Buttery Fragment in the Femoral Shaft Fractures Treated by Intramedullary Nail: Do We Need to Reduce It or Not?

Background: Limited data is available regarding the effect of gap size, fragment size and position of fragment on union in fragmented femur shaft fracture. The aim of the study is to analyze the effect of these parameters on union and determine cut-off values that necessitate reduction of fragment. Methods: Total of 59 patients with fragmented femur shaft fracture were reviewed and the effect of fragment size, gap size and fragment position on union were evaluated on 3 rd , 6 th and 12 th month graphies from medical record of patients. Union rate was determined due to Radiographic Union Score of Femur. Patients were divided into small gap ( (cid:0) 10 mm) or large gap ( (cid:0) 10 mm), reverse fragment or non-reverse fragment and small fragment ( (cid:0) 50 mm) or large fragment ( (cid:0) 50mm) groups. Kruskal–Wallis Variance Analysis and Mann Whitney U test was used for comparisons among groups. The post-hoc Mann Whitney U-Test with Bonferroni Correction was used when the Kruskal Wallis Variance Analysis determined a signicant difference. Results: In comparison of union scores of groups in regarding gap size and fragment size, there was not signicant difference between groups at 3rd, 6th and 12th months. Union scores in the 3 rd month (p=0,011) and 6 th month (p=0,039) were lower in the reverse group than non-reverse group. But there was not statistically signicant difference between two groups at 12 th month (p=0.819). There was no signicant difference in union rates in respect to intramedullary nail types. There was an indirect correlation between age and mean union score at 12 th month. Conclusions: we could not determine any cutoff value regarding gap width and fragment size in the treatment of the femoral shaft fractures by IMN. Although there was a tendency to decrease in union scores due to the increase in the gap size, any signicant difference was not determined. Only the reverse position of fragment had an adverse effect on the union until 6 th months but the union score of reverse group reached the result of the non-reverse group by 12 th months.

Results: In comparison of union scores of groups in regarding gap size and fragment size, there was not signi cant difference between groups at 3rd, 6th and 12th months. Union scores in the 3 rd month (p=0,011) and 6 th month (p=0,039) were lower in the reverse group than non-reverse group. But there was not statistically signi cant difference between two groups at 12 th month (p=0.819). There was no signi cant difference in union rates in respect to intramedullary nail types. There was an indirect correlation between age and mean union score at 12 th month.
Conclusions: we could not determine any cutoff value regarding gap width and fragment size in the treatment of the femoral shaft fractures by IMN. Although there was a tendency to decrease in union scores due to the increase in the gap size, any signi cant difference was not determined. Only the reverse position of fragment had an adverse effect on the union until 6 th months but the union score of reverse group reached the result of the non-reverse group by 12 th months.

Background
The incidence of femoral shaft fractures has been reported as 1-1.3 fractures per 10,000 people per year.
While it usually occurs due to high-energy injuries in the young population, it may occur due to low-energy injuries in elderly [1]. The most common causes of femoral shaft fracture were pedestrian accidents (37%), falls (24%) and motorcycle accidents (15%) [2]. Accepted treatment modality for the femoral shaft fracture is the intramedullary nailing performed with minimal soft tissue disruption however, some unclear points still exist [3]. The management of the butter y fragment is one of these challenging problems while treating femur shaft fractures by intramedullary nailing. In a morphologic study, Salminen reported the incidence of femur shaft fracture types classi ed according to AO/OTA (Arbeitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association) classi cation system as follows: 48% Type A, 39% Type B, and 13% Type C. He also reported that AO 32 B2 subgroup with a wedge was the second most common femur shaft fracture subtype [5,6]. It means that this is a quite often fracture pattern. In the literature, but, there is little evidence to use in the management of the butter y fragment. Winquist proposed that it was reasonable to perform an open reduction and cerclage wiring of these fractures to gain anatomic restoration in young person [7]. Lin reported that greater than 1 cm displacement of fragment after nailing had adversely affected the bone healing and also proposed that an unreduced reversed fragment indicated poor prognosis [8]. Lee reported that the degree of fragment displacement had a greater effect on bone union than did the fragment size [9]. He suggested that any attempt such as reinforcement with percutaneous wiring might be required during the initial surgery. Some other authors also reported that the presence of a post-surgical fracture gap was also an important factor that might deteriorate the fracture healing [10,11,12]. On the other hand, open reduction is a risk factor for delayed union, non-union or infection.
In this study, we aimed to determine the relationship between gap size, fragment size, position of the fragment with the union in the AO/OTA 32B and AO/OTA 32C type femur shaft fractures and compared the results with the literature.

Methods
Total of 59 patients with AO/OTA type 32-B and 32-C type femoral shaft fractures treated by intramedullary nailing from 2014 to 2017 were included in the study. Exclusion criteria were: Periprosthetic fracture, reoperation by intramedullary nail for nonunion after plate xation, pathological fracture, subtrochanteric fracture, fracture extending to the knee joint and the fracture with follow-up period less than 12 months.
The area accepted as the femur shaft in the study is the region extending from trochanter minor distally to the level of eight cm proximal to the adductor tubercle. Fractures were classi ed according to AO Classi cation System [6] and Winquist and Hansen (W-H) Classi cation System [7]. IM Nail type and entry point were recorded as locked antegrade trochanteric, locked antegrade piriformis or locked retrograde nails. Open fracture was classi ed according to Gustilo-Anderson (G-A) Classi cation [13]. In all cases, closed reduction techniques were applied. Fragment displacement was determined by measuring the distance from the farthest point of the fragment perpendicular to the cortex of the femoral shaft on AP and lateral radiographies ( Figure 1). The gap distance was grouped as ''large gap'' or ''small gap''. If the gap distance was 10 mm it was classi ed as small gap. If the gap distance was 10 mm it was classi ed as large gap. The position of the butter y fragment described as follows: reverse or non-reverse, if the butter y fragment was on the cortex to cortex position on AP or lateral radiographies, ıt was described as reverse. If not, it was described as non-reverse. Size of butter y fragments were divided into two groups: large fragment group and small fragment group. It was called large fragment when the size of the butter y fragment was larger than 50 mm. When the size of the butter y fragment was smaller than 50 mm it was called small fragment ( Figure 2, 3).
Anteroposterior and lateral radiographies at 3 rd , 6 th and 12 th months were evaluated for union rate.
Evaluation of union rate was made due to Radiographic Union Score of Femur (RUSF) described by Lin [8]. This scoring system is a modi cation of the tibia union scoring system reported by Whelan [14].
The mean union time, re-operation, implant failure and other complication rates were recorded for each group. Nonunion was de ned as ''persistent pain at the fracture site combined with the absence of progressive healing on 6 th month radiography'' [15]. Delayed union is an ununited fracture which continues to progress toward healing or that has not been present for long enough to satisfy an arbitrary time standard for nonunion [16].
Statistical Analysis: Continuous variables were expressed as mean ± standard deviation (SD), median (minimum-maximum values) and categorical variables as number and percent. Shapiro-Wilk tests were used for testing normality. Kruskal-Wallis Variance Analysis and Mann Whitney U test was used for comparisons among groups. The post-hoc Mann Whitney U-Test with Bonferroni Correction was used when the Kruskal Wallis Variance Analysis determined a signi cant difference. In addition, the relationships between continuous variables were analyzed by Spearman correlation analysis. All statistical analyses analyzed by SPSS, 24.0 and p-value less than 0.05 was considered statistically signi cant.

Results
The average age of the patients was 31, 44 years (range 16 to 61 years). Nine of 59 patients were female (15.3%) and 50 of 59 patients were male (84.7%). 34 out of 59 patients had been operated for right femur fractures (57.6%), 21 out of 59 patients (35.6%) had been operated for left femur fractures and 4 patients had been operated for bilateral femoral fractures (6.8%). Patients being operated for bilateral femoral fractures were included only for one side because the other side was inappropriate for our study ( xation by plate or external xator) ( Average gap size in large gap group and small gap group were 23.4 mm (range 11 to 62 mm) and 7.05 mm (range 4 to 10 mm) respectively ( Table 2).
Mean union score of 59 patients at 3 th month was 7.68 (range 4 to 12), mean union score at 6 th month for 58 patients was 10.09 (range 4 to 12) and mean union score at the end of 12 th month for 49 patients were 11.84 (range 9 to 12) according to RUSF scale ( Table 2).
There were two implant failures, three non-union and all had been treated by exchange nailing. There were ve delayed union. One delayed union case was treated by plating+ grafting (Figure 4), one delayed union was treated by repositioning and cerclage wiring of the fragment and three delayed unions was treated by dynamization between six to 12 months. Ten of the 59 patients had re-operation. Four out of ten patients were in non-reverse group and six were in reverse group. But there was not statistically signi cant difference between two groups in terms of reoperation (p=0.144).
We determined implant failure in two patients. Both were in the reverse group. There isn't statistically signi cance between reverse and non-reverse groups related to implant failure (p=0,144). However, we thought that this was due to the low number of implant failure.
The union scores at the 3 rd month (p=0,011) and 6 th month (p=0,039) were lower in reverse group than the non-reverse group and it was statistically signi cant. However, when the union scores of two groups at the 12 th month were compared, there was no statistically signi cant difference (p=0.819) ( Table 3).
In comparison of union rates in related to gap size and fragment size, there wasn't any statistically signi cant difference between groups at 3rd, 6th and 12th months ( Table 4, 5). Although two patients could not get the perfect healing RUSF score at 12th month follow-up period. They don't have pain during weight-bearing. We didn't plan any intervention. When the union rates were compared in related to fracture type classi ed according to AO Classi cation and W-H Classi cation, we didn't' nd any statistically signi cant difference between groups at 3rd, 6th and 12th months ( Table 6,7).
Mean RUSF score of 31 isolated femur fractures (8 ± 1.15) was higher than that of 24 cases with additional extremity trauma (6.96 ± 1.81). It was statistically signi cant (p=0.048). In comparison of union rates in related to IMN type, there was no statistically signi cance difference. Evaluation of continuous variables by Spearman Correlation Analysis has shown that there was a direct correlation between the fragment size and age. In addition, there was an indirect correlation between age and mean RUSF score at 12 th month (Table 8).

Discussion
Although high success rates have been reported for the treatment of femoral shaft fractures by closed reduction and intramedullary nailing, nonunion is still a problem [17,18]. Patient with femoral nonunion is faced with signi cant functional and economic problems, including persistent disability, gait abnormality, and prolonged physical and psychological disability [19,20]. The presence of a gap at the fracture site has been reported as one of the 4 main factors resulting in delayed union or nonunion [19][20][21]. In AO/OTA 32 B or 32C type fracture a residual gap is usually persistent after closed reduction which may result in nonunion. However, there is limited data in the literature about the effects of the gap size and the butter y fragment size on the fracture healing. Lin reported that a gap size greater than 10 mm had an adverse effect on fracture healing [8]. He proposed that excessive motion resulting from larger gap might lead to poor callus formation and also presence of a reversed fragment resulted in unfavorable outcome that need to reoperations. On the contrary, there was not any signi cant difference between small and large gap groups in present study, even in the early healing period. We thought that this might have resulted from the difference in fracture subtypes. While W-H type 3 and type 4 fractures account for 31.25% of total cases in Lin's study, it was 66.1% in our study. High rates of more fragmented fractures and fractures with longer fracture line might have resulted in this difference. In the article explaining the basics of fracture healing, Perren has reported that "the repair cell does not see the amount of fragment movement nor the width of the gap but senses its own deformation, called strain" [22]. This means that even the gap size is wider, depending on the amount of movement, strain may be less than the limit of strain tolerance and vice versa. So more fragmentation and longer fracture line might reduce the strain and keep it within strain tolerance as in bridging plate technique.
When the effect of the IM nail type on the union have been evaluated there was no signi cant difference.
There was a direct correlation between the fragment size and the age and an indirect correlation between age and mean RUSF score in 12 th month.
Evaluation of the effect of the fragment position on healing has shown that the union scores at 3 rd month (p=0,011) and 6 th month (p=0,039) were signi cantly lower in the reverse group than the non-reverse group but there was no statistically signi cant difference at 12 th month (p=0.819). Lin also didn't nd any correlation between the mean union scores of reverse and non-reverse fragment groups at 3 rd , 6 th and 12 th months [8]. But he still suggested the percutaneous and minimally invasive reduction techniques because of increased risk of reoperation. In terms of clinical practice, ıt may be considered that in the presence of a reverse fragment, delayed weight-bearing (after 6 th month) will be more reliable to prevent implant failure according to our results. There were two implant failures in our study and both were in reverse group. Although It was not statistically signi cant (p=0.123), we thought that it was an underestimation due to the small size of the group.
In our study, we didn't nd any effect of fragment size on the median RUSF score on the 12 th month. It was 12 and 12 respectively for the large fragment group and small fragment group (p = 0.588). In their study, Lee et al. [9] reported that fractures with fragments 8 cm or longer might have more negative outcomes when intramedullary nailing was performed. However, they also underlined that as fragments increased in size, the degree of displacement also tended to increase, whereas the union rate did not decrease as much and so the degree of fragment displacement has a greater effect on bone union than does fragment size. When they compared the union rates according to the fragment size and degree of fragment displacement by logistic regression analysis, no signi cant effect of fragment size on union has been observed. Vicenti et al. [23] reported a prognostic algorithm about how to manage the third fragment of the femoral shaft fractures treated by intramedullary nail and determined 40 mm fragment size as the cut-off value and the leading parameter that in uences the fracture healing in rst six months or more. They also reported that the amount of third fragment displacement may result in delayed union.

Conclusions
We could not determine any cutoff value regarding gap width, fragment size or position of fragment in the treatment of the femoral shaft fractures by IMN. Our results do not match the literature information in all aspects. The only suggestion of this study is that delayed weight bearing after 6 months will be bene cial to prevent implant failure in the reverse fragment fractures. We think that larger series are needed to determine precisely the effect of the gap width, fragment size on the union.

Competing interests
The authors declare no con ict of interest.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors

Figure 3
Twenty-three years old patient. post-operative x-rays; a,b: the butter y fragment; reverse, large gap and large fragment. c and d shows perfect RUSF healing score at the end of 12th month.