Medial Femoral Trochlea Osteochondral Graft: A Quantitative Anatomic Comparison to the Proximal Pole of the Scaphoid

Abstract

Background  Treatment of scaphoid proximal pole (SPP) nonunion with a vascularized osteochondral graft from the medial femoral trochlea (MFT) has been described, with positive outcomes thus far. However, our understanding of the congruency between the articular surfaces of these structures is incomplete.

Objective  Our purpose was to evaluate the congruency of the MFT and SPP using a quantitative anatomical approach.

Methods  The distal femur and ipsilateral scaphoid were dissected from 12 cadavers and scanned with computerized tomography. Three-dimensional models were created and articular surfaces were digitally “dissected.” The radius of curvature (RoC) of the radioulnar (RU) and proximodistal (PD) axes of the SPP and MFT, respectively, as well as the orthogonal axes (SPP, anteroposterior [AP]; MFT, mediolateral [ML]) were calculated. The RoC values were compared using the Wilcoxon signed-rank test.

Results  The RoC values for the SPP and MFT were not significantly different in the RU–PD plane ( p  = 0.064). However, RoC values for the SPP and MFT were significantly different in the AP-ML plane ( p  = 0.001).

Conclusions  For most individuals, the RU curvature of the SPP was similar to the PD curvature of the MFT. For nearly all individuals, the AP curvature of the SPP and the ML curvature of the MFT shared less congruence.

Clinical Relevance  Articular surface congruity may not be a critical factor associated with improvements in wrist function following this procedure.

The scaphoid is the most frequently injured bone of the carpus, representing 80 to 90% of all carpal fractures. ¹ Due to its predominately retrograde intraosseous blood supply, fractures of the scaphoid are predisposed to poor healing, especially those involving the proximal pole. Nonunion, malunion, and osteonecrosis of the scaphoid proximal pole (SPP) are well-recognized complications following acute fracture. ² Left untreated, scaphoid pathology can progress to carpal collapse, with altered biomechanics that lead to significant pan-carpal degenerative changes. ³

Management of refractory SPP nonunions often requires the use of an autologous bone graft. In the presence of osteonecrosis, vascularized bone grafting (VBG) has demonstrated significantly higher rates of union compared with non-VBG. ⁴ Common sources for VBG include pedicles from the pronator quadratus muscle, ⁵ ulnar artery, ⁶ and distal radius, ⁷ as well as the first to second intercompartmental supraretinacular artery pedicle. ⁸ An alternative technique utilizes the descending genicular artery (DGA) about the knee to create a free vascularized osteochondral flap from the medial femoral trochlea (MFT). This method of VBG allows the use of a volar approach for humpback deformity correction, unlike pedicled grafts from the distal radius. ⁹ It also permits greater manipulation of the graft without devascularization, which is particularly advantageous when treating small proximal pole nonunion fragments. ¹⁰ Evidence thus far suggests union rates near 100% and minimal donor site morbidity. ^{11 12 13 14}

In their anatomic study, Hugon et al demonstrated that the MFT was a suitable osteochondral graft source for SPP nonunion. ¹⁵ Among several suggested advantages of this technique, the authors reported that the contours of the MFT and SPP articular surfaces are convex in three planes and therefore have greater potential for congruity than other grafts. However, detailed anatomic comparisons were limited by the use of computed tomography (CT) wrist images from a retrospective, nonmatched sampling of patients. No further quantitative studies have investigated the relationship between the curvatures of the SPP and MFT. Accordingly, our understanding of the true congruity between these surfaces is incomplete. The aim of this study was to provide a detailed morphologic comparison of the articular surfaces of the MFT and SPP using a quantitative anatomical approach in matched cadaveric specimens.

Methods

Specimens

This study was performed using cadaveric specimens from the Keck School of Medicine of the University of Southern California (Los Angeles, CA). Donors were sampled only if they were previously registered for research purposes, and if they had no known history of prior articular pathology. Specimens with significant marginal osteophytes, cartilage thinning, or eburnation of the articular surface were not included for analysis. Ultimately, 12 matched femur-scaphoid pairs were selected for analysis—six females, six males; average age 78.5 years (range, 62–91 years).

Image Acquisition and Processing

Digital surface models of the distal femur and complete scaphoid were created from CT scans. Each femur-scaphoid pair was scanned in the same field of view using an Aquilion One 320 detector row 0.5 mm detector scanner (Toshiba Medical Systems; Otawara, Japan) under the scan parameters: energy, 120 kVP; current, 120 mAs; slice thickness, 0.5 mm; field-of-view, 180 mm; reconstruction algorithm, bone kernel.

Images were saved as 16-bit DICOM stacks, then converted to three-dimensional (3D) polygon surface models with Amira v.5.6 software (FEI; Hillsboro, OR). Using Geomagic v.12 software (Geomagic, Inc.; Cary, NC), 3D models were cleaned of any artificial defects and refined to increase triangle density in the surface mesh (>250,000 triangles). The articular surfaces of the SPP and the proximal medial quadrant of the MFT were then manually isolated ( Fig. 1A ). Following prior description, ¹¹ the radioulnar (RU) axis of the resected SPP (rSPP) was manually aligned with the proximodistal (PD) axis of the MFT and used as a guide to resect the MFT (rMFT), mimicking the operative harvest of a graft ( Fig. 1B, C ).

Fig. 1.

(Color online) Resection of the scaphoid proximal pole (SPP) and medial femoral trochlea (MFT). ( A ) SPP (green) and proximal-most medial quadrant of the distal articular surface of the femur (blue). ( B ) Resected SPP superimposed on the MFT to guide MFT resection. ( C ) Axes for radius of curvature (RoC) measurements of the resected MFT (proximodistal, PD; mediolateral, ML) and SPP (radioulnar, RU; anteroposterior, AP).

Measurements

Using Amira software, digital landmarks were placed on the boundaries of the RU and PD axes for each rSPP and rMFT, respectively, as well as their corresponding orthogonal axes—anteroposterior (AP) for the rSPP and mediolateral (ML) for the rMFT. From the resulting x-y-z coordinates of these landmarks, radius of curvature (RoC) values were calculated using simple geometry ( Fig. 1C ). The percent difference between RoC values of the rSPP and rMFT was determined for each donor, using the average of the femur and scaphoid RoCs in each plane as the denominator.

To further visualize 3D congruence, the articular surfaces were aligned according to previous studies ¹¹ using functions in Amira software that minimizes the overall distance between two surfaces. Each specimen was scaled to its total surface area to facilitate comparisons across donors of varying size (e.g., larger males vs. smaller females). With the rSPP and rMFT in the optimized superimposed position, scaled distances were calculated between the surface triangles of each and visualized as color heat maps superimposed on the SPP. The scaled mean distance between the rSPP and rMFT was recorded to demonstrate correspondence with RoC values. Scaled measurements are unitless and represent relative congruence within the study cohort, not absolute distances.

Single-plane RoC values (in mm) for each donor were compared between the rSPP and rMFT using the Wilcoxon signed-rank test, with significance set at p  < 0.05.

Results

Visual inspection revealed areas of incongruity between the rMFT and rSPP in all donors. The convex curvature of the SPP was relatively conserved in all planes across specimens. However, the MFT exhibited a large amount of variation, with a nearly flat surface in some donors and a notably convex curvature in others.

The RoC data for each specimen along with descriptive statistics are presented in Table 1 . The mean PD RoC for the rMFT was 23.6 mm (standard deviation [SD], 10.1 mm). The mean RU RoC for the rSPP was 17.1 mm (SD, 3.7 mm). Donors 2, 5, 9, and 10 were the least congruent in this plane, exhibiting the highest absolute and percent differences between RoC values. Analysis using the Wilcoxon signed-rank test revealed that the two surfaces were not significantly different in this plane ( p  = 0.064; Fig. 2 ).

Table 1. Measures of congruence.

Specimen	PD/RU axis RoC (mm)				ML/AP axis RoC (mm)				Mean 3D distance a
	rMFT (PD)	rSPP (RU)	Absolute difference	% Difference	rMFT (ML)	rSPP (AP)	Absolute difference	% Difference
1	21.1	22.2	1.1	5.3	52.0	12.2	39.8	123.8	0.0027
2	23.7	10.8	12.9	74.7	52.6	9.3	43.4	140.1	0.0043
3	24.4	23.6	0.9	3.6	16.0	10.4	5.6	42.5	0.0018
4	16.8	19.9	3.2	17.2	64.7	11.8	52.9	138.3	0.0038
5	50.9	17.1	33.8	99.3	142.9	27.0	115.9	136.4	0.0026
6	18.2	14.1	4.1	25.3	26.2	9.5	16.7	93.5	0.0011
7	18.1	15.6	2.5	15.0	35.5	9.1	26.4	118.7	0.0031
8	17.6	16.3	1.2	7.3	23.4	10.4	13.0	77.0	0.0026
9	32.2	15.4	16.8	70.7	18.9	10.8	8.1	54.4	0.0022
10	28.4	17.5	11.0	47.8	146.1	11.2	134.8	171.5	0.0031
11	15.4	13.5	1.8	12.5	33.6	9.8	23.8	110.0	0.0031
12	16.4	19.5	3.1	17.2	30.5	9.5	21.0	105.3	0.0032
Average	23.6	17.1	7.7	33.0	53.5	11.7	41.8	109.3	0.0028
SD	10.0	3.7	9.8	32.1	44.9	4.9	41.7	37.5	0.0009

Abbreviations: 3D, three-dimensional; AP, anteroposterior; ML, mediolateral; PD, proximodistal; rMFT, resected medial femoral trochlea; RoC, radius of curvature; rSPP, resected scaphoid proximal pole; RU, radioulnar; SD, standard deviation.

^aScaled, unitless.

Fig. 2.

Radius of curvature (RoC) measurements for the resected medial femoral trochlea (rMFT) and resected scaphoid proximal pole (rSPP). Bars indicate mean values and whiskers indicate one standard deviation. The Wilcoxon signed-rank test was used to compare the RoC of the rMFT in the proximodistal (PD) axis with the RoC of the rSPP in the radioulnar (RU) axis, and the RoC of the rMFT in the mediolateral (ML) axis with the RoC of the rSPP in the anteroposterior (AP) axis.

The mean ML RoC for the rMFT was 53.5 mm (SD, 44.9 mm). The mean AP RoC for the rSPP was 11.7 mm (SD, 4.9 mm). Only donors 3, 6, and 8 showed some degree of congruence in this plane, with lower absolute and percent differences compared with the other specimens. Accordingly, the Wilcoxon signed-rank test revealed that the two surfaces were significantly different in this plane ( p  = 0.001; Fig. 2 ).

Figure 3 illustrates distance comparisons between matched pairs of rMFT and rSPP in representative donors with minimal, intermediate, and maximal 3D scaled mean distances (donors 6, 1, and 2, respectively). Scaled 3D distances between the two surfaces were variable, both within ( Fig. 3 ) and across donors ( Table 1 ). Areas of poor congruence include the center and surrounding edges of the resections, reflecting the greater curvature of the SPP relative to the MFT in all planes. The continuity of color in the PD–RU axis on the heat maps indicates greater congruity in this plane.

Fig. 3.

(Color online) Visual representations of congruence between the scaphoid proximal pole (SPP) and medial femoral trochlea (MFT) showing minimal, intermediate, and maximal three-dimensional scaled mean distances between the two surfaces (donors 6, 1, and 2, respectively). Blue stars indicate areas of maximal congruence, while red stars indicate areas of minimal congruence. ( A ) Overlay of the resected SPP (rSPP, green) and resected MFT (rMFT, blue), indicative of the position for graft harvest. ( B ) Overlay of rSPP and rMFT in profile, demonstrating similarities and differences in the curvatures of the two resections. ( C ) Color heat map (superimposed on the SPP) representing the scaled distance between the rSPP and rMFT. Cooler colors indicate stronger congruence, while warmer colors indicate weaker congruence. The scaphoid is presented with its radioulnar (RU) axis oriented approximately in the vertical plane of the page and the distal pole toward the top, to mimic alignment with the proximodistal (PD) axis of the femur. ( D ) Scaphoid with superimposed color heat map, oriented in its anatomic position within the carpus and the proximal articular surface facing the viewer. RoC, radius of curvature; ML, mediolateral; AP, anteroposterior.

Discussion

Nonunion is a recognized sequela of SPP fractures, and can be a challenge to manage. ² While traditional treatment options produce varying results in the setting of osteonecrosis, VBG has emerged as a promising solution. ^{4 16} Harvest of vascularized osteochondral flaps from the MFT has demonstrated excellent short- and intermediate-term outcomes in the treatment of difficult SPP nonunions. In the clinical series conducted by Bürger et al, 15 of 16 patients reached union, 75% reported complete pain relief, and all patients reported symptomatic improvement at an average follow-up of 14 months. ¹¹

The results of our quantitative anatomical study suggest that the relationship between the curvatures of the SPP and MFT is more complex than previously highlighted. In general, congruence appears to vary greatly, both between individuals and within each matched pair of articular surfaces. The curvatures of the resected SPP and MFT (assessed in a single 2D plane) were similar in the RU and PD axes, respectively, in eight of twelve individuals examined. In contrast, the two surfaces shared little congruence in the orthogonal axes (AP axis for the SPP; ML axis for the MFT) in 9 of 12 individuals.

In their anatomical study of the MFT and its vascular supply, Hugon et al used 3D reconstructions from CT scans of 11 nonmatched femora and wrists to compare the curvature of the MFT with that of the proximal carpal row and capitate. ¹⁵ Their analysis was conducted in two orthogonal planes: the transverse and sagittal axes of the MFT, and the corresponding sagittal and frontal axes, respectively, of the carpus. The authors reported that the mean RoC of the sagittal axis of the femur was similar to that of the frontal axis of the proximal carpal row. This plane is roughly analogous to the PD axis of the MFT and RU axis of the SPP analyzed in the present study, and our data are in agreement with the prior findings. However, direct comparison of these results is difficult because the authors treated the proximal carpal row as a single entity for their RoC calculations in this plane. Hugon et al also reported similarity between the RoC of the transverse axis of the MFT and the sagittal axis of the SPP. The corresponding axes (ML axis for the MFT; AP axis for the SPP) in the present study were found to be significantly different. There are several factors that may explain this difference. First, our ability to manipulate the scaphoid in isolation enabled us to position the SPP in a specific manner, closely simulating the established protocol for MFT graft harvest. ¹¹ As a result, the axes we chose for RoC measurements may differ slightly from those used by Hugon et al. Second, the virtual 3D environment utilized in this study enabled the placement of landmarks at precise locations, allowing for a more accurate identification of our desired axes and subsequent calculation of RoC values. This is especially important for the MFT due to the significant variability in the morphology of its articular surface.

Our study has several notable strengths. First, each MFT was compared with the ipsilateral SPP from the same donor. This increases the translational relevance of our results, especially considering the variation in congruity between subjects. Second, by utilizing a virtual 3D environment to freely manipulate bones with complex morphology, we were able to replicate the in vivo surgical harvest of the MFT as a vascularized graft for the SPP. Moreover, this approach allowed us to calculate the mean scaled distances between each MFT–SPP pair, which provides additional characterization of the overall congruity between the two surfaces.

The findings of this study must be interpreted within the scope of its limitations. First, our data were derived from only 12 donors, and future studies with larger sample sizes drawn from different populations would provide additional insight into the relative morphological congruency between the MFT and the SPP. Second, the average age of the donors used in this study was ∼79 years, and may represent individuals that would not typically undergo MFT grafting for SPP nonunion. Although we visually examined each specimen for gross articular defects, the effects of degenerative change at this age cannot be ignored. Subtle arthritic degeneration may differentially influence the curvature of the articular surfaces of the MFT and SPP, leading to age-related variations in RoC values. Ideally, further investigations would utilize curvature data from the matched femora and wrists of young adults—the demographic for whom this procedure is most commonly used. ^{11 13} Finally, further refinement of curvature analysis could be accomplished with higher resolution imaging, using micro-CT scans of each bone individually in a smaller field of view.

The clinical results of MFT grafting for cases of difficult SPP nonunion are promising thus far. This quantitative anatomical study expands our understanding of the articular structures involved in the treatment of SPP nonunion with vascularized grafts from the MFT. Our data suggests that the MFT and SPP may share less morphological congruency than previously thought. This suggests that the vascularized and osteochondral characteristics of the MFT graft are the primary factors responsible for clinical improvement, and that recreating the curvature of the SPP when grafting may be of lesser importance.