Minimally invasive plate osteosynthesis of the humeral shaft with regard to adjacent anatomical characteristics

The study goal was to evaluate the distances from the radial (RN), the musculocutaneous (MN) and axillary nerves (AN) and the medial neurovascular bundle of the upper arm to a minimally invasive applied plate and to define its relation to the RN during different degrees of malrotation during MIPO. The sample involved ten upper extremities. Application of a PHILOS plate was performed through a Delta-split. Intervals between the AN, MN, RN and the medial vascular bundle were defined at various positions. The humeral shaft was artificially fractured at a height of about the mean of the plate. The distal fragment was brought into 15° and 30° internal (IR) as well as external rotation (ER) and here, the plate’s relation to the RN was evaluated. The medial neurovascular bundle intersected the plate at its distal part in two specimens. Regarding the distances from the RN to the plate during different rotation positions the distances became significantly longer during ER, respectively shorter during IR. The medial neurovascular bundle and the RN were identified as the main structures at risk. Care must be taken during distal screw placement and malrotation exceeding 15° must be avoided during MIPO.


Materials and methods
Study sample. The study sample involved ten unpaired upper extremities from ten human adult cadavers embalmed using Thiel's solution 11 . All investigated corpses were donated to the Division of Macroscopic and Clinical Anatomy of the Medical University of Graz under the approval of the Anatomical Donation Programme and accordingly to the Austrian curial law. Since these bodies are dedicated to research and scientific teaching, the donors accepted participation in research during their lifetime. All experimental protocols and methods were carried out in accordance with relevant guidelines and regulations. The sample consisted of three limbs from female and seven upper extremities from male donors. Their age ranged from 49 to 92 years with a median of 82. Via inspection of the skin and soft tissues as well as through radiographic control with the C-arm, extremities with former fractures, surgical interventions or malformations in the region of interest were excluded.
Plate application. Primary, the humeral length (HL), which was defined as the distance between the lateral humeral epicondyle (LE) and the tip of the greater tubercle, was evaluated. For implantation, we used a 12-hole PHILOS plate (Synthes GmbH, Oberdorf, Switzerland) which's middle third had been twisted in a ventral direction in an angle of about 70°-90° over a distance of two plate holes. Access to the proximal part of the humerus was gained through an about six centimetres long skin incision and a Delta-split approach in typical manner.
Here, the AN was depicted and countermined with a raspatory. Care was taken not to manipulate the nerve's original course. Next, the plate was inserted and advanced epiperiosteally to the deltoid tuberosity and the ventral humeral surface. The PHILOS plate was applied to the bone through K-wires about 4 mm posterior to the bicipital grove. The approach to the distal portion of the plate was gained through an about 5 cm skin incision at the distal antero-lateral part of the upper arm which was conducted directly superficial to the palpable plate.
Here, after splitting of the fascia, the biceps was retracted in a medial direction following depiction of the MN. The brachial muscle was split at its median part directly superficial to the plate. Here, fixation was conducted through one conventional screw. rule. The distance between the proximal margin of the PHILOS plate and the AN as well as the interval between the tip of the greater tubercle and the AN were measured. Additionally, we observed the location of the nerve with regard to the plate's screw holes. After plate application, the relationship between the MN and the plate was evaluated. Therefore, the distal skin incision was advanced in a proximal direction. Primary, the ventral crossing point of the plate and the nerve was evaluated and its height with regard to the LE and the distal tip of the plate was measured. Further the distance between the medial margin of the plate and the exit point of the MN from the coracobrachialis was evaluated.
For defining the distances between the plate and the RN, respectively the medial neurovascular bundle three distances were defined: (1) The interval between the medial/lateral border of the plate and the structures at the height of the RN's exit from the lateral brachial septum, (2) the distance between the plate and the structures at the median point between 1 and 3 and (3) the interval between the medial/lateral edge of the plate and the characteristics at the height of the distal tip of the plate (Figs. 2, 3). For measurement of the RN, the brachialis muscle was split at its median part and the two halves were retracted. For a schematic depiction see Fig. 4.

Rotation tests of the RN.
After collection of the measurements, the humeral shaft was artificially fractured at a height of about the mean of the plate by use of a circular saw. The distal fragment was brought into 15° internal (IR) and external rotation (ER) as well as 30° IR and ER. In each rotation position, the plate was re-fixed to the distal humeral surface and the distances to the RN were measured at the (above described) positions 1-3.
Statistical analysis. Statistical analysis was conducted using SPSS statistical software (version 24.0; IBM Corp, Armonk, NY, USA). Regarding descriptive statistics, continuous variables are presented as mean, standard deviation (SD), minimum and maximum, whereas categorical data are described as frequencies and percentages. All continuous data were normally distributed. Further, a repeated measurement ANOVA including degrees of distal fracture fragment rotation and radial nerve position as within subject factors was conducted. For comparison of the three different distances between the RN and the plate at different degrees of rotation, t-tests for paired samples were used. Bonferroni adjustment for multiple testing was performed (alpha = 0.05/12 = 0.004).   Table 1. Regarding the relation between the AN and the screw holes, the nerve's most common location was at the proximal of the so-called calcar screws (5 cases/50%).  Here, the bundle intersected the plate in two specimens. The distance between the MN and the plate was on average 14.9 mm (SD 4.5; range 4.2-20.6) at its exit point from the coracobrachialis. Regarding its intersection point with the plate, this was at a height of 52.5 mm (SD 16.0; range 28-80) measured from the distal border of the plate, and 82.9 mm (SD 19.4; range 59-120 mm) with regard to the LE. Further, the plate-MN distance was 13.7 mm (SD 9.9 mm; range 0-29) at the median of the two measurement points. Details see Table 1.

RN measurements and rotation tests.
Repeated measurement ANOVA revealed statistically significant differences regarding the distances from the RN to the plate during the neutral position against all of the different rotation positions of the distal fracture fragment (p = < 0.001). Thereof, distances became significantly longer during ER, respectively shorter during IR (see Fig. 5).    Table 2.
Further, the RN was located on the plate in one specimen at position 2 (two millimetres medial to the lateral border of the plate; value: − 2), respectively in two cadavers at position 3 (both three millimetres medial to the lateral edge; value: − 3).

Discussion
MIPO by use of curved implants has recently become a more frequent technique in the treatment of humeral shaft fractures and has resulted in satisfactory clinical outcomes in several studies 2,6,7,9,10 . The proximity of single nerves to the osteosynthesis material during MIPO has been described in the current literature 12,13 . However, the intervals between the nerves and the plate at different positions have not been delineated in detail so far.
Gardner and colleagues 13 have declared the MN to be the main structure at risk during MIPO. Therefore, they defined a safe zone for its crossing point of the anterior humeral shaft at an interval between 12.2 and 14.8 cm measured from the greater tubercle. We in contrast, found its intersection point with the plate at a height of 52.5 mm starting from the distal border of the plate, respectively 82.9 mm with regard to the LE.
In Křivohlávek et al. 's 14 dissection study, a PHILOS plate was applied in 24 specimens through the minimally invasive anterolateral approach. Here, the AN did not suffer injuries in any of the cases. As we performed the plate insertion through a typical Delta-split approach including depiction of the nerve, evaluation of possible iatrogenic lesions was not possible. The mean distance from the AN to the tip of the greater tubercle was at a mean of 48.2 mm with a range from 40.5 up to 55.2 mm in our survey. In comparison, Křivohlávek et al. 's 14 intervals were shorter, ranging from 37 to 44 mm. Further, our evaluated mean distance from the AN to the proximal margin of the plate proved to be 38.9 mm which was also slightly greater than in Ninck et al. 's 15 dissections (mean of 36.1 mm). With regard to the AN in relation to the proximal screw holes, we evaluated the nerve's most common location (50%) on the calcar screws. This was also the most frequent position in Křivohlávek et al. 's 14 study, however, Ninck and colleagues 15 found the AN in most of the cases more distal at region F.
Regarding the medial vascular bundle of the upper arm, we observed the location of this structure on the plate in one specimen at position 2, respectively in two cases at position 3-a fact that needs to be considered during distal screw placement.
With respect to tolerated malrotation during conservative treatment of humeral shaft fractures, this was suggested not to exceed 15°1 6,17 . Further, a postoperative malrotation exceeding 20° was described to be associated with secondary shoulder arthritis 18,19 . Wang et al. 3 observed a significant increase of postoperative malrotation after MIPO in comparison to their ORIF control group. Additionally, the concomitant rotation of the RN during IR and ER of the distal fracture fragments needs to be considered. Our conducted repeated measurement ANOVA showed a statistically significant increase of the intervals between the RN and the lateral border of the plate at the three defined positions during ER, whereas they became significantly shorter during IR. During IR of 30°, we observed the location of the RN directly on the plate in one case at position 2 and in two specimens at position 3.
In conclusion we have examined the relation of a minimally invasive inserted plate to adjacent neurovascular characteristics of the upper arm. We have identified the medial neurovascular bundle and the RN (during IR) www.nature.com/scientificreports/ as the main structures at risk. Therefore, care must be taken during distal screw placement and malrotation exceeding 15° needs to be avoided. To prohibit the latter, we recommend to place the upper extremity under continuous longitudinal tension and to perform regular X-ray controls during surgery.