Critical factors in cut-out complication after gamma nail treatment of proximal femoral fractures

Results were tabulated and statistically analysed by using the IBM SPSS (version 17, SPSS Inc. Chicago, Illinois, USA). The SPSS database was uniquely designed for our patient population and the questions adapted with the help of patient notes samples before starting the study. Frequency and relative distribution were presented in tabular form for categorical variables. Comparative analysis was performed with the Chi-Square or Fisher Exact (for 2x2 tables) test for categorical variables. Before analysing continuous variables, the data sets were assessed for normality by performing the Shapiro-Wilk test. When distribution was considered to be normal, two-sided Student’s t-test was performed; otherwise the Mann–Whitney test was used (confidence level for all tests = 95%). P-value <0.05 was considered statistically significant.

Cut-out was defined as being primary when the reason for this failure was of biomechanical origin, i.e. it was determined by fracture geometry before and after fracture reduction and by the lag screw positioning.

Fifty-seven patients (median age 82.6 years, 12/45 males/females) were identified. 97% of the factures were caused by low-energy trauma and two fractures by high-energy trauma. Six patients had an associated injury.

In 45 cases (79%) cut-out occurred within first 12 weeks after surgery (range 8 to 670 days). Twenty-one patients (37%) received no surgical treatment of this complication due to advanced age, major medical co-morbidities and low functional demands. Thirty-six patients underwent revision procedure. In 13 cases a total hip replacement and in four cases unipolar hip replacement were performed. The nail was exchanged in seven patients. Removal of the nail after consolidation was carried out in seven patients. Five patients, for whom the surgical treatment was planned, were lost to follow up.

A number of different configurations of cut-out were observed in relation to the primary position of the lag screw, its migration and approximate penetration point in the femoral head (Figure 2).

We observed that the majority of the lag screws migrated anteriorly-superiorly relatively to its intraoperative position. Only three lag screws migrated posteriorly (Figure 2). Central cut-out (along the lag screw axis) occurred in eight patients. In six cases the lag screw was prevented from sufficient lateral sliding. In two cases the lag screw migrated medially relative to the nail. In another six cases, the assessment of the lag screw migration was not possible because of insufficient quality of radiographs.

Cut-out frequency varied slightly over the twelve years. After introduction of the Trochanteric Gamma Nail in 1997, the cut-out rate fell from 2.5% to 1.1% (p=0.031) (Figure 3). The distribution of the fracture types did not differ over the years for the cut-out cases.

An unstable and complex fracture pattern, non-anatomical reduction and non-optimal positioning of the lag screw were found to be critical factors contributing to the cut-out complication. The interrelations between these factors are presented in Figure 7. All cut-out cases but two had at least one of the mentioned features. 31 patients presented a typical fracture type. These were basicervical (AO 31-B2.1) or trochanteric with comminute “pantrochanteric” fracture pattern (AO 31-A3.3). In 44 cases reduction of the fracture was assessed as non-anatomical. The lag screw was malpositioned (outside of the zones 3,6,9, Figure 6) in 42 patients. The combination of all three factors or the combination of non-anatomical reduction and non-optimal lag screw positioning was found strongly predictive for the cut-out. There were two patients without any contributing factors: a 78-year-old woman and an 80-year-old man. These patients had fractures classified respectively as AO/ASIF 31-A1.2 and 31-A2.1 with corresponding cut-outs occurring at 7 and at 8 months after surgery. Whilst they were assessed to have had stable fractures primarily, they both had non-unions at the base of the neck at the time of cut-out failure, suggesting that they had been incorrectly classified initially.

In previous reports, cut-out has been evaluated on two-dimensional radiographs, which show varus collapse of the femoral head and superior cut-out of the lag screw [1, 13]. However, in vivo telemetry study of hip implants shows surprisingly large rotational moments acting on the femoral head during gait caused by the A-P force [14, 15]. The authors suggest that these forces should be taken into consideration in the implant design process. Further, recent biomechanical studies direct attention to a multiplanar mechanism of cut-out, due to combined axial loads and rotational moments acting during walking gait [16, 17]. Under these conditions the mechanism of cut-out exhibits combined varus collapse and backward rotation provided that the lag screw have been placed in the central position in the femoral head [17]. In the present study, the cranial migration of the lag screw and varus collapse of the femoral head were accompanied by predominantly anterior movement. This observation supports the findings in the studies mentioned above. Central cut-out (along the lag screw axis) was associated with the failure of the sliding mechanism and simultaneous instability of the fracture. This can be explained by the inaccurate use of the set screw either by over-tightening preventing the lag screw from sliding or by failure to engage the set screw in the lag screw notch, thus allowing uncontrolled lag screw rotation and central migration toward acetabulum. This cut-out pattern has already been described in the literature [18].

The optimal position of the lag screw has been widely discussed in the literature, particularly the aspect of central [11, 19, 20] or inferior [21‐25] placement of the screw in the femoral head as seen on the A-P view .

The importance of a central placement of the screw on the lateral radiograph has been emphasised in the literature [3, 12, 26]. Central placement of the screw reduces the risk of rotation of the femoral head and neck around the screw (small torsional moment) that can occur with eccentric placement [27, 28].

The present study emphasizes the importance of placing the lag screw in the centre of the femoral head on the lateral radiograph. However, it was not possible to define a single optimal zone (inferior, central or even slightly superior) on the A-P view. This finding correlates precisely with the statement of Davis [3] that the cut-out rate was not significantly affected by either a superior or an inferior placement as seen on the A-P view if the implants were centrally positioned on the lateral radiograph. This indirectly supports the concept of tip-apex distance [1]: the placement of the lag screw close to the apex of the femoral head on A-P view with the central placement on the lateral view is essential for the outcome.

The majority of the cut-out cases displayed non-anatomical reduction (44/57) and a number of studies show a statistically significant relationship between non-anatomical reduction and cut-out complication [29‐31]. However, in the comparison with the matching controls in our study, no clear difference could be shown perhaps due to limited sample size. Nevertheless, the basicervical fractures with cut-out had a slight overrepresentation of non-anatomical reduction (p=0.089). A possible explanation for the poor reduction could be the complexity of these fractures (AO/ASIF 31-A3.3).

In the present study, the cut-out rate for AO/ASIF 31- A3.3 fracture was 6.5%. The complication rate for basicervical fracture was even higher - 9%, while overall cut-out rate was 1.0% (AO/ASIF 31- A3.3 and B2.1 excluded).

The fracture type (its complexity and concomitant stability) has been recognised as an important factor contributing to the osteosynthesis failure [3, 32, 33]. The treatment of fractures that have two fragments is usually associated with fewer complications [30, 34]. The increased rate of mechanical failure in complex inter-trochanteric fractures with subtrochanteric or cervical extension has been described in the literature. Haidukiewych [32] reported on high cut-out rate of 12.7% in 47 reverse obliquity fractures (AO/ASIF 31-A3.1 and A3.3) regardless of the type of internal fixation devices. The author also observed worse results for fractures with poor reduction or poor implant position in the femoral head. The treatment of this type of fracture with a Gamma Nail can also result in a high risk of cut-out regardless of the lag screw location [26, 35].

Only few reports focus on basicervical fractures as a separate entity [36, 37], these demonstrate a high incidence of local complications including cut-out leading to re-operations.

Clinical studies have consistently failed to find statistically significant differences between implant designs with regard to lag screw cut-out [38‐40].

In the present study, cut-out rates decreased significantly when the SGN was replaced by TGN (2.7% vs. 1.1%, p= 0.006). This might have been caused by the improved design of the second-generation nail: reduced valgus bend of 4°, reduced length of 180 mm and only one distal locking hole (the lag screw itself was not modified). These design changes eliminated the concept of 3-point contact of the nail in the femoral shaft [4] and subsequently could have optimised the positioning of the lag screw in the femoral head. The influence of the improved implant design on the decreasing complication rate in hip fractures has been recently shown by Bhandari et al. [41]. The meta-analysis suggests that newer intramedullary nail designs have reduced the risk of femoral shaft fracture.

The influence of the neck-shaft angle of an implant on cut-out has been controversial. The clinical study by Kukla [4] showed a significant increase in cut-outs in higher angle implants. On the contrary, biomechanical studies show that higher angles, in order to enhance sliding of the screw and fracture site impaction, result in less cut-out [42‐45]. In the present study, the analysis of the CCD-nail angle did not reveal any difference for this parameter.

The present study supports the finding that age does not determine the cut-out rate [3]. There was no statistically significant age difference between the non-cut-out and the cut-out group, although this complication did not occur in patients younger than 50 years. On the other hand, there are clinical studies where increasing age of the patient is found to be predictive of implant failure [1, 29, 46].

The opinion that poor bone quality increases the mechanical failure rate of an osteosynthesis is widely represented [30, 44, 47]. However, some authors have shown with help of Singh index or quantitative computed tomography (QCT) that cut-out is not influenced by this factor [3, 26, 46]. In the present study the bone quality was not assessed because of lack of reliable methods, however the patients with cut-out complication were considered to have osteoporotic bone based on age, gender and the presence of the low-energy fracture [13, 48, 49]. We were unable to identify the osteoporosis as a contributing factor and all but two cut-outs could be explained by the presence of other factors (Figure 7).

The review of large consecutive series of proximal femoral fractures enables statistically supported statements. On the other hand, the retrospective nature leads to some limitations in interpretation caused by loss to follow-up, non-standardised methods of data collection or inconsistent quality of radiographs. However, the prospective record of all Gamma Nails kept at the hospital since the introduction of this implant in the late 80’s allowed demographic and technical intra-operative data to be completed.

We therefore believe that the results and particularly the cut-out rate in the present study are reliable. This complication is associated with an important pain and reduction of function, forcing the patient to look for help at the CTO.