Keywords:
Tissue characterisation, Cancer, Segmentation, Decision analysis, Computer Applications-Detection, diagnosis, Image manipulation / Reconstruction, CT-Quantitative, CT, Thorax, Oncology, Computer applications
Authors:
G. Ficarra1, E. Barabino1, C. Genova1, S. Mennella1, M. Verda2, G. Pittaluga1, F. Grossi1, G. Cittadini1; 1Genova/IT, 2Imperia/IT
Methods and Materials
A retrospective analysis of all consecutive patients receiving anti-PD-1 antibodies (nivolumab) for advanced NSCLC after failure of at least one line of chemotherapy,
was conducted at the San Martino Policlinic,
University of Genoa,
Italy,
between January 2015 and April 2017 (n=74).
Informed consent was obtained from all patients.
Patients with a concomitant second cancer,
who had received radiation therapy,
without baseline or follow up CT scan or with inadequate CT evaluation (mainly non contrast enhanced CT scans due to renal function impairment) were excluded,
thus leading to a study cohort of twenty-three patients.
Follow-up scans were performed periodically according to study protocols (after 4 cycles of nivolumab).
All CT scans were performed by MDCT scanners (Lightspeed 16s and Lightspeed 64s,
General Electric Healthcare,
Milwaukee,
Wisconsin).
Iodinated contrast has been administered at a flow rate of 2-3 ml/s via a needle cannula placed in the antecubital vein using an automatic injector,
followed by a bolus of saline solution.
Venous phase acquisitions were performed through a “smart-prep” technique,
placing a ROI on descending thoracic aorta,
with a 60s delay time after reaching the peak (100HU).
If requested a preliminary angio-CT scan to study the pulmonary arteries has been performed,
also through a “smart-prep” technique,
placing a ROI on pulmonary artery,
with minimal (6s) delay time after reaching the peak (100HU).
CT scan data were acquired using the following parameters: 120 kVp,
100-300 mA,
5mm slice thickness with 1.25mm collimation.
CT scans reconstructed by using a filtered back projection technique with a soft (b40) kernel,
were then reformatted on an axial plan with a 3mm thickness on a dedicated workstation (Advanced Workstation 7,
GE Healthcare,
Milwaukee,
Wisconsin,
USA) and each lesion was manually segmented on the slice with the largest cross-sectional area by a radiology resident experienced in CTTA,
blinded to clinical and temporal data; then 295 texture analysis parameters were extracted from each image using an open source software (MaZda,
version 4.6)19,20,21.
Absolute variations of these parameters at baseline and follow up examination were calculated,
and statistical analysis was performed using a dedicated software (MedCalc Statistical Software version18); a p-value of 0.05 or lower has been considered statistically significant.