Observer Variability in Scoring Abdominal Aortic Calcifications and Vertebral Morphometry

Calcify2D: Assessment of Observer Variability of a New Software for the Quantitative Evaluation of Abdominal Aortic Calcifications, Vertebral Morphometry and Their Relationships in Fragile Patients, Compared to Current Visual Scoring Methods

BACKGROUND In the context of a progressively aging population, monitoring the status of Vascular Calcifications (VC) and Vertebral Fractures (VF) over time would be of primary importance, as VC and VF are recognized to be hallmarks of severe cardiovascular events (hospitalization and/or death) and hip fractures respectively, and VF represent an under-diagnosed cause of progressive disability and pain on its own. Moreover, there is an acknowledged relationships between VC and VF.

However, data about the emergence/progression of VC and the emergence/worsening of VF over time are lacking. This is likely due to the absence of monitoring instruments for VC and VF that are both precise and easily accessible/applicable.

OBJECTIVE This study aims to define the observer variability of a new software developed by the study sponsor and collaborators, called Calcify2D. Calcify2D offers physicians a computer-assisted procedure to simultaneously score vascular calcifications at the abdominal aorta and lumbar vertebral fractures (according to Quantitative Vertebral Morphometry principles) based on a latero-lateral thoracolumbar spine radiography. Secondary aims are the validation of the scores obtained from latero-lateral thoracolumbar spine radiography with more invasive and/or costly gold-standard imaging modalities (Computed Tomography for VC, Magnetic Resonance for VF) that may have been acquired near-simultaneously to radiographs on the patients enrolled for the study.

STUDY DESIGN Not-for-profit monocentric observational study to be conducted on the diagnostic images of the thoracolumbar spine already collected at Istituto Ortopedico Rizzoli (IOR) within a previous interventional study.

Scoring of VC and VF will be performed by four clinicians from four relevant specialties, chosen among those who may often see VC and VF and are already familiar with the traditional scoring systems for both VC and VF (one radiologist and one spine orthopaedics from IOR, one nephrologist from the National Research Council and one internist from University of Padua).

Each clinician will assess all radiographs to score VC and QVM, both via computer assisted procedures and via traditional visual inspection. To avoid bias, an interval of at least one week will be left between the computer assisted and visual scoring. To define intra-observer variability (i.e. repeatability), the whole dataset will be re-assessed three times.

Inclusion Criteria:

• Patients included in a previous interventional study (VTP vs KYPHO) of the sponsor.
• Availability of diagnostic images (Dorso-lumbar radiographies taken in L-L projection).
• Rx images will be screened for the presence of VC, retaining only those showing VC.
• Informed consent obtained prior to any study analysis-evaluation.

Exclusion Criteria:

• Images showing severe artefacts (usually due to presence of metallic devices) that alter the grayscale range and hinder correct identification of VC and vertebral fractures