SEARCH

SEARCH BY CITATION

Keywords:

  • OSTEOPOROSIS;
  • FRACTURE;
  • ATYPICAL;
  • SUBTROCHANTERIC;
  • FEMORAL;
  • DIAPHYSIS;
  • BISPHOSPHONATES;
  • FRACTURE ANGLES;
  • TRANSVERSE;
  • OBLIQUE;
  • LATERAL CORTEX

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

In 2010, the American Society for Bone and Mineral Research (ASBMR) task force defined major and minor features to assist in the case finding and reporting of atypical femoral fractures (AFFs). One major feature that was proposed was a “transverse or short oblique configuration.” Our primary aim was to compare the conventional overall fracture morphology (OFM) with its associated angle (OFMA) and our proposed lateral cortical fracture angle (LCFA) in the assessment of fracture configuration in suspected AFFs and non-AFFs. The radiographs of 79 patients with AFFs and 39 patients with non-AFFs were each analyzed by two blinded reviewers to obtain the OFM, OFMA, and LCFA. Using the overall fracture morphology to assess the suspected AFFs resulted in discordance between reviewers in 18 cases (22.8%), of which 5 (6.3%) were discordant between short oblique (>30° to 60°) and long oblique (>60° to 90°) configurations, therefore affecting their classifications as AFFs. By assessing only the critical component within the lateral cortex, all the suspected AFFs fell well within the classification as transverse fractures with a mean LCFA of 4.8° (range 0.3 to 18.0, SD = 4.23). The inter-reader variability was also lower for LCFA versus OFMA (4.1° versus 6.9°, p = 0.001) when used to assess AFFs. Fracture angles were significantly different in AFFs versus non-AFFs regardless of whether the OFMA or LCFA methodology was employed, but the greater difference associated with LCFA suggests its greater discriminating power. When LCFA was used in conjunction with 0° to 30° as the criteria for transverse morphology, all the AFFs and non-AFFs were correctly classified. By using a standardized and precise method in measuring the fracture angle, specifically using only the component of the lateral cortex and limiting to truly transverse fractures, ie, between 0° and 30°, the LCFA is a robust and accurate method to assess the fracture morphology in suspected AFFs. © 2014 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Atypical femoral fractures (AFFs) have garnered much recent attention. These are femoral fractures, mostly in the subtrochanteric and diaphyseal region that have occurred spontaneously or under circumstances of minimal trauma in patients, the majority of whom were receiving bisphosphonate therapy. In 2010, the American Society for Bone and Mineral Research (ASBMR) convened a task force who defined major and minor features to assist in case finding and reporting.[1] All major features were required to be present in order to designate a fracture as atypical and to distinguish it from more common hip fractures (ie, femoral neck and intertrochanteric), which are considered “typical” of fragility fractures arising from osteoporosis. One major ASBMR feature is a “transverse or short oblique configuration.” The lack of precision of this definition may affect its specificity in classifying AFFs. Based on convention, a “transverse” fracture line is generally taken to be at 0° to 30° deviation from the perpendicular to the long axis of the femoral shaft; a short oblique fracture is taken to be at >30° to 60° deviation; and a long oblique fracture is taken to be >60° to 90° from this same reference.

However, unpredictable crack progression resulting in variable lengths of the medial cortical spike can affect the assessment of the obliquity of the overall fracture angle. Our primary hypothesis is that determination of fracture angles in suspected AFFs is best assessed using only the component that is restricted to the lateral cortex, the most likely site of fracture initiation in AFFs.[2]

Our aims are: 1) to compare lateral cortical fracture angle (LCFA) versus overall fracture morphology/angle (OFM/OFMA) in determining the transverse morphology of AFFs; 2) to compare the interindividual measurement variability of LCFA versus OFMA when assessing AFFs; and 3) to explore fracture angle differences between AFFs and non-AFFs using both LCFA and OFMA methodologies to determine the best cut-off angles for distinguishing between fracture types.

Materials and Methods

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

We studied a total of 79 consecutive patients who were admitted to Singapore General Hospital from May 1, 2004, to March 31, 2012, with suspected AFFs (meeting the ASBMR criteria with the exception of the criteria for noncomminution; minimally comminuted fractures were included). Their mean age was 68.3 years (range 47 to 91) and all 79 were female. Six had never been previously exposed to bisphosphonates. The mean duration of bisphosphonate use was 4.9 years (range 1 to 10 years). The most common bisphosphonate used was alendronate. Sixty-five were on alendronate, 2 were on risedronate, 2 were on sequential alendronate-risedronate, 1 was on sequential etidronate-alendronate, 1 was on zoledronic acid, 1 was on pamidronate, and 1 was on an unknown bisphosphonate. Other clinical features of these patients have been described in our earlier published series.[3-6]

We reviewed our hospital fracture database over the same period of time to assemble a control group of non-AFFs. This was a random selection of 39 patients who were admitted for nonpathological fragility fractures of the subtrochanteric region and shaft of femur (excluding neck of femur and intertrochanteric fractures), which occurred under minimal or low-energy trauma. These included 32 female and 7 male patients. They were significantly older, with a mean age of 73.8 years when compared with the AFFs (p < 0.005). History of bisphosphonate use could not be reliably determined because of the retrospective nature of data collection in this control group. Based on the best available information, 2 patients were noted to have been on bisphosphonate therapy.

Each set of radiographs was reviewed by 2 readers (designated as R1 and R2) using both techniques as described below. Readers were blinded to the cases' background information. Each reader assessed their randomly assigned set of radiographs twice based on each technique. Two senior radiologists (MAP and PCM), 2 senior orthopedic surgeons (JSBK and TSH) and 1 endocrinologist (ACN) participated in this exercise. Digital DICOM files of plain X-rays of the pelvis, hip and femur were accessed using the institutional Amalga Picture archiving and communication system and measurements were performed using the program's digital toolbox. Statistical analysis was performed using SPSS 11.0 (SPSS Inc., Chicago, IL, USA).

Overall fracture morphology and overall fracture morphology angle (OFM and OFMA)

This was assessed by the following method. A line was drawn perpendicular to the long axis of the femoral shaft and another line of best fit was drawn across the fracture line, and the overall fracture morphology (OFM) was classified as transverse (0° to 30°), short oblique (>30° to 60°) or long oblique (>60° to 90°) based on the angle (OFMA) between the aforementioned two lines. If a medial spike was present in these fractures, the line of best fit included this medial spike (Fig. 1A).

image

Figure 1. (A) This illustrates the method by which the overall fracture morphology (OFM) was assessed. A line was drawn perpendicular to the long axis of the femoral shaft and another line of best fit was drawn across the fracture line (indicated by arrow) and fractures were classified as transverse (0° to 30°), short oblique (>30° to 60°), or long oblique (>60° to 90°) based on the angle between the aforementioned two lines. (B) This illustrates the method by which the lateral cortical fracture angle (LCFA) was assessed. A line is drawn perpendicular to the long axis of the femoral shaft and another line was drawn across the fracture line that was confined to the lateral cortex (as indicated by the arrow). The angle between the aforementioned two lines was recorded.

Download figure to PowerPoint

Lateral cortical fracture angle (LCFA)

The LCFA was assessed by the following method. A line was drawn perpendicular to the long axis of the femoral shaft and another line was drawn across the fracture line that was confined to the lateral cortex (Fig. 1B). The angle between the aforementioned two lines was measured and recorded.

Results

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

AFF Cohort

OFM and OFMA in the AFF cohort

The mean OFMA of AFFs was 34.1° (range 1.1 to 63.6, SD = 16.5). The mean inter-reader variation was 6.9° (range 0.1 to 39.2, SD = 7.24).

AFFs were classified as transverse in 44.3% (R1) and 43.0% (R2), short oblique in 46.8% (R1) and 49.4% (R2), and long oblique in 8.9% (R1) and 7.6% (R2). This was concordant in 61 cases (77.2%). It was discordant in 18 cases (22.8%). Thirteen cases (16.5%) were discordant between transverse and short oblique, but this did not affect their classification as AFF by ASBMR criteria (designated as Type A discordance as shown in Fig. 2A). Five cases (6.3%) were assessed as short oblique by one reader and long oblique by the other (designated as Type B discordance as shown in Fig. 2A). No cases were assessed as transverse by one and long oblique by the other (designated as Type C discordance as shown in Fig. 2A). Therefore, in 5 cases (6.3%), the discordance affected their classification as AFF by ASBMR criteria.

image

Figure 2. (A) This illustrates the agreement between readers in assessing the OFM in AFFs. It was concordant in 61 cases (77.2%). It was discordant in 13 cases (16.5%) between transverse and short oblique (Type A). In 5 cases, it was assessed as a short oblique by one and long oblique by the other (Type B). None of the cases were assessed as a transverse by one and long oblique by the other (Type C). (B) This illustrates the agreement between readers in assessing the OFM in non-AFFs. This was concordant in 32 cases (82.1%). It was discordant in 7 cases (17.9%). One case (2.6%) was discordant between transverse and short oblique, ie, Type A. Six cases (15.3%) were discordant between short oblique by one reader and long oblique by the other, ie, Type B discordance. There were no cases of Type C discordance.

Download figure to PowerPoint

LCFA in the AFF cohort

For the AFFs, the mean degree of the deviation (from the perpendicular to the long axis of the femoral shaft) of the component of the fracture line that was confined to the lateral cortex, ie, the LCFA, was 4.8° (range 0.3 to 18.0, SD = 4.23). The mean inter-reader variation was 4.1° (range 0 to 29.0, SD = 4.81). This was significantly lower than the mean inter-reader variation of 6.9° when OFMA was used to assess suspected AFFs (p = 0.001). Using the LCFA, all the AFFs were classified as transverse fractures.

Non-AFF cohort

OFM and OFMA in the non-AFF cohort

The mean OFMA of non-AFFs was 65.0° (range 29.4 to 77.5, SD = 10.8). The mean inter-reader variation was 6.6° (range 0.3 to 26.2, SD = 6.34).

Non-AFFs were classified as oblique in 76.9% (both R1 and R2), short oblique in 23.1% (R1) and 20.5% (R2), and transverse in 0% (R1) and 2.6% (R2). This was concordant in 32 cases (82.1%). It was discordant in 7 cases (17.9%). One case (2.6%) was discordant between transverse and short oblique (designated as Type A discordance as shown in Fig. 2B). Six cases (15.3%) were discordant between short oblique as assessed by one reader and long oblique by the other (designated as Type B discordance as shown in Fig. 2B). There were no cases of Type C discordance.

LCFA in the non-AFF cohort

For the non-AFFs, the mean LCFA was 61.3° (range 30.5 to 82.0, SD = 12.45). The mean inter-reader variation was 7.5° (range 0.4 to 43.2, SD = 7.94). Using the LCFA, none of the non-AFFs were classified as transverse fractures.

OFMA and LFCA in AFFs versus non-AFFs

The mean OFMA was significantly different between AFFs and non-AFFs (95% confidence interval [CI] 25.9 to 36.0). Similarly, the mean LCFA was also significantly different (95% CI 52.3 to 60.8) between AFFs and non-AFFs. However, as shown in Fig. 3, the difference appears to be much greater when LCFA was employed, suggesting that it was a much better discriminating measurement.

image

Figure 3. This illustrates the distribution of the angles assessed by the two methodologies, ie, LCFA and OFMA in both AFFs and non-AFFs.

Download figure to PowerPoint

Fracture comminution

Ten AFFs (12.7%) were deemed to be comminuted, which in most cases was minimal and consisted of a small solitary bony fragment (less than 10% of the shaft width), whereas 15 non-AFFs (38.4%) were considered to be comminuted (p = 0.009), most of which were of moderate to severe degrees of comminution.

Discussion

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Earlier literature on AFFs relied mostly on conventional methodologies largely similar to the OFM to classify these fractures.[2] Indeed, in clinical practice, this would have been the most likely modus operandi, albeit in an even more imprecise fashion, because the attending physician or surgeon would simply “eyeball” the fracture line on the radiograph rather than rely on actual measurements to estimate the approximate morphology, ie, whether the fracture line appears transverse/short oblique as opposed to oblique.

Even if the angle of the overall fracture morphology, ie, OFMA was objectively measured, there were important problems with this approach of assessing the fracture line beyond the confines of the lateral cortex. Depending on the path of fracture propagation, the medial portion of the fracture may be of variable lengths and shapes. Furthermore, the variable nature of the falls with the varying combinations of twisting, bending, and shearing forces contribute to the myriad possible configurations of the medial spike.[7] Therefore, by incorporating the medial spike in the measurement of the fracture angle and consequently the fracture morphology, the assessment can be skewed quite dramatically in some cases.

Based on more recent literature, there is some consensus that AFFs likely originate from stress lesions in the lateral femur cortex.[2] This is in line with our hypothesis that bisphosphonate therapy may predispose patients to tension side fatigue fractures of the lateral femur cortex. We postulate that the fracture morphology within the lateral cortex is most reflective of the underlying pathophysiology of AFFs and hence our focus within this specific area of interest.

Our study confirmed that using overall fracture morphology to assess fracture morphology in suspected AFFs results in considerable discordance between readers. This occurred in approximately 1 in 4 cases (22.8%). In 6.3%, the discordance had a potential for misclassification if the ASBMR criteria of requiring either a transverse or short oblique morphology was adhered to. Indeed, measurement of the angle using overall fracture morphology was shown to be associated with a significantly higher inter-reader variation when compared with measuring the angle of the fracture line confined to the lateral cortex. By assessing only the critical component within the lateral cortex, ie, the LCFA, all the suspected atypical femoral fractures in this study had fracture angles that fell well within the classification as transverse fractures with a mean of 4.8° (range 0.3 to 18.0, SD = 4.23). The inter-reader variability was also extremely low, with a variation of <5°. These results show that the LCFA has better precision than OFMA. Furthermore, although the angles derived from the overall fracture morphology and that confined to the lateral cortex were both significantly different between AFFs and non-AFFs, the difference was much greater when LCFA was employed, thereby supporting its discriminating value. Our results also indicated that when LCFA was used in this study, together with the criteria of transverse morphology, ie, 0° to 30° as the threshold, all the AFFs were correctly classified, whereas none of the non-AFFs were mistaken as atypical. Conversely, when OFMA was used, although more than 90% of the AFFs were found to be transverse or short oblique, less than half (47%) were truly transverse in the strictest sense, ie, 0° to 30°. Also, with OFMA, as many as 10 of 39 (25%) non-AFFs could have been classified as transverse or short oblique and therefore mistaken as AFFs. It must be remembered that these differences between AFFs versus non-AFFs were not meant to test the hypothesis about a difference between them because that difference was the criteria in defining the groups in the first place. Rather, the primary purpose was to evaluate the differences in the measurement techniques to test the hypothesis that the fracture line within the lateral cortex is the critical element when assessing the fracture angle. Our results suggested that using the LCFA and a cut-off of 0° to 30° to determine transverse morphology was the more precise and accurate method when assessing suspected AFFs.

In a recent study by Aspenberg and colleagues[8] of 227 fractures comprising 59 atypical and 218 ordinary fractures, frequency distribution analysis of the fracture angle showed a distinct subgroup whose fracture angle was ± 15° (converted to our reference line) with mean of 1° and SD of 10° that had a specificity of 0.93 (95% CI 0.88 to 0.96) for bisphosphonate exposure. However, in their methodology, they assessed their fracture angle using a line that was parallel to the fracture line, extending from the lateral cortex into the shaft, approximately one-third of its width. This is consistent with our own data, although we are of the view that specificity can be further improved by measuring only the lateral cortex.

There are limitations to our study. The ideal control population would have been a group of age- and sex-matched individuals who had experienced fragility fractures while currently or previously on bisphosphonate therapy but did not meet the ASBMR criteria as AFFs. Our controls were mostly not exposed to bisphosphonates and included male patients. Our AFFs were also uniformly female and of Chinese descent, so therefore these results might not be applicable to subjects of others descents. Our study was limited to assessment of two-dimensional radiographs, wherein lies the inherent problem of possible distortion of the radiographic image dependent upon the angle between the fracture and the incident X-ray beam.

The results of this study support the inclusion of the transverse morphology confined to the lateral cortex for classifying a fracture as an AFF. By using a standardized and precise method in measuring the fracture angle, specifically using only the component of the lateral cortex, ie, LCFA, and limiting to truly transverse fractures, ie, between 0° to 30°, we can reduce misclassification of AFFs and improve the quality of ongoing data collection.

Acknowledgments

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Authors' roles: Study conceptualization: TSH. Study design: TSH, ACN, MAP, PCM, JSK. Study conduct: TSH, ACN, MAP, PCM, JSK. Data collection: JSK, TSH. Radiographic analysis: Major readers- MAP, PCM, JSK; Minor readers- TSH, ACN. Data analysis: ACN, MAP. Data interpretation: TSH, ACN, MAP, PCM, JSK. Drafting manuscript: ACN. Revising manuscript content: TSH, ACN, MAP, PCM, JSK. Approving final version of manuscript: TSH, ACN, MAP, PCM, JSK. ACN takes responsibility for the integrity of the data analysis.

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References
  • 1
    Shane E, Burr D, Ebeling PR, et al. Atypical subtrochanteric and diaphyseal femoral fractures: report of a task force of the American Society for Bone and Mineral Research. J Bone Miner Res. 2010; 25:226794.
  • 2
    Koh JS, Goh SK, Png MA, Kwek EB, Howe TS. Femoral cortical stress lesions in long-term bisphosphonate therapy: a herald of impending fracture? J Orthop Trauma. 2010; 24:7581.
  • 3
    Goh SK, Yang KY, Koh JS, et al. Subtrochanteric insufficiency fractures in patients on alendronate therapy: a caution. J Bone Joint Surg Br. 2007; 89:34953.
  • 4
    Koh JS, Goh SK, Png MA, Ng AC, Howe TS. Distribution of atypical fractures and cortical stress lesions in the femur: implications on pathophysiology. Singapore Med J. 2011; 52:7780.
  • 5
    Kwek EB, Goh SK, Koh JS, Png MA, Howe TS. An emerging pattern of subtrochanteric stress fractures: a long-term complication of alendronate therapy? Injury. 2008; 39:22431.
  • 6
    Ng AC, Koh JS, Howe TS. Vitamin D and atypical femoral fractures. Osteoporos Int. 2013; 24:1767.
  • 7
    Evans FG, Pedersen HE, Lissner HR. The role of tensile stress in the mechanism of femoral fractures. J Bone Joint Surg. 1951; 33A:4854501.
  • 8
    Schilcher J, Koeppen V, Ranstam J, Skripitz R, Michaelsson K, Aspenberg P. Atypical femoral fractures are a separate entity, characterized by highly specific radiographic features. A comparison of 59 cases and 218 controls. Bone. 2013; 52:38992.