Distribution of coronal plane alignment knee phenotypes in Japanese patients with knee osteoarthritis: a comparison of two-dimensional and three-dimensional measurements

Gai Kobayashi; Yohei Naito; Shine Tone; Masahiro Hasegawa

doi:10.1016/j.jor.2026.04.012

View/Download PDF

Buy Reprints

PDF

Translate this page into:

76 (); 341-346

doi:

10.1016/j.jor.2026.04.012

Distribution of coronal plane alignment knee phenotypes in Japanese patients with knee osteoarthritis: a comparison of two-dimensional and three-dimensional measurements

Gai Kobayashi, Yohei Naito, Shine Tone, Masahiro Hasegawa

Department of Orthopaedic Surgery, Mie University Graduate School of Medicine, 2-174 Edobashi, Tsu, Mie, 514-8507, Japan

⁎Corresponding author: Masahiro Hasegawa. masahase@clin.medic.mie-u.ac.jp

Received: 2026-4-1, Accepted: 2026-4-27, Published: 2026-4-28

The Authors

Disclaimer:
This article was originally published by Reed Elsevier India Pvt. Ltd. and was migrated to Scientific Scholar after the change of Publisher.

Abstract

This study aimed to determine whether the distribution of coronal plane alignment of the knee (CPAK) classification differs in Japanese patients with knee osteoarthritis (OA) when assessed using two-dimensional (2D) standing long-leg radiographs versus three-dimensional (3D) supine computed tomography-based measurements.

This retrospective study included 100 knees of 86 patients who underwent primary total knee arthroplasty for OA (mean age, 72.9 years; 27 men, 73 women; body mass index, 27.3 kg/m²). The lateral distal femoral angle, medial proximal tibial angle, arithmetic hip–knee–ankle angle (aHKA), and joint line obliquity (JLO) were measured on 2D and 3D images. Corresponding parameters were compared, and correlations were evaluated. Using aHKA and JLO, differences in CPAK phenotypes between 2D- and 3D-based classifications were assessed.

All parameters differed significantly between 2D and 3D measurements. Mean lateral distal femoral angle was 88.2° on 2D and 87.9° on 3D (p < 0.01), medial proximal tibial angle was 85.3° and 83.8° (p < 0.01), aHKA was −2.9° and −4.1° (p < 0.01), and JLO was 173.5° and 171.7° (p < 0.01), respectively. The distribution of CPAK phenotypes did not differ significantly between 2D- and 3D-based classifications, with good agreement (Cohen's kappa = 0.615). All parameters demonstrated strong correlations between 2D and 3D measurements (r > 0.80).

In Japanese patients with knee OA, the distribution of CPAK phenotypes showed good agreement between 2D- and 3D-based assessments. Although 2D measurements may underestimate varus alignment and JLO, they appear sufficient for CPAK-based phenotypic assessment in clinical practice.

Abstract

Highlights

•Good agreement in CPAK phenotype distribution was observed between 2D radiographs and 3D CT in Japanese knee OA.•2D evaluation appeared sufficient for CPAK phenotyping despite underestimating varus alignment and joint line obliquity.

Keywords

Knee osteoarthritis

Coronal plane alignment of the knee

Two-dimensional

Three-dimensional

Japanese

Show Related Articles from PubMed

1 Introduction

Total knee arthroplasty (TKA) is an effective treatment for knee osteoarthritis (OA). However, it has been reported that approximately 10% to 20% of patients are dissatisfied with the outcome.¹^,² Recently, personalised alignment targets have been developed to reduce dissatisfaction and improve patient satisfaction and are attracting increasing interest.³^,⁴ The coronal plane alignment of the knee (CPAK) classification proposed by MacDessi et al..⁵ has been used as a simple and practical system to discuss alignment strategies in TKA. In recent years, the distribution of CPAK phenotypes has been reported in various populations worldwide, revealing geographic and ethnic differences; CPAK has become increasingly important in the debate on personalised TKA.⁶^,⁷ Conversely, CPAK classification is based on radiography-derived two-dimensional (2D) measurements⁵; the validity of radiographic assessment and its agreement with three-dimensional (3D) measurements remain important issues. Although the interobserver reliability of long-leg radiographs (LLRs) is generally good,⁸ radiographic measurements are highly sensitive to lower limb positioning⁹ and may underestimate varus alignment and joint line obliquity (JLO) compared with computed tomography (CT)-based measurements,¹⁰ with discrepancies in alignment parameters also being reported.¹¹ Moreover, the reliability of CPAK classification has been shown to vary across different imaging modalities, highlighting that the choice of imaging modality is a critical issue.¹² Furthermore, CPAK distributions have been reported using proprietary CT-based 3D planning systems and semi-automated software.¹⁰^,¹³^,¹⁴ However, the definition of anatomical landmarks in these systems depends on specific vendor- and engineer-defined settings; the measurement algorithms are not necessarily standardised. ZedKnee (LEXI, Tokyo, Japan) is a CT-based 3D planning system compatible with multiple implant systems, and its reliability has been validated in previous studies.¹⁵ To the best of our knowledge, no study has evaluated coronal alignment parameters and CPAK distributions using supine CT-based 3D measurements obtained from a generic CT-based planning system that is not tied to a specific platform. Although the distribution of CPAK phenotypes in Japanese patients with knee OA has been reported based on 2D measurements from LLRs,¹⁶^,¹⁷ no study has assessed CPAK using supine CT-based 3D measurements. Therefore, the purpose of this study was to compare the lateral distal femoral angle (LDFA), medial proximal tibial angle (MPTA), JLO, and arithmetic hip–knee–ankle angle (aHKA) derived from 2D measurements of standing LLRs with those obtained from supine CT-based 3D measurements in Japanese patients with knee OA and to investigate the resulting CPAK phenotype distributions.

2 Methods

2.1

2.1 Patient selection

This retrospective study included 138 consecutive patients who underwent primary TKA at a single institution between October 2019 and March 2023. After excluding 24 knees with rheumatoid arthritis, two knees with osteonecrosis, five knees with prior intra- or extra-articular fractures of the femur or tibia, five knees with prior total hip arthroplasty, and two knees from non-Japanese patients, 100 knees were available for analysis. Patient demographic characteristics are presented in Table 1. This study was approved by our facility's Institutional Review Board (approval no. H2018-083; approval date: December 21, 2018). Informed consent was obtained from all patients for the use of their data. All procedures were performed in accordance with the principles of the Declaration of Helsinki.

Table 1 Patient demographics.

Parameters
Mean age, years (mean ± SD)	72.9 ± 6.9
Sex	Male: 27 knees
Sex	Female: 73 knees
Mean weight, kg (mean ± SD)	66.0 ± 16.2
Mean height, cm (mean ± SD)	155.0 ± 9.5
Mean BMI, kg/m² (mean ± SD)	27.3 ± 5.4
Right-to-left ratio	Right: 47 knees
Right-to-left ratio	Left: 53 knees
K-L grade (3:4)	22: 78

2.2

2.2 Radiological evaluation (2D measurements)

Radiographic measurements were performed according to previously published methods.⁵^,¹⁷ All patients underwent digital LLRs. The distance between the cassette and tube was 180 cm, tube voltage was 75 kV, and tube current was 400 mA. Patients stood barefoot in a closed-leg stance with the patellae facing anteriorly; image stitching was subsequently performed using FUJIFILM software (FUJIFILM, Tokyo, Japan). The appropriate rotational position was defined as the patella facing forward symmetrically on both sides, lesser trochanters with similar shapes, and overlapping proximal tibiofibular joints with similar shapes. The 2D radiographic measurements included LDFA and MPTA. The mechanical femoral axis was defined as the line from the centre of the femoral head to the centre of the knee. The mechanical tibial axis was defined as the line from the centre of the knee to the centre of the ankle. LDFA was defined as the lateral angle between the mechanical femoral axis and line crossing the articular surface of the distal femur at its distal points on the lateral and medial sides. MPTA was defined as the angle between the mechanical tibial axis and line drawn through the proximal articular surfaces of the medial and lateral tibial plateaus.

2.3

2.3 Radiological evaluation (3D measurements)

Preoperative CT scans from the pelvis to the ankle were obtained for all patients in the supine position. Scans were acquired with a tube voltage of 120 kV, tube current of 150 mA, and slice thickness of 1.0 mm. The CT data were imported into the Zed Knee System (LEXI Co., Tokyo, Japan), a validated CT-based 3D preoperative planning system for TKA.¹⁴ For the 3D measurements, anatomical landmarks were defined with reference to a previously published report.¹⁸ The mechanical femoral axis was defined as the line connecting the centre of the femoral head to the midpoint of the surgical epicondylar axis (SEA). LDFA was measured as the lateral angle between the mechanical femoral axis and line connecting the most distal points of the medial and lateral femoral condyles. The mechanical tibial axis was defined as the line connecting the centre of the intercondylar eminence to the centre of the ankle joint. MPTA was measured as the medial angle between the mechanical tibial axis and line connecting points located at the deepest part of the medial and lateral tibial compartments, positioned two-thirds of the way from anterior to posterior on the tibial plateau.

2.4

2.4 CPAK classification

Following the original description of the CPAK classification,⁵ lower limb alignment was determined using the aHKA; aHKA was calculated as MPTA minus LDFA; negative values indicated varus alignment, and positive values indicated valgus alignment. JLO was defined as the sum of LDFA and MPTA. A JLO of 180° was interpreted as an approximately neutral joint line, values > 180° as an apex proximal joint line, and values < 180° as an apex distal joint line. Knees were classified into nine CPAK phenotypes based on two independent variables: aHKA (varus, aHKA < −2°; neutral, aHKA = 0° ± 2°; and valgus, aHKA >2°) and JLO (apex distal, JLO >183°; neutral, JLO = 180° ± 3°; and apex proximal, JLO <177°). All 2D and 3D radiological parameters and CPAK classifications were analysed by the authors.

2.5

2.5 Statistical analysis

Based on a previous study,¹⁰ a minimum of 61 paired measurements was required to detect whether a true difference exists (α = 0.05, power = 0.9). For both the 2D and 3D measurements, reliability was assessed as follows: for intraobserver reliability, each parameter was measured twice on 20 knees at an interval of 4 weeks by a single orthopaedic surgeon; for interobserver reliability, two orthopaedic surgeons measured each parameter twice on 20 knees at an interval of 4 weeks. Intraclass correlation coefficients were calculated to assess intra- and interobserver agreement. The mean values of LDFA, MPTA, aHKA, and JLO derived from the 2D measurements were compared with those derived from the 3D measurements using the Wilcoxon signed-rank test. Spearman's rank correlation coefficient was used to examine associations between the 2D and 3D measurements. Differences in the distribution of CPAK phenotypes between the 2D and 3D measurements were analysed using Fisher's exact test. Agreement between the 2D and 3D CPAK classifications was evaluated using Cohen's kappa coefficient (κ). Cohen's κ was interpreted as follows: 0–0.40, poor; 0.41–0.60, moderate; 0.61–0.80, good; 0.81–1.00, excellent agreement.¹⁹ Statistical analyses were performed using IBM SPSS Statistics (software version 29.0; IBM Corp., Tokyo, Japan). Statistical significance was set at p < 0.05.

3 Results

In the 2D measurements, intra- and interobserver reliabilities were 0.940 and 0.934 for LDFA and 0.936 and 0.923 for MPTA, respectively. In the 3D measurements, intra- and interobserver reliabilities were 0.950 and 0.952 for LDFA and 0.960 and 0.951 for MPTA. The measurement results are presented in Table 2. All alignment parameters (LDFA, MPTA, aHKA, and JLO) were significantly smaller when measured using 3D imaging than when measured using 2D imaging. Fig. 1 shows the CPAK phenotype distributions derived from the 2D and 3D measurements. There were no significant differences in the frequency of any phenotype; the Cohen's κ for agreement between the 2D and 3D CPAK distributions was 0.615, indicating good agreement. Fig. 2 illustrates correlations between the 2D and 3D measurements for each parameter. Strong positive correlations were observed for all parameters (r > 0.80).

Table 2 Radiographic data.

Parameters	2D measurement	3D measurement	p value
LDFA (°)	88.2 ± 2.9	87.9 ± 2.9	<0.01∗
MPTA (°)	85.3 ± 3.9	83.8 ± 4.4	<0.01∗
aHKA (°)	−2.9 ± 5.8	−4.1 ± 6.3	<0.01∗
JLO (°)	173.5 ± 3.6	171.7 ± 3.9	<0.01∗

Coronal plane alignment of the knee phenotype distribution measured by two-dimensional and three-dimensional methods. aHKA, arithmetic hip–knee–ankle angle; JLO, joint line obliquity; LDFA, lateral distal femoral angle; MPTA, medial proximal tibial angle; 2D, two-dimensional; 3D, three-dimensional.

Correlations between two-dimensional and three-dimensional measurements. (a) LDFA, (b) MPTA, (c) aHKA, and (d) JLO. All parameters showed significant positive correlations between 2D and 3D measurements. LDFA, lateral distal femoral angle; MPTA, medial proximal tibial angle; aHKA, arithmetic hip–knee–ankle angle; JLO, joint line obliquity; 2D, two-dimensional; 3D, three-dimensional.

4 Discussion

The most important finding of this study was that the distribution of CPAK phenotypes in patients with knee OA did not differ significantly between 2D measurements based on standing LLR and 3D measurements based on supine CT; a good level of agreement was observed between the two modalities (κ = 0.615). For the mean values of LDFA, MPTA, aHKA, and JLO, significant and consistent differences were observed between values derived from the 2D measurements and those derived from the 3D measurements for all parameters, demonstrating a discrepancy between 2D and 3D measurements, similar to those of previous studies.¹⁰^,¹¹^,¹³ Parameters measured in 2D and 3D showed strong positive correlations, as previously reported.¹⁸ These findings suggest that although individual coronal alignment parameters may not be directly interchangeable between 2D and 3D assessments, the CPAK classification derived from aHKA and JLO remains largely stable across imaging modalities. Accordingly, 2D evaluation based on LLR, which is widely available and routinely used in clinical practice, appears to be sufficiently practical for CPAK phenotypic characterisation of coronal knee alignment in patients with knee OA.

In this study, all alignment parameters (LDFA, MPTA, aHKA, and JLO) were significantly smaller when assessed using 3D measurements than when assessed using 2D measurements. The mean difference in LDFA between the 2D and 3D images was modest at 0.3°, whereas the mean difference in MPTA was larger at 1.5°. The posterior tibial slope is approximately 3° greater on the medial tibial plateau than on the lateral tibial plateau²⁰; a CT-based analysis in Japanese patients with knee OA similarly demonstrated that the medial posterior tibial slope is approximately 2° greater than the lateral slope.²¹ Therefore, use of points located at two-thirds of the anteroposterior depth on the tibial plateau in the 3D protocol is likely influenced by this mediolateral difference in posterior slope, raising the possibility that tibial varus alignment is underestimated in 2D LLR. This interpretation is consistent with the findings of Tarassoli et al.,¹⁰ who reported that coronal lower limb alignment determined by LLR underestimated the magnitude of both constitutional varus alignment and JLO compared with observations using CT. Despite the use of similar anatomical landmarks, the mean MPTA in this study was 83.8°, which is markedly smaller than the value of 86.1° reported by Tarassoli et al..¹⁰ This discrepancy likely reflects geographic or ethnic differences in coronal tibial morphology. Indeed, a CT-based study of Japanese patients with knee OA using a central tibial landmark reported a mean MPTA of 83.9°,²² which is almost identical to the present findings and supports the consistency of the results within this geographical cohort.

When the proportion of CPAK Type I in our cohort was compared with that in previous studies that compared 2D and 3D CPAK classifications, geographic differences became apparent. In an Australian cohort,¹⁰ CPAK Type I accounted for 24.1% of knees on 2D LLR and increased to 35.3% on 3D CT-based measurements, indicating that CT identified a larger subset of varus and distal apex phenotypes than 2D LLR. In a Slovenian cohort,¹⁴ CPAK Type I represented only a minority of cases, accounting for 14.9% on 2D LLR and 12.3% on 3D CT-based measurements; no significant differences were observed between the 2D and 3D assessments for any alignment parameter within this phenotype. However, in Japanese patients with knee OA, CPAK Type I has been reported as the predominant phenotype on 2D LLR, with proportions of 53.8% and 64.7% in studies.¹⁶^,¹⁷ In this study, CPAK Type I accounted for 65.0% of the knees on 2D LLR and further increased to 70.0% on 3D CT-based measurements. These observations are consistent with those of a previous report,⁷ which demonstrated an enrichment of Type I phenotypes in Asian patients with knee OA than in Western populations. Collectively, these data indicate that the high proportion of CPAK Type I seen in our cohort is more likely to reflect underlying geographic and ethnic differences in coronal knee morphology than a measurement artefact, while the modest increase from 2D to 3D is compatible with the tendency of CT-based assessment to reveal varus and apex distal alignment patterns more clearly. Although a European cohort study¹³ did not provide the exact proportion of CPAK Type I, the results are largely consistent with our findings. In that study, the mean aHKA and JLO values were highly similar between the 2D and 3D assessments, whereas the complete agreement rate for the CPAK classification was 60.8%, indicating that phenotypes can change between modalities in a certain proportion of cases. This pattern is consistent with the good but not perfect agreement observed in the present study (κ = 0.615) and suggests that, while 2D and 3D assessments generally produce comparable CPAK distributions at the population level, individual knees may not always be assigned to the same phenotype.

Changes in CPAK classification before and after TKA are associated with patient-reported outcome measures, with shifts in aHKA classification and the presence of an apex proximal JLO potentially exerting a detrimental effect on postoperative outcomes.²³ Other authors have found that changes in CPAK classification do not necessarily translate into differences in functional outcomes²⁴; specific CPAK phenotypes may interact differently with various alignment strategies.²⁵ Collectively, these findings suggest that the CPAK classification should not be viewed as a system for determining a single optimal alignment value, but rather as an index for understanding the coronal alignment phenotype of each knee. The results of this study support this interpretation by indicating that, for the purpose of CPAK phenotypic assessment, 2D evaluation using LLR may be sufficiently practical in patients with knee OA. However, as Strahovnik et al..¹⁴ demonstrated, measurement errors in LLR can be substantial in certain phenotypes. Reports have shown a non-negligible proportion of mismatches between 2D- and 3D-based CPAK classifications,¹³ whereas weight-bearing upright CT can further alter coronal alignment parameters.¹¹ Therefore, it would be inappropriate to conclude that 2D evaluation is adequate for all cases. In knees with severe deformities, marked valgus alignment, or suspected rotational abnormalities, CT-based 3D analysis is likely to play an important complementary role.

Our study has several limitations. First, the results may have been influenced by the choice of tibial reference points used for the 3D measurements. However, the landmarks adopted in this study have previously been shown to provide good intra- and interobserver reliability and are useful for coronal alignment assessment.¹⁰^,¹⁸ Our findings are consistent with those of a CT-based analysis performed in the same geographic cohort.²² Second, our cohort exclusively consisted of Japanese patients. As coronal alignment morphology is known to vary substantially across geographic and ethnic populations,⁶^,⁷ the generalisability of our findings to a global population may be limited. Nevertheless, this underscores the need for further accumulation of CPAK-related data in diverse populations. We believe that this study provides meaningful information from an Asian cohort.

5 Conclusion

In Japanese patients with knee OA, the distribution of CPAK phenotypes showed good agreement between 2D measurements based on LLR and 3D measurements based on CT. Although 2D measurements tend to underestimate varus alignment and JLO, they appear sufficient for CPAK-based phenotypic assessment in routine clinical practice.

Data availability statement

The datasets analysed during the current study are available from the corresponding author upon reasonable request.

Ethical statement

This study was approved by our facility's Institutional Review Board (approval no. H2018-083; approval date: December 21, 2018). Informed consent was obtained from all patients for the use of their data. All procedures were performed in accordance with the principles of the Declaration of Helsinki.

Guardian/patient's consent statement

This manuscript is an original article. Guardian/patient consent is not mandatory for this article type. No identifiable patient information or images are included in this manuscript.

Authorship contribution: CRediT

Gai Kobayashi: Formal analysis, Investigation, Methodology, Writing – original draft. Yohei Naito: Writing – review & editing., Shine Tone: Writing – review & editing. Masahiro Hasegawa: Writing – review & editing.

Funding statement

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

References

Bourne R.B., Chesworth B.M., Davis A.M., Mahomed N.N., Charron K.D., . Patient satisfaction after total knee arthroplasty: who is satisfied and who is not? Clin Orthop Relat Res. 2010;468(1):57-63.
[Google Scholar]
DeFrance M.J., Scuderi G.R., . Are 20% of patients actually dissatisfied following total knee arthroplasty? A systematic review of the literature. J Arthroplast. 2023;38(3):594-599.
[Google Scholar]
Karasavvidis T., Pagan Moldenhauer C.A., Haddad F.S., Hirschmann M.T., Pagnano M.W., Vigdorchik J.M., . Current concepts in alignment in total knee arthroplasty. J Arthroplast. 2023;38(7 Suppl 2):S29-S37.
[Google Scholar]
Elmajee M., Mersal M., Ben Nafa W., et al . Exploring knee alignment: demystifying traditional and emerging approaches. Cureus. 2025;17(9)
[Google Scholar]
MacDessi S.J., Griffiths-Jones W., Harris I.A., Bellemans J., Chen D.B., . Coronal Plane Alignment of the Knee (CPAK) classification. Bone Joint J. 2021;103-B(2):329-337.
[Google Scholar]
Pagan C.A., Karasavvidis T., Lebrun D.G., Jang S.J., MacDessi S.J., Vigdorchik J.M., . Geographic variation in knee phenotypes based on the coronal plane alignment of the knee classification: a systematic review. J Arthroplast. 2023;38(9):1892-1899.e1.
[Google Scholar]
Giurazza G., Tanzilli A., Franceschetti E., et al . Coronal plane alignment of the knee phenotypes distribution varies significantly as a function of geographic, osteoarthritic and sex-related factors: a systematic review and meta-analysis. Knee Surg Sports Traumatol Arthrosc. 2025;33(10):3592-3605.
[Google Scholar]
Boonen B., Kerens B., Schotanus M.G., Emans P., Jong B., Kort N.P., . Inter-observer reliability of measurements performed on digital long-leg standing radiographs and assessment of validity compared to 3D CT-scan. Knee. 2016;23(1):20-24.
[Google Scholar]
Sun X., Yang B., Xiao S., et al . Effect of limb rotation on radiographic alignment measurement in mal-aligned knees. Biomed Eng Online. 2021;20(1):119.
[Google Scholar]
Tarassoli P., Corban L.E., Wood J.A., Sergis A., Chen D.B., MacDessi S.J., . Long leg radiographs underestimate the degree of constitutional varus limb alignment and joint line obliquity in comparison with computed tomography: a radiographic study. Knee Surg Sports Traumatol Arthrosc. 2023;31(11):4755-4765.
[Google Scholar]
Sasaki R., Niki Y., Kaneda K., et al . Discrepancies in long-leg alignment and knee joint line obliquity between two- and three-dimensional measurements under weight-bearing conditions: effects on coronal plane alignment of the knee classification. Knee. 2025;53:237-245.
[Google Scholar]
Bouché P.A., Kutaish H., Gasparutto X., Lubbeke A., Miozzari H.H., Hannouche D., . Coronal plane alignment of the knee classification in osteoarthritic knees: poor to moderate reliability and implications for imaging choice. J Arthroplast. 2026;41:725-732.
[Google Scholar]
Itou J., Gazali I., Pandit H., Okazaki K., Ascani D., Peersman G., . The CPAK classification in three-dimensional measurements is consistent with those in two-dimensional measurements. Arch Orthop Trauma Surg. 2025;145:160.
[Google Scholar]
Strahovnik I., Strahovnik A., Fokter S.K., . Rotational projection errors in coronal knee alignment on weight-bearing whole-leg radiographs: a 3D CT reference across CPAK morphotypes. Bioengineering (Basel). 2025;12(8):794.
[Google Scholar]
Yoshino K., Hagiwara S., Nakamura J., Tsukeoka T., Tsuneizumi Y., Ohtori S., . Intra- and interobserver reliability and agreement in three-dimensional computed tomography measurements of component positions after total knee arthroplasty. Knee. 2019;26(5):1102-1110.
[Google Scholar]
Toyooka S., Osaki Y., Masuda H., et al . Distribution of coronal plane alignment of the knee classification in patients with knee osteoarthritis in Japan. J Knee Surg. 2023;36(7):738-743.
[Google Scholar]
Kobayashi G., Hasegawa M., Yamabe Y., Tone S., Naito Y., Sudo A., . Classification of coronal plane alignment of arthritic and healthy knees in Japan. J Joint Surg Res. 2025;3:26-30.
[Google Scholar]
Gieroba T.J., Marasco S., Babazadeh S., Di Bella C., van Bavel D., . Arithmetic hip knee angle measurement on long leg radiograph versus computed tomography-inter-observer and intra-observer reliability. Arthroplasty. 2023;5(1):35.
[Google Scholar]
Kundel H.L., Polansky M., . Measurement of observer agreement. Radiology. 2003;228(2):303-308.
[Google Scholar]
Degen N., Sass J., Jalali J., et al . Three-dimensional assessment of lower limb alignment: reference values and sex-related differences. Knee. 2020;27(2):428-435.
[Google Scholar]
Nishitani K., Kuriyama S., Nakamura S., et al . A plain radiography may underestimate the steep posterior tibial slope of the medial condyle in elderly Japanese with varus knee osteoarthritis undergoing arthroplasties. J Joint Surg Res. 2023;1:139-144.
[Google Scholar]
Ogawa H., Nakamura Y., Sengoku M., et al . Medial proximal tibial angle at the posterior tibial plateau represents the pre-arthritic constitutional medial proximal tibial angle in anterior cruciate ligament-intact, advanced osteoarthritis of the knee. Knee Surg Sports Traumatol Arthrosc. 2022;30(9):2941-2947.
[Google Scholar]
Konishi T., Hamai S., Tsushima H., et al . Pre- and postoperative Coronal plane alignment of the knee classification and its impact on clinical outcomes in total knee arthroplasty. Bone Joint J. 2024;106-B(10):1059-1066.
[Google Scholar]
Agarwal S., Ayeni F.E., Sorial R., . Impact of change in coronal plane alignment of knee (CPAK) classification on outcomes of robotic-assisted TKA. Arthroplasty. 2024;6(1):15.
[Google Scholar]
Young S.W., Tay M.L., Kawaguchi K., et al . Insall award: functional versus mechanical alignment in total knee arthroplasty: a randomized controlled trial. J Arthroplast. 2025;40(7S1):S20-S30.e2.
[Google Scholar]