Peripheral Site TScore Equivalents for the Diagnosis of Osteopenia and Osteoporosis

Attempts to determine site, device, and manufacturer-specific T-scores or T-score ranges at peripheral sites to identify women who are osteopenic or osteoporotic at the spine or proximal femur generally involve the calculation of sensitivity, specificity and area under the receiver operating characteristic curve (AUC) as discussed in Chapter 3.

Fig. 9-1. The T-score regression on age for women by site and device, from the reference databases of the various manufacturers. The expected T-score for an average 60-year-old woman ranges from -0.7 at the heel by QUS to -2.5 at the spine by QCT. Reproduced with permission of the publisher from ref. 15.

Fig. 9-1. The T-score regression on age for women by site and device, from the reference databases of the various manufacturers. The expected T-score for an average 60-year-old woman ranges from -0.7 at the heel by QUS to -2.5 at the spine by QCT. Reproduced with permission of the publisher from ref. 15.

In a study from Pouilles et al. (16), 234 women, ranging in age from 45 to 60 years, underwent DXA studies of the PA lumbar spine and proximal femur (Lunar DPX-IQ) and pDXA forearm studies (Norland pDEXA). The average peak BMD and SD for each skeletal site was determined from studies of a group of young, healthy women, rather than from the manufacturers' reference databases. At the PA lumbar spine, the young-adult mean BMD and SD were 1.180 ±0.12 g/cm2. The values at the femoral neck were 0.990 ±0.11 g/cm2. At the forearm, the values for the distal region of interest were 0.347 ±0.05 g/cm2 and at the proximal region, 0.816 ±0.06 g/cm2. Using the WHO Criteria of less than or equal to -2.5 SD below the young-adult mean bone density for a diagnosis of osteoporosis, 17.5% of the women were osteoporotic at the spine and femoral neck combined. Fourteen percent were osteoporotic based on measurements at the proximal forearm and only 3.4% were osteoporotic based on measurements at the distal forearm. The authors then determined the number of SD below the young-adult mean BMD or T-score that would be required at the distal and proximal forearm for 95% sensitivity in diagnosing osteoporosis or a combination of osteopenia and osteoporosis at the spine and proximal femur. At the distal forearm, a T-score of -0.74 was required for 95% sensitivity for the diagnosis of osteoporosis at the central sites. The corresponding specificity was 59.5% and the AUC was 0.80. At the proximal forearm, the T-score cutpoint was -0.73 with a corresponding specificity of 46.7% and an AUC of 0.75. To detect both osteopenic and osteoporotic bone densities at the central sites, a T-score of+1.06 was required at the distal forearm and +0.53 at the proximal forearm for 95% sensitivity. The corresponding specificities were 23.3% and 23.4%, respectively.

In 115 women (average age 61 years), Varney et al. (14) measured bone density at the PA spine and proximal femur with DXA (Hologic QDR 4500A), at the phalanges with DXA (Schick accuDEXA), and at the heel with QUS (Hologic Sahara). Using the WHO Criteria of a T-score of -2.5 to diagnose osteoporosis, 28% of the women were osteoporotic based on measurements at the spine, total hip, and femoral neck combined. Forty-nine percent were osteopenic and 23% were normal. The authors determined that a T-score of -1.9 or less was necessary to diagnose a similar percentage of women as osteoporotic using QUS of the heel, using the estimated heel BMD in g/cm2 as provided by the software. A T-score of -1.4 or less for phalangeal DXA was required to diagnose a similar percentage of women as osteoporotic. At these T-score thresholds, the sensitivity and specificity were 56 and 78% for QUS and 56 and 80% for phalangeal DXA, respectively. The authors also determined the equivalent T-score range for osteopenia for the two devices. For QUS of the heel, a similar percentage of women were diagnosed as osteopenic when osteopenia was defined as a T-score between -1.9 and +0.3. For DXA of the phalanges, the equivalent osteopenic range was between -1.4 and +0.4. The sensitivities and specificities for the two ranges were 86% sensitivity and 48% specficity and 89% sensitivity and 59% specificty, respectively.

T-score thresholds for phalangeal DXA were also determined by Mulder et al. (17). One hundred twenty-three women (average age 64.6 years) underwent PA spine, proximal femur, and forearm DXA measurements (Hologic QDR-4500) and phalangeal DXA measurements (Schick accuDEXA). In contrast to the definition used by Varney et al. (14), the authors of this study defined osteoporosis as a T-score of -2.5 or poorer at the femoral neck only. A T-score of -2.5 at the finger had a sensitivity of only 35% for osteoporosis at the femoral neck although the specificity was 88%. Better sensitivity was obtained with a phalangeal T-score of -1.0. At this T-score, the sensitivity for femoral neck osteoporosis was 85% although specificity fell to 49%. This phalangeal T-score threshold also had excellent sensitivity for femoral neck T-scores of -2 or poorer. Here the sensitivity was 82% with a specificity of 56%.

Fiter et al. (18) suggested that a T-score of -1.65 provided a better combination of sensitivity and specificity for phalangeal DXA for the diagnosis of osteoporosis at the PA lumbar spine or femoral neck. In a study of 230 postmenopausal women (average age 58.4 years), BMD was measured at the PA lumbar spine and femoral neck with DXA (Hologic QDR 1000) and at the middle phalanx with DXA (Schick accuDEXA). Applying the WHO Criteria, the prevalence of osteoporosis was 33% based on measurements of the PA lumbar spine and femoral neck. At the middle phalanx, however, the prevalence of osteoporosis using the WHO threshold of 2.5 SD below the young-adult mean value was only 18%. Based on an AUC analysis, the authors determined that the optimum sensitivity and specificity for the middle phalangeal DXA measurement for diagnosing osteoporosis at the PA lumbar spine and femoral neck was a T-score of -1.65. At this T-score, the sensitivity was 75% with a specificity of 77%. The AUC was 0.822.

Optimum T-score thresholds for a different heel QUS device were determined by Damilakis et al. (19). Bone density was measured at the PA lumbar spine, femoral neck, and total hip with DXA (Hologic QDR 1000) and at the heel with QUS (Ubis 3000, DMS, France) in 333 women, 42 to 79 years of age. Using the WHO Criteria of a T-score of -2.5 or poorer, osteoporosis was found in 14% of the women at the spine, 13% at the femoral neck, and 10% at the total femur, for an average prevalence of 12%. Using a T-score of -2.5 or less based on the QUS measurement at the heel for BUA

resulted in only 1.5% of the women being diagnosed as osteoporotic. The SOS T-score of -2.5 resulted in a diagnosis of osteoporosis in only 8.4%. In order to achieve a similar percentage of women diagnosed as osteoporotic as seen at the central sites by DXA, the T-score thresholds for BUA and SOS had to be adjusted to -1.8 and -2.2, respectively. The authors also used sensitivity, specificity and AUC analyses to determine the optimum T-score thresholds for BUA and SOS for detecting osteoporosis at the femoral neck. In that analysis, the optimum T-score for BUA was -1.3. The sensitivity and specificity at this T-score was 68% and 83%, respectively with an AUC of 0.82. The optimum T-score threshold for SOS was -1.5. The sensitivity and specificity for the T-score of -1.5 was 63% and 79%, respectively. The AUC was 0.75. The difference between the AUC for the optimum BUA and SOS T-score thresholds was statistically significant, suggesting that BUA was a better predictor of low femoral neck bone density than was SOS.

Fordham et al. (20) attempted to define an optimum T-score threshold for the heel measured by DXA in 443 women with a mean age of 60 years. Bone density was measured at the PA lumbar spine and femoral neck with DXA (Lunar DPX-L) and with DXA at the heel (Lunar PIXI). Using the manufacturer's reference database, osteoporosis was considered to be present if either the PA lumbar spine or femoral neck T-score was -2.5 or less. Based on sensitivity, specificity, and AUC analyses, the optimum heel T-score threshold for identifying women who were osteoporotic at the spine or femoral neck was -1.3 or poorer for the entire group. At this T-score, the sensitivity was 69.6% and the specificity was 82.6%. The AUC was 0.836. The authors also noted that the sensitivity and specificity varied by age. In women age 50 to 64, a heel DXA T-score of -1.3 or poorer had a sensitivity and specificity of 68% and 85%, respectively, with a false-negative rate of 13%. In women age 65 and older, however, the same T-score threshold resulted in a sensitivity and specificity of 71% and 68%, respectively, with a false-negative rate of 40%.

Pacheco et al. (21) also attempted to determine the optimum T-score threshold for BMD measured at the calcaneus by DXA for the diagnosis of osteoporosis at the PA lumbar spine, total femur, or femoral neck. Bone density was measured in 204 Caucasian women age 20 and older at the heel (Lunar PIXI) and at the PA lumbar spine and proximal femur (Lunar DPX-L and Hologic QDR 4500). Osteoporosis was defined using the WHO Criteria for the diagnosis of osteoporosis and femoral neck T-scores were calculated using the NHANES III reference database. At a T-score of -2.5 at the calcaneus, the sensitivity for osteoporosis at any of the three central sites was only 20.3% although the specificity was 97.1%. As in the study from Fordham et al. (20), the optimum T-score threshold at the calcaneus for the diagnosis of central osteoporosis based on a ROC analysis appeared to be a T-score of -1.3, which the authors noted coincided with a calcaneal BMD of 0.390 g/cm2 on the Lunar PIXI.

Using a different heel DXA device (Norland Apollo), Sweeney et al. (22) measured bone density in the calcaneus in 119 women with an average age of 51.6 years. Bone density was also measured at the PA lumbar spine and femoral neck with DXA (Norland Eclipse). Using the established WHO Criteria for the diagnosis of osteopenia and osteoporosis at the spine, 30% of the women were osteopenic and 17% were osteoporotic. At the femoral neck, 41% were osteopenic and 21% were osteoporotic. At the heel, 32% were osteopenic and only 2% were osteoporotic. The authors attempted to identify the heel DXA T-score threshold that would identify women with spine or femoral neck

Table 9-2

Equivalent T-Scores and T-Score Ranges at Peripheral Sites by Specific Devices for Osteopenia and Osteoporosis at the PA Lumbar Spine and/or Proximal Femur

Osteopenia Osteoporosis

Table 9-2

Equivalent T-Scores and T-Score Ranges at Peripheral Sites by Specific Devices for Osteopenia and Osteoporosis at the PA Lumbar Spine and/or Proximal Femur

Osteopenia Osteoporosis

Device

Site/ROI

Equal Prevalence

Optimum Sensitivity

Equal Prevalence

Optimum Sensitivity

pDEXAa

Distal Proximal

Was this article helpful?

0 0

Post a comment