Relative validity of Dual-energy X-ray Absorptiometry in the assessment of visceral adipose tissue. A validation study in colorectal cancer patients, using Computed Tomography as reference method.

Abstract

Objective: Visceral adipose tissue (VAT) is mainly found around the abdominal organs, and is an established risk factor for colorectal cancer (CRC). High levels of VAT increase the risk of CRC and are shown to predict poor prognosis and decreased survival in CRC patients. There is a considerable need for accurate tools to detect VAT. Dual-energy X-ray Absorptiometry (DXA) is a safe, quick and accurate method for measuring body composition. However, the method is not commonly used for body composition purposes in the clinic. The aim of this current thesis was to validate VAT and other body compartments measured by DXA, against superior method Computed Tomography (CT). Subjects and methods: CRC patients with a maximum of 45 days between the DXA and CT scans from the CRC-NORDIET study were included in this study. The abdominal area of each subject was individually tailored and generated by DXA, and thereby matched against corresponding CT images. Automated segmentation was performed by a software based on artificial intelligence (BodySegAI), followed by semi-automated corrections in sliceOmatic. Validity and associations were examined for VAT, subcutaneous adipose tissue (SAT), anthropometric methods and fat-free mass (FFM). Results: VAT measured by DXA explained 89 % of the variation of VAT measured by CT for the 66 subjects included. The mean difference of VAT between the two methods was 189 cm3, equal to 178 g (p<0.001). The mean overestimation of VAT by DXA was significantly lower for the subjects with higher amounts of VAT in comparison with subjects with lower amounts of VAT (5% vs. 37% overestimation, p= 0.01). Anthropometric measures showed similar association to VAT compared to other studies (R2=0.39-0.65), where waist circumference (WC) was found to be the superior method. DXA demonstrated an underestimation of SAT, which increased with higher SAT values, but not with increasing BMI. The mean difference for SAT between the two methods was 516 cm3, equal to 487 g (p<0.001). SAT measured by DXA explained 78 % the variance in SAT measured by CT. There was poor agreement between DXA and CT in the measurement of FFM (R2=0.56). Conclusion: This study confirms that DXA is an acceptable tool for measuring VAT, but provides less accurate measurements of SAT and FFM, compared to CT. The amounts of fat may influence the DXA’s accuracy. The anthropometric methods did not predict VAT accurately, which highlights the importance of using more sophisticated tools such as DXA in the measurement of VAT and body composition in general.

Objective: Visceral adipose tissue (VAT) is mainly found around the abdominal organs, and is an established risk factor for colorectal cancer (CRC). High levels of VAT increase the risk of CRC and are shown to predict poor prognosis and decreased survival in CRC patients. There is a considerable need for accurate tools to detect VAT. Dual-energy X-ray Absorptiometry (DXA) is a safe, quick and accurate method for measuring body composition. However, the method is not commonly used for body composition purposes in the clinic. The aim of this current thesis was to validate VAT and other body compartments measured by DXA, against superior method Computed Tomography (CT). Subjects and methods: CRC patients with a maximum of 45 days between the DXA and CT scans from the CRC-NORDIET study were included in this study. The abdominal area of each subject was individually tailored and generated by DXA, and thereby matched against corresponding CT images. Automated segmentation was performed by a software based on artificial intelligence (BodySegAI), followed by semi-automated corrections in sliceOmatic. Validity and associations were examined for VAT, subcutaneous adipose tissue (SAT), anthropometric methods and fat-free mass (FFM). Results: VAT measured by DXA explained 89 % of the variation of VAT measured by CT for the 66 subjects included. The mean difference of VAT between the two methods was 189 cm3, equal to 178 g (p<0.001). The mean overestimation of VAT by DXA was significantly lower for the subjects with higher amounts of VAT in comparison with subjects with lower amounts of VAT (5% vs. 37% overestimation, p= 0.01). Anthropometric measures showed similar association to VAT compared to other studies (R2=0.39-0.65), where waist circumference (WC) was found to be the superior method. DXA demonstrated an underestimation of SAT, which increased with higher SAT values, but not with increasing BMI. The mean difference for SAT between the two methods was 516 cm3, equal to 487 g (p<0.001). SAT measured by DXA explained 78 % the variance in SAT measured by CT. There was poor agreement between DXA and CT in the measurement of FFM (R2=0.56). Conclusion: This study confirms that DXA is an acceptable tool for measuring VAT, but provides less accurate measurements of SAT and FFM, compared to CT. The amounts of fat may influence the DXA’s accuracy. The anthropometric methods did not predict VAT accurately, which highlights the importance of using more sophisticated tools such as DXA in the measurement of VAT and body composition in general.