JORT Journal

Introduction

Accurate visualization of knee alignment, particularly in the coronal plane, is crucial for preoperative planning for total knee arthroplasty (TKA). Preoperative weight-bearing anteroposterior (AP) and lateral radiographs of the knee with a field of view from mid-femur to mid-tibia, so-called “short-leg” radiographs (SLRs), are commonly used for preoperative planning. However, full-length (long-leg) AP weight-bearing radiographs (LLRs) have been deemed the gold standard for visualizing many of the radiographic measurements of knee alignment [1,2]. The Coronal Plane Alignment of the Knee (CPAK) system uses joint line obliquity (JLO) and arithmetic hip-knee-ankle angle (aHKA) to determine coronal knee alignment, which has been validated on LLRs [3].

While seemingly more accurate, the use of LLRs as opposed to SLRs involves significantly increased radiation exposure, carries additional costs, and requires more time to capture a clear image [2,4,5]. However, SLRs do not contain all of the anatomic landmarks to exactly reproduce the angles that comprise the CPAK classification. While previous studies have concluded that standard (short-leg) AP radiographs inaccurately assess several factors of lower limb alignment and cannot adequately replace LLRs [2,6], some authors have advocated for their use anyway, including a “correction factor” based on differences in anatomic tibiofemoral angle and mechanical tibiofemoral angle [5,7-12]. This technique, however, has not been previously validated. Kraus et al employed SLRs with a correction factor of 5° added to the anatomical LDFA to determine CPAK classification in a recent publication investigating the relationship between CPAK retention and patient-reported outcome measures [7].

We therefore asked the following questions: 1) How does interrater and intrarater agreement compare between LDFA and MPTA measurements on short-leg and long-leg radiographs? 2) Is there a systematic difference for LDFA when measured on short-leg radiographs versus long-leg radiographs? 3) What LDFA correction for short-leg radiographs provides the optimal CPAK agreement?

Materials and methods

Study Design

This study included a retrospective radiographic review of 271 patients who underwent staged bilateral TKA with mechanical alignment approach at an ambulatory surgery center by a single orthopedic surgeon. Patients were included if they had both SLRs and LLRs from before and after surgery. Patients were excluded if they had a fracture malunion, fracture fixation, osteotomy, or other surgery altering bony alignment. Patients who underwent additional surgery (other than a TKA) on the operative leg in the time between preoperative and postoperative imaging were also excluded (n=3).

Radiographic Assessment and Calculations

Preoperative SLRs and LLRs were utilized for measurement. Two independent raters reviewed all patients’ short- and long-leg AP weight-bearing radiographs to assess knee alignment. Long-leg radiographic measurements included the lateral distal femoral angle (LDFA) and the medial proximal tibial angle (MPTA). From these, the arithmetic hip-knee-ankle angle (aHKA) was calculated as MPTA minus LDFA, and the joint line obliquity (JLO) was calculated as the sum of LDFA and MPTA. For short-leg radiographs, measurements included the short-leg LDFA (slLDFA) and short-leg MPTA (slMPTA). Corresponding calculations included the short-leg arithmetic HKA (slaHKA), defined as slMPTA minus slLDFA, and the short-leg JLO (slJLO), defined as the sum of slLDFA and slMPTA. Long-leg radiographic measurements were obtained following the methodology described by Graden et al and others [2,13-15]. For SLRs, in cases where the hip and ankle were not visualized, lines were drawn through the center of the femoral and tibial diaphysis (Figure 1).

Figure 1: Long-leg (A) and short-leg (B) radiographs of a knee showing the lateral distal femoral angle (LDFA) and the medial proximal tibial angle (MPTA) measurements to determine Coronal Plane Alignment of the Knee (CPAK) classifications.

Measurements were completed in a blinded manner using OpalRad software (Version 2.4.4.7), which ensures accurate magnification calibration via a standard 25.4 mm orthosphere. After completing all analyses and re-measuring 50 cases, inter- and intra-rater reliability were calculated. CPAK classifications were determined with the algorithms described by MacDessi et al. [3,16].

Data Analysis

For all alignment angles (LDFA, MPTA, aHKA, JLO), we first assessed intra- and inter‐rater reliability using the intraclass correlation coefficient (ICC) and Kappa Statistics [17]. To compare short-leg and long-leg measurements, we conducted paired t-tests and calculated Pearson correlation coefficients with simple linear regression. Differences exceeding the minimal clinically important difference (MCID) of 2°, as defined by Graden et al, were considered both statistically and clinically significant [2]. Agreement in CPAK classification was quantified in three ways: JLO agreement, aHKA agreement, and exact CPAK agreement, summarized in contingency tables. Finally, to identify the optimal LDFA correction for SLRs, we plotted JLO agreement, aHKA agreement, and CPAK agreement versus LDFA corrections from 2° to 10° by .001° increments. The maximum composite agreement (defined by the average of the JLO and aHKA agreements) was identified, and the range of LDFA correction values that demonstrated a composite agreement within 0.2% of the maximum composite agreement was marked. Statistical analyses were performed using R version 4.4.2 (Vienna, Austria).

Results

Patient characteristics

A total of 268 preoperative SLRs and LLRs, each from a unique patient, were examined. The patients’ mean ± SD (range) age was 70.51 ± 7.53 (42-91). Of 268, male patients made up 102 (38%) of the limbs, the rest were female (166 (62%)). Laterality was recorded with right knees comprising 143 (53%) and left knees comprising 125 (47%) of the 268 radiographs. In the sample of 268 knees, preoperatively, 42.5% of knees were classified as CPAK I, 22.4% CPAK II, and 18.7% CPAK III, aligning with common phenotypes of an arthritic population [3].

Interrater and Intrarater Agreement

The intra-observer reproducibility (ICCs) ranged from 0.82 to 0.97 for long-leg and 0.78 to 0.96 for short-leg measurements. The inter-observer reliability (ICCs) ranged from 0.90 to 0.91 for longleg and 0.79 to 0.91 for short-leg measurements (Table 1).

Table 1: Inter- and intra-rater correlation coefficients for lateral distal femoral angle (LDFA) and medial proximal tibial angle (MPTA).

Systematic Differences in LDFA Measurements between SLR and LLR

The measured LDFA was assessed for the entire cohort and within each CPAK classification (Table 2). On SLRs, the mean (range) LDFA was 81.4° (73.3-97.9°). On corresponding LLRs, the mean LDFA was 87.7° (78.1-96.3°). The average difference between long-leg and short-leg LDFA measurements was 6.3°(2.9-9.3°).

Table 2: Descriptive Statistics of short-leg and long-leg LDFA measurements by corresponding CPAK classification, and the difference (delta) between them.

Optimal LDFA correction factor for SLRs

To identify the slLDFA correction that maximizes agreement between SL and LL CPAK phenotypes, we plotted JLO agreement, aHKA agreement, and CPAK agreement versus slLDFA corrections from 2° to 10° (Figure 2). The optimal correction threshold fell between 4.8° and 5.1°, yielding ≥66% agreement. Applying the 5° slLDFA correction to the SLRs and then comparing the predicted CPAK classifications against the gold standard LLR classifications (Table 3) yielded a JLO agreement of 237/268 (88%), while aHKA agreement was 201/268 (75%).

Figure 2: Agreement between short-leg and long-leg radiographic classifications across a range of slLDFA correction factors. JLO, aHKA, and CPAK agreements are plotted as functions of slLDFA correction from 2° to 10°. The shaded region denotes the range of maximal composite agreement. Abbreviations: JLO, joint line obliquity; aHKA, arithmetic hip-knee-ankle angle; CPAK, coronal plane alignment of the knee; slLDFA, short-leg lateral distal femoral angle.

Table 3: Agreement between Short-Leg and Long-Leg CPAK classifications after applying a 5° slLDFA correction.