Emphysema as a 'radiomic-biomarker' using low-dose computed tomography scans in highrisk smokers in an Australian cohort

Aim: Automated software can quantify emphysema in low dose spiral computed tomography chest scans (LDCT) but its utility in lung cancer risk prediction is unknown. We aimed to quantify the amount and distribution of emphysema using LDCT in a high risk Australian lung cancer screening population. Methods: Current or former smokers aged 55-80 years were recruited at St Vincent’s Hospital, Sydney as part of the International Lung Screening Trial www.clinicaltrials.gov (NCT02871856) and performed spirometry and LDCT. Commercial emphysema software (Philips IntelliSpace Portal CT COPD) calculated the amount and distribution of emphysema (using a standardised threshold of -950 Hounsfield Units, HU) on 0.65 -2mm thin scan slices. Emphysema amount (%) was determined by the volume of lung with <-950HU as a ratio of the corresponding lung volume. Lung nodules were characterised by size, type and location. Results: 83 patients (51 male; 44 former & 39 current smokers; mean+SD age 65+years; smoking pack-pack years 48+23) with normal spirometry (FEV1/FVC 0.73+0.10, %predicted FEV1 91+11%) had a small amount of total emphysema (4.6+ 6.5%). There was no difference in the total amount of emphysema between the right (R) (4.6+ 6.9%) & left (L) lung(4.5+6.3%) and between the R upper (UL 4.5+8.9%) and lower lobes (LL 3.9+6.4%). The LUL (5.3+7.1%) had more emphysema compared to the LLL (3.5+5.6%) and RUL (p<0.05). Lung nodules were detected in 47/83 patients and 36 nodules were >5mm. Of these nodules, 10/36 had suspicious features (8 in UL; 2 in LL) Conclusion: Analysis has shown that LDCT detected a modest degree of emphysema in high risk smokers in the presence of normal spirometry.