Correlation of Plain Radiographs and 3-Dimensional CT With Coronal and Sagittal Measurements in Patients Undergoing Corrective Osteotomies

Abstract

Background:

The patient-specific instrumentation (PSI) used during corrective high tibial osteotomies and distal femoral osteotomies is based on 3-dimensional computed tomography (3D CT). Plain radiographs are typically used preoperatively to determine the need for an osteotomy; however, it is unclear how well measurements on plain radiographs correlate with 3D CT.

Purpose/Hypothesis:

The purpose of this study was to evaluate the correlation between coronal and sagittal alignment measurements on plain radiographs and 3D CT. It was hypothesized that there would be high agreement in the measurement of the mechanical medial tibial width ratio (mMTWr) and the medial posterior tibial slope (PTS) between both modalities.

Study Design:

Cohort study (diagnosis); Level of evidence, 3.

Methods:

Patients who underwent hip-to-ankle CT as part of the preoperative workup before a corrective osteotomy from October 2020 to November 2023 were reviewed. Coronal (mMTWr) and sagittal alignment (medial PTS) were evaluated preoperatively by 2 raters on standing whole-leg radiographs and a lateral radiograph of the knee, respectively, and by semi-automated PSI software on 3D CT. Intraclass correlation coefficients (ICC) were calculated to assess interrater reliability for each measurement and to evaluate agreement between raters and the PSI software.

Results:

Complete data sets were obtained for 91 cases. The ICC between raters for preoperative mMTWR was 0.99. The ICC between the raters’ measurements and the PSI software measurements of mMTWr was 0.99. The ICC between raters for preoperative PTS was 0.82. The ICC between the raters’ measurements and the PSI software's PTS measurements was 0.63.

Conclusion:

This study found that coronal measurements performed on whole-leg radiographs and 3D CT were highly correlated, with near-perfect agreement, and that medial PTS measurements showed moderate agreement between modalities. These data suggest that measurements on plain radiographs are reproducible and accurate for evaluating coronal alignment and PTS preoperatively. Surgeons can confidently use plain radiographs to assess whether or not a patient is a candidate for a knee osteotomy.

There has been increasing interest in the use of 3-dimensional (3D) patient-specific instrumentation (PSI) for high tibial osteotomies (HTOs) and distal femoral osteotomies (DFOs), with several studies published in the past few years.^12,14,27 Many of these systems require a preoperative hip-to-ankle computed tomography (CT) scan, from which a 3D model can be generated for coronal and sagittal measurements.^12,14,27 Based on these measurements and the surgeon's planned correction, 3D PSI cutting guides may be produced and used intraoperatively to perform the osteotomy in the desired axis and orientation, and custom-contoured plates specific to the patient's bony morphology can be used for fixation.

While creating a surgical plan for the osteotomy using the PSI, the coronal and sagittal corrections are based on measurements performed on the 3D CT model in combination with the surgeon's plan; however, the decision to perform an osteotomy is typically made in the clinical setting based on plain radiographs. While previous studies have evaluated the correlation between CT imaging and plain radiographs for assessing coronal alignment and posterior tibial slope (PTS),^11,18 they did not specifically assess the semi-automated commercial software (Bodycad), which is increasingly used for HTOs and DFOs in multiple recent studies.^14,25 Previous studies have also focused on 3D CT for planning total knee arthroplasty and its correlation with full-length lower extremity plain films, but have not specifically examined how 3D planning software for knee osteotomy correlates with full-length lower extremity plain films.^4,23

Given the increasing interest in this system and 3D-PSI for osteotomies,^2,30 data supporting the correlation between measurements made on plain radiographs and 3D CT would help clinicians confirm that the decision to perform an osteotomy can be reliably made on plain radiographs. Conversely, if there is poor correlation between measurements on plain radiograph and 3D CT, it would suggest that perhaps surgeons should consider routinely obtaining CT imaging before indicating patients for osteotomies.

Although there is a large amount of data to support the benefits of HTOs and DFOs in many patients who only had plain radiographs performed before surgery,^1,9,15,21,31 further understanding of the 3D anatomy in this population will help surgeons refine their indications and preoperative evaluation. This may, in turn, increase the likelihood of patients achieving clinically meaningful improvement after surgery.

Thus, the goal of this study was to evaluate the correlation between preoperative measurements of coronal mechanical alignment and PTS on preoperative radiographs and 3D CT. It was hypothesized that measurements of coronal alignment and PTS on plain radiographs would correlate significantly with those obtained from 3D CT using the PSI software.

Methods

After obtaining institutional review board approval, a retrospective review was performed on a prospectively maintained database of patients who were indicated for HTO or DFO utilizing 3D PSI by 2 sports medicine fellowship-trained senior orthopaedic surgeons (M.T.P. and A.F.V.) at a single institution from October 2020 to November 2023.

Patients were included if they underwent supine whole-leg CT imaging as part of the preoperative workup before an opening wedge (oW) or closing wedge (cW) HTO and had complete preoperative radiographs—including a standing whole-leg radiograph and a standard lateral radiograph of the knee. Patients were excluded if preoperative imaging—including CT or radiographs—was unavailable. Lateral radiographs were excluded from measurements and analyses if they were malrotated, defined as a distance >5 mm between the posterior aspects of the medial and femoral lateral condyles, as described in previous literature. ¹⁷ To standardize the limb rotation for the whole leg radiograph, the patella was centered on the knee joint to minimize the effect of limb malrotation on the measured mechanical axis. Two patients were indicated for an osteotomy but did not undergo surgery; however, they were included in this study as they had met all inclusion criteria with preoperative imaging available. The decision was made to utilize standard lateral knee radiographs and not whole tibia lateral radiographs to measure the PTS because whole tibia radiographs were not available for all patients, were not routinely obtained by all surgeons, and much of the previous literature examining the effect of the PTS on outcomes has utilized standard lateral knee radiographs.

Radiographic Evaluation

Radiographic measurements were performed using the Merge Picture Archiving and Communication System (IBM) by 2 blinded, independent raters (R.L.A. and C.R.), orthopaedic sports medicine fellows. To assess coronal and mechanical alignment, the mechanical medial tibial width ratio (mMTWr), defined as the distance from the medial aspect of the tibia to the point on the tibia where the mechanical axis intersects the tibial plateau, divided by the entire width of the tibial plateau, was measured on standing whole-leg radiographs (Figure 1). The PTS was measured on standing lateral knee radiographs (Figure 2) as the angle between a line tangent to the medial tibial plateau and the tibial proximal anatomical axis, defined as a line passing through the center of 2 circles, 5 cm and 10 cm distal to the joint line, which are aligned with the anterior and posterior tibial cortices.^13,33 The medial PTS was utilized because it has demonstrated a greater number of studies showing association with anterior cruciate ligament (ACL) graft failure than the lateral PTS, and to remain consistent with measurement throughout the study. ²⁶

Figure 1.

Measurement of the mechanical medial tibial width ratio (mMTWr). A preoperative bilateral whole-leg weightbearing radiograph is shown. The mechanical axis weightbearing line (blue line) is drawn from the center of the femoral head to the center of the talus. The distance from the medial aspect of the tibial plateau to the intersection point between this line and the tibial plateau is measured (red line) and divided by the entire width of the tibial plateau (green line) and multiplied by 100 to calculate the mMTWr, which is reported as a percentage.

Figure 2.

Measurement of the posterior tibial slope (PTS). A lateral radiograph of a left knee is shown. A line is drawn tangential to the medial tibial plateau (blue line), and a second line is drawn along the tibial proximal anatomic axis (TPAA) (solid green line), which is represented by a line connecting the center of 2 circles 5 cm and 15 cm distal (red circles) to the tibial plateau that are aligned with the anterior and posterior tibial cortices. The PTS is calculated by subtracting the angle formed by these 2 lines from 90°.

Evaluation and Measurements Based on 3D CT

Whole-leg CT imaging was obtained preoperatively in all patients, which is a prerequisite for the 3D PSI system used (Fine Osteotomy), and 3D digital models were generated for all cases using semi-automated PSI software (Bodycad). To obtain the coronal measurements, the 3D models are superimposed upon the patients’ standing whole-leg radiographs, and the software identifies various bony landmarks, including the medial and lateral femoral epicondyles, distal articular surfaces, as well as the tibial spines, tibial tubercle, and distal tibial plafond, and performs measurements based on these landmarks (Figure 3). This process is semi-automated, meaning that the user can intervene with the measurements and adjust the initial landmarks and best fit. The mMTWr is measured from this. The software also measures the medial and lateral PTS as detailed in Figure 4. Surgeons are not involved in creating these 3D models or in the measurements.

Figure 3.

Example of a preoperative plan based on whole-leg 3D CT imaging. 3D CT, 3-dimensional computed tomography. (A) Preoperative initial measurements performed with the semi-automated commercial software (Bodycad) include the mMTWr and anatomic medial and lateral tibial slope angles (aMTsA and aLTsA), which are shown in the left column. The final planned parameters are shown in the middle column, and the difference between the 2 values is shown in the delta column on the right (in A). (B) The preoperative mMTWr (blue box, A) is reported as a percentage. (C) The aMTsA (green box, A) is reported in degrees of flexion, equates directly to degrees of the PTS. (E) The whole-leg 3D reconstruction of the right lower extremity superimposed on the patient's standing whole-leg radiograph, along with the mechanical axis (teal line) of the entire leg, as well as the mechanical axes of the femur and tibia (purple lines).

Figure 4.

Example of measurement of the PTS based on whole-leg 3D CT imaging. PTS, posterior tibial slope; 3D CT, 3-dimensional computed tomography. Measurements of the medial and lateral PTS are shown. (A) The semi-automated commercial software (Bodycad) generates two 3D discs representing the best fit for each plateau. (B) The anatomic axis of the tibia (blue line, in C and D) is generated from a line running from the proximal anatomic point on the proximal tibia (red circle and green arrow), defined as the point which between the medial and lateral tibial spines and one-fifth of the distance from the anterior to the posterior aspect of the proximal tibia, to the center of the tibia at a point 25% of the proximal to distal length of the tibia distal to the tibial plateau. The software calculates the angle between each disc and the tibial anatomic axis to independently determine the medial and lateral PTS.

Statistical Analysis

Descriptive statistics were calculated for all continuous variables, and frequency counts and percentages were determined for categorical variables. Intraclass correlation coefficients (ICC) were calculated to determine interobserver reliability for radiographic measurements and to assess agreement between the mean of the raters’ measurements on plain radiographs and the semi-automated software using 3D CT. To qualify the ICC, the following commonly used guidelines were set: 0-0.2, slight agreement; 0.21-0.40, fair agreement; 0.41-0.60, moderate agreement; 0.61-0.80, substantial agreement; and 0.81-1, almost perfect agreement. ²² The Pearson correlation coefficient was used to evaluate the correlation between the mMTWr and PTS measurements obtained by the raters and by the software. A paired t test was used to assess whether there was a significant difference between the mean mMTWr and PTS measurements obtained by the raters on plain radiographs and those obtained by the software.

The Shapiro-Wilk test was used to determine whether continuous variables were normally distributed. The appropriate nonparametric test was used in cases where variables were not normally distributed, specifically the Wilcoxon-rank sum test in place of the paired t test, the Mann-Whitney U test in place of an independent t test, and the Spearman rho coefficient in place of the Pearson correlation coefficient. Statistics were performed using JASP Version 0.17.3. (University of Amsterdam). The significance level was set at P < .05. The methodology used here is similar to previous studies evaluating the relationship between CT and plain radiographs in assessing alignment. ¹¹

An a priori power analysis was performed using G*Power 3.1.9.2 (Franz Paul) to estimate the sample size necessary to identify a modest correlation between the 2 measurement methods. To achieve a statistical power of 0.9 for detecting a correlation coefficient of 0.3 at a significance level of .05, a minimum sample size of 88 paired observations was required.

Results

Complete data sets, including all radiographs, were available for 91 patients (45 women, 46 men) who underwent preoperative radiography and CT imaging. Preoperative lateral radiographs of 21 patients were malrotated and were excluded from analyses of PTS measurements; however, their whole-leg standing radiographs were used for coronal measurements. Descriptive data for the cohort are shown in Table 1. The distribution of osteotomy types was as follows: 43 oW-HTO, 17 cW-HTO, 24 oW-DFO, 4 cW-DFO, 1 rotational HTO, 1 combined cW-HTO and oW-DFO, and 1 combined cW-HTO and cW-DFO.

Table 1.

Patient Characteristics (n = 91) ^a

Age, years	37.3 ± 11
Sex, female/male	45/46
Knee, right/left	52/39
Concomitant ACLR or planned ACLR	27
Osteotomy type, n	43 oW-HTO, 17 cW-HTO, 24 oW-DFO, 4 cW-DFO, 1 combined cW-HTO and oW-DFO, 1 combined cW-HTO and cW-DFO, 1 rotational HTO

^aAll continuous variables are reported as a mean ± SD. All categorical variables are reported as n unless otherwise noted. ACLR, anterior cruciate ligament reconstruction; cW, closing wedge; DFO, distal femoral osteotomy; HTO, high tibial osteotomy; oW, opening wedge.

The ICC between raters for preoperative mMTWr measurements (n = 91) was 0.99. The ICC between the rater measurements and the PSI software mMTWr measurements was 0.99. There was no significant difference between the mean mMTWR measured by the raters (40.8% ± 23.5) and the software (41.3% ± 22.1; P = .69). There was a significant correlation between the mMTWr measured by the software and by the raters (r = 0.99; P < .001). The absolute difference in mMTWr between the raters and the PSI software was not correlated with osteotomy type (tibial versus femoral), laterality, or the absolute value of the mMTWr on either modality (P > .05).

The ICC between raters for preoperative PTS (n = 70) measurements was 0.82. The ICC between the rater measurements and the PSI software PTS measurements was 0.63. There was a significant difference between the mean PTS measured by the raters (10.4°± 3.4°; P < .001) and the PSI software (12°± 3.2°). There was a significant correlation between the PTS measured by the raters and the PSI software (r = 0.72; P < .001). The raters’ measurement of PTS was significantly less, 1.6° (95% CI, –3.3 to 6.5), than the PSI software's measurement. A summary of the measurements from each modality is shown in Table 2.

Table 2.

Comparison of Rater Versus PSI Software Measurements of mMTWr (n = 91) and PTS (n = 70) ^a

	Mean	Interrater Reliability, ICC	Agreement With Software	Correlation With Software, r (P)
Rater mMTWr, %	40.8 ± 23.5	0.99	0.99	0.99 (<.001)
Software mMTWr	41.3 ± 22.1	NA	NA
Rater PTS, deg	10.4 ± 3.4	0.82	0.63	0.72 (<.001)
Software PTS, deg	12 ± 3.2	NA	NA	NA

^aAll continuous variables are reported as a mean ± SD. The ICC was used to determine interrater reliability between the 2 raters for mMTWr and PTS measurements and to determine agreement between the raters and the PSI software for mMTWr and PTS measurements. Correlation is reported as the Pearson correlation coefficient (r). Statistical significance was set at P < .05. ICC, intraclass correlation coefficient; mMTWr, mechanical medial tibial width ratio; NA, not applicable; PSI, patient-specific instrumentation; PTS, posterior tibial slope.

The PTS in patients undergoing osteotomy as part of a revision anterior cruciate ligament reconstruction (ACLR) after previous failed ACLR (n = 23) was significantly greater than in patients undergoing osteotomy for chondral/meniscal pathology (n = 46), based on both the rater and the PSI software measurements of the PTS. This comparison is shown in Table 3.

Table 3.

Comparison of the PTS Between Patients Undergoing Slope-Reducing Osteotomy for Prior Failed ACLR Versus Osteotomy for Chondral/Meniscal Pathology (n = 69) ^a,b

	Failed ACLR (n = 23)	Chondral/Meniscal (n = 46)	Student t test, P
Software PTS, deg	14.5 ± 2.3	10.8 ± 2.9	<.0001
Rater PTS, deg	12.5 ± 2.1	9.4 ± 3.6	<.0001
Rater mMTWr, %	40.4 ± 17	41.1 ± 26.6	.92

^aAll continuous variables are reported as mean ± SD. Statistical significance was set at P < .05. ACLR, anterior cruciate ligament reconstruction; mMTWr, mechanical medial tibial width ratio; PTS, posterior tibial slope; 3D CT, 3-dimensional computed tomography.

^bMalrotated radiographs were excluded from the rater data. While 3D CT is not affected by malrotation, to allow for comparisons, data for the same set of patients are included. There was no difference in the mean PTS on 3D CT between patients included and those excluded based on plain radiograph malrotation.

The absolute difference in PTS between the raters and the PSI software was not correlated with osteotomy type (tibial versus femoral), the amount of radiograph malrotation (measured as the mm difference between the posterior aspects of the medial and lateral femoral condyles), laterality, or the absolute value of the PTS on either modality (P > .05). In other words, patients with larger PTS or more radiograph malrotation did not have larger discrepancies between the PTS measured on their plain radiographs and 3D CT.

Discussion

The primary finding of this study was that preoperative evaluation of coronal alignment and PTS on plain radiographs is significantly correlated with software measurements on 3D CT, suggesting that surgeons may use preoperative radiographs reliably to indicate patients for HTOs and DFOs, thus confirming the hypothesis under study. This is important as it provides evidence to support surgeons discussing and planning osteotomies with patients in the clinical setting before obtaining a CT.

Measurements of coronal alignment and the mechanical axis on standing whole-leg radiographs had a high interrater reliability of 0.99, which aligns with previous studies suggesting that coronal alignment can reliably and reproducibly be measured on standing whole-leg radiographs.^5,32 Moreover, measurements on radiographs had a high level of agreement with measurements performed by semi-automated commercial software on 3D CT (ICC, 0.99), and the mean difference between the 2 measurements was 0.5%, which was not significant. This is important because multiple factors can affect the evaluation of coronal alignment on standing whole-leg, including limb rotation and knee flexion.^19,28 This study's data suggest that the measured mMTWr on postoperative standing whole-leg radiographs can be directly compared with the measured mMTWr on 3D CT without the need for any mathematical conversion. However, whole-leg standing films are obtained while weightbearing, whereas 3D CT is performed supine. While no significant differences were seen in this study between weightbearing whole-leg films and CT, one could imagine the possibility of more substantial differences if the knee evaluated had marked deformity present when weightbearing but not while supine (eg, posterolateral laxity in a markedly varus knee).

The interrater reliability for the measurement of PTS on plain lateral radiographs was 0.82. This is similar to previously reported^10,16,29 interrater reliability for the measurement of PTS, which ranged from 0.62 to 0.92. Medial PTS measurements are susceptible to increased variation based on radiograph malrotation as well as how the articular surfaces of the medial and lateral tibial plateaus overlap, which is affected by subtle differences in limb positioning. ³⁴

Measurement of the PTS on plain lateral radiographs by the raters showed moderate agreement with the PTS measured on 3D CT by the PSI software (ICC = 0.63). The measurement of PTS on 3D CT was, on average, 1.6° greater than on plain lateral radiographs. This is likely because the proximal aspect of the anatomic axis on the 3D CT is based on the proximal tibial anatomic point defined by Paley, which is set at one-fifth of the distance from the anterior to the posterior aspects of the tibial plateau, which likely makes the anatomic axis with this method more anterior compared with the TPAA, which tends to posteriorize the anatomic axis given the posterior flare of the tibia proximally as well as given the apex anterior bow of the proximal tibia. ⁷ Posteriorizing the anatomic axis reduces the measured slope, which likely explains why the PTS utilizing the TPAA axis is less than the PTS on 3D CT.

While there appear to be differences in the absolute value of the PTS between radiographs utilizing the TPAA and 3D CT, it is important to emphasize that much of the literature evaluating PTS as a risk factor for ACL injuries has only utilized plain lateral radiographs of the knee. ⁸ Several studies have identified a PTS of >12° as a risk factor for ACL rupture and rerupture after ACLR.^8,20,24 Based on the current data, PTS on 3D CT tends to be greater than that on plain radiographs. Thus, if 3D CT scans were performed on patients in prior studies of PTS, the threshold for increased risk of ACL rupture and rerupture on 3D CT would be >12°. PTS on 3D CT is a different measurement from PTS on plain radiographs.

Nevertheless, a moderate correlation was found between the PTS measured by the raters and by the software (r = 0.72; P < .001). The correlation between the 2 suggests that PTS on plain radiographs can continue to be used as an indicator of the patient's 3D PTS. In other words, these data indicate that surgeons may reliably use preoperative radiographs when considering an osteotomy for both coronal and sagittal alignment correction.

While there is limited reporting of nationwide trends in the utilization of 3D PSI for osteotomies, given the increasing number of studies being published on this topic,^14,25,27 interest in 3D PSI appears to be growing. This parallels the rising use of computer-navigated arthroplasty and the growing patient interest in technology-assisted arthroplasty.^3,6 As more patients undergo osteotomies with 3D PSI and obtain preoperative 3D CT imaging, it is important to confirm that the anatomy on 3D CT correlates with the plain radiography used to evaluate these patients daily. The results of this study suggest that plain radiographs can continue to be used; however, it is important to be aware that, while coronal alignment on the 2 modalities appear to correlate highly, clinicians should be aware that there are more likely to be discrepancies between PTS on both modalities, and there may be patients initially indicated for an isolated coronal correction who may benefit from a biplanar correction.

Limitations

This study had several limitations. Postoperative CT scans were not obtained for patients, which would likely have provided more accurate assessments of postoperative alignment. We attempted to address this by comparing preoperative to postoperative radiographs. The PSI software measurements are semi-automated and may be adjusted by the implant manufacturer's engineering team during preoperative planning. While these measurements appear to be predominantly based on the software assessment of the anatomic landmarks, the possibility of variation in these measurements between different engineers using the semi-automated software cannot be excluded, and no interrater reliability analysis was performed between multiple engineers utilizing the PSI software to perform measurements on the same patient. Measurements of coronal alignment on standing alignment radiographs and the PTS can both be affected by radiograph malrotation. However, malrotated lateral knee radiographs were excluded to attempt to mitigate this concern. Additionally, PTS measurement can be affected by the use of a short lateral knee radiograph versus a full-length lateral tibia, and although this study consistently utilized short knee laterals, it is possible that full-length images might yield different measurements. While this study establishes a correlation between plain radiographs and 3D CT scans, which may give surgeons confidence in preoperative planning based on plain radiographs, it does not obviate the need for 3D CT scans when using this PSI system.

Conclusion

This study found that coronal measurements performed on whole-leg radiographs and 3D CT were highly correlated, with near-perfect agreement, and that medial PTS measurements showed moderate agreement between modalities. These data suggest that measurements on plain radiographs are reproducible and accurate for evaluating coronal alignment and PTS preoperatively. Surgeons can confidently use plain radiographs to assess whether or not a patient is a candidate for a knee osteotomy.