Variation of Femoral Component Size by Reference Points and Target Alignments in Total Knee Arthroplasty


 Selecting appropriately sized components is important in total knee arthroplasty because they can affect postoperative knee function and pain. This study investigated size differences of 19 different femoral component placements from the standard position by three-dimensional virtual surgery using three-dimensional bone models of 101 varus osteoarthritic knees. Distal femoral bone was cut perpendicular to the femoral mechanical axis (MA) in the coronal plane. Twenty different component placements consisting of five cutting directions (perpendicular to MA, 3° and 5° extension/flexion relative to MA in the sagittal plane), two rotational alignments (clinical and surgical epicondylar axes), and two rotational types of anterior reference guide (central and medial) were simulated. The mean anteroposterior dimension of the standard position was 55.5 mm which means that the difference compared to 19 different methods ranged from -1.2 ± 0.2 mm to 7.1 ± 1.3 mm. Multiple regression analysis revealed that flexion cutting direction, surgical epicondylar axis, and central were associated with smaller component size. In conclusion, the femoral component size can be affected easily by not only cutting direction but also the reference guide type and the target alignment. Our findings could provide surgeons with clinically useful information to fine-tune for unintended loose or tight joint gaps by adjusting the component size.


Introduction
Selecting appropriately sized components is important in total knee arthroplasty (TKA) because they can affect postoperative knee function and pain 1 . In the measured resection technique, optimal alignment can be achieved using a detailed preoperative planning system and several jigs. However, it is di cult to accurately determine proper component sizes because cutting surface conditions vary among intraoperative techniques despite accurate preoperative planning 2 . Larger components result in tighter exion gaps, which can decrease range of motion, increase wear of inserts, and elevate patellofemoral joint pressure [3][4][5] . Smaller component sizes cause looser exion gaps, which can cause knee instability and implant loosening 1,6,7 . The nal decisions regarding component size were obtained by measuring the joint gap via a tensioning device using the gap balance technique, whereas the measured resection technique was obtained from a joint gap that consisted of the results of all intraoperative procedures. Several studies showed that femoral component size can be changed by preoperative planning 9,10 and intraoperative alignment 11,12 . However, most used a two-dimensional (2D) or three-dimensional (3D) templating system that evaluates component size on an inappropriately derived plane that is affected by limb position and selection of reference points 13 . Virtual surgery via computer simulation is useful for precisely evaluating the effects of different component placements during TKA, because it should reduce the effects of inaccurate alignment or inadequate observation by 2D or 3D templating system 14 . The purpose of this study was to evaluate the sizes of femoral components with 20 different component placements and investigate the size differences from the standard position in TKA using 3D virtual surgery with computer simulation. In addition, we sought to compare the size difference among cutting directions, rotational alignments, and rotational types of reference guide. The hypothesis is as follows.
(1) Size can easily change from the standard position based on cutting directions, rotational alignments, and rotational types of the anterior reference guide. (2) Size signi cantly differs among cutting directions, rotational alignments, and reference guide rotation types.

Results
The ICC (1,1) and ICC (2,1) of this coordinate system were 0.97 and 0.96, respectively, suggesting excellent agreement for both.  Table 3. The mean component sizes were not smaller in any knees using CEA than in knees using SEA. The AP dimension and mean component size with CEA as the target rotational alignment had a signi cantly larger size than that with SEA when a MR reference guide was used (p < 0.0001; Table 3). By contrast, about half of all knees (49.5 -60.4 %) had a same component size with CEA compared with SEA when a CR reference guide was used ( Table 3).
The mean femoral AP dimension and component size with two different reference guides are shown in Table 4. The mean AP dimension and component size with MR yielded a signi cantly larger size than with CR (p < 0.0001), and mean component sizes were not smaller in any knees using MR than in knees using CR when CEA was the target rotational alignment (Table 4). By contrast, fewer than half of knees (38.6 -47.5 %) had a larger component size with MR than with CR, when SEA was the target rotational alignment ( Table 4).

Discussion
The most important nding of this study was that intraoperative surgical techniques, including surgeons' selection of the reference guide type and the target alignment, readily affected the femoral component size. This suggests that it is di cult to achieve an accurate component size despite establishing a presumably accurate preoperative plan. The mean difference of AP dimension showed 8.2-mm ranged from 54.3 mm to 62.5 mm, which corresponded to 3.9-size difference in terms of the femoral component size (4.9 to 8.8). The largest difference from the standard position was 7.1-mm in the femoral AP dimension, equivalent to 3.4 in the component size. Surgeons should be aware that inaccurate surgical techniques using conventional alignment guides and cutting blocks can not only cause di culty in acquiring optimal alignment, but also readily change femoral component size. Few studies have examined the detailed effects of surgical techniques on component size 9,13 . Therefore, we accurately compared size differences resulting from 20 different component placements performed using virtual surgery and computer-assisted design software.
In all knees, cutting in extension resulted in a greater femoral AP dimension than cutting in exion. Our accurate measurements showed a similar trend with a previous study using 3D templating software 12 . Furthermore, multiple regression analysis showed that cutting in exion affected the femoral component size most negatively. The size changed signi cantly when the difference in cutting direction was more than 3°. In previous studies using conventional methods, only 45-65% of cases exhibited values within 3° of the target angle in the sagittal plane 20,21 . To prevent unexpected component size changes, surgeons should be aware of several pitfalls related to cutting direction (for instance, inappropriate spacing between the distal femoral cutting guide and bone saw, and improper exure of the bone saw edge).
Regardless of rotational alignment, no knees had a smaller component size when CEA was used as opposed to SEA. Both CEA and SEA were used for the target rotational alignment in this study because posterior condylar axis may not be suitable due to variety of bone deformities and cartilage thickness. Based on intraoperative measurements, Koninckx et al. reported that the AP dimension of the distal femur increased by 2.3 mm and 3.8 mm with 3° and 5° of external rotational alignment, respectively, relative to the posterior condylar axis 11 . Our study obtained similar results using the CEA and SEA as major target rotational alignments, because the CEA is usually externally aligned relative to the SEA. One reason for the larger component size when using the CEA can be explained based on the measurements shown in Fig. 3. The AP dimension was calculated using the following equations: AP dimension (CEA) = r1×sinα + r2×sinβ, and AP dimension (SEA) = r1×sin(α−θ) + r2×sin(β−θ). The distance between the knee center and the attachment of the anterior boom (r1), and between the knee center and the rotation center of the reference guide (r2) on the XY-plane, are the same when using the CEA and SEA. AP dimension (SEA) is smaller than AP dimension (CEA) due to the low value of the sine of θ (the angle between the CEA and SEA).
Regarding the effects of reference guide rotation types, almost no knees had a smaller component size when using MR rather than CR due to the longer AP dimension with MR. No studies have investigated how the rotation type of the anterior reference guide affects femoral component size. The reasons why the AP dimension is longer with MR than CR is simply because the rotation center is located more posteriorly with MR (Figs. 2A and 2B). When using the CEA, the component size is larger with MR than CR in most knees, but this is the case in only about half of all knees when using the SEA; this can be explained by calculations based on the measurements in Fig. 4. The difference in AP dimension when using MR versus CR was calculated using the following equations: difference in AP dimension when using MR versus CR (CEA) = r3×sinδ, and difference in AP dimension when using MR versus CR (SEA) = r3×sin(δ−θ) (δ: angle between the CEA and the posterior condylar axis). The distance between the rotation center on the XY-plane when using MR versus CR (r3) is the same for both the CEA and SEA. The difference in AP dimension when using MR versus CR (SEA) is smaller than the difference in AP dimension when using MR versus CR (CEA) due to the low value of the sine of θ. with rheumatoid arthritis, a history of knee injuries or infections, or severe bone defects in the distal femur were excluded. The preoperative alignments and progression of osteoarthritic knees (determined using the Kellgren-Lawrence (K-L) osteoarthritis knee scale) 15 were measured on full-length, weight-bearing anteroposterior (AP) radiographs using a digital measurement software 2D template (Japan Medical Materials Corp., Osaka, Japan). The mean preoperative femorotibial angle was 185.4 ± 5.5° and hip-kneeankle angle was 193.2 ± 8.0°. Ninety-one knees were classi ed as grade 4 on the K-L scale, and 10 knees as grade 3.

Three-dimensional Bone Model and the Coordinate System
A computed tomography (CT) scan of the lower extremity that was scheduled to undergo TKA was obtained from each patient within 3 months preoperatively (Aquilion 64-slice CT Scanner; Toshiba, Tochigi, Japan). CT slices were 2 mm thick. A 3D femoral bone model was reconstructed from preoperative CT data using MIMICS (Materialise, Leuven, Belgium). The bony geometry was imported into a computer-assisted design software program (Rhinoceros; Robert McNeel and Associates, Seattle, WA, USA) in stereolithography format. The coordinate system consisted of the femoral mechanical axis (MA) and the functional transepicondylar axis (TEA), which was projected onto the plane perpendicular to femoral MA (Fig. 1A) 16 . The center of the hip was determined by tting a sphere to the femoral head. The center of the knee joint was identi ed as the midpoint of the TEA, which consisted of the surgical epicondylar axis (SEA) and clinical epicondylar axis (CEA). The femoral MA was de ned as the line connecting the center of the knee and the center of the hip. The SEA was de ned as the line connecting the most prominent point of the lateral epicondyle with the deepest point of the sulcus on the medial epicondyle. The CEA was de ned as the line connecting the most prominent point of the lateral epicondyle with the most prominent point anterior to the medial sulcus of the medial epicondyle. The Zaxis of the knee (proximal-distal) was de ned as the extension of the femoral MA. The plane normal to the Z-axis at the center of the knee was de ned as the XY-plane. The Y-axis (medial-lateral) was de ned as the extension of the functional TEA. The X-axis (anterior-posterior) was de ned as the line normal to the coronal plane (YZ-plane) at the center of the knee (Fig. 1A).

Virtual Surgery
Bone cutting and implantation were performed on the femoral bone model using the established 3D coordinates. Cutting of the distal femur was performed to be perpendicular to the femoral MA (Z-axis) at the level of intercondylar notch in the coronal plane (Fig. 1B)

Statistical analysis
Data are presented as means and standard deviations. Data analysis was performed using JMP Pro software version 12.0 (SAS Institute, Cary, NC, USA). To investigate the reliability and reproducibility of this coordinate system for measuring AP dimension, intraobserver and interobserver reliabilities were assessed using intra-and interclass correlation coe cients, i.e., ICC (1,1) and ICC (2,1), respectively 19 . All measurements were obtained by two orthopedic surgeons (SI, HM) at an interval of more than 1 week. Data were blinded and included no patient information. Student's t-test was used to compare differences between the CEA and SEA and between MR and CR. Different cutting directions in sagittal alignment and deviations from the standard position were compared between groups using ANOVA. Post hoc analysis was conducted using the Steel-Dwass test. Multiple regression analysis was performed to determine which factors had the greatest effect on component size, using the following factors: ve cutting directions, two rotational alignments, and two rotational types of the reference guide (effect size: β coe cient). Statistical signi cance was set at a p value < 0.05. [1] The conception and design, [2] Acquisition or analysis of data, [3] Interpretation of data, [4] Drafting of the article, [5] Critical revision of the article for important intellectual content, [6] Final approval of the article Compliance with ethical standards

Competing interests
The authors declare no competing interests.

Ethical approval
The study was approved by the institutional review board.

Informed consent
Informed consent was obtained from all individual participants included in this study.  The values are given as the mean with standard deviation.   The values are given as the mean with standard deviation.