Introduction

Kinematically aligned total knee arthroplasty (KA TKA) is an innovative surgical technique with high 10-year implant survival and patient function that is an alternative to conventional mechanically aligned total knee arthroplasty (MA TKA) [13]. The ‘calipered’ kinematic alignment surgical technique uses serial verification checks incorporating ten caliper measurements of bone resections and positions [25, 32] (Fig. 1). The assumption is that these verification checks insure that the surgical alignment goal of restoring joint lines to native is satisfied.

Fig. 1
figure 1

Worksheet for intraoperatively recording serial verification checks based on caliper measurements of bone resections and positions for a femoral component with 9 mm thick distal femoral condyles and 8 mm thick posterior femoral condyles. The order of the bone cuts progresses from distal femoral, posterior femoral, anterior femoral, chamfer femoral, and tibial resection. The thicknesses of the distal and posterior femoral resections are adjusted so that they equal the thickness of the component within 0 ± 0.5 mm after compensating for 2 mm of cartilage wear when present and a ~ 1 mm kerf from the saw cut

Full size image

Focusing on femoral joint lines, the surgical alignment goal in calipered KA TKA should be closely satisfied in theory. The thickness of each femoral resection should equal that of the corresponding region of the femoral component after correcting for sawblade kerf and any wear on the articular surfaces (Fig. 1). By measuring the resection thickness with calipers, the thickness can be controlled within ± 0.5 mm [12]. Holding this tolerance could translate into alignment deviations as small as 0.5 mm and approximately 1° for the proximal–distal (P–D) and anterior–posterior (A–P) positions and the varus–valgus (V–V) and internal–external (I–E) axial orientations, respectively. These positions and orientations are termed alignment variables. The P–D position and V–V orientation determine the alignment of the distal femoral joint line and the A–P position and I–E orientation determine the alignment of the posterior femoral joint line.

However, sources of alignment deviation other than the resection thickness affect the final position and orientation of the femoral joint lines in practice. Sources include cementation and compaction of the femoral component [4] and assuming a worn articular cartilage thickness of 2 mm for all patients whereas the actual thickness can be as great as 4 mm [17, 19]. Because these alignment deviations can stack up leading to deviations as large as 4° in V–V orientation, it is of interest to determine alignment deviations from the surgical alignment goal.

A method for determining these deviations is to shape match the articular surfaces of a mirrored 3D model of the femoral component in the KA TKA knee to the articular cartilage surfaces of a 3D model of the native femur in the contralateral knee. Because the shape-matching process optimizes the alignment of a femoral component for KA TKA by closely matching the distal and posterior articular cartilage surfaces [10, 11], the joint lines of the shape-matched femoral component model can serve as surrogates for those of the native femur and thus provide a viable reference from which deviations in KA TKA femoral joint lines can be determined.

Also it is of interest to determine whether the alignment deviations affect clinical outcomes. Alignment deviations less than 2 mm in position and 2° in orientation from the native femoral joint lines reduce the risks of stiffness, loss of extension, loss of flexion, and higher tibial compartment forces than those of the native knee [27, 28, 31]. Thus, the proportion of patients within these limits is a meaningful metric. In addition, patient-reported outcome scores should be documented and assessed for any significant association between the alignment deviations in each alignment variable.

A prior study determined alignment deviations between the planned alignment of the femoral component and the achieved alignment [29]. However, the results do not translate into alignment deviations in the four alignment variables mentioned above and the effect of alignment deviations on clinical outcomes was not studied.

Accordingly, the purposes of the work described by this paper were twofold. One was to determine alignment deviations in the two femoral component positions and two orientations from those of a femoral component virtually planned to the contralateral native knee in 36 patients with a primary calipered KA TKA. A second was to determine whether the resulting alignment deviations affect clinical outcome scores. Our hypotheses were that the alignment deviations for most knees would be less than 2 mm and/or 2° and that larger alignment deviations would not be associated with lower clinical outcome scores. If these hypotheses were supported, then this result would validate calipered KA TKA as a surgical technique which largely achieves the surgical alignment goal.

Methods and materials

Patients

An institutional review board approved this retrospective study (IRB 1362165-1) and the study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. From January 2017 to July 2017, 251 primary KA TKAs were performed using ten caliper measurements and serial verification checks (Fig. 1) [25, 32]. All patients fulfilled the Centers for Medicare & Medicaid Services guidelines for medical necessity for TKA treatment and pre-operatively completed the Oxford Knee Score. The indications for TKA included disabling symptoms not resolved after conservative knee treatment, radiographic evidence of Kellgren–Lawrence Grade 2–4 arthritic changes or osteonecrosis, and any severity of flexion, varus, and valgus deformity.

Inclusion criteria were patients having KA TKA performed with asymmetric, fixed bearing, posterior cruciate-retaining (PCR) components (Persona CR, Zimmer-Biomet, Warsaw, IN), a native contralateral limb with no evidence of degenerative joint disease, no skeletal abnormalities or prior surgery in either limb except for the KA TKA, no history of rheumatic or traumatic arthritis, age between 40 and 85 years, a body mass index less than or equal to 40. Note that patients were selected with no restriction on preoperative varus–valgus or flexion-contracture deformity.

Patients considered for inclusion were those operated on between January 2017 and July 2017 by one surgeon who performed primary calipered KA TKA on 251 consecutive patients. Patients were winnowed down to those meeting the inclusion criteria in a three-step process. The first step entailed a review of medical records to identify those patients within the age range and BMI range and those with acceptable medical history. The second and third steps involved a review of CT images. Postoperative AP and lateral CT scanograms of both limbs and CT axial images of both knees were obtained with the Perth protocol. The second step involved a review of the scanograms to identify those patients with a unilateral calipered KA TKA without a skeletal abnormality in either limb. The third step involved a review of multiplane reconstructions of the axial images to identify those patients without subchondral sclerosis, joint space narrowing, marginal osteophytes, and subchondral cysts of the tibiofemoral and patellofemoral joints in the unoperated knee. From these three steps, 36 patients were identified who met the inclusion criteria (Table 1). Patients completed questionnaires to determine the Oxford Knee and Forgotten Joint scores at an average follow up of 20 months (range 18–23 months).

Table 1 Preoperative patient demographics, clinical characteristics, pre- and postoperative Oxford Knee scores, and Forgotten Joint score
Full size table

The average age of the 36 patients was 69 ± 8 years, 22 were women, and the body mass index averaged 29 ± 4 kg/m2 (Table 1). The preoperative Kellgren–Lawrence classification of osteoarthritis was II in 3%, III in 55%, and IV in 42% as determined by standing full extension and 45° flexion knee radiographs. The mean postoperative OKS was not different between patients with a Kellgren–Lawrence classification of II (40 ± 12.4 points), III (40 ± 10.9 points), and IV (41 ± 9.8 points) (p = 0.9107). The clinical varus or valgus deformity as measured non-weight bearing ranged from 25° valgus to 15° varus.

Surgical technique

Using ten sequential caliper measurements and a series of verification checks with manual instruments, KA TKA was performed by a single surgeon using a mid-vastus approach following a previously described technique [25]. Asymmetric, fixed bearing, PCR retaining components and a patella button were implanted with cement (Persona CR, Zimmer-Biomet, Warsaw, IN). For the femoral component, the varus–valgus orientation and proximal–distal location were set to restore the native distal femoral joint line by adjusting the thickness of the distal femoral resections as measured with a caliper to within 0 ± 0.5 mm of the thickness of the femoral component condyles after compensating for cartilage wear and saw blade kerf. The internal–external orientation and anterior–posterior position were set to restore the native posterior joint line by adjusting the thickness of the posterior femoral resections as for the distal femoral joint line. These steps set the femoral component with a bias of 0.3° and precision of ± 1.1° with respect to the flexion–extension plane of the knee [24].

For the tibial component, the varus–valgus orientation was set to restore the native joint line by ensuring that the thicknesses measured with a caliper at the base of the tibial spines medially and laterally were within 0 ± 0.5 mm. With the knee in full extension, the varus–valgus angle of the tibial resection was fine-tuned working in 1° to 2° increments until the varus–valgus laxity was negligible as in the native knee [30]. The internal–external rotation of the tibial component was set using a kinematic tibial template with a negligible bias of 0.1° external and a precision of ± 3.9° [26]. With the knee in 90° of flexion, the slope was set to restore the native joint line in the medial compartment by working in 1° to 2° increments until the offset of the anterior tibia from the distal medial femoral condyle with trial components matched that of the knee at exposure after adjusting for cartilage wear on the femur and ensuring that the internal–external laxity approximated 14° as in the native knee [30]. Ligament releases were not performed. This surgical technique restores the hip–knee–ankle angle, distal lateral femoral angle, and proximal medial tibial angle to native within ± 3° with frequencies of 95%, 97%, and 97%, respectively [25].

Method for determining deviations in alignment variables

CT images of the operated femur and contralateral native femur were segmented using the automatic tools and refined manually using a combination of thresholding and manual segmentation (Mimics® v20.0, Materialise, Belgium). The classic ‘marching cubes algorithm’ was used to reconstruct 3D models of the operated femur with the implanted femoral component and the contralateral native femur (Fig. 2). Using the iterative closest point (ICP) algorithm, a CAD model of the femoral component was best-fit to the 3D model of the implanted femoral component (Geomagic®, 3D Systems, Cary, NC). A 2 mm thick surface was applied to the 3D model of the contralateral native distal femur to create virtual cartilage surfaces comparable to the average thickness of the distal femoral cartilage (Fig. 2) [23]. Twenty-five cm in length mid-sections of the femoral shafts of the operated femur and the contralateral native femur were formed by cutting the femur 12 cm from the most proximal point of the femoral head and 25 cm distal to the first cut (Fig. 2). A mirror of the mid-section of the operated femur was best-fit to the mid-section of the contralateral native femur to yield a transformation matrix. The root mean squared deviation (RMSD) of the best-fit of the midsections was 0.7 mm (interquartile range 0.6 to 0.8 mm). The transformation matrix was applied to a mirror of the CAD (mCAD) model of the femoral component best-fit to the 3D model of the implanted femoral component thus superimposing this mCAD model on the native distal femur model. The articular surfaces of the mCAD model of the femoral component also was best-fit to the virtual cartilage surfaces of the contralateral native distal femur model (Fig. 2). Differences in positions and orientations of the superimposed mCAD model of the femoral component from the mCAD model of the femoral component best fit to the virtual cartilage surfaces (Fig. 3) provided values for computing deviations in alignment variables of the distal (P-D and V-V) and posterior (A-P and I-E) femoral joint lines. 

Fig. 2
figure 2

Diagram illustrating the process for determining deviations in alignment variables of the distal and posterior femoral joint lines following kinematically aligned TKA from virtually planned joint lines on the contralateral native knee

Full size image
Fig. 3
figure 3

Diagram illustrating coordinate system in a left knee for determining differences in position and orientation of the mirrored CAD (mCAD) model of the femoral component superimposed on the native distal femur model from the mCAD model best fit to the virtual cartilage surfaces of the native distal femur model. The origin was the midpoint of the line connecting the base of the lugs of the mCAD model of the femoral component. The X-axis was parallel to the posterior condyles and pointed in the medial–lateral direction with medial being positive for a left knee. The Y-axis was mutually perpendicular to the X-axis and the lugs and pointed in the anterior–posterior direction with anterior being positive. The Z-axis was mutually perpendicular to the X- and Y-axes and pointed in the proximal–distal direction with proximal being positive. Differences in position were expressed in this coordinate system whereas differences in orientation were given in a 3-2-1 Euler angle sequence. Note that for a right knee, the lateral direction was positive which required that negative values in this direction be taken and used in further analyses

Full size image

Statistical analysis

To quantify repeatability and reproducibility, three observers independently processed the 3-D bone and femoral component models three times with at least 24 h between trials from the same seven randomly selected patients and determined the differences in position and orientation for each femoral joint line  for each trial and each observer. A two-factor mixed-model analysis of variance (ANOVA) was performed for each of the four alignment variables where the two factors were observer at 3 levels and patient at 7 levels. Observer and patient were modelled as random effects. The resulting variance components for observer, patient, and error were used to compute the intraobserver and interobserver ICCs [3]. An ICC value of > 0.9 indicated excellent agreement, 0.75–0.90 indicated good agreement, and 0.5–0.75 indicated moderate agreement [14]. The ICC values for repeatability (i.e. intraobserver) and reproducibility (i.e. interobserver) for the A–P position were 0.98 and 0.98, 0.80 and 0.81 for the P–D position, 0.96 and 0.96 for the M–L position, 0.98 and 0.97 for the V–V orientation, and 0.98 and 0.98 for I–E orientation, respectively, confirming previously established good to excellent repeatability and reproducibility [33].

The systematic deviation (mean difference), random deviation (± 1 standard deviation of difference), and root mean squared deviation (RMSD which combines the systematic and random deviations) of the mCAD model superimposed on the native distal femur model from the best fit of the mCAD model to the cartilage surfaces of the native distal femur model were computed for alignment variables of the distal (P-D and V-V) and posterior (A-P and I-E) femoral joint lines.  To determine whether deviations affected clinical outcomes, the proportion of femoral components within 1, 2, and 3 mm or degrees was determined for each alignment variable. Also a simple linear regression was performed with the absolute deviation in each alignment variable for each patient as the independent variable and the corresponding outcome score for that patient for both the FJS and OKS (4 alignment variables × 2 outcome scores = 8 total simple linear regressions). The latter analysis was to determine whether larger deviations correlate with lower FJS and OKS. Significance was set at p < 0.05.

Results

The RMSDs were 1.4 mm for the P–D and A–P positions and were bounded by 1.5° for the V–V and I–E orientations (Table 2). The absolute systematic alignment deviations (i.e. mean differences) were bounded by 0.2 mm and 0.2° whereas the random alignment deviations (i.e. standard deviations of the differences) were much greater at 1.4 mm and 1.4°; thus the random alignment deviations dominated the RMSDs for all alignment variables.

Table 2 Descriptive statistics of deviations in femoral joint line alignment variables and percentages of knees with deviations within specified limits for kinematically aligned implanted femoral components from virtually planned femoral components in the native contralateral knee
Full size table

For the P–D and A–P positions, the proportion of implanted femoral components within 2 mm of the contralateral distal femur was at least 91%. For the V–V and I–E orientations, the proportion of femoral components within 2° from the contralateral distal femur was at least 83% (Table 2).

For the patient-reported outcome scores, the FJS median (interquartile range) was 79 (48–95) and the OKS median (interquartile range) was 45 (39–47) (Table 1). There was no significant correlation between the deviations in any of the four alignment variables and either the FJS or the OKS (Table 3).

Table 3 Summary of results from simple linear regressions
Full size table

Discussion

The most important findings of the present study were that calipered KA TKA with use of serial verification checks (1) restored the femoral component positions to within 2 mm and orientations to within 2° of the planned femoral joint lines in the contralateral native knee with a frequency of at least 91% and 83%, respectively, and (2) achieved relatively high median FJS and OKS and without larger alignment deviations correlating with lower scores.

For calipered KA TKA, the RMSDs of the four alignment variables reported as a worst case and the percentage of patients within 2 mm or 2o of planned femoral joint lines were comparable to, if not better than, those reported for mechanical alignment with the use of robotic-arm, navigation, patient-specific, and manual instrumentation. The RMSDs for the femoral component positions and orientations ranged from 1.4 to 1.5 (mm or degrees) and were comparable to or lower than those reported for robotic-arm mechanical alignment TKA (2.0° for V–V using optical motion capture) [16]. For the V–V orientation of the femoral component, calipered KA TKA oriented 88% of patients within 2° of the contralateral planned distal femoral joint line. This result was comparable to the navigation alignment of 92% of patients and greater than manual instrumentation alignment of 67% of patients within 2° of the mechanical axis of the femur as measured directly on CT images [21]. For the I–E orientation, calipered KA TKA oriented 83% of patients within 2° of the planned posterior femoral joint line. This result was comparable to the navigation alignment of 91% of patients and greater than manual instrumentation alignment of 68% of patients within 2° of the transepicondylar axis as measured directly from CT images [21]. Calipered kinematic alignment of the femoral component within 2° of the planned femoral joint lines of 88% and 83% of patients in the V–V and I–E orientations, respectively, was comparable to or higher than patient-specific alignment of 79% and 68% of patients in these orientations within 2° of the pre-operative plan based on measurements using a navigation system [20]. Calipered KA TKA achieved this consistency with use of two simple, inexpensive sequential intraoperative verification checks (Fig. 1). The two checks were the caliper measurement and adjustments of the distal femoral resections and posterior femoral resections until the thickness of each resection matched the thickness of the corresponding condyle of the femoral component within 0 ± 0.5 mm after compensating for cartilage wear and kerf of the saw cut.

Considering the expense involved in investigating and treating the painful and poorly functioning TKA [15], a surgical technique that is associated with high patient satisfaction and improvement in function is important from both an individual patient and a global health system perspective. The latest and most compelling support for use of kinematic or an ‘individualized’ alignment philosophy in place of mechanical alignment is based on the new systematic classification of the phenotypes of the native limb and knee joint line [9]. 3D-reconstructed CT images confirmed the great variability of the coronal alignment of the lower limb and joint lines in both non-osteoarthritic [22] and osteoarthritic knees [8]. The currently used classification system (neutral, varus, valgus) oversimplifies the coronal alignment and should be replaced by the use of femoral and tibial phenotypes. The detailed phenotype assessment of a patient’s individual anatomy justifies the individualized approach to TKA of restoring the native joint lines and limb alignment, which is the goal of KA TKA.

The patient-reported postoperative FJS and OKS indicated that calipered KA TKA restored relatively high patient satisfaction and function. To appreciate that these scores are relatively high, a comparison to those reported for mechanically aligned TKA is useful. Since the FJS has not been reported as commonly as the OKS, this comparison will be restricted to postoperative OKS from three previous studies. One study with a 1-year follow-up on 578 patients reported a mean OKS of 35.0 [5], a second study with a 1-year follow-up on 602 patients reported a mean OKS of 34.3 [34], and a third study with a 2-year follow-up on 222 patients reported a mean OKS of 35.9 [2]. In comparison, the mean OKS in the present study was 41 or at least 14% higher.

The calipered kinematic alignment technique resulted in a high proportion of patients with deviations less than 2 mm in position and 2° in orientation from the virtually planned femoral joint lines. As demonstrated by several biomechanical studies, meeting these alignment deviation limits is needed to reduce the risks of stiffness, loss of extension, loss of flexion, and tibial compartment forces higher than those of the native knee [27, 28, 31]. The fact that these alignment deviation limits were largely met may explain in part the relatively high FJS and OKS achieved with calipered KA TKA.

Finally larger alignment deviations did not correlate with lower patient-reported outcome scores. Given that (1) the FJS and OKS scores were relatively high; (2) alignment deviations were less than 2 mm and 2° for most patients; (3) larger alignment deviations did not correlate with lower FJS or OKS, and (4) the alignment deviation analysis was a worst case as explained above, it can be reasonably concluded that the alignment deviations in the femoral joint lines resulting from calipered KA TKA do not adversely affect clinical outcomes.

Five limitations should be discussed. First, the study was performed on a subset of 36 of 251 (14%) patients as skeletal deformities or arthroplasties prevented the use of the 3-D computational processes in the others. However, the reported deviations should be generalizable as calipered kinematic alignment of the femoral component is not affected by skeletal deformities or arthroplasties outside the treated knee. Second, the actual alignment deviations are likely less than those reported because using the operated femur as the reference was not possible and using the contralateral native femur and virtual cartilage thickness of 2 mm might have affected the deviations. Side-to-side differences (i.e. asymmetry) in femur morphology exist [6, 7] which could contribute to alignment deviations. Likewise, the use of a 2 mm thick virtual cartilage surface could contribute to alignment deviations because the cartilage thickness varies 1.4 to 4.3 mm between native knees [18]. Third, the difference in flexion-extension (F–E) of the mCAD model of the femoral component from the mCAD model best-fit to the virtual cartilage surfaces of the native distal femur model was not measured because the virtual cartilage surface is cylindrical and does not control random F–E of the best-fit mCAD model. Uncertainty in F–E rotation would not affect our results markedly because the cylindrical nature of the articular surfaces insured that the distal and posterior femoral surfaces were invariant regardless of small variations in F–E rotation. Fourth, although the sample size was not sufficiently large to confidently conclude that the slopes in the regression analyses were not significantly different from zero, the issue is moot because even the highest R-squared value (Table 3) accounted for only 8% of the variability thus rendering none of relationships clinically meaningful [1].

Finally, clinical outcome scores might be affected not only by deviations in the alignment of the femoral component but also by deviations in the alignment of the tibial component so that analyzing the effect of deviations in the alignment of the femoral component on clinical outcomes may seem confounded. However, three of the six alignment variables of the tibial component (I–E orientation, A–P position, and M–L position) are independent of the alignment of the femoral component so that any alignment deviations would randomly affect clinical outcomes. Since the remaining two alignment variables (V–V and F–E orientations), which are not adjusted by the thickness of the tibial insert, are both coupled to alignment of the femoral component, any systematic effect on patient-reported outcomes scores will be due primarily to deviations in alignment variables of the femoral component.

Conclusion

In summary, calipered kinematic alignment of the femoral component with serial verification checks restored the distal and posterior femoral joint lines with sufficiently small deviations from those planned on the native femur of the contralateral knee that clinical outcomes were unaffected. This was accomplished without the use and expense of preoperative radiographs or high-cost imaging modalities such as MRI and CT. Thus, surgeons can use the calipered KA TKA technique with confidence that the surgical alignment goal will be satisfied with sufficient accuracy that high patient-reported outcomes are achieved.