Skeletal Radiology

, Volume 40, Issue 7, pp 905–911

Association of the presence of bone bars on radiographs and low bone mineral density

Authors

    • Department of Radiology, School of MedicineUniversity of Alabama at Birmingham
    • Department of Radiology, School of MedicineUniversity of Alabama at Birmingham
  • Sarah L. Morgan
    • Departments of Nutrition Sciences and MedicineSchools of Health Professions, Medicine, and Dentistry
  • Robert Lopez-Ben
    • Department of Radiology, School of MedicineUniversity of Alabama at Birmingham
  • Rebecca E. Steelman
    • University of Alabama
  • Nancy Nunnally
    • The UAB Osteoporosis Prevention and Treatment Clinic
  • Leandria Burroughs
    • The UAB Osteoporosis Prevention and Treatment Clinic
  • Naomi Fineberg
    • Department of Biostatistics, School of Public HealthUniversity of Alabama at Birmingham
Scientific Article

DOI: 10.1007/s00256-010-1087-x

Cite this article as:
Pitt, M.J., Morgan, S.L., Lopez-Ben, R. et al. Skeletal Radiol (2011) 40: 905. doi:10.1007/s00256-010-1087-x
  • 123 Views

Abstract

Objective

Bone bars (BB) are struts of normal trabecular bone that cross the medullary portions of the metaphysis and diaphysis at right angles to the long axis of the shaft. The purpose of this investigation was to determine whether the presence of bone bars (BB) identified on radiographs of the proximal femurs and tibia, predict lower bone mineral density (BMD) as evaluated with dual-energy x-ray absorptiometry (DXA) in the lumbar spine, total hip, or femoral neck.

Materials and methods

A total of 134 sequential DXA patients underwent radiography of the pelvis, hips, and both knees. The radiographs were evaluated for the presence of BB by two musculoskeletal radiologists who were blinded to DXA results. A t test was used to evaluate the relationship of BB to BMD and a Chi-square test was used to determine if BB were equally distributed among the categories of normal BMD, low bone mass (osteopenia), and osteoporosis.

Results

BB were associated with lower BMD at all measured sites. BB at the intertrochanteric and proximal tibial sites were the most predictive of low BMD while supraacetabular and distal femur BB were less predictive. Osteoporosis or osteopenia is seen in 60–91% of those with BB depending on the side and reader. It is only seen in about 40% of those without BB.

Conclusions

We conclude that the presence of BB suggest decreased BMD and when correlated with other clinical information, might support further evaluation of BMD.

Keywords

Bone barsDual-energy x-ray absorptiometryBone mineral density

Introduction

Bone bars (BB) are struts of normal trabecular bone that cross the medullary portions of the metaphysis and diaphysis at right angles to the long axis of the shaft [1]. Figures 1, 2, 3 and 4 show the typical radiographic appearance of BB in the hip and knees. BB are often seen on routine radiographs of older adult patients, most frequently in the intertrochanteric proximal femur and proximal tibial diametaphyses. They are often overlooked or misdiagnosed as enchondromas or bone infarcts.
https://static-content.springer.com/image/art%3A10.1007%2Fs00256-010-1087-x/MediaObjects/256_2010_1087_Fig1_HTML.jpg
Fig. 1

AP radiograph of hip. Bone bars are seen as grouped punctuate intertrochanteric densities (solid arrow). There are also supraacetabular bone bars (open arrow). The bone bars are prominent

https://static-content.springer.com/image/art%3A10.1007%2Fs00256-010-1087-x/MediaObjects/256_2010_1087_Fig2_HTML.gif
Fig. 2

AP frog leg projection of hip from Fig. 1. Bone bars now project as multiple linear densities (solid arrow). The bone bars are prominent

https://static-content.springer.com/image/art%3A10.1007%2Fs00256-010-1087-x/MediaObjects/256_2010_1087_Fig3_HTML.jpg
Fig. 3

AP radiograph of knee showing bone bars as grouped punctuate metaphyseal densities (solid arrow). The bone bars are prominent

https://static-content.springer.com/image/art%3A10.1007%2Fs00256-010-1087-x/MediaObjects/256_2010_1087_Fig4_HTML.jpg
Fig. 4

Lateral radiograph of knee from Fig. 3 shows bone bars projecting as multiple linear densities (solid arrow). The bone bars are prominent

It has been hypothesized that BB develop in slowly developing osteopenia in mature skeletons as a response to mechanical stress [2]. Alternatively, BB may have formed during skeletal growth but remain hidden and are unmasked by surrounding osseous resorption; a combination of the two hypotheses may be causative [1, 2].

Although BB have been postulated to be associated with the osteopenic skeleton, to our knowledge, there has not been an investigation that has evaluated bone mineral density (BMD) in subjects with BB [1, 2]. The purpose of this project was to determine if the presence of BB on radiographs is predictive of low BMD assessed by dual-energy x-ray absorptiometry (DXA). If it is possible to confirm that BB are predictive of low BMD, this could provide an indication for requesting a screening DXA scan and assist in the diagnosis of osteoporosis and metabolic bone disease.

Materials and methods

Human use approval was obtained from the UAB Institutional Review Board (IRB). Approval was obtained to waive the waiting period prior to signing the informed consent form. The protocol was a cross-sectional study of 150 patients who had standard-of-care DXA scans of the lumbar spine, total hip, and femoral neck because their ordering physician felt they were at risk of having low bone mass. An IRB-approved flyer was placed in the DXA scan waiting area asking individuals having standard of care DXAs to participate in the trial by having radiographs of the pelvis and hips, and both knees. After obtaining informed consent, the subjects had radiographs of the pelvis and hips and both knees to document the presence of BB so that the relationship between BB and BMD could be determined. The subjects were recruited between 06/19/08 to 08/07/08.

Inclusion criteria for the trial included: 1) willing to participate in the protocol and able to give informed consent; and 2) postmenopausal women. Exclusion criteria included: 1) Indwelling metallic objects which overlie the spine and hip and render the spine and hip DXA data useless or laminectomy; 2) recent radiographic contrast (such as barium) within the past week and 3) non-English speakers who could not complete the consent form in English.

DXA scans were completed using standardized protocols at the lumbar spine (L1-L4) and the left hip (total hip and femoral neck). The scans were completed on Hologic Discovery W scanners using 12.7:7 software. Standard quality control measures at our institution’s DXA machines are: short-term precision at the 95% confidence interval is 0.040 g/cm2 (4.34%), 0.034 g/cm2 (4.87%), and 0.035 g/cm2 (3.86%) at the lumbar spine, femoral neck, and total hip, respectively. Standard clinical radiographic techniques of the pelvis and hips were performed and included neutral and frog leg position with the central ray of the x-ray beam centered midline, 2 cm below the anterior superior iliac spine and antero-posterior and lateral views of both knees. Our protocol is two orthogonal views of skeletal sites.

The radiographs were independently evaluated by two experienced MSK radiologists (MJP – 39 years of experience and RLB – 11 years of experience) who were blinded to DXA results. The presence of BB at four sites (hip: intertrochanteric, supraacetabular and knee: distal femur and proximal tibia) were scored as present or not present. The two observers met prior to the study and agreed on the identification and grading of bone bars.

The demographic characteristics of the study population were tabulated. Data analysis was completed using the categories of BB absent (not present or subtle/faint) or BB present (present or prominent) and the presence of BB was correlated with BMD (g/cm2) and T-scores. A kappa statistic was calculated to measure the agreement of the presence of BB between the two readers and between the hip and knee within the same reader. A t test was used to evaluate the relationship of BB to BMD. A Chi-square test was used to determine if BB were equally distributed among the categories of normal BMD, low bone mass (osteopenia), and osteoporosis using the World Health Organization diagnostic criteria [3, 4]. Statistical significance was assigned at the p <0.05 level. A t test was used to compare body mass index (BMI) and age in women with and without bone bars as determined by both radiologists at all four sites.

Results

A total of 134 individuals completed the radiographs. The demographic characteristics of the cohort are presented in Table 1. The population was predominantly Caucasian with a mean age of 61.5 ± 8.9 years.
Table 1

Demographic characteristics of the study population

Demographic characteristic

 

Gender n (%)

 

 Female

134 (100)

Ethnicity n(%)

 

 Caucasian

102 (76)

 African-American

29 (21)

 Hispanic

1 ( 1)

 Asian

2 ( 2)

Postmenopausal

134 (100)

Mean age years (range)

61.5 ± 8.9 (40–85)

BMI (kg/m2)

27.9 ± 6.3 (17.6–54.9)

The prevalence of intertrochanteric BB on radiographs in this patient cohort, as noted independently by the two observers, was 30.3% and 16.5% of evaluated radiographs. BB were unilateral in 25-30% of patients in this location. When evaluating the knee radiographs for proximal tibial BB, the prevalence was similar, with a range of 35–19%. BB were unilateral in 21-40% of patients in this location.

A kappa statistic was calculated to measure the agreement on the presence of BB between the two readers and between the hip and the knee within the same reader. The kappa statistics and p values were 0.449 (p < 0.001), 0.411 (p < 0.001), 0.421 (p < 0.001), 0.300 (p < 0.001) between readers at the right hip intertrochanteric, left hip intertrochanteric, right knee proximal tibia, and left knee proximal tibial sites, indicating fair (κ = 0.300) to moderate (κ > 0.400) agreement between readers [5]. Each observer still had a strong correlation of the presence of BB with bone mineral density.

There was poor agreement between BB at the knee and hip in the same person for both radiologists, indicating that bone bars in the same individual may not be present at all sites.

Regardless of location, BB are associated with lower BMD in patients compared to BMD in patients without BB. BB at the intertrochanteric and proximal tibial sites were the most predictive of low BMD and supraacetabular and distal femur BB were less predictive. Tables 2 and 3 display the relationship between intertrochanteric and proximal tibial BB and BMD at all sites. BMD is less at all sites in patients with hip intertrochanteric BB compared to patients without BB and is significantly lower at the femoral neck and total hip DXA sites. Proximal tibial BB were less predictive of low BMD at all sites.
Table 2

Relationship of BMD and the presence of intertrochanteric bone bars*

MSK radiologist #1 - Intertrochanteric bone bars

MSK radiologist #2 - Intertrochanteric bone bars

Side/location

Absent

Present

p value

Absent

Present

p value

Right hip (n)

97

37

 

109

25

 

Lumbar spine BMD (g/cm2)

0.99 ± 0.15

0.94 ± 0.12

0.058

0.99 ± 0.15

0.91 ± 0.11

0.006*

Femoral neck BMD(g/cm2)

0.74 ± 0.12

0.66 ± 0.09

<0.001*

0.74 ± 0.12

0.64 ± 0.07

<0.001*

Total hip BMD (g/cm2)

0.88 ± 0.13

0.79 ± 0.08

<0.001*

0.87 ± 0.13

0.78 ± 0..07

<0.001*

Left hip (n)

89

44

 

114

19

 

Lumbar spine BMD (g/cm2)

0.99 ± 0.16

0.94 ± 0.12

0.058

0.99 ± 0.15

0.90 ± 0.11

0.015*

Femoral neck BMD (g/cm2)

0.74 ± 0.12

0.69 ± 0.11

0.013*

0.73 ± 0.12

0.65 ± 0.06

<0.001*

Total hip BMD (g/cm2)

0.89 ± 0.13

0.81 ± 0.10

0.001*

0.87 ± 0.13

0.77 ± 0.04

<0.001*

*Indicates a significant difference

Table 3

Relationship of BMD and the presence of proximal tibia bone bars*

MSK radiologist #1 - Proximal tibia bone bars

MSK radiologist #2 - Proximal tibia bone bars

Side/location

Absent

Present

p value

Absent

Present

p value

Right knee (n)

84

50

 

97

37

 

Lumbar spine BMD (g/cm2)

1.00 ± 0.15

0.94 ± 0.13

0.016*

0.99 ± 0.15

0.94 ± 0.12

0.058

Femoral neck BMD(g/cm2)

0.74 ± 0.12

0.69 ± 0.10

0.012*

0.73 ± 0.12

0.69 ± 0.12

0.106

Total hip BMD (g/cm2)

0.88 ± 0.13

0.81 ± 0.11

0.001*

0.87 ± 0.13

0.82 ± 0.13

0.071

Left knee (n)

88

46

 

108

26

 

Lumbar spine BMD (g/cm2)

0.99 ± 0.14

0.95 ± 0.15

0.136

0.98 ± 0.15

0.94 ± 0.10

0.063

Femoral neck BMD (g/cm2)

0.73 ± 0.11

0.70 ± 0.12

0.143

0.73 ± 0.12

0.67 ± 0.10

0.027*

Total hip BMD (g/cm2)

0.87 ± 0.12

0.82 ± 0.13

0.023*

0.87 ± 0.13

0.79 ± 0.10

0.001*

*Indicates a significant difference

Tables 4 and 5 show an analysis where T-scores were coded as osteoporosis (≤-2.5), osteopenia (-2.5 to -1.0) and normal (> -1) cross-tabulated with the presence or absence of BB in the femoral neck and total hip. Lumbar spine classification was found not to be related to the presence of BB (p > 0.05). The analysis by diagnostic group shows a clear difference by DXA site.
Table 4

The relationship of bone bar presence to WHO osteoporosis diagnostic categories at the femoral neck*

MSK radiologist #1

MSK radiologist #2

Intertrochanteric bone bars

Intertrochanteric bone bars

 

Absent

Present

p value

Absent

Present

p value

Right side

 Osteoporosis n (%)

8 (8)

5 (14)

 

9 (8)

4 (16)

 

 Low bone mass n (%)

38 (39)

27 (73)

0.001*

47 (43)

18 (72)

0.004*

 Normal n (%)

51 (53)

5 (14)

 

53 (49)

3 (12)

 

Left side

 Osteoporosis n (%)

6 (7)

6 (14)

 

10 (9)

3 (16)

 

 Low bone mass n (%)

38 (43)

27 (61)

0.016*

51 (44)

14 (74)

0.012*

 Normal n (%)

45 (51)

11 (25)

 

54 (47)

2 (11)

 

MSK radiologist #1

MSK radiologist #2

Proximal tibia bone bars

Proximal tibia bone bars

 

Absent

Present

p value

Absent

Present

P value

Right side

 Osteoporosis n (%)

5 (6)

8 (16)

 

7 (7)

6 (6)

 

 Low bone mass n (%)

40 (48)

25 (50)

0.108

48 (50)

17 (46)

0.287

 Normal n (%)

39 (46)

17 (34)

 

42 (43)

14 (38)

 

Left side

 Osteoporosis n (%)

6 (7)

7 (15)

 

8 (7)

5 (19)

 

 Low bone mass n (%)

43 (49)

22 (48)

0.273

54 (50)

11 (42)

0.186

 Normal n (%)

39 (44)

17 (37)

 

46 (43)

10 (39)

 

*Indicates a significant difference

Table 5

The relationship of bone bar presence to WHO osteoporosis diagnostic categories at the total hip*

MSK radiologist #1

MSK radiologist #2

Intertrochanteric bone bars

Intertrochanteric bone bars

 

Absent

Present

P value

Absent

Present

P value

Right side

 Osteoporosis n (%)

2 (2)

1 (3)

 

3 (3)

0

 

 Low bone mass n (%)

27 (29)

21 (60)

0.005*

28 (27)

20 (87)

<0.001*

 Normal n (%)

64 (69)

13 (37)

 

74 (70)

3 (13)

 

Left Side

 Osteoporosis n (%)

2 (2)

1 (2)

 

3 (3)

0

 

 Low bone mass n (%)

25 (29)

22 (52)

0.040*

33 (30)

15 (88)

<0.001*

 Normal n (%)

58 (68)

19 (45)

 

75 (68)

2 (12)

 

MSK radiologist #1

MSK radiologist #2

Proximal tibia bone bars

Proximal tibia bone bars

 

Absent

Present

P Value

Absent

Present

P value

Right Side

 Osteoporosis n (%)

1 (1)

2 (4)

 

1 (1)

2 (6)

 

 Low bone mass n (%)

25 (31)

23 (50)

0.035*

32 (34)

16 (47)

0.085

 Normal n (%)

56 (68)

21 (46)

 

61 (65)

16 (47)

 

Left side

 Osteoporosis n (%)

1 (1)

2 (5)

 

1 (1)

2 (9)

 

 Low bone mass n (%)

27 (32)

21 (49)

0.058

35 (33)

13 (56)

0.005*

 Normal n (%)

57 (67)

20 (46)

 

69 (66)

8 (35)

 

* Indicates a significant difference

Femoral neck BMD status is related to hip intertrochanteric BB but not knee proximal tibia BB. Approximately 85% of patients with hip intertrochanteric BB have either osteoporosis or osteopenia as compared to 50% of those without BB. The relationship is even stronger for total hip measurements. Osteoporosis or osteopenia is seen in 60–91% of those with BB depending on the side and reader. It is only seen in about 40% of those without BB. There was no association between the presence of BB and either BMI or age (data not shown).

Discussion

BB are frequently seen in the radiographs of adults and have been postulated to be associated with the osteopenic skeleton [1]. In slowly developing osteoporosis, there is resorption at the endosteal cortical surface resulting in cortical thinning. BB may be a mechanical response to strengthen or support the thinned cortical tube; hence the term “bone reinforcement lines” was introduced in a subsequent publication [2]. The authors prefer the designation BB, which describes the morphology of these lines without ascribing a putative etiology (but likely) term like bone reinforcement lines.

BB should be differentiated from “growth arrest” lines (GAL) seen in children or young adults. These transversely oriented linear densities, located near or close to the unfused physis, are templates of the subjacent physis, and result from either local injury or reflect a systemic insult (for example infection) [6]. GAL are a misnomer, since they do not form until growth at the affected physis has resumed. GAL gradually disappear with bone remodeling as the individual matures. In contrast, BB typically appear after skeletal maturation.

Although it has been noted that BB are more often associated with the osteopenic skeleton, however, to the best of our knowledge, there has not been a prior investigation evaluating BMD in subjects with BB. BB were evaluated in a previous study of 29 prehistoric adult skeletal specimens from the San Diego Museum of Man and 20 specimens of adult cadavers or surgical specimens [2]. Gross morphology of the BB showed bars or wafers of bone extending partially or completely across the medullary canal, generally oriented at right angles to the cortex in the diaphysis or slightly obliquely in the metaphyses [1, 2]. Histological examination of BB shows mature lamellar bone without recent bone deposition [2]. In that study, the subjects with radiographs showing BB were reported to have osteopenia from disuse, immobilization, neurological injury, or amputation [2].

The assessment of BMD on radiographs is imprecise; a loss of approximately one-third of the spinal trabecular bone is necessary before osteopenia can be diagnosed [7]. In our study, the presence of femoral intertrochanteric (and to a lesser degree proximal tibial metaphyseal BB) had a high predictive association for low BMD. This is a unique finding since other signs on radiographs suggestive of low bone mass (such as the Singh index in the hip) do not correlate as well with BMD measurements [8].

Low BMD at a specific anatomic site is generally more predictive of fracture at that site than at other anatomic sites [9]. We saw a similar finding in that the presence of BB at the pelvis, hip, and knees were more predictive of low BMD at hip sites than in the spine, and the strongest correlation with low BMD of the total hip was seen with intertrochanteric proximal femur BB.

We acknowledge the bias in this study since the population sent for DXA scanning is generally related to a concern about a prevalent fracture or clinical condition that predisposes to low BMD, which we attempted to minimize by blinding the interpreters to DXA results. Our results mostly showed osteopenia with only a few patients showing DXA scores of osteoporosis. While BB may be seen at other sites in more advanced osteopenia (i.e., phalanges) we have not studied their predictive association for low BMD. We cannot explain why BB are more common and more pronounced in the proximal femur and tibia than other sites. Given some discrepancy in observation of BB between both observers, BB determination on radiographs can be subjective, especially when the findings are subtle.

The radiographic identification of intertrochanteric and proximal tibial BB is thus a useful and objective sign of low bone mass, seen in 60–91% of those patients with DXA correlated osteopenia and osteoporosis. We believe that BB are an additional valuable radiographic finding, along with cortical thickness on radiogrammetry, breast arterial calcifications, vertebral fracture deformity, distal radial deformity, and hand osteoarthritis, which are shown to correlate with decreased BMD by DXA. Depending on other clinical correlations, a radiographic finding of BB supports a recommendation for a DXA scan to further assess BMD.

Acknowledgements

This study was supported by an unrestricted educational grant from Procter & Gamble.

Copyright information

© ISS 2011