1. Abrar Adil DO – Oklahoma State University Medical Center
2. Mark Calder MD – OU-Tulsa School of Community Medicine, The University of Oklahoma-Tulsa
3. Nicholas Dobrasevic MD – Summit Orthopaedic Group
4. Jake Checketts DO – Oklahoma State University Medical Center
5. Chad Hanson DO – Oklahoma State University Medical Center
6. Byron Detweiler DO – Oklahoma State University Medical Center
7. Arjun Reddy BA – Oklahoma State University Center for Health Sciences
8. Nasir Mushtaq Ph.D – Office of Research Development at the University of Oklahoma Tulsa
9. Brent Norris MD – Oklahoma State University Medical Center

Abstract

Objectives
Fluoroscopic imaging of the distal femur has been nicely described in the literature and is easily reproducible in the operating room. The purpose of this study is to determine if fluoroscopy can be used to accurately assess femoral anteversion in the operating room setting. 

Methods
Ten Symbone models were obtained and osteotomized in the middle 1/3 shaft. CT scans of these specimens were then obtained and read by a MSK radiologist to determine the new version. Six orthopedic surgeons of varying levels of experience were asked to define femoral anteversion of the specimens.  Measurements were taken at 90 and 135 degrees relative to the specimens. 

Results
In the orthogonal position, all investigators’ fluoroscopic measurements were statistically significantly different than the measurements made by CT. In comparison, all investigator’s fluoroscopic measurements in the abducted position did not differ significantly in comparison with the CT measurements. Inter-rater reliability for fluoroscopic measurements by investigators was found to be good for orthogonal fluoroscopic measurements (ICC 0.929, 95%CI: 0.799, 0.979) and for abducted fluoroscopic measurements (ICC 0.928, 95%CI: 0.836, 0.977). Furthermore, in both  the orthogonal and abducted positions the measurements made by senior in comparison to junior investigators were non-significant (p=0.232 and p=0.475). In the abducted position the mean number of measurements correctly made within 10 degrees of the CT measurements were 7.3/11 (66.67%, range 45.45-81.82%). 

Conclusion
The variability of our data indicates the difficulty of obtaining appropriate femoral anteversion using C-arm fluoroscopy. Establishing a more accurate method of obtaining femoral anteversion is needed to provide the most robust surgical outcomes. 

Keywords: Fluroscopy, Femoral Version

Introduction
The anatomy of the proximal femur is quite complex, most likely prompted by the evolution of the hip joint to allow for bipedal gait¹. Various changes have occurred over time including, but not limited to, elongation of the femoral neck (increasing lateralization or offset of the hip) and development of anteversion of the hip to bring the femur more posterior in the coronal plane (from a relatively anterior hip joint)². This adaptation likely evolved to allow for prolonged standing with less muscle fatigue. The complex proximal femur anatomy make the reconstruction of it during hip arthroplasty and repair of fractures in this region much more challenging.

Techniques to assess and calculate intraoperative anteversion of the proximal femur have not been well defined. Only a few studies discuss methodology to assist surgeons in locating landmarks that would allow for accurate reconstruction of the femoral anteversion in the setting of a proximal femoral or shaft fracture^3–5. Furthermore, localizing and standardizing proximal femoral landmarks that can assist in determining true anteversion will likely be compromised by implants placed in the region of the proximal femur to stabilize the fracture. 

For these reasons, we will attempt to better define the proximal femoral anatomy, specifically femoral anteversion, using a synthetic femoral model and a standard intraoperative fluoroscopy machine. By using specific radiographic landmarks, anteversion of the proximal femur can be reliably and consistently determined with standard fluoroscopic imaging techniques. The purpose of this study is to answer whether femoral anteversion can be consistently measured using fluoroscopy in an idealized experimental intraoperative condition as compared to the version assessed by a CT rotational profile. Secondarily, we will compare the accuracy of version assessment at 2 different c-arm positions at the proximal femur one with the beam orthogonal to the long axis of the femur and one with the beam angled approximately 135 degrees to better delineate the femoral neck.  

Materials and Methods
Eleven synthetic femora were obtained from Sawbones (Pacific Research Laboratories, Vashon Island, Washington USA). A single transverse osteotomy was made in the mid-diaphysis of each femur to allow for the introduction of rotation, introducing relative anteversion or retroversion. A marking and cutting guide was used to mark the osteotomy level which constrained each femur distally along the posterior and distal condylar surfaces. The marking guide was used to place longitudinal indexing marks at fixed intervals of 10 degrees at the osteotomy site. The interval length was selected using the arc length formula.  

The transverse osteotomy was then made using a bandsaw. Ten out of eleven femora had osteotomies performed. Out of the ten osteotomized femora, nine had a relative anteversion or retroversion introduced by advancing to an indexed mark placed based on the arc length formula. One femur was re-fixed in its native position and one was left uncut. The osteotomies were then fixed with a Synthes 4.5 mm LCP Plate (Synthes, Inc., West Chester, PA) with 4 bicortical 4.5 mm cortical screws on each side of the fracture placed under compression. 

The ten osteotomized femora with introduced version as well as one un-osteotomized femur were each marked with an arbitrary identifying number and a CT scan was obtained of each femur to establish a rotational profile. A musculoskeletal radiologist blinded to the version set on each femur then performed a rotational CT scan per our institutional protocol. The anteversion calculated by the radiologist remained blinded from the orthopedic department until completion of the fluoroscopic portion of the study. 

Intraoperative assessment of version was performed by obtaining a true lateral of the distal femur where the fluoroscopic beam was aligned orthogonal to the transepicondylar axis and the posterior condylar axis (Figure 1). The proximal femur was imaged with the beam orthogonal to the long axis of the femur (90 degree view). The jig was then rotated to the abducted view (defined as 135 degree view) and another image was obtained (Figure 2). 

In calculating the version, the coarse angular guide markings on the C-arm are used but these markings are imprecise and cumbersome. To improve the accuracy and efficacy of data collection, a digital protractor was affixed to the C-arm tube. The protractor was calibrated to zero degrees at the perfect lateral of the distal femur (Figure 2,3). Data points were obtained from two senior fellowship-trained orthopaedic traumatologists, one junior fellowship-trained orthopaedic traumatologist, an orthopaedic trauma fellow and two senior level orthopaedic residents. A perfect lateral image of each proximal femur was obtained at each of the two aforementioned positions of neutral and relative beam abduction (Figure 3) by each investigator. The relative version was then recorded from the digital protractor for each femur. Each of the observed data points was then recorded by an independent observer and kept blinded to the investigator.

Statistical Analysis
The recorded measurements from the CT data and fluoroscopic data were then entered into a spreadsheet for statistical analysis. For the primary analysis, two measurements on each specimen from one doctor will be averaged to obtain one value. The paired t-test will be used to compare degrees of anteversion between each doctor and CT reading.  In addition, average degrees of anteversion from all three doctors will be further averaged as an overall measurement for each specimen from three doctors. Then, the paired t-test will be used to compare degrees of anteversion between three doctors and the CT reading.  

A total of 10 specimens were used in this study.  This sample provides at least 90% power to detect a difference of 10 degrees of anteversion between each doctor and the CT reading using a two-sided paired t-test.  A standard deviation of 6 degrees of anteversion was used in the sample calculation.  Bonferroni’s correction is used for the multiplicity adjustment and a p-value of less than .0125 is considered statistical significant for the primary analysis.

Interclass correlation (ICC) and corresponding 95% confidence intervals (95%CI) were calculated based on single rating (k = 2  for ICC between fluoroscopic measurement by each investigator and CT readings, and k = 6 for ICC between investigators’ fluoroscopic measurements), absolute-agreement, and two-way random effects model. ICC was calculated using SPSS version 24, whereas all other analyses were performed using SAS version 9.4.

Results
Each specimen had a CT version measurement, an orthogonal lateral version measurement (fluoroscopic) and a 135-degree abducted version measurement (fluoroscopic). Each specimen was evaluated by the aforementioned six investigators. A total of 143 measurements were obtained consisting of; 11 CT version (set as the standard reference version), 66 orthogonal fluoroscopic measurements, and 66 abducted fluoroscopic measurements. 

Initial intraoperative femoral version measurements in the orthogonal lateral (90 degrees) and abducted (135 degrees) positions revealed differing reliability between measurement systems. With orthogonal lateral version, the mean fluoroscopic measurement was 16.556 (SD +/- 27.628), while the mean CT measurement was -4.545 (SD +/- 28.762). Alternatively, with abducted version, the results between measurement systems were more similar. The mean fluoroscopic measurement in the abducted position was 7.768 (SD +/- 31.161), while the mean CT measurement was 6.182 (SD +/- 27.433) (Table 1). 

When comparing the mean difference between orthogonal and abducted fluoroscopic measurements by each investigator and CT measurements there were also profound differences depending on position. In the orthogonal position, all investigator’s fluoroscopic measurements were statistically significantly different than the measurements made by CT (p-values (range) <0.0001-0.0008) (Table 2). In comparison, all investigator’s fluoroscopic measurements in the abducted position did not differ significantly in comparison with the CT measurements (p-values (range) 0.0609-0.7107) (Table 3). 

The results of Interclass correlation when evaluating reliability of fluoroscopic measurement by each investigator and CT measurement, indicate a poor level of reliability for the orthogonal lateral version measurement, however, for abducted version measurement the reliability ranged from moderate to good (Table 4). Inter-rater reliability for fluoroscopic measurements by investigators was found to be good for orthogonal fluoroscopic measurements (ICC 0.929, 95%CI: 0.799, 0.979) and for abducted fluoroscopic measurements (ICC 0.928, 95%CI: 0.836, 0.977). Furthermore, in the orthogonal position the measurements made by senior in comparison to junior investigators was non-significant (p=0.232) (Figure 4). Additionally, measurements in the abducted position made by senior in comparison the junior investigators was also non-significant (p=0.475) (Figure 5). 

When comparing the accuracy of each investigator similar discrepancies regarding orthogonal and abducted measurements were noted. In the orthogonal position, the mean number of measurements correctly made within 10 degrees of the CT measurements were 1/11 (9.09%, range 0.00-36.36%). In the orthogonal position, the mean number of measurements correctly made within 15 degrees of the CT measurements were 2.83/11 (25.76%, range 0.00-63.64%). In contrast, in the abducted position the mean number of measurements correctly made within 10 degrees of the CT measurements were 7.3/11 (66.67%, range 45.45-81.82%). Furthermore, the mean number of measurements correctly made within 15 degrees of the CT measurements were 8.5/11 (77.27%, range 63.64-100.00%) (Table 5) (Figure 6). 

Discussion
This study examines how well surgeons can measure relative version with intraoperative fluoroscopy using a sawbone model. Our study found interobserver reliability to be high even among orthopaedic trauma surgeons of varying levels of experience. However, when excessive anteversion or excessive retroversion was placed in sawbone model (>25 degrees either way) the ability to accurately interpret the anteversion is greatly diminished and the intraobserver reliability decreases as well.   

Intraoperative clinical assessment of version involves internally and externally rotating the hip and comparing with the contralateral limb. This method cannot be applied to the injured limb until the fracture has been stabilized and thus cannot be readily used prior to or during reduction and fixation. Other methods of restoring alignment based on the use of topographical landmarks such as the alignment of the patella and the skin-folds about the thigh are imprecise and do not provide a finite objective end-point⁶.

The cortical step sign as an indicator of rotational malreduction was proposed by Krettek³. This method is based on the variance in cortical thickness about the circumference of the cross section of the femoral shaft. Tornetta et al⁵ attempted to estimate femoral version by using computed tomography (CT) scans of both the affected and unaffected side of each patient. The greater trochanter-head contact method by Kenawey and Krettek⁴ offers the ability to calculate version by measuring the angular variance from the true lateral of the distal femur to the point on the AP projection of the hip where the greater trochanter contacts the lateral margin of the femoral head. 

The methods above are not without limitations. First, the cortical step sign is of limited use in severe comminution and does not allow for a numerical, objective measure of version. Additionally, the cortical step sign has been found to be inaccurate in determining whether version is due to external or internal rotation⁷. Tornetta et al’s method of using CT scans to assess femoral version has proven to be a simple method with high interrater reliability⁸. However, it is impractical to obtain an intraoperative rotational CT scan of both hips during the procedure precluding same surgery intervention if needed. Lastly, the greater trochanter-head contact method has limitations in that the technique is difficult to employ after hardware has been placed around the trochanter and in the setting of trochanteric comminution.

Similar to the greater trochanter-head contact method, we utilize a method which assesses version using the angular variance between two discrete, finite radiographic targets. However, we feel the method investigated in this study is advantageous because it is applicable to multifocal femoral fractures, and can be used in the setting of comminution. It does require obtaining a ‘true lateral’ view of the hip and knee but allows same surgery intervention if necessary.

In this study, we have shown that intraoperative femoral version obtained by comparing proximal and distal femoral anatomy in the true lateral projections can be obtained within 10 degrees of the CT standard version measurement at a high percentage in the abducted (135 degree position) (Table 5). Additionally, the difference between junior and senior attendings was negligible, indicating that the methods described herein are efficient and reproducible among orthopaedic surgeons with a wide range in operative experience level. Thus, because the use of fluoroscopy is readily available, used intraoperatively, and well known to most orthopaedic surgeons, it is an ideal resource to measure femoral version. Additionally, it will save time and resources while preventing the need for excessive postoperative CT scans and subsequent re-operations due to femoral malrotation. 

Our findings suggest that the 135 degree abducted view is the preferred beam orientation as it improves the accuracy of version assessment to within 10-15 degrees of the CT gold standard. This correlates to being within the clinically well-tolerated 15 degrees of malrotation.  As a comparison, the measurements obtained in the 90 degree orthogonal view more frequently varied significantly from the CT standard. We speculate that the increased accuracy is likely related to the fact that the 135 degree view is a better orthogonal view of the femoral neck versus the 90 degree view. Additionally, while the abducted view improves the accuracy of measuring the version, extremes of version markedly limit the accuracy of these measurements in both views. This may represent a clinical pearl whereby difficulty encountered in obtaining these images may be suggestive of significant malrotation. 

Definition of True Lateral of the hip
The complex anatomy of the proximal femur presents a challenge to defining a single standardized true lateral view. We propose that the true lateral is obtained in the abducted view when lines drawn along the anterior and posterior cortices of the femoral neck are parallel (collinear) with lines drawn along the anterior and posterior cortices of the proximal diaphysis (figure 3). In the orthogonal view,  a line starting at the apex of the femoral head should evenly bisect the visualized portions of the femoral neck and proximal shaft (figure 2).

The strengths of the study included the samples being blinded to the observer and the reference rotational data from the CT being blinded until the study was complete. The digital angle finder allowed a more accurate and precise assessment by the physician compared to using the marks on the fluoroscopy machine. The use of a rotating fixture helped to decrease error in measurement by minimizing movement of the c-arm base which is difficult to constrain during iterative imaging between specimens. Additionally, by minimizing manual movement of the c-arm, error introduced by variation in the beam position is minimized and fluoroscopic exposure to the examiners was markedly reduced.

The limitations of the study include the fact that our imaging was performed of synthetic femora in a fixture. Inherently, such a setup differs from the multitude of real-world scenarios in which the femur is imaged. Patient body habitus, various local implants, operating table type, patient positioning, fluoroscope dimensionality and aperture, and many other factors impact the way in which the femur is imaged. We feel our model approximates all of these conditions well by evaluating a true lateral taken at the proximal femur in two positions which represent two extremes of a spectrum of beam angles which may be encountered in clinical practice based on the aforementioned factors. 

Conclusion
This method of intraoperative version assessment offers an accurate, useful tool in the avoidance of clinically significant malrotation in the treatment of diaphyseal femur fractures. Furthemore, this method has a higher interrater reliability among orthopaedic surgeons of varying experience levels. It remains unanswered exactly to what degree malrotation is tolerable and the sequelae of malrotation. With the advent of computer control, navigation, and digital measurements in the operating room, we hope such technologies will be used for improving accuracy in the restoration of the complex femoral geometry.

Figure 1 | Figure 2 | Figure 3 | Figure 4 | Figure 5 | Figure 6

References

1. Kozma EE, Webb NM, Harcourt-Smith WEH, et al. Hip extensor mechanics and the evolution of walking and climbing capabilities in humans, apes, and fossil hominins. Proc Natl Acad Sci U S A. 2018;115(16):4134-4139.
2. Hogervorst T, Vereecke EE. Evolution of the human hip. Part 1: the osseous framework. J Hip Preserv Surg. 2014;1(2):39-45.
3. Krettek C, Miclau T, Grün O, Schandelmaier P, Tscherne H. Intraoperative control of axes, rotation and length in femoral and tibial fractures. Technical note. Injury. 1998;29 Suppl 3:C29-C39.
4. Kenawey M, Krettek C, Ettinger M, Hankemeier S, Breitmeier D, Liodakis E. The greater trochanter-head contact method: a cadaveric study with a new technique for the intraoperative control of rotation of femoral fractures. J Orthop Trauma. 2011;25(9):549-555.
5. Tornetta P 3rd, Ritz G, Kantor A. Femoral torsion after interlocked nailing of unstable femoral fractures. J Trauma. 1995;38(2):213-219.
6. Deshmukh RG, Lou KK, Neo CB, Yew KS, Rozman I, George J. A technique to obtain correct rotational alignment during closed locked intramedullary nailing of the femur. Injury. 1998;29(3):207-210.
7. Langer JS, Gardner MJ, Ricci WM. The cortical step sign as a tool for assessing and correcting rotational deformity in femoral shaft fractures. J Orthop Trauma. 2010;24(2):82-88.
8. Georgiadis AG, Siegal DS, Scher CE, Zaltz I. Can femoral rotation be localized and quantified using standard CT measures? Clin Orthop Relat Res. 2015;473(4):1309-1314.

Required Disclosures and Declaration

• Copyright Information: No Copyright Information Added
• IRB Approval Information: Not applicable
• Disclosure Information: BN has the following disclosures: Consulting: J and J, Acumed, Wishbone Grant Support: AONA COTA Committee Member: OTA, AAOS Section Editor: Injury Journal Principle: Boys of Summer Enter. Norris Surgical OR Ingenuity