Intended for Healthcare Professionals
You are here
Home > Blog > Endocrinology > Bone Mineral Density Screening Combined with Mammogram

Bone Mineral Density Screening Combined with Mammogram

Improving Bone Mineral Density Screening by Using Digital X-Radiogrammetry (DXR) Combined With Mammography

Bone Mineral Density: THE STUDY

Osteoporosis is often underdiagnosed and undertreated, especially in women.  In Australia, 2 in 5 women aged 50 or over will experience fragility fractures. Fractures associated with reduced bone mineral density (BMD) may cause disability and even result in death. Severe fractures involving hip, spine, or femur are estimated to cause up to 30 % additional mortality. Low BMD is the single best predictor of fracture risk in asymptomatic postmenopausal women.  Dual-energy x-ray absorptiometry (DXA) of the hip and spine is considered the “gold standard” for measurement of BMD. However, due to cost and accessibility not all eligible Australian women ( >65 years, or <65 years ) are evaluated with DXA. One accessible technique trialed was Digital X-ray radiogrammetry (DXR). DXR estimates bone mineral density in the hand (DXR-BMD) through an automated radiogrammetric analysis.  This means that DXR can be opportunistically integrated in existing mammography screening with minimal impact on the workflow. Previous studies have shown that DXR of the nondominant hand and femoral neck DXA has significant correlation. 

The purpose of this study is to evaluate the performance of DXR and DXA at different sites in Australian women over 50 years presenting for mammography as well as to find the optimal threshold for DXR to screen osteoporosis as diagnosed by central DXA.


Between August 2014 and February 2016, 216 women aged >50 years and were postmenopausal or ≥ 55 years with a prior hysterectomy were recruited from the Breast Cancer Institute of Westmead Hospital for participation in the study.  Briefly, at the baseline visit, DXR images were acquired immediately after mammography. The nondominant hand was examined except for participants with a hand fracture or functional injury of ≥ 6 months duration. All participants then underwent a second examination that included measurement of the forearm, both hips, and lumbar spine BMD by DXA.

Clinical risk factors 

Participants completed a questionnaire and were interviewed at the baseline examination about medical history, falls, history of fracture since the age of 50, age at menopause, and factors affecting BMD. Radiography staff was also given a questionnaire at the end of the study recruitment.

Statistical analysis 

A linear regression model was adjusted for body composition data, TBS and DXR, and DXA T-scores and BMD values. The concordance of DXR and DXAT scores at each site was determined using Lin’s concordance correlation coefficient, where values of 1 reflect perfect concordance. The Bland-Altman method was used to plot Limits of agreement (LOA). Receiver operator characteristic (ROC) curve analysis was used to illustrate the relationships between DXR and DXA. Deming regression was used to assess the agreement between DXR and DXA. Youden’s method in conjunction with ROC curve analysis was used to identify the optimal threshold DXR to diagnose osteoporosis. All statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC, USA).


A total of 200 women were included in the analysis. Average (SD) age was 64 ± 7 years and 163 (82%) had a diagnosis of breast cancer with 3.7 years median period since diagnosis. About 28% of women used menopausal hormonal therapy. By DXA, 157 (78.5%) participants had a T-score ≤ -1 at any spine, hip, or forearm site, whereas 61 (30.5%) had a T-score ≤ -2.5 at any site. Of 72 participants under 60 years of age, 35 (48.6%) had T-scores at the hip or spine in the osteopenic range whereas 6 (8.3%) had T-scores in the osteoporotic range. Of 128 participants aged 60 and over, 66 (51.6%) had T-scores in the osteopenic range and 22 (17.2%) in the osteoporotic range. By DXR, about 132 women (66%) had a T-score ≤ -1, and 52 (26%) had a T-score ≤ -2.5 by DXR compared to DXA. 

Lin’s concordance correlation coefficient between DXR and DXA at the spine or hip was significantly low.  However, there was a significant concordance coefficient between DXR and the one-third radius by DXA, with Lin’s concordance coefficient 0.64 (95% CI, 0.56–0.71; P < 0.001) and 95% LOA assessed using Bland-Altman plots of -1.32 to 2.35. 

For DXR and DXA T-scores ≤ -2.5, the AUC ROC was 0.87 (95% CI, 0.81–0.94) at the 1/3 radius, and 0.74 (95% CI, 0.64–0.84) for hip or spine. DXR T-scores > -1.98 provided a high probability of not having osteoporosis (range 88%, 98%) by central DXA. 


In the present report, it was demonstrated that DXR correlates well with DXA. A DXR value of < -1.98 was established as the optimal threshold for DXR to screen for DXA-defined osteoporosis. Although some women with a DXR value of < – 1.98 would be misclassified to have osteoporosis, those women would be identified if this threshold was the basis for proceeding to DXA evaluation. DXR value of > -1.98 indicated a relatively lower risk of DXA osteoporosis at the hip or spine, with a negative predictive value of 94% (range 88%, 98%). Nevertheless, the major concern for a DXR screening is that some women with a DXR T-score ≥ – 1.98 may be misclassified to have a BMD above the osteoporotic threshold, despite having osteoporosis at the hip or spine by DXA.

The participants in this study had multiple risk factors for low BMD. First, all of them were postmenopausal, with a mean age at menopause of 49.8 ± 4.7 years, and the mean age at the time of examination was 64 ± 7 years. Second, the majority had a breast cancer diagnosis, and many had undergone chemo or postmenopausal hormone therapy. Also, about 37% had a prior fracture. 

Of the 200 women included in the study, 14% were found to have a T-score ≤ – 2.5 at the hip or spine by DXA. About 50.5% of the participants were osteopenic at the hip or spine. For the participants under the age of 60 years (n = 72), rates of osteoporosis and osteopenia based on DXA at the hip and spine were 8.3% and 48.6%, respectively. This was per the results of previous studies done on Australian women. For women aged 60 years and over (N =128), osteopenia was present in 51.6%, which is within the expected range of 51.4% – 48% for Australian women aged 60–69 years. Only 17.2% of these women had DXA hip or spine T-scores in the osteoporotic range, despite apparent increased risk factors. 

The study has limitations, with a cohort consisting of women who had a diagnosis of breast cancer and some were on medications that may have reduced BMD. The data may not apply to younger women or other ethnic groups since the mean age of the cohort was 64 ± 7 years and the participants were predominantly white. Most fractures occur in the participants with BMD in the osteopenic range, and a DXR value of > – 1.98 does not exclude fracture risk despite a low risk for osteoporosis by DXA. DXR results aren’t considered for the calculation of absolute fracture risk (AFR). Finally, cost benefit analysis was not conducted in this study.  


Oncology Related Tools


Latest Research

bone mineral density

Similar Articles

Leave a Reply