During the last two decades, bisphosphonates (BPs) have been the most commonly prescribed anti-resorptive agents for the prevention of hip and vertebral fractures in osteoporotic patients. However, prolonged BPs therapy was reported to be associated with atypical femoral fractures (AFFs) in subtrochanteric and diaphyseal area.1,2,3)

The task force of the American Society for Bone and Mineral Research (ASBMR) reported that a causal relationship between BPs and AFFs had not been established.4) However, recent reports have suggested that prolonged BPs therapy may severely suppress bone turnover and impair bone remodeling, eventually leading to accumulation of micro-damage, which results in AFFs.5,6,7,8) The updated case definition of AFFs is a fracture located along the femoral diaphysis from just distal to the lesser trochanter to just proximal to the supracondylar flare. These insufficiency fractures can be seen on plain radiographs as simple transverse patterns with unicortical beaking or flaring and hypertrophy of the lateral cortex.4,9)

There is no established management protocol with strong evidence for the AFFs.10) However, a number of reports suggest a multidisciplinary approach, including discontinuation of BPs, adequate consumption of calcium and vitamin D, and consideration of teriparatide, a recombinant form of parathyroid hormone.11) Banffy et al.12) and Egol et al.13)suggested that non-operative management of AFFs has a high likelihood of failure with the majority of impending fractures progressing to complete fractures. However, surgical treatment of AFFs is associated with a high rate of intraoperative and post-operative complications. Compared to the traditional femoral fractures, many literatures report that AFFs are more frequently associated with delayed union, nonunion, implant failure and reoperations.

We hypothesized that endosteal thickening of the lateral cortex, can interfere with intramedullary reaming and nailing. It consequently results in the varus reduction of the fracture, resulting in postoperative complications such as delayed union and nonunion. Therefore, we assessed whether endosteal thickening of the lateral cortex affects fracture reduction and surgical outcomes in subtrochanteric AFFs treated with intramedullary devices.

There were 8 cases in group I and 19 cases in group II. Baseline characteristics such as age, height, weight, BMI, BMD, duration of BPs, presence of BPs holiday, prodromal symptoms and follow-up period did not differ significantly between the two groups (Table 2).

Table 2
Baseline Characteristics Included in This Study

• Download Table

Implant type (p=0.655) and length (p=0.558) did not differ significantly between the two groups. The coronal plane femoral bowing angle was 3.2°±8.2° in group I and 2.9°±4.6° in group II (p=0.933). The sagittal plane femoral bowing angle was 11.4°±4.6° in group I and 10.4°±3.3° in group II (p=0.570). Varus reduction occurred in total 16 cases, with 1 patient in group I and 15 cases in group II. Varus reduction occurred significantly more in group II (Group I: 12.5% vs. Group II: 78.9%; p<0.001), and they all showed more than 10° at the fracture site. Nonunion occurred in total 9 cases, all of them were group II and this result statistically significant (Group I: 0% vs. Group II: 47.4%; p=0.017). All patients with nonunion were continuously observed in postoperative radiographs. Revision surgery was performed on all the patients. Delayed union occurred in total of 7 group II patients, and this result also statistically significant (Group I: 0% vs. Group II: 70.0%; p=0.003). The mean union time was observed to be 5.5±1.1 months in group I and 8.3±1.6 months in group II. The bone healing period was 2.8 months longer on average in group II and it was statistically significant (p<0.001) (Table 3).

Table 3
Intra- and Postoperative Variables

• Download Table

Univariate analyses were used to assess the individual effects of nonunion in all patients. A total of nine nonunions occurred (33.3%). Age (p=0.231), BMI (p=0.153), BMD (p=0.143), ASA classification (p=0.532), duration of BPs (p=0.854), presence of BPs holiday (p=0.675), prodromal symptoms (p=0.551), implant type (p=0.865), implant length (p=0.550), and femur bowing (coronal plane; p=0.626, sagittal plane; p=0.925) did not have statistically significant associations. However, statistically significant differences were observed for endosteal cortical thickening (p=0.026) (Table 4).

Table 4
Univariate Comparison between the Union and Nonunion Groups

• Download Table

Multiple logistic regression analyses performed to identify independent factors associated with nonunion with variables found that endosteal cortical thickening was the independent predictor of nonunion (odds ratio: 7.198; 95% confidence interval: 1.718–30.154; p=0.007) (Table 5).

Table 5
Logistic Regression Analysis of Nonunion

• Download Table

Cox proportional hazard model shows the difference in survival curves between two groups for the nonunion (p=0.025) (Fig. 5).

Fig. 5
Cox proportional hazard survival curve (solid line: Group I, dotted line: Group II).

• Download Figure

Increased bone turnover and insufficient bone healing due to long-term BP use are the keys to understanding the pathogenesis of AFFs.3) Repeated micro-damage to the lateral aspect of the femur, where withstands the maximal tensile stress, also plays an important role. In a physiologic femur model, Koch demonstrated how the highest tensile stress falls upon on the lateral cortex in the subtrochanteric region of femur.22) These features of the subtrochanteric AFFs cause poor union rates and numerous postoperative complications.

The average bone union time of subtrochanteric AFFs is reported to be more than 8 months,4,23,24) longer than that of the typical femoral fractures at an average of 3 to 6 months.25,26) Prasarn et al.27) has analyzed factors affecting healing time in subtrochanteric AFFs and demonstrated a 44% fracture healing complication rates in his series of subtrochanteric AFFs, but most of those complications were treated with plate fixation. In addition, a systematic review by Koh et al.2) has demonstrated that there was a significantly greater need of revision surgery in plate fixation group over intramedullary nailing group (31.3% vs. 12.9%). The current surgical intervention of choice that allows endochondral ossification for AFFs is intramedullary nailing.10,11) However, no specific intramedullary device was advocated in current literature for the treatment of subtrochanteric AFFs. In our study, we used three models of intramedullary nails, which fixation the femoral head with single blade, single lag screw, and two lag screws. Spinelli et al.28) has demonstrated that in patients with osteoporosis and a high risk of fracture in femur neck and intertrochanteric area, it may be advantageous to use a cephalomedullary nail for prevention.

Cephalo-medullary nails allow for static and dynamic locking options on distal screws. Dynamically locked screws have the advantage of providing rotational stability while allowing for axial compression of the fracture. The dynamic screw position can also be used to enhance the formation of the callus if the fracture gap remains after nail insertion. However, it has been proved that because the process of callus formation takes longer for AFFs, it may be safer to select a static screw position. No large-scale studies have been conducted to demonstrate superiority between distal locking options, and there is no univocal consensus.29,30)

Yang et al.31) have analyzed other factors that can potentially affect healing time in AFFs. Subtrochanteric AFFs commonly have disproportionate medullary space and intramedullary device size when compared to AFFs that occur in the femoral diaphysis. This can aggravate rotational instability, and subtrochanteric AFFs are at a higher risk of delayed union and nonunion. In our study, the total nonunion rate was 33.3%, similar to the reported rate of revision surgery due to nonunion or implants failure in various studies.32,33,34) Subtrochanteric AFFs tend to show much higher rates of nonunion compared to two percent reported in typical femoral fractures.35)

Another possible variable for predicting a fixation failure of the subtrochanteric AFFs is the reduction status. Egol et al.13) reported that varus reduction of the fracture site required an average of 3.7 months or more for bone union compared to that of acceptable reduction. In our study, the radiological union time for group I was average of 5.5 months, and for group II was 8.3 months, indicating similarity to that of the previous study. In addition, considering that the varus reduction rate was significantly higher in group II, it is logical to assume that the long union period in group II was directly related to varus reduction. Lim et al.36) also reported the importance of acceptable reduction in the healing of AFFs. The reduction status on the coronal plane and the residual gap in the fracture area were important factors causing delayed union and nonunion. They concluded that the remaining gap of more than 20% of the cortical thickness on the anterior and lateral sides of the fracture could cause problems with the bone union.

As demonstrated in this study, endosteal thickening of the lateral cortex in subtrochanteric AFFs is a factor that interferes with acceptable reduction and negatively affects surgical outcomes. However, there are not many studies that try to identify the factors affecting cortical thickness.32,33) It has been postulated that cortical thickening is due to long-term BPs use, but the study of Lenart et al.37) reported that cortical thickening was observed in AFF patients who had never used BPs. In addition, Giusti et al.38) demonstrated that cortical thickness is not different in AFF patients with or without exposure to BP therapy and that cortical thickness does not increase with prolonged use of BPs.

There was no mention in the literature about the direction of thickening the cortex. There have also been no previous studies on factors affecting endosteal cortical thickening. However, our study results show that the direction of cortical thickness is closely related to the fracture reduction and surgical results. The interesting fact is that one of the patients in this study had a different direction in cortex thickening between the two sides (Fig. 6).

Fig. 6
Thickening of the lateral cortex in subtrochanteric regions is observed on both sides. On the right side, both endosteal and periosteal cortical thickening are observed (white arrows), whereas only periosteal cortical thickening is observed on the left side (white arrowhead).

• Download Figure

There are several studies that have investigated on factors that promote fracture union and improve the results of surgery in subtrochanteric AFFs. The ASBMR recommended that the medullary canal should be over-reamed to a minimum diameter of at least 2.5 mm larger than the intramedullary nail diameter in order to prevent varus reduction and compensate for the narrow medullary diameter.4) A retrospective study of Lovy et al.39) has reported that percutaneous injections of bone marrow aspirate concentrate and demineralized bone matrix on fracture sites can be used to avoid delayed union and nonunion. Another study by Yeh et al.40) demonstrated that teriparatide treatment in patients with subtrochanteric AFFs may help in fracture healing, hip function recovery, and pain relief.

The advantage of our research is that it is a single center research. All operations were performed by one skilled orthopedic surgeon. This will produce consistent results in surgery compared to studies involving multiple orthopedic surgeons in multi center research. It can also reduce the number of variables that can occur during the treatment period, thereby increasing the reliability of this study. Another advantage of our research is use of inferential statistical analysis to enhance the validity of the conclusions made.

However, there are some limitations that need to be acknowledged and addressed regarding this study. First, the main limitation of the current study is its retrospective design. There would be observer bias, including no standardized quality and duration of follow-up, missing data, and inability to control confounding variables. Second, the follow-up period is relatively short, leading to an inability to represent long-term outcomes of subtrochanteric AFFs. Third, due to the uncommon nature of these fractures, there is a relatively small study population. Due to such limitations, it was difficult to retain sufficient power for each variable used for analysis in this study; consequently, future study needs to compensate for such short comings to provide more concrete correlations.

This study suggests that femoral endosteal thickening of the lateral cortex in the fracture site is an obstacle in intramedullary nail fixation for subtrochanteric AFFs. Endosteal callus acts as a ‘malpositioned poller screw’, interrupts reduction and contributes to varus deformation during nail advancement. As a result, it is necessary to make every effort to correct this as it is an independent factor in increasing the delayed union and nonunion. The only strict acceptable reduction is the key to successful management of subtrochanteric AFFs. As such, intentionally excessive overreaming during operation is essential.