Abstract
Background:
The purpose of this study was to analyze factors that affect healing time after operative treatment of complete femoral fractures associated with long-term use of bisphosphonates. In particular, we sought to determine surgically controllable factors related to fracture-healing time.
Methods:
Ninety-nine consecutive patients (109 fractures) who had been surgically treated for a complete atypical femoral fracture were enrolled. All patients had a documented history of bisphosphonate therapy at the time of presentation, with an average duration of 7.4 ± 3.5 years (range, 3 to 20 years). Baseline demographic data, characteristics of the fracture and surgery, and radiographic findings including femoral neck-shaft angle, coronal and sagittal bowing of the femur, and thickness of the femoral cortex were examined. Univariate and multivariate logistic regression analyses were performed to identify predictive factors associated with delayed union or nonunion.
Results:
Of the 109 fractures, 76 (69.7%) showed osseous union within 6 months after the index surgery and were assigned to the successful healing group. The remaining 33 fractures (30.3%), which showed delayed union or nonunion, were assigned to the problematic healing group. There were differences in body mass index (BMI), bisphosphonate therapy duration, and the rate of prodromal symptoms between the 2 groups. Supra-isthmic fracture location, femoral bowing of ≥10° in the coronal plane, and a lateral/medial cortical thickness ratio of ≥1.4 were predictive of problematic healing but were uncontrollable factors. Iatrogenic cortical breakage around the fracture site as well as a ratio of ≥0.2 between the remaining gap and the cortical thickness on the anterior and lateral sides of the fracture site were controllable predictive factors associated with problematic healing.
Conclusions:
Intramedullary nailing without cortical breakage around the fracture site and decreasing the anterior and lateral fracture gaps (avoidance of distraction) as much as possible are recommended to reduce healing time in complete femoral fractures associated with long-term use of bisphosphonates.
Level of Evidence:
Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.
Bisphosphonates are widely used for treatment of patients who have, or are at elevated risk of developing, osteoporosis1. Their efficacy in reducing the risk of fragility fractures has been established in clinical trials2-4.
Although bisphosphonates prevent typical osteoporotic fractures, several studies have raised concern that these agents cause atypical femoral fractures5-7. As a result of inhibition of osteoclast-mediated bone resorption, bisphosphonates can cause accumulation of trabecular microdamage8. This may compromise the mechanical and regenerative properties of bone, resulting in fractures and delayed bone-healing9,10.
Several recent studies have addressed the operative treatment, complications, and healing rates of atypical femoral fractures11-16. Although the precise prognosis is still unknown, there is a growing consensus that the altered bone metabolism caused by long-term use of bisphosphonates would adversely affect bone-healing even after osteosynthesis12,15,17. Delayed or failed fracture-healing is becoming a major concern after fracture stabilization in patients taking bisphosphonates13.
It is crucial to determine which factors have a positive or negative impact on healing of atypical femoral fractures. This study was conducted to analyze factors that affect healing time in the operative treatment of complete femoral fractures associated with long-term use of bisphosphonates. In particular, we sought to determine operatively controllable factors related to fracture-healing time.
Materials and Methods
Subjects
A consecutive series of 130 patients (140 fractures) who had been operatively treated for a complete atypical femoral fracture from July 2006 to February 2014 at a single academic center (Samsung Changwon Hospital) were identified. Atypical femoral fractures were defined by characteristic radiographic features including a transverse or short oblique fracture line, medial spike, focal lateral cortical thickening, and no or minimal comminution, according to the criteria of a 2013 American Society for Bone and Mineral Research task force18.
All fractures were located along the femur from just distal to the lesser trochanter to just proximal to the supracondylar flare, and all were associated with long-term (≥3-year) use of bisphosphonates and a history of minimal or no trauma. Minimal trauma was defined as a slip or fall from a standing height or less19. Of these 130 patients, 18 patients (18 fractures) who had a history of fracture, infection, or surgery involving the affected femur or underwent fixation with a plate were excluded from this study, as were 5 patients (5 fractures) who had a history of corticosteroid use, hormonal treatment, rheumatoid arthritis, metabolic bone disease other than osteoporosis such as hyperparathyroidism, or paraneoplastic syndrome. No pathologic fractures associated with metastatic malignancies were found. Four patients (4 fractures) were lost to follow-up before osseous union, and 4 patients (4 fractures) were excluded as a result of insufficient data. The final cohort comprised 99 patients (109 fractures). The mean follow-up duration (and standard deviation) was 24.2 ± 8.4 months (range, 6 to 60 months). Patients who were lost or who died during the follow-up period are summarized in Table I. This study was approved by our institutional review board, and informed consent was waived for this retrospective study.
TABLE I.
Follow-up According to Time Interval
| Time Since Surgery (yr) | No. Entering Interval | Deaths | Losses to Follow-up | Withdrawals | Follow-up Rate (%) |
| 0.5 to 1 | 109 | 1 | 3 | 14 | 96.3 |
| >1 to 2 | 91 | 2 | 5 | 23 | 92.3 |
| >2 to 3 | 61 | 3 | 6 | 15 | 85.2 |
| >3 to 4 | 37 | 1 | 7 | 22 | 78.4 |
| >4 to 5 | 7 | 0 | 2 | 5 | 71.4 |
All patients had a documented history of bisphosphonate therapy at the time of presentation. Of the 99 patients, 28 (28.3%) had been prescribed anti-osteoporotic medication at our hospital and the other 71 (71.7%) had been prescribed such medication at other hospitals. Alendronate was prescribed to 43 patients (43.4%); risedronate, to 27 (27.3%); ibandronate, to 11 (11.1%); pamidronate, to 3 (3.0%); and zoledronate, to 3 (3.0%). Two or more of these drugs had been sequentially used in 12 patients (12.1%). Six patients had taken a drug holiday from bisphosphonates for at least 1 year. The average duration of bisphosphonate treatment was 7.4 ± 3.5 years (range, 3 to 20 years). Long-term use was defined as ≥3 years in accordance with a prior study16. Prodromal symptoms had developed in 31 (28.4%) of the 109 fractures, and the mean duration of symptoms was 7.0 ± 8.5 months (range, 3 to 18 months). All patients discontinued bisphosphonate medications at the time of admission to our hospital. Teriparatide was not used.
Data Collection and Radiographic Measurements
For each patient, we retrospectively determined detailed demographic data including age, sex, body mass index (BMI), bone mineral density (BMD), and residence type. Comorbidities and functional status20 of the study population are shown in Table II. Characteristics of the fracture and surgery (including injury mechanism, fracture location, and time interval between fracture and surgery) were obtained from our fracture database. Fracture location was categorized into 3 groups (supra-isthmic including subtrochanteric, isthmic, and infra-isthmic). All of the fractures were fixed with a statically locked intramedullary nail (diameter of the proximal portion, 12 mm).
TABLE II.
Baseline Patient Characteristics
| Successful Healing (N = 76) | Problematic Healing (N = 33) | P Value | |
| Age* (yr) | 78.00 (73.75, 83.25) | 78.00 (75.00, 85.00) | 0.889 |
| Female sex† | 76 (100%) | 33 (100%) | 1.000 |
| Affected side (right:left) | 56:20 | 14:19 | 0.001 |
| Body mass index* (kg/m2) | 20.50 (19.50, 21.59) | 21.48 (20.00, 23.56) | 0.024 |
| Femoral bone mineral density* (T-score) | −3.5 (−2.9, −3.8) | −3.4 (−2.7, −4.2) | 0.523 |
| Comorbidities† | |||
| Cardiovascular disease | 11 (14.5%) | 4 (12.1%) | 0.690 |
| Cerebrovascular disease | 8 (10.5%) | 3 (9.1%) | 0.738 |
| Chronic pulmonary disease | 9 (11.8%) | 4 (12.1%) | 0.868 |
| Chronic renal disease | 2 (2.6%) | 1 (3.0%) | 0.575 |
| Chronic liver disease | 3 (3.9%) | 1 (3.0%) | 0.461 |
| Diabetes mellitus | 17 (22.4%) | 8 (24.2%) | 0.775 |
| Smoking | 2 (2.6%) | 1 (3.0%) | 0.575 |
| Parkinsonism | 3 (3.9%) | 1 (3.0%) | 0.461 |
| Cognitive impairment | 10 (13.2%) | 4 (12.1%) | 0.893 |
| Charlson comorbidity index† | 0.645 | ||
| 1 or 2 | 34 (44.7%) | 15 (45.5%) | |
| 3 or 4 | 31 (40.8%) | 14 (42.4%) | |
| ≥5 | 11 (14.5%) | 4 (12.1%) | |
| ASA classification†‡ | 0.586 | ||
| I or II | 72 (94.7%) | 31 (93.9%) | |
| III or IV | 4 (5.3%) | 2 (6.1%) | |
| Koval score20† | 0.795 | ||
| 1 | 63 (82.9%) | 27 (81.8%) | |
| 2 or 3 | 13 (17.1%) | 6 (18.2%) | |
| Residency type† | 0.716 | ||
| With family | 51 (67.1%) | 26 (78.8%) | |
| Single | 15 (19.7%) | 4 (12.1%) | |
| Care facility | 10 (13.2%) | 3 (9.1%) | |
| Bisphosphonate medication duration* (yr) | 6.15 (4.20, 9.83) | 8.00 (4.90, 12.00) | 0.041 |
| Presence of prodromal symptoms† | 0.026 | ||
| Yes | 17 (22.4%) | 14 (42.4%) | |
| No | 59 (77.6%) | 19 (57.6%) | |
| Presence of drug holiday† | 0.586 | ||
| Yes | 4 (5.3%) | 2 (6.1%) | |
| No | 72 (94.7%) | 31 (93.9%) |
The values are given as the median with the interquartile range in parentheses.
The values are given as the number of fractures with the percentage in parentheses.
ASA = American Society of Anesthesiologists.
Fracture-healing was defined as osseous bridging of 3 or 4 cortices on anteroposterior and lateral radiographs21. Fractures that showed osseous union within 6 months after the index surgery were considered to have achieved successful healing, whereas fractures that did not were considered to have problematic healing.
Standing femoral anteroposterior radiographs with correct patellar alignment with no rotation and lateral radiographs with the medial and lateral condyles of the femur overlapping were selected for analysis. All radiographs were assessed independently by 1 orthopaedic surgeon and 1 independent musculoskeletal radiologist, both blinded to the patient information, twice within an interval of 1 week. Preoperatively, the femoral neck-shaft angle22, coronal and sagittal bowing of the femur, and thickness of the femoral cortex were evaluated. Femoral bowing was measured using the curvature of the contralateral femur as previously described19. In patients with bilateral fractures, bilateral radiographs made before the second fracture were available. The thickness of the femoral cortex was measured at 4 areas of the fracture site (medial, lateral, anterior, and posterior) on both anteroposterior and lateral views. With respect to the reduction status of the fractures, coronal alignment of the femur was defined as valgus, neutral, or varus, and sagittal alignment was defined as flexion, neutral, or extension. The remaining gap at the fracture site was measured postoperatively. The presence of iatrogenic cortical breakage around the fracture site during nail insertion was also investigated.
Statistical Analyses
Basic descriptive statistical analyses (SPSS version 18.0) were used to describe the study population. Values are reported as the median and interquartile range or as the number and percentage. The Mann-Whitney U test was used for continuous variables, and the Fisher exact test or Pearson chi-square test was used for categorical variables. Significance was defined as p ≤ 0.05. Univariate logistic regression was performed on each variable to determine differences between the 2 groups. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Variables with p ≤ 0.05 were incorporated into a forward, stepwise multivariate logistic regression to calculate adjusted ORs. Intraclass correlation coefficients were used to assess intrarater and interrater agreement with regard to continuous scales of radiographic measurements.
Results
Demographic Data and Clinical Course
Of the 109 fractures (in 99 patients) in the final cohort, 76 (69.7%) showed osseous union within 6 months after the index surgery and were assigned to the successful healing group. The remaining 33 fractures (30.3%), which showed delayed union or nonunion, were assigned to the problematic healing group. In the latter group, 27 fractures underwent delayed union at a mean of 10.6 months (range, 8 to 18 months). The other 6 fractures went on to oligotrophic nonunion. At the time of the final follow-up, 2 of these 6 fractures had undergone revision surgery (Fig. 1) and the 4 patients with the remaining fractures had declined revision. Iatrogenic cortical breakage occurred intraoperatively in 5 fractures during nail placement (Fig. 2).
Fig. 1.
Figs. 1-A and 1-B Initial radiographs of a 63-year-old woman showing an atypical fracture of the subtrochanteric region of the femur (supra-isthmic location). Figs. 1-C and 1-D The size of the remaining gap at the fracture site is larger on the lateral side than on the medial side as a result of varus malreduction. Figs. 1-E and 1-F The patient had nonunion even 18 months after surgery. Figs. 1-G and 1-H Complete union was achieved 12 months after revision surgery using an angled blade plate.
Fig. 2.
Figs. 2-A and 2-B Initial radiographs of a 67-year-old woman who had been treated with bisphosphonates for 12 years, showing a complete femoral fracture. Figs. 2-C and 2-D Lateral cortical breakage occurred during the index surgery. Figs. 2-E and 2-F The lateral osseous fragment was not healed 6 months after surgery. Figs. 2-G and 2-H Complete healing had occurred 14 months after surgery.
Demographic data for the successful and problematic healing groups are summarized in Table II. There were significant differences in the bisphosphonate medication duration (6.15 compared with 8.00 years, respectively; p = 0.041) and the rate of prodromal symptoms (17 of 76 compared with 14 of 33, p = 0.026) between the 2 groups.
Fracture and Surgery Characteristics
As shown in Table III, the distribution of fracture locations differed significantly between the 2 groups (p < 0.001).
TABLE III.
Fracture and Surgery Characteristics
| Successful Healing (N = 76) | Problematic Healing (N = 33) | P Value | |
| Injury mechanism* | 0.738 | ||
| Fall down | 8 (10.5%) | 3 (9.1%) | |
| Slip | 68 (89.5%) | 30 (90.9%) | |
| Fracture location* | <0.001 | ||
| Supra-isthmic (including subtrochanteric) | 4 (5.3%) | 16 (48.5%) | |
| Isthmic | 51 (67.1%) | 8 (24.2%) | |
| Infra-isthmic | 21 (27.6%) | 9 (27.3%) | |
| Fracture on the contralateral side* | 0.622 | ||
| Complete or incomplete | 21 (27.6%) | 7 (21.2%) | |
| None | 55 (72.4%) | 26 (78.8%) | |
| Time interval between fracture and surgery† (d) | 4.18 (2.32, 5.40) | 4.28 (2.45, 5.20) | 0.737 |
The values are given as the number of patients with the percentage in parentheses.
The values are given as the median with the interquartile range in parentheses.
Radiographic Parameters
Radiographic parameters before, during, and after surgery are shown in Table IV. Preoperative coronal bowing of the femur was significantly more severe in the problematic healing group (12.00°) compared with the successful healing group (8.00°, p = 0.003). The lateral/medial cortical thickness ratio showed significant differences between the 2 groups (1.50 compared with 1.30, p = 0.012), as did the anterior/posterior ratio (1.29 compared with 1.46, p = 0.040). In addition, the intraoperative reduction in the coronal plane differed significantly between the 2 groups (p = 0.015).
TABLE IV.
Radiographic Characteristics
| Successful Healing (N = 76) | Problematic Healing (N = 33) | P Value | |
| Preoperative | |||
| Femoral neck-shaft angle* (deg) | 132.00 (126.00, 137.70) | 132.00 (130.50, 134.70) | 0.522 |
| Coronal bowing of the femur* (deg) | 8.00 (2.00, 12.00) | 12.00 (5.00, 15.10) | 0.003 |
| Sagittal bowing of the femur* (deg) | 7.00 (3.90, 12.00) | 8.40 (6.00, 12.00) | 0.140 |
| Cortical thickness* (mm) | |||
| Anterior | 9.23 (8.30, 10.28) | 9.93 (7.96, 11.85) | 0.620 |
| Posterior | 7.13 (6.27, 7.68) | 7.58 (6.21, 8.82) | 0.327 |
| Medial | 5.41 (4.74, 6.18) | 5.05 (4.20, 6.01) | 0.179 |
| Lateral | 7.20 (6.08, 10.25) | 7.87 (7.03, 10.05) | 0.059 |
| Cortical thickness ratio* | |||
| Lateral/medial | 1.30 (1.20, 1.55) | 1.50 (1.30, 1.65) | 0.012 |
| Anterior/posterior | 1.46 (1.30, 1.59) | 1.29 (1.21, 1.74) | 0.040 |
| Intraoperative | |||
| Reduction in the coronal plane† | 0.015 | ||
| Valgus | 27 (35.5%) | 9 (27.3%) | |
| Neutral | 45 (59.2%) | 15 (45.5%) | |
| Varus | 4 (5.3%) | 9 (27.3%) | |
| Reduction in the sagittal plane† | 0.723 | ||
| Flexion | 4 (5.3%) | 2 (6.0%) | |
| Neutral | 43 (56.6%) | 19 (57.6%) | |
| Extension | 29 (38.1%) | 12 (36.4%) | |
| Postoperative | |||
| Gap at fracture site* (mm) | |||
| Anterior | 1.50 (1.00, 2.00) | 1.90 (1.00, 3.00) | 0.041 |
| Posterior | 2.00 (1.50, 3.40) | 2.00 (0.80, 4.00) | 0.474 |
| Medial | 2.45 (1.40, 3.60) | 2.60 (1.50, 4.00) | 0.113 |
| Lateral | 1.50 (1.00, 2.20) | 2.00 (1.30, 4.00) | 0.001 |
| Gap/cortical thickness* | |||
| Anterior | 0.15 (0.10, 0.22) | 0.38 (0.14, 0.56) | 0.004 |
| Posterior | 0.34 (0.22, 0.44) | 0.34 (0.20, 0.45) | 0.709 |
| Medial | 0.56 (0.32, 0.80) | 0.59 (0.28, 0.90) | 0.665 |
| Lateral | 0.18 (0.13, 0.29) | 0.57 (0.17, 0.84) | <0.001 |
| Cortical breakage at fracture site† | 0.008 | ||
| Yes | 1 (1.3%) | 4 (12.1%) | |
| No | 75 (98.7%) | 29 (87.9%) |
The values are given as the median with the interquartile range in parentheses.
The values are given as the number of patients with the percentage in parentheses.
After surgery, the size of the remaining gap on the anterior side of the fracture site was significantly different between the 2 groups (1.90 compared with 1.50, p = 0.041), as was the gap on the lateral side (2.00 compared with 1.50, p = 0.001). The ratio of the remaining gap to cortical thicknesses on the anterior side of the fracture site (0.38 compared with 0.15, p = 0.004) and on the lateral side (0.57 compared with 0.18, p < 0.001) also differed significantly between the 2 groups. The presence of iatrogenic cortical breakage around the fracture site was associated with significant differences in fracture-healing (p = 0.008).
Intraclass correlation coefficients for interobserver and intraobserver variability averaged 0.68 and 0.76 when calculated for all radiographic parameters. Thus, there was good intraobserver and interobserver agreement in the radiographic measurements.
Factors Affecting Fracture-Healing
Of the 46 variables tested for association with problematic healing, 13 reached significance in the univariate analysis. Seven of those 13 variables had already been predetermined at the time of injury or initial evaluation and therefore represented uncontrollable factors associated with problematic healing (Table V). The 6 remaining variables were directly related to the operative intervention and were defined as controllable factors associated with problematic healing (Table VI). Of the 7 uncontrollable factors, supra-isthmic fracture location, femoral bowing of ≥10° in the coronal plane, and lateral/medial cortical thickness ratio of ≥1.4 were identified as independent predictors of problematic healing on the basis of multivariate analysis. Of the 6 controllable factors, cortical breakage around the fracture site and a ratio of the remaining gap size to cortical thickness of ≥0.2 on the anterior and lateral sides of the fracture site were independent predictive factors associated with problematic healing.
TABLE V.
Uncontrollable Variables Associated with Delayed Union or Nonunion
| Successful Healing* (N = 76) | Problematic Healing* (N = 33) | Crude OR (95% CI) | P Value | Adjusted OR (95% CI) | P Value | |
| Body mass index | 0.014 | 0.240 | ||||
| ≥21.0 kg/m2 | 32 (42.1) | 22 (66.7) | 2.7 (1.2-6.1) | 2.0 (0.6-6.5) | ||
| <21.0 kg/m2 | 44 (57.9) | 11 (33.3) | 1 | 1 | ||
| Prodromal symptoms | 0.024 | 0.070 | ||||
| Yes | 17 (22.4) | 14 (42.4) | 2.6 (1.1-5.9) | 3.4 (0.9-13) | ||
| No | 59 (77.6) | 19 (57.6) | 1 | 1 | ||
| Bisphosphonate duration | 0.051 | 0.100 | ||||
| ≥7 yr | 30 (39.5) | 19 (57.6) | 2.2 (1.0-4.8) | 3.8 (0.7-18.9) | ||
| <7 yr | 46 (60.5) | 14 (42.4) | 1 | 1 | ||
| Fracture location | <0.001 | 0.020 | ||||
| Supra-isthmic | 4 (5.3) | 16 (48.5) | 17.3 (4.9-61.3) | 38.9 (6.7-224) | ||
| Isthmic | 51 (67.1) | 8 (24.2) | 1 | 1 | ||
| Infra-isthmic | 21 (27.6) | 9 (27.3) | 2.2 (0.8-5.6) | 6.7 (1.3-33.7) | ||
| Femoral bowing in the coronal plane | 0.004 | 0.002 | ||||
| ≥10° | 25 (32.9) | 20 (60.6) | 2.9 (1.3-6.4) | 10.8 (2.5-46.6) | ||
| <10° | 51 (67.1) | 13 (39.4) | 1 | 1 | ||
| Cortical thickness, lateral/medial | 0.015 | 0.036 | ||||
| ≥1.4 | 32 (42.1) | 22 (66.7) | 2.9 (1.3-6.4) | 11 (2.9-42.7) | ||
| <1.4 | 44 (57.9) | 11 (33.3) | 1 | 1 | ||
| Cortical thickness, ant./post. | 0.033 | 0.163 | ||||
| ≥1.4 | 21 (27.6) | 14 (42.4) | 2.5 (1.2-5.7) | 6.2 (0.4-24.3) | ||
| <1.4 | 55 (72.4) | 19 (57.6) | 1 | 1 |
The values are given as the number of patients with the percentage in parentheses.
TABLE VI.
Controllable Variables Associated with Delayed Union or Nonunion
| Successful Healing (N = 76)* | Problematic Healing (N = 33)* | Crude OR (95% CI) | P Value | Adjusted OR (95% CI) | P Value | |
| Reduction in the coronal plane | 0.013 | 0.413 | ||||
| Valgus | 27 (35.5) | 9 (27.3) | 1.0 (0.4-2.4) | 1 (0.3-3.4) | ||
| Neutral | 45 (59.2) | 15 (45.4) | 1 | 1 | ||
| Varus | 4 (5.3) | 9 (27.3) | 6.4 (1.8-23) | 3.1 (0.3-37.3) | ||
| Cortical breakage at fracture site | 0.010 | 0.039 | ||||
| Yes | 1 (1.3) | 4 (12.1) | 8.8 (1.9-81.6) | 19.7 (1.5-261) | ||
| No | 75 (98.7) | 29 (87.9) | 1 | 1 | ||
| Gap at fracture site, ant. | 0.226 | 0.260 | ||||
| ≥1.5 | 42 (55.3) | 20 (60.6) | 1.4 (0.7-1.8) | 0.7 (0.4-1.3) | ||
| <1.5 | 34 (44.7) | 13 (39.4) | 1 | 1 | ||
| Gap at fracture site, lat. | 0.010 | 0.340 | ||||
| ≥1.5 | 50 (65.8) | 28 (84.8) | 2.3 (1.5-3.4) | 0.7 (0.3-1.9) | ||
| <1.5 | 26 (34.2) | 5 (15.2) | 1 | 1 | ||
| Gap/cortical thickness, ant. | <0.001 | 0.035 | ||||
| ≥0.2 | 34 (44.7) | 25 (75.8) | 3 (1.3-6.8) | 34.1 (2.7-437.8) | ||
| <0.2 | 42 (55.3) | 8 (24.2) | 1 | 1 | ||
| Gap/cortical thickness, lat. | 0.005 | 0.012 | ||||
| ≥0.2 | 40 (52.6) | 25 (75.8) | 3.2 (1.4-7.3) | 11.6 (1.4-94.1) | ||
| <0.2 | 36 (47.4) | 8 (24.2) | 1 | 1 |
The values are given as the number of patients with the percentage in parentheses.
Discussion
Although atypical femoral fractures have been reported to be at risk of failure of healing12,15-17,19, few studies have analyzed factors affecting healing time in long-term bisphosphonate users. The present study sought to determine factors predictive of delayed union or nonunion in a consecutive series of 109 complete femoral fractures associated with long-term use of bisphosphonates. Specifically, we focused on operatively controllable factors related to fracture-healing time.
In our series of 109 fractures, 76 (69.7%) showed osseous union within 6 months after the index surgery, whereas the remaining 33 (30.3%) showed delayed union or nonunion. The overall union rate was 94.5% (103 of 109). In a previous study12 that evaluated 41 atypical low-energy femoral fractures associated with ≥5 years of bisphosphonate use, 98% (40 of 41) showed radiographic union at a mean of 8.3 months (range, 2 to 18 months). The average healing time of almost 8 months for those fractures appeared to be longer than that for typical femoral fractures, which heal at an average of 3 to 6 months.
We identified a difference in the duration of bisphosphonate use between the successful healing and problematic healing groups. A longer duration of bisphosphonate use was significantly associated with a higher rate of problematic healing, which is consistent with the results of other studies17,23. Prodromal symptoms can serve as an indicator of impending fracture associated with long-term use of bisphosphonates24. However, there are limited data regarding the relationship between prodromal symptoms and the healing time of these fractures. Moreover, it is uncertain whether longer prodromal symptoms are associated with the changes in radiographic features, including increase in the cortical thickness.
Mechanical stability after intramedullary nailing depends on fracture location25. Femoral fractures occurring in the non-isthmic region are prone to lack of a tight-fitting contact. This can aggravate rotational instability, the main mechanical problem in failed intramedullary nailing. In this study, the healing time of bisphosphonate-associated femoral fractures differed according to the fracture location. Fractures occurring in the supra-isthmic region including the subtrochanteric region were at higher risk of delayed union or nonunion.
Bisphosphonate-associated complete femoral fractures have been reported to have high complication rates with operative fixation13,16,17. Prasarn et al.15 concluded that the higher complication rate in bisphosphonate-associated fractures was primarily attributable to cortical breakage during nail insertion (5 of 17, 29%) and postoperative plate failure (3 of 10, 30%). In our series, iatrogenic cortical breakage around the fracture site occurred in 5 fractures, of which 4 showed delayed union. Asians have differences in femoral geometry, with higher rates of bowing, compared with Western populations17. Some authors have suggested that increased femoral bowing might be an important causative factor for low-energy diaphyseal femoral fractures19,26. In the present study, coronal bowing of the femur was more severe in the problematic healing group compared with the successful healing group, suggesting that the severity of femoral bowing may contribute to the high rate of delayed union.
With intramedullary nailing of atypical fractures, there are concerns regarding the later occurrence of a femoral neck fracture or cortical perforation distal to the fracture site. We believe that cephalomedullary screw insertion through the proximal portion of the nail could be advantageous, not only to prevent femoral neck fracture but also to obtain more rigid fixation. However, this type of nail was not used in our series. We attempted to insert nails that were as long as possible and statically locked to secure sufficient working length and thus acquire firm fixation. However, in an excessively bowed femur, there may be a risk of cortical perforation distal to the fracture site.
There has been no definitive conclusion regarding the optimal entry point of the nail (piriformis versus trochanteric starting point) in surgical treatment of atypical femoral fractures. The mismatch of curvature between the implant and femur makes it difficult to pass the nail into the medullary canal. In a case report27, a good result was achieved after fixation with a trochanteric starting nail of the opposite curvature. In our series, piriformis nails starting in the tip of the greater trochanter were used to reduce the risk of cortical perforation.
It has become widely accepted that tensile strain increase in focally thickened anterolateral cortex locations precedes an atypical femoral fracture26,28. In the present study, delayed union or nonunion occurred primarily at anteriorly or laterally thickened cortex at the fracture site. Thus, we believe that reducing the anterior and lateral fracture gaps as much as possible during surgery is critical to achieving osseous union. Intraoperative anatomical or valgus reduction of the femur is required to reduce the lateral fracture gap. In a retrospective series of surgically treated atypical fractures, varus malreduction at the fracture site negatively affected fracture-healing12. In our study, the remaining fracture gap was measured on radiographs in more detail in 4 locations (anterior, posterior, medial, and lateral). We focused not only on the absolute gap size but also on the relative ratio of gap size to cortical thickness because the cortical thickness at the fracture site was found to be associated with healing time. Anterior and lateral fracture gap sizes divided by cortical thickness were independent predictive factors for delayed union or nonunion.
This study has some limitations. First, the number of patients was limited. It is not easy to recruit a large sample of patients with a relatively rare injury. However, investigation of the available studies on the surgical outcome of such an injury revealed that the current study is one of the largest single-center series. Second, the retrospective nature of this study results in an inherent risk of observer bias, including the potential for missing data and inability to control confounding variables. We performed 46 statistical comparisons. Although an alpha level of 5% is known to represent an appropriate value for a single comparison, it may not be appropriate when numerous comparisons are performed. Third, we did not evaluate functional outcomes. Egol et al.12 reported that 66% of patients with surgically treated complete atypical fractures became pain-free, and pain combined with apprehension was a major cause of functional limitations in the remaining patients. Lastly, PTH (parathyroid hormone) 1-34 was not used in this study. The termination of the bisphosphonates and the addition of that agent could have changed some of the healing results. In addition, we did not investigate calcium and vitamin-D treatment and the possible role of low vitamin D in those with poor healing.
In conclusion, 30.3% of all patients with operatively treated complete femoral fractures associated with long-term use of bisphosphonates had delayed union or nonunion. Supra-isthmic fracture location, femoral bowing of ≥10° in the coronal plane, and a lateral/medial cortical thickness ratio of ≥1.4 were predictive of problematic healing but were uncontrollable. Cortical breakage around the fracture site and a ratio of remaining gap to cortical thickness of ≥0.2 on the anterior and lateral sides of the fracture sites were operatively controllable predictive factors associated with problematic healing. Our data suggest that intramedullary nailing without cortical breakage around the fracture site and decreasing the anterior and lateral fracture gaps as much as possible (minimal fracture distraction) could be helpful in reducing healing time in complete femoral fractures associated with long-term bisphosphonate use.
Footnotes
Investigation performed at the Department of Orthopaedic Surgery, Samsung Changwon Hospital, Sungkyunkwan University School of Medicine, Changwon, South Korea
A commentary by Edward Joseph Harvey, MDCM, MSc, is linked to the online version of this article at jbjs.org.
Disclosure: There was no external funding for this study. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of this article.
References
- 1.Hollick RJ, Reid DM. Role of bisphosphonates in the management of postmenopausal osteoporosis: an update on recent safety anxieties. Menopause Int. 2011. June;17(2):66-72. [DOI] [PubMed] [Google Scholar]
- 2.Das De S, Setiobudi T, Shen L, Das De S. A rational approach to management of alendronate-related subtrochanteric fractures. J Bone Joint Surg Br. 2010. May;92(5):679-86. [DOI] [PubMed] [Google Scholar]
- 3.Girgis CM, Seibel MJ. Guilt by association? Examining the role of bisphosphonate therapy in the development of atypical femur fractures. Bone. 2011. May 1;48(5):963-5. Epub 2011 Feb 22. [DOI] [PubMed] [Google Scholar]
- 4.Somford MP, Draijer FW, Thomassen BJ, Chavassieux PM, Boivin G, Papapoulos SE. Bilateral fractures of the femur diaphysis in a patient with rheumatoid arthritis on long-term treatment with alendronate: clues to the mechanism of increased bone fragility. J Bone Miner Res. 2009. October;24(10):1736-40. [DOI] [PubMed] [Google Scholar]
- 5.Cheung RK, Leung KK, Lee KC, Chow TC. Sequential non-traumatic femoral shaft fractures in a patient on long-term alendronate. Hong Kong Med J. 2007. December;13(6):485-9. [PubMed] [Google Scholar]
- 6.Lenart BA, Neviaser AS, Lyman S, Chang CC, Edobor-Osula F, Steele B, van der Meulen MC, Lorich DG, Lane JM. Association of low-energy femoral fractures with prolonged bisphosphonate use: a case control study. Osteoporos Int. 2009. August;20(8):1353-62. Epub 2008 Dec 9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Neviaser AS, Lane JM, Lenart BA, Edobor-Osula F, Lorich DG. Low-energy femoral shaft fractures associated with alendronate use. J Orthop Trauma. 2008. May-Jun;22(5):346-50. [DOI] [PubMed] [Google Scholar]
- 8.Einhorn TA. The cell and molecular biology of fracture healing. Clin Orthop Relat Res. 1998. October;355(Suppl):S7-21. [DOI] [PubMed] [Google Scholar]
- 9.Mashiba T, Mori S, Burr DB, Komatsubara S, Cao Y, Manabe T, Norimatsu H. The effects of suppressed bone remodeling by bisphosphonates on microdamage accumulation and degree of mineralization in the cortical bone of dog rib. J Bone Miner Metab. 2005;23(Suppl):36-42. [DOI] [PubMed] [Google Scholar]
- 10.Odvina CV, Zerwekh JE, Rao DS, Maalouf N, Gottschalk FA, Pak CY. Severely suppressed bone turnover: a potential complication of alendronate therapy. J Clin Endocrinol Metab. 2005. March;90(3):1294-301. Epub 2004 Dec 14. [DOI] [PubMed] [Google Scholar]
- 11.Chan SS, Rosenberg ZS, Chan K, Capeci C. Subtrochanteric femoral fractures in patients receiving long-term alendronate therapy: imaging features. AJR Am J Roentgenol. 2010. June;194(6):1581-6. [DOI] [PubMed] [Google Scholar]
- 12.Egol KA, Park JH, Rosenberg ZS, Peck V, Tejwani NC. Healing delayed but generally reliable after bisphosphonate-associated complete femur fractures treated with IM nails. Clin Orthop Relat Res. 2014. September;472(9):2728-34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Grady MK, Watson JT, Cannada LK. Treatment of femoral fracture nonunion after long-term bisphosphonate use. Orthopedics. 2012. June;35(6):e991-5. [DOI] [PubMed] [Google Scholar]
- 14.Nieves JW, Cosman F. Atypical subtrochanteric and femoral shaft fractures and possible association with bisphosphonates. Curr Osteoporos Rep. 2010. March;8(1):34-9. [DOI] [PubMed] [Google Scholar]
- 15.Prasarn ML, Ahn J, Helfet DL, Lane JM, Lorich DG. Bisphosphonate-associated femur fractures have high complication rates with operative fixation. Clin Orthop Relat Res. 2012. August;470(8):2295-301. Epub 2012 Jun 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Weil YA, Rivkin G, Safran O, Liebergall M, Foldes AJ. The outcome of surgically treated femur fractures associated with long-term bisphosphonate use. J Trauma. 2011. July;71(1):186-90. [DOI] [PubMed] [Google Scholar]
- 17.Kang JS, Won YY, Kim JO, Min BW, Lee KH, Park KK, Song JH, Kim YT, Kim GH. Atypical femoral fractures after anti-osteoporotic medication: a Korean multicenter study. Int Orthop. 2014. June;38(6):1247-53. Epub 2014 Jan 25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shane E, Burr D, Abrahamsen B, Adler RA, Brown TD, Cheung AM, Cosman F, Curtis JR, Dell R, Dempster DW, Ebeling PR, Einhorn TA, Genant HK, Geusens P, Klaushofer K, Lane JM, McKiernan F, McKinney R, Ng A, Nieves J, O’Keefe R, Papapoulos S, Howe TS, van der Meulen MC, Weinstein RS, Whyte MP. Atypical subtrochanteric and diaphyseal femoral fractures: second report of a task force of the American Society for Bone and Mineral Research. J Bone Miner Res. 2014. January;29(1):1-23. Epub 2013 Oct 1. [DOI] [PubMed] [Google Scholar]
- 19.Sasaki S, Miyakoshi N, Hongo M, Kasukawa Y, Shimada Y. Low-energy diaphyseal femoral fractures associated with bisphosphonate use and severe curved femur: a case series. J Bone Miner Metab. 2012. September;30(5):561-7. Epub 2012 May 19. [DOI] [PubMed] [Google Scholar]
- 20.Koval KJ, Aharonoff GB, Rosenberg AD, Bernstein RL, Zuckerman JD. Functional outcome after hip fracture. Effect of general versus regional anesthesia. Clin Orthop Relat Res. 1998. March;348:37-41. [PubMed] [Google Scholar]
- 21.Whelan DB, Bhandari M, McKee MD, Guyatt GH, Kreder HJ, Stephen D, Schemitsch EH. Interobserver and intraobserver variation in the assessment of the healing of tibial fractures after intramedullary fixation. J Bone Joint Surg Br. 2002. January;84(1):15-8. [DOI] [PubMed] [Google Scholar]
- 22.Kay RM, Jaki KA, Skaggs DL. The effect of femoral rotation on the projected femoral neck-shaft angle. J Pediatr Orthop. 2000. Nov-Dec;20(6):736-9. [DOI] [PubMed] [Google Scholar]
- 23.Thompson RN, Phillips JR, McCauley SH, Elliott JR, Moran CG. Atypical femoral fractures and bisphosphonate treatment: experience in two large United Kingdom teaching hospitals. J Bone Joint Surg Br. 2012. March;94(3):385-90. [DOI] [PubMed] [Google Scholar]
- 24.Wang K, Moaveni A, Dowrick A, Liew S. Alendronate-associated femoral insufficiency fractures and femoral stress reactions. J Orthop Surg (Hong Kong). 2011. April;19(1):89-92. [DOI] [PubMed] [Google Scholar]
- 25.Yang KH, Kim JR, Park J. Nonisthmal femoral shaft nonunion as a risk factor for exchange nailing failure. J Trauma Acute Care Surg. 2012. February;72(2):E60-4. [DOI] [PubMed] [Google Scholar]
- 26.Oh Y, Wakabayashi Y, Kurosa Y, Ishizuki M, Okawa A. Stress fracture of the bowed femoral shaft is another cause of atypical femoral fracture in elderly Japanese: a case series. J Orthop Sci. 2014. July;19(4):579-86. Epub 2014 May 1. [DOI] [PubMed] [Google Scholar]
- 27.Seo JG, Moon YW, Lim JS, Park SJ, Kim SM. Mechanical axis-derived femoral component rotation in extramedullary total knee arthroplasty: a comparison between femoral transverse axis and transepicondylar axis. Knee Surg Sports Traumatol Arthrosc. 2012. March;20(3):538-45. Epub 2011 Jul 13. [DOI] [PubMed] [Google Scholar]
- 28.Saita Y, Ishijima M, Mogami A, Kubota M, Baba T, Kaketa T, Nagao M, Sakamoto Y, Sakai K, Kato R, Nagura N, Miyagawa K, Wada T, Liu L, Obayashi O, Shitoto K, Nozawa M, Kajihara H, Gen H, Kaneko K. The fracture sites of atypical femoral fractures are associated with the weight-bearing lower limb alignment. Bone. 2014. September;66:105-10. Epub 2014 Jun 14. [DOI] [PubMed] [Google Scholar]