J Bone Miner Metab (2004) 22:496–499 DOI 10.1007/s00774-004-0513-3
© Springer-Verlag 2004
Range of change of measured BMD in the femoral neck and total hip with rotation in women Leonard Rosenthall McGill University Health Center, 1650 Cedar Avenue, Montreal, Canada H3G 1A4
Abstract To obtain a quantitative assessment of the range of variation of bone mineral density (BMD) measurements in the femoral neck and total hip with rotation of the hip, 200 women, ages 21–86 years, were scanned by DXA in the neutral position and with 25% internal rotation of the leg. The difference in BMDs (neutral minus internal rotation) was ⱖ0 in about 65% of the patients, whereas the remaining 35% were ⬍0. In terms of absolute change, the femoral neck median BMD of 0.025 g/cm2 was significantly greater than the total hip, 0.016 g/cm2. Percent absolute femoral neck change was 3.13% compared to 1.79% for the total hip. Absolute change of the total hip correlated positively with age, while the femoral neck showed a positive trend that was not statistically significant at the P ⫽ 0.05 level. Despite the total hip’s lower precision error and smaller BMD change with rotation, the number of patients exceeding the 95% confidence limit for each site was virtually the same, 11% for the femoral neck and 13% for the total hip. This finding underlines the need to achieve confirmed repositioning accuracy in longitudinal studies, whether at zero femoral neck axis anteversion or otherwise, to appreciate the relative advantages of total hip measurement. The question that remains is what method can be employed to obtain this confirmation simply and economically in a busy clinical service facility with the usual turnover in personnel. Key words
the foot is strapped to a foot block to maintain the femoral neck axis parallel to the scanning table top. The DXA procedure provides bone mineral density measurements of the femoral neck and total hip; the latter includes the femoral neck, trochanteric region, and a segment of the proximal shaft of the femur. Opinions differ as to which site should be used to aid in the initial clinical assessment. Some prefer the total hip because its precision error is significantly lower than that of the femoral neck, varying from 1.7% to 2.5% compared to 2.3% to 3.6% for the femoral neck [3–7]. It has been stated that the femoral neck is adequate for diagnosis because the errors of malpositioning are small compared to intrapopulation bone mineral density (BMD) variability, but the variability may be too large to be useful for monitoring treatment unless there is diligent repositioning to minimize precision error [5]. The lowest of the femoral neck and total hip T scores has been recommended for clinical evaluation, but with no opinion on follow-up selection [8]. The range of BMD error due to malpositioning of the femoral neck and total hip has not determined for a large series of patients. This report addresses the question in a cohort of 200 women.
bone mineral density · femoral neck · hip
Methods and patients Introduction Anteversion of the femoral head and neck is defined as the angle between the axis of the femoral neck and the coronal plane of the body. In adults, it varies from 5° to 15° [1,2]. For dual-energy X-ray absorptiometry (DXA) of the hip, the lower extremity is rotated internally and
Offprint requests to: L. Rosenthall (e-mail:
[email protected]) Received: January 14, 2004 / Accepted: March 24, 2004
Hip BMD measurements were performed on the Lunar DPX (GE Lunar, Madison, WI, USA) utilizing the software version 4.6f. Three experienced technicians shared the scanning procedures. There were two cohorts of women in the study. Sixtyfive postmenopausal women (median age, 63 years; range, 48–71 years) were screened as potential candidates for bisphosphonate therapy in a clinical drug trial. They had same-day paired measurements of their hips with repositioning between scans. The foot and knee were internally rotated with the foot strapped to the
L. Rosenthall: Femoral neck/hip BMD change with rotation
manufacturer-supplied positioner that was angled 25° from the vertical. Root mean square (RMS) average of the precision errors were calculated from the equation 1 SD ⫽ (Σd2/2n) /2, where SD denotes standard deviation, d is the paired differences in BMD, and n is the number of pairs. The coefficients of variation percent (CV%) 1 were derived from the expression 100 · (Σd2/2n) /2/M, where M is the mean BMD of the 2n measurements [9,10]. The second cohort consisted of 200 randomly chosen women, ages 21–86 years (mean, 58.9 ⫾ 12.9), who were referred to the clinical service for bone densitometry. Patients with orthopedic hip appliances or gross hip disorders that prevented femoral neck bone measurement were excluded. To obtain range and direction of BMD change with rotation of the femoral neck axis, the patients were scanned twice; first with internal rotation of the leg and then in the neutral position, i.e., with the foot and patella pointing upward or no rotation. The patients were not repositioned between measurements by getting off the table, but rather the foot and leg were turned to the neutral position while the patient remained supine.
Results Precision errors for the femoral neck and total hip in the 65 patient cohort were significantly different. The CV% was 2.65% for the femoral neck and 1.7% for the hip. Bone mineral density SD (RMS) was 0.019 g/cm2 and 0.013 g/cm2 for the femoral neck and total hip, respectively. The difference in BMD, calculated as the neutral position BMD minus the internal rotation BMD, varied in degree and sign. Table 1 shows that the median difference of the femoral neck was not significantly different than that of the total hip. In contrast, the median difference of the absolute BMD values, i.e., the magnitude of
497
change regardless of whether it is positive or negative, of the femoral neck was significantly greater than the median of the absolute differences of the total hip. Table 2 details the relationships between the two sites with regard to gains and losses of BMD between the neutral and internal rotation positions. In 64%–68% of the patients there was an increase in BMD in the neutral position, whereas 13% of the patients manifested a femoral neck increase and a concomitant total hip decrease; the reverse occurred in 16.5%, for an overall 29.5% disagreement. A difference between two measurements exceeding 2.77 times the precision CV% is considered a significant change with 95% confidence; these are 7.3% and 4.8% for the femoral neck and total hip, respectively, when derived from our precision error determinations. Applied to the absolute percent change, no significant change was registered between internal rotation and neutral positions in 178 patients (89%) for the femoral neck and 174 patients (87%) for the total hip. Similarly, using 2.77 times the precision SD, no significant absolute change occurred in 173 patients (86.5%) for the femoral neck and total hip. Thus, 11%–13.5% of the patients will sustain a significant change between measurements in the internal rotation and neutral positions. Spearman correlations between the various parameters were calculated (Table 3). There were significant negative correlations between the femoral neck and total hip BMDs at internal rotation with age, and positive correlations with body mass index (BMI). Similar results were obtained for the neutral position. The absolute and percent absolute change in the total hip BMD correlated positively with age, whereas that of the fe-
Table 2. Number of patients (n) with gains or losses in BMD with change (∆) between neutral position and internal rotation of the leg in the femoral neck, total hip, and combinations thereof Site (s)
Table 1. BMD differences and absolute differences between neutral and internal rotation of the leg at the femoral neck and total hip Differences Femoral neck ∆ Total hip ∆ abs femoral neck ∆ abs total hip ∆ abs femoral neck %∆ abs total hip %∆
Median BMD 0.011 0.009 0.025 0.016 3.129 1.795
P*
BMD range
0.844
⫺0.086 to 0.122 ⫺0.069 to 0.137 0 to 0.122 0 to 0.137 0% to 19.2% 0% to 20.4%
0.001 0.000
∆, difference of neutral BMD minus internal rotation BMD; abs, absolute * Wilcoxon signed-rank test
Femoral neck ∆ ⱖ 0 Total hip ∆ ⱖ 0 Femoral neck ∆ ⬍ 0 Total hip ∆ ⬍ 0 Femoral neck ∆ ⱖ 0 and total hip ∆ ⱖ 0 Femoral neck ∆ ⬍ 0 and total hip ∆ ⬍ 0 Femoral neck ∆ ⱖ 0 and total hip ∆ ⬍ 0 Femoral neck ∆ ⬍ 0 and total hip ∆ ⱖ 0 No agreement
n
Percent
129 136 71 64 103
64.5 68.0 35.5 32.0 51.5
38
19.0
26
13.0
33
16.5
59
29.5
498
L. Rosenthall: Femoral neck/hip BMD change with rotation
moral neck did not, nor was there a significant association between these variables and BMI. An analysis of the absolute percent change as a function of age tertile is detailed in Table 4. The medians of the femoral neck changes are significantly greater than the total hip in each of the three groups. Within the femoral neck group there is a progressive increase in median percent difference with age, but they are not significant by the Kruskal–Wallis multiple comparison test. In contrast, there is a significant increase in the total hip median changes with age.
Discussion Internal rotation of the leg functions to overcome the anteversion of the femoral neck axis and present the smallest cross section of the neck to the incident scanning X-rays. Although it is readily attained and confirmed with ex vivo phantom studies, in clinical practice there is no confirmatory evidence that the plane of the Table 3. Spearman correlations (rho, r) Site
Age rho (P*)
BMI rho (P*)
Femoral neck BMD (internal rotation) Total hip BMD (internal rotation) Absolute femoral neck difference Absolute total hip difference Absolute femoral neck %difference Absolute total hip % difference
⫺0.354 (0.000) ⫺0.315 (0.000) 0.015 (0.826) 0.258 (0.000) 0.082 (0.246) 0.314 (0.000)
0.223 (0.001) 0.306 (0.000) 0.024 (0.727) 0.041 (0.564) ⫺0.011 (0.874) ⫺0.001 (0.989)
* Probability
femoral axis is truly parallel to the table top in patient measurements with internal rotation of the leg. In addition to the fact that anteversion varies in adults from 5° to 15°, such factors as hip disorders, patient frailty, and unnoticed patient movement, even with the foot firmly strapped to the foot positioner, may detract from the desired parallelism. The reproduction of the same rotation of the femoral neck axis in longitudinal studies is also subject to the same uncertainties plus the factor of intra- and interoperator variability [3,11]. The latter variance can be reduced by having the same operator perform all the examinations in follow-up, but this is not realistic in most institutions because of absenteeism and personnel turnover with time. Distribution of bone mineral in the femoral neck can be quite heterogeneous, and a minor displacement of the small region of interest can yield a remarkably different bone mineral density [12]. The intent of this study was to evaluate the range of difference that might accrue from errors of repositioning in the context of the various mitigating factors that compromise accuracy in a routine clinical service facility. The transition from the internal rotation to neutral positions is associated with an increased BMD difference in about 65% of the patients for both the femoral neck and total hip, but it should be emphasized that there is a decrease in BMD in 35%. Furthermore, in about 30% of the patients the femoral neck and total hip changes are of opposite sign (see Table 2). This difference can be a source of confusion in serial monitoring of patients for their bone mineral status and emphasizes the need for accurate repositioning. In terms of absolute difference and absolute percent difference, the medians of the femoral neck were significantly greater than the total hip by 56% and 74%, respectively. This result is a reflection of the greater variability in the magnitude of the femoral neck changes compared to the total hip for the regions of interest employed. However, using the femoral neck 95% confidence intervals of
Table 4. Percent change as a function of age tertile Age tertile ranges (years) 21.0 to ⱕ55.4 ⬎55.4 to ⱕ66.2 ⬎66.2 to 86 Kruskal–Wallis**
Absolute %femoral neck change, median (95% CIL)
Absolute %total hip change, median (95% CIL)
2.05 (T1) (1.73, 3.23) 3.26 (T2) (2.04, 3.69) 3.99 (T3) (2.59, 4.52)
1.21 (T1) (0.78, 1.44) 1.95 (T2) (1.08, 2.44) 2.92 (T3) (2.00, 3.67)
T3 ⫽ T2 ⫽ T1
T3 ⬎ T2 ⬎ T1
CIL, confidence interval limits * Wilcoxon signed-rank test for medians ** Kruskal–Wallis multiple comparison Z test for medians (ANOVA)
P* 0.001 0.000 0.040
L. Rosenthall: Femoral neck/hip BMD change with rotation
7.3% and 4.8% for the total hip, which are based on our precision study, the number of absolute percent change values exceeded the limit for the femoral neck in 22 patients (11%) and 26 patients (13%) for the total hip. It would seem that the number of false positives is relatively small and approximately equal for both sites, and this raises the question as to which site is preferable for longitudinal studies in clinical practice. If confirmed accurate repositioning can be achieved, whether at zero anteversion or otherwise, then the total hip is preferable because its associated precision error is 35% smaller than that of the femoral neck and smaller changes in BMD can thereby be disclosed. Positioners more sophisticated than the simple foot block have been designed and reported to be more precise, but they were investigated in a research setting and definitive improvement is yet to be proved in a large unselected cohort [7,13]. It can be argued that the hip should not be internally rotated, but rather scanned in the neutral position for the total hip measurement. This method simply requires that the patient lie supine with the foot fixed perpendicular to the table top, and thereby potentially reduces the repositioning inaccuracies of internal rotation that prevail in longitudinal studies. To be determined are the precision errors for the femoral neck and total hip in the neutral position to establish the confidence limits peculiar to each service facility. It is anticipated that the precision error will remain considerably smaller for the total hip than the femoral neck. There was no correlation of femoral neck and total hip differences with age, because of the mix of gain and loss of BMD with rotation from internal rotation to neutral position. When expressed as absolute difference, there was a significant positive Spearman correlation (rho, r) between total hip and age, which was also shown by age tertile analysis of the BMD medians (see Tables 3,4). r was not significant for absolute femoral neck difference, but the tertile analysis demonstrated a trend in that there was a progressive BMD median increase with age that did reach significance by the Kruskal–Wallis multiple comparison Z test.
499
BMD of the femoral neck and total hip decreased with age as expected (see Table 3), but the reason for the total hip increase in absolute difference with age is speculative. References 1. Tonnis D, Heinecke A (1999) Acetabular and femoral anteversion: relationship with osteoarthritis of the hip. J Bone Joint Surg [Am] 81:1747–1770 2. Mesgarzadeh M, Revesz G, Bonakdarpour K (1987) Femoral neck torsion angle measured by computed tomography. J Comput Tomogr 11:799–803 3. Phillipot G, Seaborn CJ, Phillips PJ (2001) Reproducibility of DXA: potential impact on serial measurements and misclassification of osteoporosis. Osteoporos Int 12:49–54 4. Wahner HW, Looker A, Dunn WL, Walters LC, Hauser MF, Novak C (1994) Quality control of bone densitometry in a national health survey (NHANES III) using three mobile examining centers. J Bone Miner Res 9:951–960 5. Wilson CR, Fogelman I, Blake GM, Rodin A (1991) The effect of positioning on dual energy X-ray bone densitometry of the proximal femur. Bone Miner 13:69–76 6. Maggio D, McCloskey, Camilli L, Cenci A, Cherubini A, Kanis JA, Senin U (1998) Short-term reproducibility of proximal femur bone mineral density in the elderly. Calcif Tissue Int 63:296–299 7. Goh JCH, Low SL, Bose K (1995) Effect of femoral rotation on bone mineral density measurements with dual energy X-ray absorptiometry. Calcif Tissue Int 57:340–343 8. The International Society for Clinical Densitometry (2004) Official positions. J Clin Densitometry 7:1–6 9. Svendsen OL, Marlow U, Hassager C, Christiansen C (1992) Measurements of bone mineral density of the proximal femur by two commercially available x-ray absorptiometric systems. Eur J Nucl Med 19:41–46 10. Gluer C, Blake G, Lu Y, Blunt BA, Jerges M, Genant HK (1995) Accurate assessment of precision errors: how to measure the reproducibility of bone densitometry techniques. Osteoporos Int 5:262–270 11. Fuleihan GE, Testa MA, Angell JE, Porrino N, Leboff MS (1995) Reproducibility of DXA absorptiometry: a model for bone loss estimates. J Bone Miner Res 10:1004–1014 12. Blake G, Tong CM, Fogelman I (1991) Intersite comparison of the Hologic QDR-1000 dual energy X-ray bone densitometer. Br J Radiol 64:440–446 13. Hans D, Duboeuf F, Schott AM, Horn S, Avioli LV, Drezner MK, Meunier PJ (1997) Effects of a new positioner of hip bone mineral density measurement. J Bone Miner Res 12:1289– 1294