDFA in 78.1% of patients.

Conclusions

There is considerable variability in femoral anatomy within the osteoarthritic Spanish population. The 6.7% of patients had AA–MA values outside the 5–7° range. The 41.8% of patients showed PCA–TEA angles smaller than 2° or greater that 4°. The 30.6% of LDFA were outside the 85–90° interval. Simulating a femoral component positioned using only AA and PCA references was associated with deviations from the intended mechanical alignment, most commonly resulting in varus deviation and internal rotation from the TEA.

Resumen

Objetivo

El objetivo principal es analizar la distribución de tres parámetros anatómicos femorales en una población española con gonartrosis mediante mediciones basadas en TC: 1) ángulo anatómico–mecánico (AA-MA), 2) ángulo cóndilos posteriores – transepicondíleo (PCA-TEA) y 3) ángulo distal lateral femoral (LDFA). El objetivo secundario es evaluar la relevancia quirúrgica de estos hallazgos en la ATR.

Métodos

Se realizó un estudio observacional transversal y multicéntrico. Se analizaron las TC preoperatorias de 318 pacientes de una cohorte española para determinar los valores de AA-MA, PCA-TEA y LDFA. Posteriormente, se simuló la posición del componente femoral con 5° de valgo respecto al AA y 3° de rotación externa respecto al PCA. En cada paciente, esta posición simulada se comparó con el objetivo teórico de alineación mecánica. Finalmente, la oblicuidad de la interlínea femoral nativa (LDFA) de cada paciente se comparó con el objetivo de alineación mecánica (90°).

Resultados

La relación media AA-MA fue de 6.0° ± 0.9°, el ángulo medio PCA-TEA fue de 3.1° ± 1.8° y el LDFA nativo medio fue de 87.3° ± 2.9°. Posicionar sistemáticamente el componente femoral a 5° de valgo respecto al AA y 3° de rotación externa respecto al PCA daría lugar a que el 74.7% de los implantes quedaran en varo y el 46.1% en rotación interna respecto al TEA. Además, la alineación mecánica produciría una oblicuidad de la interlínea femoral con mayor varo que el nativo en el 78.1% de los pacientes.

Conclusiones

Existe una variabilidad considerable en la anatomía femoral de la población española con gnartrosis. Un 6.7% de los pacientes presentó valores AA-MA fuera del rango 5°–7°. Un 41.8% mostró ángulos PCA-TEA menores de 2° o mayores de 4°. El 30.6% de los LDFA estuvieron fuera del intervalo 85°–90°. Simular un posicionamiento del componente femoral utilizando únicamente las referencias AA y PCA se asoció con desviaciones respecto el objetivo de alineación mecánica, resultando con mayor frecuencia en varo y en rotación interna respecto al TEA.

Introduction

Anatomical features show considerable variability among population, generally following a normal distribution.¹ In particular, a wide range of native three-dimensional femoral alignment has been reported.^2–5 Furthermore, Pagan et al.⁶ observed significant differences in the distribution of knee phenotypes across geographic areas. Some of these femoral anatomical parameters are critical for accurate planning total knee arthroplasty (TKA).

The relationship between the femoral anatomical axis (AA) and the mechanical axis (MA) is clinically important, as this angle is frequently used to guide coronal alignment in manual TKA.⁷ Similarly, the relationship between the posterior condylar axis (PCA) and the transepicondylar axis (TEA) can be used to determine the rotational alignment of the femoral component.^8,9 The lateral distal femoral angle (LDFA) informs about the native femoral joint line obliquity (JLO). Mechanical alignment in TKA aims to position the femoral component perpendicular to the MA in the coronal plane and with a neutral JLO,^9,10 while the femoral rotation is suggested to find the parallelism between the prosthetic posterior condyles and the TEA.^8,11 However, because intraoperative identification of the femoral MA and TEA is poorly reproducible, these axes are typically estimated using the more accessible AA and PCA.^12–14 These approximations rely on demographic averages.^10,15 Consequently, a detailed understanding of population-specific anatomical characteristics of the femur is essential.¹⁶

The primary objective of this study is to analyze the distribution of three femoral anatomical parameters in a Spanish osteoarthritic population using CT-based measurements: (1) AA–MA relationship, (2) PCA–TEA relationship and (3) LDFA. The secondary objective is to assess the surgical relevance of these findings in TKA.

Methods

Study design

After obtaining the local Ethical Committee approval (24/043), a multicentric cross-sectional observational study was performed in three medical institutions. Patients with knee osteoarthritis treated with a TKA between 2022 and 2024 were analyzed. Inclusion criteria were: (a) patients with primary knee osteoarthritis who underwent a TKA, (b) surgery performed using the Mako robotic system (Stryker Corporation, Kalamazoo, USA) and with preoperative lower limb CT-scan available and (c) intraoperative robotic data accessible. Exclusion criteria were: (a) TKA secondary to infection, fracture, osteotomy or malignancy, (b) revision surgery or (c) severe bone loss.

Outcomes variables

Demographic information as age, gender and diagnosis to indicate a TKA was collected.

CT-based data was collected according to the following protocol: Mako assisted TKA systematically requires a preoperative CT-scan to create a 3D model that guides the preoperative planning and intraoperative steps. The CT-based scanning protocol (PN 200004) focuses on minimizing radiation dose and involves a spiral continuous scan of the hip, knee and ankle.¹⁷ The scan is processed to perform 1 mm slices segmentation, anatomical landmarks identification and measurements (Protocol 214928). Anatomical bone landmarks were identified in the scan. The PCA is defined by a line connecting the two most posterior points of the medial and lateral condyles and the TEA by a line connecting the “surgical” medial epicondyle (bony sulcus) to the lateral epicondyle (bony prominence). The femoral AA is a line that bisects the medullary canal of the femur until the femur knee centre and the femoral MA is a line connecting the hip centre to the femur knee centre. The LDFA is the lateral angle between the femoral MA and the joint line of the distal femur. The tibial MA connects the knee centre with the mid-width of the talus. The HKA (hip–knee–ankle angle) is defined as the medial angle between the femoral and tibial MA. The PCA, TEA, AA, MA, LDFA, PCA–TEA, AA–MA and HKA of each patient were measured in the scan and obtained within the robotic system software. Neutral alignment was defined for limbs with HKA measuring 180° ± 3°, values smaller than 177° were considered varus while values higher than 183° were considered valgus.¹⁸ By their nature and methodology, the procedure is identical each time, so independent measurements by separate observers were not necessary.¹⁹ However, all the measurements were reviewed by a specialist knee orthopaedic surgeon.

Coronal and rotational femoral positioning analysis

First, the relationships between AA–MA and PCA–TEA were analyzed. For each patient, femoral positioning was simulated intending a conventional mechanical aligned TKA based solely on the AA and PCA references. Coronal alignment was systematically defined as 5° of valgus relative to the AA and rotational alignment as 3° of external rotation relative to the PCA. Coronal alignment in manual TKA is commonly guided by the AA; the femoral component is routinely positioned in 5° of valgus to approximate the MA.⁷ For axial alignment, according to measured resection technique, femur is externally rotated by 3° relative to the PCA, intending to compensate for the usual varus orientation of the tibial JLO and to approximate the TEA.^8,9 These approximations rely on previously reported demographic averages.^10,15 Then, for each patient, this simulated positioning was compared to the theoretical mechanical alignment target, defined as coronal alignment perpendicular to the MA and rotational alignment parallel to the TEA.

The native femoral JLO (LDFA) of the cohort was also analyzed. The native LDFA of each patient was compared to the mechanical alignment target (perpendicular to the MA; 90°).

Statistical analysis

Categorical variables were summarized as frequencies and percentages. Continuous variables were reported as mean with standard deviation or as median with interquartile ranges, depending on normality. Normality was assessed using the Shapiro–Wilk test. Comparisons of categorical variables were performed with the Chi-square test or Fisher's exact test, as appropriate. For variables expressed as absolute values, the Wilcoxon signed-rank test was performed. All tests were two-tailed and a p-value < 0.05 was considered statistically significant. The statistical analysis was conducted using Stata® v.14 software (StataCorp, College Station, USA).

Results

Three hundred eighteen patients were included in the study. Mean age was 71.6 ± 7.5 years; 62.9% of them were women and 37.1% were men. Mean measured HKA was 178.7° ± 5.1°.

Mean AA–MA relationship was 6.0° ± 0.9° (range 3–8°). Most patients (93.3%) presented an AA–MA angle between 5° and 7° (Fig. 1, Table 1). There were significant differences in AA–MA according to patient limb alignment (HKA): varus (6.2° ± 0.8°), neutral (6.0° ± 0.8°) and valgus (5.7° ± 1.0°), p < 0.001.

Fig. 1.

AA–MA angle distribution in a Spanish population.

Table 1.

Relationship between the femoral AA and MA in a Spanish population.

AA–MA angle	Femoral coronal deviation (5° valgus from AA vs MA)	Patients (%)
3	2° valgus	1.0
4	1° valgus	4.1
5	0°	20.3
6	1° varus	47.3
7	2° varus	25.7
8	3° varus	1.7

In addition, the deviation from the true MA that would result if the femoral component were systematically positioned at 5° of valgus relative to the AA is analyzed in the central column.

Abbreviations: AA (anatomical axis), MA (mechanical axis).

If femoral coronal alignment is systematically defined as 5° of valgus relative to the AA, the mean implant positioning would be 1.0° ± 0.9° of varus (range: 3° varus to 2° valgus) relative to the femoral MA. In 79.7%, 28.3% and 1.7% of patients, the femoral component would be positioned ≥1°, ≥2° and ≥3° from the actual MA respectively (Table 1). In 74.7% of cases, the component would be implanted in varus.

Mean PCA–TEA relationship was 3.1° ± 1.8° (range 0–8°). Values distribution was relatively homogeneous across the 0–6° range, with most patients (58.2%) showing an angle between 2° and 4° (Fig. 2, Table 2). There were differences in PCA–TEA according to patient limb alignment: varus (2.9° ± 1.7°), neutral (3.1° ± 1.8°) and valgus (3.6° ± 1.8°), p = 0.002.

Fig. 2.

PCA–TEA angle distribution in a Spanish population.

Table 2.

Relationship between the femoral PCA and TEA in a Spanish population.

PCA–TEA angle	Femoral rotational deviation (3° ER from PCA vs TEA)	Patients (%)
0	3° ER	7.1
1	2° ER	13.2
2	1° ER	17.6
3	0°	15.9
4	1° IR	24.7
5	2° IR	12.2
6	3° IR	7.1
7	4° IR	1.4
8	5° IR	0.7

In addition, the deviation from the true TEA that would result if the femoral component were systematically positioned at 3° of external rotation relative to the PCA is analyzed in the central column.

Abbreviations: PCA (posterior condylar axis), TEA (transepicondylar axis), ER (external rotation), IR (internal rotation).

If femoral rotation is systematically defined as 3° of external rotation relative to the PCA, the mean implant positioning would be 0.1° ± 1.8° of external rotation (range: 3° external rotation to 5° internal rotation) relative to the femoral TEA. In 84.0%, 41.7% and 16.3% of patients, the femoral component would be positioned ≥1°, ≥2° and ≥3° from the actual TEA respectively (Table 2). In 46.1% of cases, the component would be implanted in internal rotation from TEA, with an internal deviation of ≥3° in 9.2% of patients.

Mean LDFA was 87.3° ± 2.9° (range 78–100°). LDFA values presented a high variability, however, 69.4% of patients presented an angle between 85° and 90° (Fig. 3). The mean LDFA differed significantly across the limb alignment groups: varus (89.6° ± 2.4°), neutral (87.3° ± 2.0°) and valgus (83.8° ± 2.0°), p < 0.001.

Fig. 3.

LDFA distribution in a Spanish population.

If the femoral JLO is systematically set at 90°, according to mechanical alignment, 78.1% of patients would have a femoral component implanted in greater varus and 13.9% in greater valgus compared with their native femoral JLO (Table 3). Moreover, in 50.0%, 36.1% and 26.3% of patients, the femoral component would be positioned ≥3°, ≥4° and ≥5° of varus from the native femoral JLO respectively.

Table 3.

LDFA distribution in a Spanish population.

LDFA	Femoral JLO deviation (native vs mechanical alignment)	Patients (%)
≤82	≥8° varus	5.4
83	7° varus	5.7
84	6° varus	5.7
85	5° varus	9.5
86	4° varus	9.8
87	3° varus	13.9
88	2° varus	14.9
89	1° varus	13.2
90	0°	8.1
91	1° valgus	7.1
≥92	≥2° valgus	6.8

Difference between patient native femoral JLO versus systematically positioning the femoral component perpendicular to the mechanical axis (90°), according to the mechanical alignment, is analyzed in the central column.

Abbreviations: JLO (joint line obliquity).

Discussion

The most important finding of this study in a Spanish osteoarthritic population was that mean AA–MA relationship was 6.0° ± 0.9°, mean PCA–TEA angle was 3.1° ± 1.8° and mean native femoral JLO (LDFA) was 87.3° ± 2.9°. Based on the population distribution of these anatomical parameters, positioning the femoral component systematically at 5° of valgus relative to the AA and 3° of external rotation from the PCA would result in 74.7% of implants being placed in varus and 46.1% in internal rotation relative to the TEA. Moreover, mechanically alignment in TKA would produce a more varus femoral JLO than the native LDFA in 78.1% of patients.

The femoral AA–MA relationship is clinically relevant, as this angle is commonly used to guide coronal alignment in manual TKA.⁷ Under the mechanical alignment concept, the “5° valgus from the AA” rule is typically applied aiming to position the femoral component perpendicular to the MA.¹² Likewise, the PCA–TEA relationship can be used to determine the rotational alignment of the femoral component, with the implant generally placed in 3° of external rotation relative to the PCA intending to achieve parallelism with the TEA.⁸ However, these approximations are based on averages obtained from other population and, moreover, they do not account for individual anatomical variability.^10,15

In our study, contrary to the classical assumption that the mean AA–MA angle is 5°, the mean value was 6°. The varus knees presented the highest AA–MA angle, 6.2° ± 0.8°. Furthermore, considerable variability in femoral anatomy was observed within the Spanish population. Although most patients (93.3%) had an AA–MA angle between 5° and 7°, the overall range was broad (3–8°). Consistent with our results, Jang et al.²⁰ analyzed 1.078 radiographs and reported a mean AA–MA angle of 6.3°. Similarly, Lampart et al.²¹ assessed CT data from 1.480 patients and found an average AA–MA value of 6°, ranged from 2.5° to 9°. In another study based on 174 preoperative hip–knee–ankle radiographs, the mean femoral AA–MA angle was 5.7° (SD 1.2°, range 2–9°).²²

On the other hand, the mean PCA–TEA angle was 3.1°, closely matching the classical value of 3°. Nevertheless, this relationship showed marked interindividual variability; the standard deviation was ±1.8°. A relatively uniform patient distribution across the 0–6° interval and an overall range of 0–8° was found. Fitz et al.²³ analyzed 3.010 preoperative CT-scans and also reported a mean PCA–TEA angle of 2.9° (range 0.5–16.5°), with 8.4% of patients falling outside the 3 ± 1° interval. In another study of 56 patients with bilateral knee CT-scans, the mean PCA–TEA angle was 3.25° ± 1.4°, ranged from 0° to 6°.

If the variability in femoral anatomy observed in this cohort were not taken into account, and the femoral component were systematically positioned at 5° of valgus relative to the AA and 3° of external rotation from the PCA, 74.7% of implants would be placed in varus and 46.1% in internal rotation relative to the TEA. Patients with varus limb alignment demonstrated the highest mean AA–MA angle (6.2° ± 0.8°) and, therefore, would be the subgroup with more risk of varus implant. In contrast, patients with valgus alignment exhibited the highest PCA–TEA relationship (3.6° ± 1.8°), suggesting a greater risk of femoral internal rotation from TEA. Nam et al.²⁴ analyzed hip-to-ankle radiographs from 493 patients undergoing TKA and reported that 28.6% exhibited a MA-AA relationship outside the range of 5° ± 2°. Similarly, Twiggs et al.²⁵ evaluated 726 CT-scans and found that applying the standard reference of 3° external rotation from the PCA would result in 36.9% of patients presenting a rotational alignment deviating by more than ±2° from the TEA. In a study of 2.128 CT-scans, Jang et al.²⁶ reported that PCA + 3° external rotation differed from the TEA by a mean of 0.6° ± 1.6°, highlighting substantial interindividual variability despite a small mean difference.

In this Spanish cohort, if the femoral component were systematically implanted at 5° of valgus relative to the AA, 1.7% of patients would have the component positioned in ≥3° of varus from the actual MA. According to Lustig et al.,²⁷ 3° of femoral varus represents the limit of the coronal safe zone. In a study comparing 128 TKAs with 2° of residual femoral varus to 128 matched controls with neutral positioning, the under-corrected group showed better short-term clinical results.²⁸ However, femoral component varus malpositioning greater than 3° was identified as a risk factor for aseptic revision (OR 14.0) in another series of 982 TKAs.²⁹

On the other hand, if a systematic 3° of external rotation from the PCA is applied, 9.2% of patients would have the femoral component positioned in ≥3° of internal rotation relative to the TEA. Berger et al.³⁰ reported a direct association between femoral component internal rotation from TEA and patellofemoral complications. Other authors have also identified internal rotation thresholds of 3° and 6° relative to the TEA as indicative of unsafe component positioning.³¹ Moreover, a literature review by Yu et al. supported a neutral or external rotation of the femoral component to reduce femoro-patellar pressure.³²

In this cohort, the mean LDFA was 87.3° ± 2.9°. Our findings highlight the wide variability in LDFA distribution, with values ranging from 78° to 100° and 30.6% of patients outside the 85–90° interval. Consequently, if the femoral JLO was systematically set at 90° according to the principles of mechanical alignment, 78.1% of patients would receive a femoral component implanted in greater varus relative to their native JLO. This finding is particularly relevant in valgus knees, which demonstrated the greatest deviation of the LDFA from the neutral 90° (83.8° ± 2.0°). A systematic review including seven studies also reported important interindividual variability in LDFA, with mean values ranging from 87.3° ± 2.7° valgus to 91.4° ± 2° varus.² Notably, Zang et al.³³ described the widest LDFA range, from 80.6° to 93°. Interestingly, Clark et al.³⁴ demonstrated that applying an individualized alignment strategy led to a final implant position with better restored JLO, which was associated with improved outcomes at 2 years post-TKA. Nonetheless, it is not clear yet that recreating the exact native JLO produces better patient satisfaction after TKA.^35,36

In this study, a considerable variability in femoral anatomy was observed within the Spanish cohort. Moreover, our results suggest that population-specific mean anatomic parameters may deviate from the conventionally accepted averages, as demonstrated by the mean AA–MA angle of 6°. Population anatomical variability may be addressed through meticulous preoperative planning and an individualized approach to TKA. Preoperative long-leg radiographs can assist in estimating the AA–MA relationship and the LDFA for each patient to adapt distal femoral cut; however, its accuracy remains limited.^20,37 In contrast, assessment of the PCA–TEA relationship requires three-dimensional imaging. Consequently, an intraoperative alternative for individualized determination of femoral rotation can be the gap-balancing technique, which adjusts femoral rotation to achieve balanced flexion gaps on a patient-specific basis.⁸ Nevertheless, this strategy ignores osseous anatomical references. On the other hand, the individualized functional alignment strategy with CT-based robotic technology enables accurate identification of patient-specific anatomical landmarks (Fig. 4), thereby reducing dependance on surrogate estimations derived from generalized anatomical references.^38,39 Furthermore, intraoperative quantification of soft-tissue laxity facilitates real-time adjustments of implant positioning to achieve balanced flexion and extension gaps.^27,40 Therefore, multiple strategies and techniques can be useful to perform individualized TKA surgery adapted to each patient-specific anatomy.^36,41

Fig. 4.

Pictures showing the preoperative hip–knee–ankle CT-based information obtained through the robotic system software. Individual anatomical landmarks can be measured for the coronal plane (AA [purple], MA [white], LDFA [purple]) and the axial plane (PCA [purple], TEA [white] and implant rotation [blue]). Patient-specific relationships can also be collected from the “Anatomical Information” screen (red).

This study has some limitations. First, sample size was relatively small for a population-based investigation, however, it is similar to other studies. Second, anatomical femoral sagittal parameters were not assessed. However, both coronal and rotational femoral alignment were evaluated. Third, the HKA was measured using the robotic system; therefore, the assessment was not performed under weight-bearing conditions. Nonetheless, the MPTA, LDFA and JLCA were taken into account using this measurement system. Fourth, clinically relevant thresholds for component positioning are not clearly stablished in the current scientific literature, which limits the extrapolation of our findings. On the other hand, a key strength of the study is the use of routine preoperative CT scans for radiological measurements. It has demonstrated to be more precise and reproducible, particularly in the axial plane.^19,42

Conclusion

There is considerable variability in femoral anatomy within the Spanish osteoarthritic population. Mean AA–MA relationship was 6.0° ± 0.9°, although 6.7% of patients had values outside the 5–7° range. Mean PCA–TEA angle was 3.1° ± 1.8°, with 41.8% of patients showing angles smaller than 2° or greater that 4°. Mean native femoral JLO (LDFA) was 87.3° ± 2.9°, and 30.6% of cases fell outside the 85–90° interval. Simulating a femoral component systematically positioned at 5° of valgus relative to the AA and 3° of external rotation from the PCA would result in 74.7% of implants being placed in varus and 46.1% in internal rotation relative to the TEA. Moreover, mechanically alignment in TKA would produce a more varus femoral JLO than the native LDFA in 78.1% of patients.

Level of evidence

Level of evidence iv.

Author's contributions

All authors contributed equally to this work. All authors contributed to the study conception and design, material preparation, data collection and analysis. The first draft of the manuscript was written by OP, and all authors commented on the versions of the manuscript. All authors read and approved the final manuscript.

Ethics approval

The study was approved by the Center Ethics Committee of the three Institutions. The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki.

Consent to participate and publication

The study was granted exemption from requiring patients written consent by the CenterEthics Committee because it a retrospective study which only utilizes radiological datasets.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial or non-profit sectors.

Conflicts of interest

None.