Patient population
For this retrospective study, informed consent was waived and permission obtained from the local institutional review board. Between January and December 2018, 92 consecutive patients with reported wrist trauma received CT imaging after multi-compartment arthrography of the wrist. All patients were at least 18 years old at the time of examination and assented to the procedure in written form. Six patients (three women, three men) had to be excluded from this study because the prestyloid recess could not be identified in CT scans due to a lack of contrast material. Therefore, the final study group consisted of 86 patients, including 40 women, with a median age of 44.5 years (max. 77 years; min. 18 years). The left wrist was examined in 48 patients.
CT arthrograms
In all patients, carpal arthrography was performed by board-certified radiologists under fluoroscopic guidance using a multi-compartment approach. Before the procedure, the contrast agent (Imeron 300, Bracco Imaging) was diluted with sodium chloride for an iodine concentration of 150 mg/ml. Arthrograms were conducted as per department protocol with standardized injections into the DRUJ and the radiocarpal joint for TFCC assessment. If concomitant intrinsic ligament injuries were suspected, i.e., tears of the scapholunate and/or lunotriquetral ligament, the midcarpal joint was additionally contrasted before the injections into the more proximal articular compartments.
Immediately after the arthrography was completed, patients were transferred to the CT suite and brought in prone position with the injured wrist placed above the head in pronation for further imaging. All scans were performed with a commercially available multidetector CT system (Optima 660, GE Healthcare) using a tube voltage of 120 kVp, current–time product of 150 mAs, detector collimation of 64 × 0.6 mm and pitch factor of 1.2. Image acquisition and reconstruction was performed in axial orientation with slice thickness and increment of 0.6 mm and 0.3 mm, respectively. Using scanner-side software (Advantage Workstation, GE Healthcare), additional planes in coronal and sagittal orientation were prepared with thickness of 1 mm, increment of 0.5 mm, matrix of 1024 × 1024 pixels and field of view of 60 mm. Window width and center were preset for this study (3000 and 1000 HU), however, readers could change settings to their own demands.
Image analysis
Two board-certified radiologists with both seven years of experience in musculoskeletal imaging retrospectively analyzed all datasets in consensus on a radiologic workstation with certified diagnostic monitor (RadiForce RX660, EIZO) and dedicated PACS software (Merlin, Phönix-PACS). First, the readers were asked to qualitatively evaluate the prestyloid recess in each CT arthrogram with regard to shape (saccular, tubular, cone-shaped, tongue-shaped), opening (no opening, narrow opening, wide opening) and position (palmar, radiopalmar, ulnopalmar or apical). Second, the observers were tasked to quantify the maximum distance between the recess tip and opening, the opening width and bulge width in the coronal plane, as well as the anterior–posterior (AP) diameter of the recess in the axial plane (Fig. 1). Third, the presence of contrast leakage along the ulnar aspect of the distal ulna should be assessed in dichotomous fashion (absence or presence). Whenever any form of leakage was identified, its maximum extent in proximal direction from the distal tip of the styloid process was documented on coronal images.
For estimation of TFCC integrity, ulnar variance, and fracture involvement of the sigmoid notch and ulnar styloid process, surgical reports (available in 45 patients) were used in combination with radiological reports by musculoskeletal imaging specialists and clinical follow-up. Thereby, the latter refers to control examinations that patients were advised to undergo approximately six weeks after the initial presentation that included the CT arthrogram. For their reads, the observers were blinded to any information from clinical and radiological reports, as well as arthroscopic findings.
Statistics
Dedicated software was utilized to conduct all statistical analyses (SPSS Statistics Version 27, IBM). Kolmogorov–Smirnov and Shapiro–Wilk tests were applied to analyze continuous variables for normal distribution. Normally distributed items are presented as means ± standard deviation, while metric data without normal distribution, as well as categorical and nominal items are reported as absolute and relative values with medians and interquartile ranges. The correlation between dichotomous variables was assessed using the mean square contingency coefficient (rɸ), whereas the correlation between binary and nominal items was tested by calculation of Cramér’s V. Results were interpreted following Kotrilik et al.: 1.0–0.8: very strong; 0.8–0.6: strong; 0.6–0.4: relatively strong; 0.4–0.2: moderate; 0.2–0.1: weak; < 0.1: no association [14]. For all tests, statistical significance was presumed for p values ≤ 0.05.