Contrast media volume is significantly related to patient lung volume during CT pulmonary angiography when employing a patient-specific contrast protocol

Aim: The purpose of this study is to investigate the relationship between contrast media volume and patient lung volume when employing a patient-specific contrast media formula during pulmonary computed tomography angiography (CTA). 
Materials and methods: IRB approved this retrospective study. CTA of the pulmonary arteries was performed on 200 patients with suspected pulmonary embolism (PE). The contrast media volume (CMV) was calculted by employing a patient-specific contrast formula. Lung volume was quantified employing semi-automated lung software that calculated lung volumes (intellispace -Philips). The mean cross-sectional opacification profile of central and peripheral pulmonary arteries and veins were measured for each patient and arteriovenous contrast ratio (AVCR) calculated for each lung segment.  Mean body mass index (BMI) and lung volume were quantified. Receiver operating (ROC) and visual grading characteristics (VGC) measured reader confidence in emboli detection and image quality respectively. Inter and intra-observer variations were investigated employing Cohen’s kappa methodology. 
Results: Results showed that the mean pulmonary arterial opacification of the main pulmonary circulation (343.88±73HU), right lung; upper (316.51±23HU), middle (312.5±39HU) and lower (315.23±65HU) lobes and left; upper (318.76±83HU), and lower (321.91±12HU) lobes. The mean venous opacification of all pulmonary veins was below 182±72HU. AVCR was observed at all anatomic locations (p<0.0002) where this ratio was calculated. Moreover, larger volumes of contrast significantly correlated with larger lung volumes (r=0.89, p<0.03) and radiation dose (p<0.03). VGC and ROC analysis demonstrated increased area under the curve: 0.831 and 0.99 respectively (p<0.02). Inter-observer variation was observed as excellent (κ = 0.71). 
Conclusion: We conclude that increased CMV is significantly correlated to increased patient lung volume and radiation dose when employing a patient-specific contrast formula. The effects patient habitus is highlighted.


Introduction
Although the gold standard in pulmonary embolism diagnosis is still angiopulmonography, computed tomography pulmonary angiography (CTPA) protocols have witnessed a surge in thrombus detection at a specificity and sensitivity approaching 100% with additional benefits of non-invasiveness and accessibility (1). Since this efficacy is largely dependent on effective contrast media volume (CMV) administration, studies have been extensively conducted on contrast administration protocols (2)(3)(4)(5)(6)(7)(8). These studies have found that visualization of the pulmonary vasculature is significantly improved when using a simple patient-specific formula (1, 7,9), which also allows CMV to be reduced along with the potential risk of contrast induced nephropathy (1).
Studies performed on adult patients with normal pulmonary functioning demonstrated an air/tissue relationship approaching a 7:3 ratio of air to lung parenchyma (10); other studies noticed that an increase in total lung volume is concomitant with an increase in the volume of pulmonary blood and pulmonary extravascular water (11). To our knowledge, there have been no studies to date that compares CMV with patient lung volume employing a reduced patient-specific formula. The aim of the study is to investigate the relationship between CMV and patient lung volume when employing a patient-specific contrast media formula during CTPA. Our results could be used as a foundation for further research on lung volume specific CTPA protocols.

Study Population
The institutional review board approved this study and written informed consent was waived since all studies were clinically indicated and patient data was evaluated retrospectively. Two hundred consecutive patients with high clinical suspicion of acute pulmonary embolism (PE) were examined over a 6-month period between August 2018 and January 2019. The indication for CTPA was suspected PE based on clinical information (chest pain, dyspnea, hypoxia, calf pain or known deep vein thrombosis (DVT), risk factors for PE) and/or laboratory information (positive d-dimer > 0.8 mg/L). All patients with a positive PE in this study received anticoagulation therapy and due to the nature of the suspected condition, all patients were scanned. Inclusion criteria were patient with suspected PE as per the criteria mentioned above, who underwent CTPA. Patient with no suspected PE, or patient with PE who did not undergo CTPA were excluded. There were no patients in the study that had renal insufficiency and/or contra-indications to iodinated contrast media.
The diagnosis and treatment were in adherence with amercan collage of cardiology (ACC) guidelines for the diagnosis and managent of pulmonary embolism.

Ethical approval:
All procedures performed in studies involving human participants were in accordance with the ethical standards of the American university of Beirut's institutional review board (IRB) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was waived since all studies were clinically indicated, patient data contained no identifiable information and was evaluated retrospectively.

Image Acquisition
CTPA was performed using a 256-channel computed tomography scanner (Philips Brilliance iCT, Philips Healthcare, The Netherlands). Patients were positioned supine with arms resting on the gantry above the head. Anterior-posterior scout were performed, with a scan range from the apices (2 cm above the 1 st rib) to the diaphragm (2 cm below the lowest costophrenic recess). Breath-hold, with a mouth open breathing technique were employed in order to reduce hyperventilation and Valsalva. Scan range was from the lung apices to the costophrenic angles. CT scan parameters employed in each protocol were: detector width of 256×0.625 mm, pitch of 0.981:1 ratio, rotation time of 0.27 sec, 100 kVp, 140 mA, with x,y and z-axis modulation (DoseRight), craniocaudal scan direction and model based iterative reconstruction (IMR2).

Contrast Bolus Geometry
Vessel opacification for all cases was measured by placing a region of interest (ROI) over the main pulmonary trunk. A time attenuation curve in Hounsfield units (HU) was calculated and the desired peak opacification was recorded. The patient-specific protocol employed the test bolus technique (12) where the ROI was plotted inside the main pulmonary trunk with a small amount of contrast material (5 mL) injected at the same rate as the main bolus. This ROI assessed the time to peak (TTP) and determined the arteriovenous circulation time for pulmonary vasculature (12)(13)(14).

Contrast Media Acquisition
Contrast and saline chaser were injected with an automated dual barrel power injector (CT Emotion, Ullrich, Germany) via a 20G venous catheter in the right arm (15,16).
The patient-specific contrast formula employed iodinated contrast (Omnipaque 350 mgI/mL; General Electric, USA) intravenously injected at a flow rate of 4.5 mL/s. Contrast media volume was calculated according to an empirically derived formula (17): CV=(ST+TTP-OVWP)×FR. Where ST is the scan time (s), TTP (s) is as described above, OVWP is the optimal venous washout phase (6s), and FR is the flow rate (mL/s). ST differs for each patient based on the distance between the apices and diaphragm of the thorax. Patients were excluded if they were unable to have a right sided injection site in the cubital fossa with a flow rate of 4.5 mL/s.

Radiation Dose Measurement
For each of the CT scans, individual effective dose (Eff [mSv]) was calculated from the dose-length products (DLP [mGy × cm]), which were recorded from the patient protocol. A normalized conversion factor (k [mSv / mGy × cm]) for the chest-0.014 mSv/mGy × cmwas used to calculate the Eff (18): Eff = DLP x k (4).

Image assessment
Technical inclusion criteria ensuring correct scan range and anatomical inclusion of the origin, pathway and termination of the pulmonary vasculature were applied to all cases by two expert radiologists (not included in the study proper). Quantitative measurements of all images were performed using a primary reporting workstation (Intellispace,Philips Healthcare, Netherlands) with a GSDF-calibrated 3 megapixel monitor. Illumination was adjusted at 25 to 32 lux (19), with a calibrated photometer (chroma meter CL-200).

Quantitative Analysis
Opacification in HU was measured for all cases in the trans-axial images within the largest circular ROI that would fit within the lumen and exclude the vessel wall. The mean cross-sectional opacification profile of 8 central segments and 11 peripheral pulmonary arteries were measured. In cases where PE was identified, care was taken not to include the emboli within the measurements. Arterial and venous measurements took place at the heart, pulmonary trunk, segmental and sub segmental pulmonary vasculature and each measurement was no less than 2 mm in diameter (

Diagnostic Efficacy
The multi-reader analysis consisted of two cardiothoracic radiologists who had been certified by the American Board of Radiology and The Royal College of Radiologists for a mean number of 8.9 years (minimum, 3 years; maximum, 18 years). All reviewers were specialists in cardiothoracic imaging and each observer was allowed to manipulate the window level of the images.

ROC Analysis
Receiver operating characteristic (ROC) methodology was employed to illustrate radiologist confidence intervals to detect pathology. A score of 1 to 2 was assigned to each image, where 1 indicates positive for pathology detection and 2 indicates negative for pathology detection of PE. All cases were randomly allotted with the number of normal (n=87) and abnormal (n=113) cases. The abnormal cases demonstrated an array of vascular disease that were defined by two radiologists' reports (based on complete patient series, previous and subsequent examinations and clinical indications). All pathology was visible on the transaxial images and the prevalence of pathology was not revealed to the observers.
Technical criteria ensuring correct scan range and anatomical inclusion were considered (not included in the study proper) to ensure that all images displayed an acceptable level of quality before they were included in the analysis.

VGC analysis
The VGC method of Bath and Mansson (21) was used to illustrate viewer assessment of image quality based on the visibility of normal anatomy. Specifically, for this work, the presence of contrast media filling was recorded for pulmonary arterial system using a fivepoint classification scale where score 1 indicated no contrast media filling within the pulmonary vasculature and 5 represented complete filling.

Inter-and intra-reader variability
The inter-and intra-observer agreements were calculated using Cohen κ analysis. A k value 0.60 to 1, 0.41 to 0.60, 0.21 to 0.40, and less than 0.20 was considered excellent, moderate, fair, and poor agreement, respectively.

Statistical Analysis
Continuous variables were described with mean ± standard deviation (SD). The analysis of variance (ANOVA) was used to compare the means across the tertiles of contrast volume. The association between contrast volume and measured variables was evaluated through simple linear regression. Analyses were conducted using SPSS 22 for Windows (SPSS Inc, Chicago, IL). Results were considered statistically significant if p≤0.05 with a 95% confidence interval and a power of 0.8. ROC and VGC were employed to measure the confidence intervals in pathology detection and image quality respectively. Jackknife Freeresponse (JAFROC) assessment of diagnostic systems continues to gain acceptance in areas related to the detection, localization and classification of one or more "abnormalities" within a subject.

Patient Demographics
There was no significant change in CMV with gender and age, however, CMV increased with increasing body weight (p<0.004) and BMI (p<0.01) (Table 1). Additionally, the effective radiation dose also increased from low contrast media volumes (0-10mL) to higher contrast media volumes (60-70 mL) (p<0.03) incrementally. Note -Data are mean ± standard deviation, p<0.05, statistically significant with ANOVA

Image acquisition and contrast media volume
There was no statistical significance in mean scan time in each contrast range; 0-10 mL: (3.22 ± 1.2 sec) compared to 60-70 mL: (4.39 ± 1.3 sec), (p>0.05)( Table 1). All patients tolerated their assigned contrast material delivery protocol without any related complication.

Lung Volume vs. Contrast volume in patient specific contrast formula
When employing the patient-specific formula, increase in CMV positively correlated with increased total lung volume (r =0.89, p< 0001). As expected, total lung 18 volume increased with increasing mean anteroposterior and lateral lung diameters. Therefore, increased lung volume, anteroposterior, and lateral chest diameters are correlated with increased CMV (Table 4).  (2) 10 -20 (6) 20 -30 (14) 30 However, we further stratified CMV groups into tertiles (3 groups) to equally distribute the number of patients across each group unlike in table 4. Regression analysis was performed for each CMV group relative to lung volume. By comparing CMV 1(< 32mL) and 2 (>33 -<50 mL) demonstrated significant changes in the right and left lungs (p<0.006) ( Table 5). Also, when comparing group 2 (>37 -<40 mL) with group 3 (> 41mL), statistical significance was seen with increasing lung volumes in both right and left lung (p<0.001) as well as an increase in effective radiation dose (p<0.04) ( Table 5).
Interestingly, after regression analysis was performed, weight and BMI had no statistical significance association with increased lung volume and CMV.

Discussion
There have been considerable studies carried out to reduce radiation dose via reduction of radiation output, fixed CMV and reduced contrast media concentrations during CTA (22)(23)(24). Additionally, weight-based contrast media protocols have been considerably used to perform consistent optimal image quality during CTPA, but at the 23 cost of larger CMV for considerably larger patients (25). However, until recently, patient-specific protocols have not been readily employed to significantly reduce contrast media dose during CTPA without compromising image quality (12,14).
Nevertheless, to our knowledge this is the first study to investigate a possible relationship between CMV and lung volumes, weight and BMI and how different CMVs could affect confidence in emboli detection and image quality when employing a patient-specific formula.
The results of this study have demonstrated a significant correlation between CMV and lung organ volumes. When this formula was applied to patients, the CMV increased with increased lung volume which might be important in order to maintain a good visualization of pulmonary vasculature. Also, there was no significant change in CMV with gender and age, however, CMV initially increased with increasing body weight and BMI when patient were unequally grouped into 7 protocol ranges. However when grouped into 3 ranges with same patient number, body weight and BMI did not corelated with CMV which is in contrast with previous findings (26)(27)(28)(29)(30). The reduced CMV did not affect pulmonary artery opacification and the sensitivity and specificity of pulmonary emboli detection as those in previously reported papers (31)(32)(33).
There still remain controversies surrounding the effect of CMV and increased radiation dose due to the photoelectric effect. Recent studies demonstrated that the addition of contrast media during CT increased double DNA strand breaks which is attributed to increased radiation dose (34,35). Additionally, future studies published should consider the effect of iodinated contrast material on the organ doses 24 administered to patients undergoing CT, as it is important in estimating radiation dose (36,37). In our study, we demonstrated that increased lung volume was correlated with increased CMV as well as radiation dose, but when compared to other studies, our CMV range was from 10 -76mLs and other studies from 40 -120mLs (38)(39)(40)(41)(42). Finally, a strong correlation between increased lung and contrast volumes, demonstrated no correlation with BMI. Therefore, the radiation dose increase can be attributed to the increased volume in the pulmonary circulation at the time of the CTA and not patient weight.
There were shortcomings in the current study. Firstly, whilst every attempt was made to have uniform inspiration between patient to patient, the exact air volume entering the lungs could not be controlled and thus potentially the increase in blood volume could not be determined. Secondly, we did not measure the effect between weight-based contrast media protocols and the patient-specific contrast media protocols in pulmonary CTA. Finally, the main limitation of this study is the fact that it is a retrospective one. A randomized clinical trial will be needed to confirm our results.

Conclusion
High CMVs present a health hazard that could be avoided if patient specific CMVs are administered. The patient specific CMVs should preserve image quality thus maintaining the diagnostic relevance of their usage. A correlation between CMV and lung volume might be important in order to maintain a good visualization of pulmonary vasculature. This study showed that administering a patient tailored CMV can preserve 25 image quality. Moreover, increased CMV is significantly correlated to increased patient lung volume when employing a patient-specific contrast formula. The main limitations of this study is the fact that it is a retrospective study. A randomized clinical trial is needed to confirm our results.