Reconstructing Proximal Humeral Fractures Using the Bicipital Groove as a Landmark

Abstract

Controversy persists in the literature regarding the use of the bicipital groove as an anatomic landmark to restore humeral head retroversion when treating complex proximal humeral fractures with arthroplasty. We quantified the three-dimensional geometry of the bicipital groove in 49 dried humeri relative to the intramedullary axis, quantified the reliability of using the bicipital groove as an anatomic landmark, and compared this reliability with that of the conventional technique that uses a fixed, average angle relative to the epicondylar axis to establish humeral head retroversion. The data show the anterior offset of the bicipital groove is nearly constant from proximal (7.3 mm ± 2.8 mm) to distal (7.2 mm ± 1.5 mm) relative to the intramedullary axis. Given the consistency, the distal bicipital groove (at the level of the surgical neck) is a reasonable landmark to establish humeral head retroversion after complex proximal humeral fractures having reliability (angular error of 7.9° ± 5.8°) as good as or better than the conventional fixed-angle technique.

Anatomic reconstruction of complex proximal humeral fractures is difficult to achieve primarily because of the involvement or disruption of the humeral head articular surface and tuberosities. Implanting a prosthesis in greater than 40° retroversion is associated with posterior migration of the greater tuberosity and poor functional out-come.^2,4 In the absence of any established, well-defined anatomic landmarks to determine the patient's anatomic humeral head retroversion, surgeons routinely use an average value for retroversion (range, 20°-40°)^{1,3,6-8,11,15} measured relative to a distal anatomic landmark: the epi-condylar axis. This method likely reduces the incidence of implanting a prosthesis in excessive retroversion. However, reconstructing humeral head retroversion using an average value is not without concern: retroversion varies widely among individuals of different age, gender, and race.^3,6,14,16,17 Therefore with this technique there exists a high probability to implant a prosthesis in a retroversion that deviates from the patient's native anatomy by greater than 10° and a possibility that it could deviate by as much as 30°.^5,9,14

A number of studies explored the possibility of using a more proximal anatomic landmark, the bicipital groove, to reconstruct humeral head retroversion.^{1,5,9,12-14,18} Controversy persists regarding its reasonableness as an anatomic landmark primarily because of its purported angular variability from proximal to distal.^{1,5,9,12-14,18} We contend the controversy persists in part because of each study using a coordinate reference system that is not relevant to reconstruction using arthroplasty. We specifically refer to measuring the bicipital groove geometry of cadaveric humeri relative to the intramedullary canal using a Cartesian coordinate system rather than a Polar coordinate system as has been used previously.^1,5,9,12,14

We evaluated the 3-D geometry of the bicipital groove along its course from proximal to distal using an accepted method widely used to quantify the geometry of the humeral head.^3,10 We quantified the angular error associated with implanting a prosthesis oriented relative to the bicipital groove at the level of the surgical neck (ie, the probable location of the fracture line) and compared it with the angular error associated with implanting a prosthesis oriented at multiple fixed angles relative to the epicondylar axis, ie, at 20°, 30°, and 40°.

MATERIALS AND METHODS

The anatomy laboratory at the Bordeaux School of Medicine (Bordeaux, France) provided 57 dried humeri. We excluded eight specimens with apparent osseous abnormalities, leaving 49 humeri (31 left, 18 right). Donor information (eg, gender and age) was not available. We used a 3-D coordinate measuring machine (MC850; Zeiss, Esslingen, Germany) to digitize the proximal humeral geometry relative to defined reference axes. The anterior and lateral offsets of the bicipital groove were calculated relative to these axes using prepackaged coordinate measuring machine software (UMESS Unix Version; Zeiss).

To create the reference axes, an orthopaedic surgeon (PHF) marked four or five points along the anatomic neck and then identified the junction of the humeral shaft and proximal metaphysis. This junction defined the proximal portion of the humeral cylinder (Fig 1A). Next, a metal ring was aligned along the points on the anatomic neck to establish a plane (Fig 1B). The humerus then was inserted into a V-shaped fixation device at the level of the proximal humeral cylinder (Fig 1C). The assembly then was placed in the coordinate measuring machine to define the humeral head equatorial plane. The humeral head equatorial plane was defined as a plane perpendicular to the anatomic neck plane passing through the inferior and superior points of the articular surface. Next, the coordinate measuring machine identified a plurality of points on the articular surface along the humeral head equatorial plane. These points were marked with a pencil bisecting the head in a mediolateral direction (Fig 1D). This method was derived from that of Hertel et al.¹⁰ The coordinate measuring machine then was used to identify approximately 30 points (ie, the point cloud) on the humeral head articular surface and between the superior and inferior edges of the proximal humeral cylinder, the axis of the cylinder (created from the point cloud data) approximating the intramedullary axis (Fig 1E).

Fig 1A: E. (A) The surgeon defines the humeral anatomic neck and junction between the humeral shaft and the proximal metaphysis. (B) The metal ring is used to establish the humeral head anatomic neck plane. (C) The humerus is inserted into a V block to establish humeral orientation. (D) A coordinate measuring machine is used to digitize the humeral head articular surface along the equatorial plane. (E) The coordinate measuring machine is used to digitize the humeral shaft to define the orientation to the intramedullary axis.

The first axis corresponded to the intramedullary axis of the humerus. The second axis was perpendicular to the first and oriented parallel to the humeral head equatorial plane. The third axis was self-defined, orthogonal to the first and second axes. The origin of these axes is located at the intersection of the intramedullary axis and a line perpendicular to the anatomic neck at the center of the humeral head in the anteroposterior plane (Fig 2).

Fig 2: The reference axes of the proximal humerus are shown.

Similar to the widely used and accepted technique of quantifying the location of the humeral head center,^3,10 we defined the location of the bicipital groove using a Cartesian coordinate system rather than a Polar system. In doing so, we hope to clarify the debate in the literature regarding the purported angular variability of the bicipital groove from proximal to distal.^1,2,12 We used linear dimensions to quantify the position of the bicipital groove relative to the intramedullary axis (Fig 3). The anterior offset of the bicipital groove was defined as the shortest distance in the transverse plane between the bicipital groove and the intramedullary axis along the anteroposterior axis (ie, the z-axis). Similarly, the lateral offset of the bicipital groove was defined as the shortest distance in the transverse plane between the bicipital groove and the intramedullary axis along the mediolateral axis (ie, the y-axis). We used the vertical distance (ie, the x-axis) to determine the length of the bicipital groove.

Fig 3: The anterior (BGAO) and lateral (BGLO) location of the bicipital groove in the coronal plane is measured relative to the intramedullary axis.

The location of the bicipital groove was digitized with a coordinate measuring machine at four levels: from proximal (H₁) to distal (H₄) and at two symmetric points (H₂ and H₃) between. The distal portion was defined as the level at which the bicipital groove becomes nearly flat. It is located approximately 10 mm below the most inferior aspect of the articular surface. Because the location of the distal slice (H₄) varies between bones, we introduced a fifth level (H_a) calculated exactly at the level of the most inferior aspect of the articular surface (Fig 4). This level was the assumed location of the surgical neck, a probable fracture location. For comparative purposes, the location of the humeral head relative to the intramedullary axis also was quantified using the previously mentioned methods.

Fig 4: A graphic definition of the bicipital groove level Hα is shown.

We also used the point cloud data to measure humeral head retroversion relative to two different anatomic landmarks: (1) the epicondylar axis (HHR_EP) and (2) the bicipital groove (HHR_BG). The humeral head retroversion relative to the epicondylar axis (ie, HHR_EPanatomic) was defined as the angle between the epicondylar axis and the humeral head equatorial plane (Fig 3). The humeral head retroversion relative to the bicipital groove (ie, HHR_BGanatomic) was defined as the angle between the plane passing through the intramedullary axis and the center of the bicipital groove at the level of H_α and the humeral head equatorial plane. We calculated HHR_BG anatomic at the level of H_α using the following equation:

where BGAO_a and BGLO_a denote the anterior and lateral offsets of the bicipital groove at height level H_a, respectively.

To assess the reasonableness of using the bicipital groove as an anatomic landmark, we performed a computational analysis to compare the reliability of reconstructing humeral head retroversion relative to the bicipital groove and relative to the epicondylar axis. This analysis used two different humeral fracture prostheses. The first simulated the reconstruction of humeral head retroversion using a conventional prosthesis in which its lateral fin was oriented at a fixed angle relative to the epicondylar axis. The second simulated the reconstruction of humeral head retroversion using a novel fracture prosthesis whose lateral fin was shifted in an anterior direction by 7.5 mm. Shifting the fin in an anterior direction creates an asymmetric tuberosity bed (ie, a smaller bed for the lesser tuberosity and a larger bed for the greater tuberosity) that can facilitate reconstruction of the tuberosity fragments around the metaphyseal portion of the fracture stem (Fig 5).

Fig 5: A novel humeral fracture prosthesis has a lateral fin anteriorly offset by 7.5 mm to create an asymmetric tuberosity bed acting as a scaffold to support the greater and lesser tuberosities after three- and four-part fractures of the proximal humerus (Equinoxe; Exactech, Inc). This figure is published with permission from Exactech, Inc, Gainesville, FL.

Equations 2, 3, and 4 were used to calculate the angular error associated with aligning the lateral fin of the conventional prosthesis at 20°, 30°, and 40° relative to the epicondylar axis to establish humeral head retroversion. Equation 5 was used to calculate the angular error associated with aligning the anterolateral fin of the novel prosthesis with the bicipital groove at the level of the surgical neck to establish humeral head retroversion.

To quantify the study repeatability, we applied the aforementioned method to measure each of the study parameters on six plastic humeri (Humerus #1028; Sawbones, Malmö, Sweden) before performing the anatomic study. The protocol successfully obtained each desired parameter. Linear measurements were reliable to ± 0.75 mm. Angular measurements were reliable to ± 1°. We did not specifically address accuracy of the tools we used because the coordinate measuring machine has a long history of use in quality inspection.

We used a two-tailed paired t test to compare all differences in means. Significance was set at p < 0.05. For the primary study question, if the technique using the bicipital groove has a similar (or better) level of reliability than the technique(s) using a fixed angle relative to the epicondylar axis, then we concluded the bicipital groove is a reasonable anatomic landmark to reconstruct humeral head retroversion.

RESULTS

Using a Cartesian coordinate system showed the bicipital groove was in an anterior and lateral position relative to the intramedullary axis. The bicipital groove occurred in a plane relatively parallel to the humeral head equatorial plane. This characteristic is defined by the nearly constant anterior offset of the bicipital groove from proximal (7.3 mm ± 2.8 mm) to distal (7.2 mm ±1.5 mm) relative to the intramedullary axis (Table 1). Additionally, the mean anterior and lateral offsets of the bicipital groove were less variable and more normally distributed from proximal (H₁) to distal (H₄) as evidenced by the 86.6% and 31.2% decreases in the standard deviation values, respectively (Tables 1 and 2; Fig 6). The bicipital groove exhibited a distinctive S-shaped curvature; in every bone observed, the lateral offset measurements at H₂ were larger (p < 0.001) than at H₁ and H₃ (Table 2). The mean bicipital groove length was 38.2 mm ± 4.0 mm (range, 28.8-44.9 mm). The mean medial and posterior offsets of the humeral head relative to the intramedullary axis were 6.3 mm ± 1.5 mm (range, −2.9 mm-9.5 mm) and 1.8 mm ±1.5 mm (range, −0.9 mm-6.0 mm), respectively.

TABLE 1: Anterior Offset of the Bicipital Groove at Five Levels

TABLE 2: Lateral Offset of the Bicipital Groove at Five Levels

Fig 6A: B. Frequency distribution histograms show the increase in normality of the anterior offset of the bicipital groove from (A) level H1 to (B) level H4.

The mean value of HHR_EPanatomic (ie, HHR_EPmean) was 20.1° ± 11.0° (range, 1.1°-41.5°). The mean value of HHR_BGanatomic (ie, HHR_BGmean) at the level of the surgical neck was 43.7° ± 9.9° (range, 21.9°-69.6°). The mean angular errors to reconstruct humeral head retroversion using the epicondylar axis as an anatomic landmark when the lateral fin of a conventional prosthesis is aligned at 20°, 30°, and 40° were 8.8° ± 6.6° (range, 0°-21.5°), 12.4° ± 8.1° (range, 0.5°-28.9°), and 20.8° ± 10.7° (range, 0.2°-38.9°), respectively. The mean angular error to reconstruct humeral head retroversion using the bicipital groove as an anatomic landmark when the anterolateral fin of the novel prosthesis is aligned with the bicipital groove at the level of the surgical neck was 7.9° ± 5.8° (range, 0°-25.9°).

Using the bicipital groove (in conjunction with a prosthesis that uses the anterior location of the bicipital groove) as an anatomic landmark was associated with less angular error than using a fixed value of 30° (p = 0.008) or 40° (p < 0.0001) relative to the epicondylar axis. The angular error associated with using the bicipital groove as an anatomic landmark was not different than using a fixed value of 20° relative to the epicondylar axis in this study population.

DISCUSSION

Controversy persists in the literature regarding the reasonableness of using the bicipital groove as an anatomic landmark to restore humeral head retroversion when treating complex proximal humeral fractures with arthro-plasty.^{1,2,9,12,13,18} We quantified the 3-D geometry of the bicipital groove relative to the intramedullary axis, quantified the reliability of using the bicipital groove as an anatomic landmark, and compared this reliability with that of the conventional technique (ie, the gold standard), which uses a fixed, average angle relative to the epicondylar axis to establish humeral head retroversion.

The application of these results is limited by the precision of the testing methods (± 0.75 mm; ± 1°), by the small sample size (49 humeri), and by the lack of donor information (eg, gender and age). However, we do not believe the small errors in testing methods would influence clinical application since it is small. The relatively small number of cadaveric specimens likely includes a reasonable range from a much larger population.

Previous cadaveric studies have defined the shape of the bicipital groove relative to the humeral head equatorial plane rather than the intramedullary axis.^9,13,18 In one study, Tillet et al¹⁸ reported the central axis of the humeral head (ie, the humeral head equatorial plane) 9.0 mm ± 2.4 mm posterior to the posterior margin of the bicipital groove. In another study, Kontakis et al¹³ found the humeral head equatorial plane 5.2 mm ± 2.6 mm posterior to the posterior margin of the bicipital groove. The results of these two studies are not directly comparable because the location of the bicipital groove was not likely quantified at the same height or location (and neither study made any comment regarding the variability of the groove from proximal to distal). Kontakis et al¹³ quantified the location of the bicipital groove at the subcapital level. It is unclear from the text at what exact height or location Tillet et al¹⁸ queried; however, the figures suggest these measurements were not taken at the same level. In yet another study, Hempfing et al⁹ reported the distance from the center of the bicipital groove to the humeral head equatorial plane (along the anteroposterior axis) did not change substantially between the proximal (8.0 mm ± 1.4 mm) and distal (8.5 mm ± 1.1 mm) portions of the bicipital groove.

We found the humeral head posteriorly offset from the intramedullary axis by 1.8 mm ± 1.5 mm; this result is comparable to those of other studies.^3,10,16,17 By summing the mean values for the posterior offset of the humeral head and the anterior offset of the bicipital groove at the level of the surgical neck (ie, the subcapital level), we conclude the bicipital groove is on average anterior to the humeral head equatorial plane by 9.2 mm. This value is comparable to those reported by Tillet et al¹⁸ and Hempfing et al.⁹ Although we did not directly measure the width of the bicipital groove, it seems reasonable to suggest the 4.0-mm mean difference between our study and that of Kontakis et al¹³ is attributable to this different reference point (particularly considering Kummer et al¹⁴ reported the average bicipital groove width to be 8 mm ± 1.5 mm). We also observed the relatively constant location of the bicipital groove from proximal to distal. The mean difference between our study and that of Hempfing et al⁹ is attributable to the different reference point (ie, the humeral head equatorial plane versus the intramedullary axis).

Some investigators have reported the bicipital groove is not a reasonable anatomic landmark because its orientation changes from proximal to distal.^1,2,12 Balg et al¹ and Itamura et al¹² reported the bicipital groove internally rotates along its course from proximal to distal. Our data are compatible with these findings as evidenced by the decreasing lateral offset of the bicipital groove from proximal to distal in association with the relatively constant anterior offset. However, we disagree with the interpretation and implication of the data. Balg et al¹ stated the variable orientation of the bicipital groove suggests it is an unreliable landmark to reconstruct humeral head retroversion using arthroplasty. Our results suggest the anterior offset of the bicipital groove is reliably located despite its orientation being variable. The use of a Cartesian system (rather than the Polar system) made this elucidation possible. Furthermore, we contend the implication of this variability is dependent on the choice of reference axes. For example, Balg et al¹ and Itamura et al¹² used the bisector of the transepicondylar axis as a reference to define the orientation of the bicipital groove. The use of the intra-medullary axis is more appropriate for accurately defining the location of the bicipital groove, because it is a more proximal landmark and more relevant to reconstruction using arthroplasty (because it shares the same axis).

Aligning the anterolateral fin of the novel prosthesis with the bicipital groove at the level of the surgical neck represents several improvements over the conventional method, which uses a fixed angle relative to the epicondylar axis. In surgery, it is often difficult to accurately align the prosthesis relative to the epicondylar axis. This statement is defended by the widespread use of fracture jigs to facilitate the alignment. These devices are often complex to assemble, time-consuming to apply, and associated with their own angular inaccuracies. Furthermore, surgeons who do struggle with alignment but do not use these devices (for the aforementioned reasons) will often simply approximate a fixed angle based on the forearm and, in doing so, fail to consider any compound version in the elbow.

The anterior offset of the bicipital groove is nearly constant from proximal to distal relative to the intramedullary axis. When this consistency is considered, the distal bicipital groove (at the level of the surgical neck) appears a reasonable anatomic landmark to establish humeral head retroversion after complex proximal humeral fractures having reliability as good or better than the conventional fixed angle technique. Therefore, aligning the anterolateral fin of the novel prosthesis with the bicipital groove at the level of the surgical neck is a simple and reasonable solution to reconstruct humeral head retroversion for treatment of proximal humeral fractures using arthroplasty.

Acknowledgment

We thank Professor Vital (Bordeaux School of Medicine, Bordeaux, France) for assistance in completion of this study.