Keywords:
Cancer, Staging, CT, Oncology
Authors:
L. Molenaar, M. A. van der Lubbe, F. van der Krieke, M. J. Lahaye, M. Maas, F. Lalezari, R. G. H. Beets-Tan, D. M. J. Lambregts; Amsterdam/NL
DOI:
10.1594/ecr2018/C-1537
Methods and materials
Patient population:
In this retrospective study 32 patients with a variety of cancer-types were included.
Patients were diagnosed with lung,
colorectal,
ovarian,
liver,
urothelial cell,
breast,
head & neck, melanoma and neuroendocrine tumors.
Data acquisition:
All patients underwent a baseline CT-scan (prior to treatment) and 2 follow-up (FU)-scans for response evaluation according to RECIST1.1 between 2012 and 2015 at our institution.
Average time between CT-scans was 51 days (range 23 to 133 days).
CT-scans were contrast-enhanced acquisitions of the neck,
chest and/or abdomen and were performed using a 16-slice or a 64-slice scanner with 5.0 mm and 1.00 mm coupes.
The routine contrast timing for the (neck)-thorax-abdomen scan was portal venous phase and chest CTs were routinely scanned in arterial phase.
Image evaluation:
Images were viewed in the routine PACS system used at our institution.
This study has 3 arms:
- A radiologist (one of 3 readers with similar reading experience) identified baseline target lesions and performed all FU-measurements.
- A radiologist identified target lesions,
FU-measurements were done by a radiographer (one of 3 readers with similar reading experience).
- A radiographer performed both baseline and FU-measurements.
The radiographers underwent a short training on the use of RECIST 1.1 before the start of the study,
but had no extensive previous reading experience.
For arm 3,
a fourth independent radiologist reviewed all target lesions as identified at baseline by the radiographers and classified these as correct,
minor corrections required or incorrect/major corrections required.
Statistical analysis:
For all reading arms response was assessed according to RECIST 1.1 and categorized as complete response,
partial response (≥30% decrease in sum of lesion diameters),
progressive disease (≥20% increase) and stable disease (<20% increase,
<30% decrease).
Reading-arms 2 and 3 (involving a radiographer) were compared to arm-1 (reading by a radiologist),
which was considered as the standard of reference.
Interobserver agreement (IOA) was assessed between the arms by using weighted kappa statistics with quadratic weighting.
Results were separately calculated for FU1 and FU2.