Study design and population
This retrospective cross-sectional study was approved by the local Ethics Committee (San Raffaele Hospital, Milan, authorisation number 160/int/2016). A series of consecutive patients who underwent a contrast-enhanced multiphasic abdominal CT or portal-venous-phase CT at our institution from June to September 2016 were reviewed.
Exclusion criteria were: history of chronic liver disease (cirrhosis, local or diffuse liver fatty infiltration, or glycogen storage disease); congestive heart failure; prior cardiac valve replacement; restrictive and/or constrictive pericarditis; implanted devices (pacemakers, defibrillators, insulin pumps). Although Hamer et al. [29] defined steatotic hepatitis when liver parenchyma has an average CT value on unenhanced images lower than 40 Hounsfield units (HU), we excluded patients with CT values below 30 HU in the unenhanced scan. As a consequence, low grades of steatosis have presumably been included in our study population.
CT protocol
All patients were studied using a 64-row CT scan (Somatom Definition, Siemens Medical Solution, Erlangen, Germany) with 120 kVp, tube load from 100 to 200 mAs depending on automatic exposure control system (CARE Dose 4D, Siemens Medical Solution, Erlangen, Germany), rotation time 0.5 s, pitch 1, B30f medium smooth for kernel reconstruction technique and abdomen window.
Patients’ TBW and height were registered in an electronic database. Moreover, a radiologist-driven dose of iopamidol (190 patients over 201) (Iopamiro 370; 370 mgI/mL; Bracco Imaging, Milan, Italy) or iomeprol (11 patients over 201) (Iomeron 400; 400 mgI/mL; Bracco Imaging, Milan, Italy) was administered. While iopamidol is the main choice in our hospital for routine abdomen and chest CT, iomeprol is used for cardiac CT. Due to practical reasons (storage lack of iopamidol, the necessity of employing an already-open CA bottle, examination acquired during a cardiac session), some patients received iomeprol. A total of eight radiologists were responsible for the examinations. The general rule established in the department for the CA dose to be administered for multiphasic abdominal CT was to use doses proportional to the TBW, multiplying the patient body weight by a constant, which varied from 1.1 to 1.3 mL/kg, with adjustments when the CA dose was considered too high. Radiologists usually adopted their own spontaneous threshold, without any agreement among them. Another heuristic rule used by some professionals was ‘patient weight in millilitres plus 10 additional millilitres of CA’.
The CA was administered intravenously through a 20-gauge needle using an automatic power injector (EmpowerCTA® Contrast Injection System, Bracco Imaging, Milan, Italy) at the rate of 3 mL/s, followed by 50 mL of saline solution at the same rate.
The scan delay was determined using an automated triggering hardware and a dedicated software (Bolus Tracking, Siemens Medical Solution, Erlangen, Germany). Specifically, low-dose monitor images were obtained in a single axial slice of the aorta after CA injection. When the descending aorta enhanced more than 100 HU, diagnostic scans of the abdomen were acquired after an additional delay of about 18 s (arterial phase), 30 s after arterial phase (portal-venous phase), and, only in specific cases, 90 s (nephrogenic phase). For the aim of this study, we considered only the portal venous phase.
LBW estimation and image analysis
According to the international classification of body mass index (BMI) from the WHO [30], patients were considered underweight when the BMI was lower than 18.5 kg/m2, normal weight when between 18.5 kg/m2 and 25 kg/m2, overweight when between 25 kg/m2 and 30 kg/m2, and obese when higher than 30 kg/m2.
According to Awai and colleagues [31] and Nyman [32], LBW was calculated using the James formula [33] (Eq. 1) or the Boer formula [33] (Eq. 2), due to better adherence of non-obese patients to the first and of obese patients to the latter:
$$ \kern5em {LBW}_{James}=\left\{\begin{array}{c}1.1\times weight\ \left(\mathrm{kg}\right)-128\times {\left(\frac{weight\ \left(\mathrm{kg}\right)}{height\ \left(\mathrm{m}\right)}\right)}^2\mathrm{men}\\ {}1.07\times weight\ \left(\mathrm{kg}\right)-148\times {\left(\frac{weight\ \left(\mathrm{kg}\right)}{height\ \left(\mathrm{m}\right)}\right)}^2\mathrm{women}\end{array}\right\} $$
(1)
$$ {LBW}_{Boer}=\left\{\begin{array}{c}0.407\times weight\ \left(\mathrm{kg}\right)+0.267\times height-19.2\ \mathrm{men}\\ {}0.252\times weight\ \left(\mathrm{kg}\right)+0.473\times height-48.3\ \mathrm{women}\end{array}\right\} $$
(2)
All images were reviewed by a radiology resident with 2 years of experience in the field of abdominal CT. Attenuation measurements were obtained by manually placing a rounded region of interest in the anterior (III or IVb Couinaud) and in the posterior (VI Couinaud) segments at the level of the main portal vein with a diameter of between 2 and 3 cm; these two values were averaged. Two different regions in anterior and posterior liver parenchyma were chosen because of subtle territorial differences in liver enhancement, probably due to different levels of fatty infiltration and vascularity. Focal hepatic lesions, blood vessels, bile ducts, calcifications, as well as artefacts, if present, were carefully avoided.
Statistical analysis
For each patient, we retrieved from the radiological report the amount and type of injected CA in millilitres and the dose was calculated both per TBW and LBW. Moreover, to account for the different concentration of iodine of the two used CAs, we converted the absolute injected amount from millilitres to grams of iodine (gI).
The CEL was calculated as the difference between the CT value measured in the portal venous phase and that measured before CA injection. To this aim, regions of interest were copy-pasted from one phase to another.
The distribution of CEL was calculated for the whole population and for the four subgroups of BMI. Bivariate correlation analysis was performed using the Pearson correlation coefficient. The comparison of the mean CA dose, as well as of the mean CEL among the four subgroups of BMI was performed using the one-way analysis of variance (ANOVA); the variance of CEL was compared using the Levene test of homoscedasticity.
Differences in the practice of CA dose calculation among radiologists were evaluated using the ANOVA.
Continuous data were presented as mean and standard deviation while categorical data were presented as counts and percentages. The coefficient of variation (CoV) was calculated as the standard deviation/mean ratio.
Statistical analysis was carried out using SPSS Statistics (SPSS v.24, IBM Inc., Armonk, NY, USA). A p value < 0.050 was regarded as statistically significant.