Assessing Thoracic Vertebral Bone Mineral Density (T8-T10) for Osteoporosis Diagnosis During CT Lung Cancer Screening in Older Adults

Figures & data

Table 1 Characteristics of the Subjects

	Overall Population (n = 701)	Women (n = 383)	Men (n = 318)
Age (years)	65.5 ± 12.5	63.0 ± 11.9	67.2 ± 12.6
Postmenopause (yes)	N/A	362	N/A
BMI (kg/m^2)	23.24 ± 2.77	22.4 ± 2.86	24.03 ± 2.63
BMD-L1+L2 (mg/cm^3)	96.7 ± 36.3	88.6 ± 33.7	102.4 ± 37.4
BMD-T8 (mg/cm^3)	113.3 ± 39.7	106.0 ± 42.5	119.5 ± 36.1
BMD-T9 (mg/cm^3)	116.8 ± 41.3	111.7 ± 44.5	121.3 ± 37.9
BMD-T10 (mg/cm^3)	117.3 ± 41.1	112.3 ± 43.5	121.6 ± 38.6
BMD-T8+T9+T10 (mg/cm^3)*	115.9 ± 39.8	105.3 ± 38.7	122.5 ± 40.6

Note: *p < 0.05, vs BMD-L1+L2 both in men and women.

Abbreviations: BMD, bone mineral density; BMI, Body mass index.

Figure 1 Role of thoracic vertebral bone mineral density (BMD-T8+T9+T10) in identifying osteopenia and osteoporosis based on one American College of Radiology (ACR) criterion. The difference between thoracic vertebral bone mineral density (BMD-T8-T10) and lumbar bone mineral density (LBMD) (A) The kappa test showed moderate agreement in identifying osteoporosis between BMD-T8-T10 and lumbar spine BMD (kappa = 0.51) (B). Compared to those in the gold standard group, the proportions of osteopenia and osteoporosis defined by BMD-T8+T9+T10 groups were significantly different (C) ***, P < 0.001.

Table 2 Performance of Low Thoracic Vertebral Bone Mineral Density (BMD-T8+T9+T10)

	Osteoporosis					Osteopenia
	Accuracy	PPV	NPV	Se	Sp	Accuracy	PPV	NPV	Se	Sp
BMDa	87.2%	63.9%	99.6%	98.7%	83.4%	77.5%	46.9%	98.9%	98.4%	56.1%
BMDb	96.0%	100%	93.9%	89.7%	100%	87.4%	81.1%	86.2%	88.3%	77.6%
TTBMD	93.5%	90.1%	95.0%	90.6%	95.2%	88.4%	87.1%	78.5%	84.3%	82.0%

Notes: a based on the thresholds of 120 mg/cm³ and 80 mg/cm³. b based on the thresholds of 138 mg/cm³ and 97 mg/cm³.

Abbreviations: PPV, positive prediction value; NPV, negative positive value; Se, sensitivity; Sp, specificity.

Figure 2 The performance of the thoracic vertebral bone mineral density (BMD-T8+T9+T10) in identifying osteopenia and osteoporosis. Receiver operating characteristic curves (ROC) were generated for the identification of osteopenia (A) and osteoporosis (B). The thresholds were 138 mg/cm³ and 97 mg/cm³. Then, we identified osteopenia and osteoporosis using these two thresholds. The kappa test showed good agreement in identifying osteoporosis between BMD-T8+T9+T10 and lumbar spine BMD (kappa = 0.74) (C). Compared to those of the gold standard group, the proportions of osteopenia and osteoporosis defined by the two thresholds were not significantly different (p > 0.05) (D). NS: not significant.

Figure 3 The Bland‒Altman plot also showed good agreement between lumbar bone mineral density (LBMD) and translated thoracic BMD (TTBMD) (A) A formula (LBMD = 0.9 × BMD-T8-T10 – 2.56) was obtained to translate the LBMD based on the thoracic BMD (BMD-T8-T10) (B) and the performance of the translated LBMD in identifying osteopenia and osteoporosis was evaluated. The kappa test showed good agreement in identifying osteoporosis between the TTBMD and LBMD (kappa = 0.74) (C) Compared to those of the gold standard group, the the proportions of osteopenia and osteoporosis defined by TTBMD were not significantly different (p > 0.05) (D) NS: not significant.