Risk factors for nonunion after intramedullary nailing of subtrochanteric femoral fractures

Our data show that:

Intramedullary nailing of diaphyseal and metaphyseal fractures is one of the most successful techniques in orthopaedic trauma surgery [23]. It promotes secondary bone healing by providing relative stability and preserves the vascularity at the fracture site, especially the periosteal blood supply [24]. Accordingly, nonunion rates after intramedullary nailing are typically in a low single-digit percent range [25, 26]. Nonunion rates after intramedullary nailing of subtrochanteric femoral fractures, however, are markedly higher [13, 17, 19, 27, 28]. Distinct biological and mechanical properties of the subtrochanteric region may account for this finding. First, the subtrochanteric medial cortex is subject to massive bending forces with values as high as 8.2 MPa [7, 8, 29] under physiological loads [10, 15, 30]. These forces need to be neutralized by the implant until osseous healing has occurred. Second, the subtrochanteric region has been described as a region with poor vascularity and with a low blood flow rate by several authors [11‐13, 19, 21]. It is well known that bone segments with distinct blood supply distribution are associated with higher nonunion rates. This applies, for example, for displaced femoral neck fractures, scaphoid fractures or Jones fractures of the fifth metatarsal bone. Santolini et al. [24], however, showed in an extensive review of multiple anatomical studies that the vascularity of the subtrochanteric region does not differ from other metaphyseal segments of long bones. Probably it is not the vascularity per se, but rather the very high bone turnover, which distinguishes the subtrochanteric region from other metaphyseal regions.

Varus malalignment (Figs. 1a, 2b) is a well-known risk factor for subtrochanteric nonunion [2, 3, 12‐14, 18]. Varus malalignment results in increased bending forces on the medial subtrochanteric cortex. These forces need to be neutralized by the implant until osseous healing has occurred. On the one hand, intramedullary nails are therefore preferable to extramedullary devices due to shorter lever arms for the countertorque of bending moments [14]. On the other hand, intramedullary nailing per se is a risk factor for varus malalignment. The medullary canal in the inter- and subtrochanteric region is broad with a relatively short proximal fragment in subtrochanteric fractures facilitating varus malalignment [10, 17]. This is even more relevant for Seinsheimer Type V fractures with intertrochanteric fracture extension [9]. Choosing a correct nail entry point is thus of crucial importance [31] and individual patient anatomy must be taken into account. According to our own experience and consistent with the findings in the literature, we recommend the consideration of two aspects for the avoidance of varus malalignment. First, the fracture should be reduced prior to the insertion of the nail [13, 31, 32]. This facilitates identification of the correct nail entry point [10]. The nail should therefore not be used as a reduction tool for subtrochanteric fractures. Second, the ideal entry point must not be lateral to the trochanteric tip for proximally bent nails. In general, we recommend an entry point, which is located slightly medial to the trochanteric tip, for these nails to avoid varus malalignment. However, due to individual variations in the ideal trochanteric entry point, thorough preoperative analysis of the individual patient anatomy is essential [31, 33]. Alternatively, the use of straight nails with a piriformis fossa entry point (collinear with the long axis of the femoral shaft) may be a valuable option to decrease the risk of entry point-related malreduction [33, 34].

Lack of medial cortical support (Figs. 1b, 2b) is an additional previously described risk factor for nonunion after subtrochanteric fracture stabilisation [7, 10, 13, 21]. Medial cortical buttress after reduction supports the implant in the counteraction of bending forces and varus torque during postoperative mobilisation. While the intramedullary nail acts more like a load-sharing device in the presence of medial cortical support, it has to act more like a load-bearing device in the absence of medial cortical support [7, 15]. Anatomic reduction results in restoration of the medial cortical support, if there is a fracture without comminution zone. In comminuted fractures, however, this is not the case [21]. Although the problem of medial comminution is widely recognized, there are, to the best of the authors´ knowledge, no recommendations in the literature for this problem. In an analogical manner, lack of medial cortical support is the most frequent cause for varus failure after surgical fixation of proximal humerus fractures [35, 36]. Strategies to prevent varus failure at the proximal humerus include shortening by impaction of the shaft into the head fragment [35], fibular cortical allografts [37] or even “mushroom” tailored cancellous allografts [36, 38]. These options, however, may not be feasible and advisable for the subtrochanteric region. Giannoudis et al. [13] demonstrated a case of subtrochanteric fracture nonunion revised using a lateral 95 degree blade plate and an additional anterior plate. Double plating or nailing with an additional anteromedial buttress plate may be as well an option for acute subtrochanteric fractures with severe medial comminution zones.

Autodynamisation of the nail was found to be a strong risk factor for nonunion in our study, with 8 out of 9 patients with autodynamisation developing nonunion. In general, a risk factor for subtrochanteric nonunion is a variable associated with an increased risk of subtrochanteric nonunion, but this does not necessarily imply causality. It is debatable whether delayed union/nonunion and weakness of the mechanical construct causes autodynamisation or vice versa, particulary if autodynamisation occurs in an early phase postoperatively (“chicken-and-egg” dilemma). Giannoudis et al. [13] stated that autodynamisation should be considered as a consequence of rather than the cause for nonunion. In their study, autodynamisation occurred after a mean of 4.4 months with nail breakage following 2 months later on average (6.5 months). In our study, autodynamisation occurred after 12 weeks at the latest and therefore much earlier, whereas nail breakage happened between 6 and 12 months after the index surgery resulting in a much longer interval between autodynamisation and nail breakage. One reason for these differences may be the inclusion of patients with atrophic nonunions only in the study by Giannoudis et al. [13]. We agree with the authors that autodynamisation is an indicator for instability of the overall mechanical construct and that it is predictive of future nonunion and nail breakage [13]. The long time period between autodynamisation and nail breakage, however, provides an opportunity for therapeutical interventions. Furthermore, Giannoudis et al. recommended revision surgery or weight-bearing restrictions, if autodynamisation is observed in patients who are still symptomatic at the fracture level. According to our experience, weight-bearing restrictions should not be supposed to be the solution in these cases. We therefore rather recommend revision surgery, if autodynamisation of the nail is observed in the early postoperative period. Accordingly, dynamisation of the nail by removing static locking screws, in our opinion, is absolutely contraindicated after intramedullary nailing of subtrochanteric fractures.

One of the main findings of our study is that all three risk factors for subtrochanteric nonunion were mechanical parameters. Another interesting, but not surprising, finding is that the nonunion rate increased with the number of risk factors (Fig. 4). The identified risk factors either result in increased load on the implant (varus malalignment), reduced intrinsic stability of the fracture (lack of medial cortical support) or reduced stability of the overall mechanical construct (autodynamisation). It is, therefore, reasonable to assume that there is a summation effect.

In contrast to the mechanical parameters, we did not find significant biological risk factors for nonunion in our study. Our subjective observations confirm this finding. First, we found timely callus formation in all nonunion cases (Figs. 2c, 3b) indicating a favourable biological environment for bone healing. Second, breakage of the nail was the characteristic failure mode, which indicates an unfavourable mechanical environment. Additionally, Fig. 3d shows osseous healing of a nonunion following nail removal and implantation of a modular revision stem without addressing the nonunion surgically at all. Our findings are in line with those reported by other authors. Although open reduction was defined as a risk factor for subtrochanteric nonunion by some authors [19, 39], there is increasing evidence that the mechanical advantages of preventing varus malalignment and restoration of the medial cortical support outweigh the biological disadvantages of open reduction [8, 10, 29]. Closed reduction is, therefore, advisable only, if it does not forfeit these mechanical advantages [12]. The same applies for the use of cerclage wires. The periostal vascular supply in the subtrochanteric region is circumferential [2, 11, 14, 19]. It has been shown that the vascular supply is preserved after using cerclage wires [2, 32, 40]. We therefore consider the application of subtrochanteric cerclage wires to be safe and valuable, if they facilitate anatomic fracture reduction and stabilization [10, 21]. Cerclage wiring also appears to be useful to reduce displaced large medial butterfly fragments (Fig. 1b, middle) to restore medial cortical support and to improve medial cortex healing. Furthermore, subtrochanteric fractures after long-term treatment with bisphosphonates have been described as prone to nonunion. Bisphosphonates inhibit osteoclastic bone remodelling and therefore lengthen the transformation of calcified callus to mature bone tissue [20, 41]. Indeed, several studies showed delayed bone union in these cases, but the nonunion rate was not increased [3, 4]. From a mechanical point of view, it is relatively simple to restore the medial cortical support using intramedullary nails, since these fractures are typically transverse fractures without comminution zones. Accordingly, prevention of varus malalignment represents the major surgical challenge in the surgical treatment of atypical subtrochanteric fractures.

Several limitations of our study should be noted. First, this is a retrospective study with all limitations associated with this study design. Second, increasing the number of included patients would have increased the power of the statistical analysis. Third, potential risk factors, such as smoking, diabetes and medication intake, were not assessed, as this study primarily focused on potential risk factors that may be influenced by the surgeon (e.g., postoperative reduction and alignment, open vs. closed reduction). Fourth, and as expected, the mean age within this consecutive series of patients was relatively high (70.2 ± 16.6 years, range 23–96 years) and about three quarters of fractures resulted from low-energy trauma (Table 1). As a consequence, the findings of this study might not be generalisable to high-energy subtrochanteric fractures in young patients. Fifth, X-rays were used for the measurement of angulation and displacement as well as for the assessment of osteoporosis. Finally, clinical outcome as well as revision strategies for the nonunion cases were beyond the topic of this manuscript.

In conclusion, the results of this retrospective study indicate that prevention of postoperative varus malalignment and restoration of the medial cortical support are the most critical factors to prevent nonunion after intramedullary nailing of subtrochanteric femoral fractures. Moreover, autodynamisation of the nail in the early postoperative period was found to be a strong predictor for failure and thus should result in revision surgery. All of these three significant risk factors identified are mechanical risk factors. In contrast, none of the investigated biological parameters, such as open reduction, did have a significant effect on the nonunion rate in this patient series. Careful soft tissue dissection, however, is deemed mandatory during open reduction.