Correlation of iodine uptake and perfusion parameters between dual-energy CT imaging and first-pass dual-input perfusion CT in lung cancer.

To investigate the potential relationship between perfusion parameters from first-pass dual-input perfusion computed tomography (DI-PCT) and iodine uptake levels estimated from dual-energy CT (DE-CT).The pre-experimental part of this study included a dynamic DE-CT protocol in 15 patients to evaluate peak arterial enhancement of lung cancer based on time-attenuation curves, and the scan time of DE-CT was determined. In the prospective part of the study, 28 lung cancer patients underwent whole-volume perfusion CT and single-source DE-CT using 320-row CT. Pulmonary flow (PF, mL/min/100 mL), aortic flow (AF, mL/min/100 mL), and a perfusion index (PI = PF/[PF + AF]) were automatically generated by in-house commercial software using the dual-input maximum slope method for DI-PCT. For the dual-energy CT data, iodine uptake was estimated by the difference (λ) and the slope (λHU). λ was defined as the difference of CT values between 40 and 70 KeV monochromatic images in lung lesions. λHU was calculated by the following equation: λHU = |λ/(70 - 40)|. The DI-PCT and DE-CT parameters were analyzed by Pearson/Spearman correlation analysis, respectively.All subjects were pathologically proved as lung cancer patients (including 16 squamous cell carcinoma, 8 adenocarcinoma, and 4 small cell lung cancer) by surgery or CT-guided biopsy. Interobserver reproducibility in DI-PCT (PF, AF, PI) and DE-CT (λ, λHU) were relatively good to excellent (intraclass correlation coefficient [ICC]Inter = 0.8726-0.9255, ICCInter = 0.8179-0.8842; ICCInter = 0.8881-0.9177, ICCInter = 0.9820-0.9970, ICCInter = 0.9780-0.9971, respectively). Correlation coefficient between λ and AF, and PF were as follows: 0.589 (P < .01) and 0.383 (P < .05). Correlation coefficient between λHU and AF, and PF were as follows: 0.564 (P < .01) and 0.388 (P < .05).Both the single-source DE-CT and dual-input CT perfusion analysis method can be applied to assess blood supply of lung cancer patients. Preliminary results demonstrated that the iodine uptake relevant parameters derived from DE-CT significantly correlated with perfusion parameters derived from DI-PCT.

Introduction

Primary lung cancer is 1 of the most common malignant tumors worldwide, comprising 17% of the total new cancer cases and 23% of total cancer mortality.[ Early surgery and radio-chemotherapy have been proven to be highly effective in most lung cancer patients. On the contrary, the prognosis of advanced lung cancer patients is still poor, with less than 10% surviving for 5 years.[ The majority of patients can only be treated by palliation. Generally speaking, early detection and correct diagnosis of lung cancer are important for the treatment, prognosis, and survival of patients with lung cancer.

Contrast-enhanced computed tomography (CT) of the chest remains the most common imaging technique in clinical diagnosis of lung cancer. The characterization of pulmonary nodules revealed by standard contrast-enhanced CT can provide information for differentiating malignancy from benignity.[ However, since the 1970s, studies have indicated that lung tumors have a dual blood supply of pulmonary circulation and systemic circulation.[ In the past, due to the limitations of detector width of traditional CT system, and the choice of acquiring time, the pulmonary artery usually could not be included in the perfusion CT imaging territory, in which the input artery was set to the aorta.[ With the progress of volume scanning technology, the increase of the detector width makes it possible to detect both pulmonary and systemic circulation using dual-input perfusion CT (DI-PCT) technique.[

Owing to the greater coverage along the z-axis provided by the 320-row CT system, it is possible to include the pulmonary artery, the aorta, and the tumor in 1 gantry rotation without table movement. DI-PCT, which can acquire more phases than traditional CT, and can simultaneously capture the tumor's first-pass perfusion, makes it possible to assess the dual blood supply in lung tumors with the dual-input maximum slope analysis method. However, there are also significant limitations to DI-PCT. First, in the scanning process, multiple consecutive scans covering the lesion may lead to increased radiation dose.[ Second, the results of postprocessing are affected by many factors, such as the scanning parameters adjustment and mathematical mode.

Dual-energy CT (DE-CT) can provide diagnostic information about the composition of the lesions by using 2 different energy x-rays. By acquiring at 2 different kVp, DE-CT is possible to distinguish different materials, because of different atomic number, energy binding ability, and absorption properties. This technique enables the discrimination of the compositions of specific tissues, such as calculus composition analysis, bone density measurement, and iodine concentration analysis.[ The main clinical benefit of DE-CT scan as compared with DI-PCT would be that the method is not as susceptible to respiratory motion as the data are acquired within a single breath-hold. Moreover, the radiation dose may be significantly reduced.

Recent studies have shown that DI-PCT and DE-CT iodine maps were significantly correlated.[ A similar study of the abdomen indicated that the correlation between DE-CT iodine concentration profile and DI-PCT perfusion was significant.[ However, similar studies of lung cancer have not been reported. The objective of this study is to assess the correlation of iodine uptake and perfusion parameters between DE-CT and first-pass DI-PCT in lung cancer patients.

Materials and methods

All CT examinations were performed with 320-row multislice spiral CT system (Aquilion ONE, Toshiba Medical Systems, Otawara, Japan). The study, including a pre-experimental and a prospective part, was approved by the ethics committee of our institution. Written informed consent was obtained from all patients before the examination. From March 2015 to October 2016, 45 consecutive patients with suspicious lesions on chest radiography or noncontrast chest CT were enrolled. Inclusion criteria were as follows: without tumor-associated treatment before CT examinations (DI-PCT and DE-CT); the interval between CT examinations (DE-CT and DI-PCT) and CT-guided puncture biopsy or surgical resection was less than 2 weeks; clinical information including pathological results was intact. Also, patients who met any of the following criteria were excluded: hypersensitivity to iodinated contrast media (n = 0); nephropathy (estimated glomerular filtration rate <30 mL/min, n = 0); non-lung cancers (n = 2) confirmed by pathological results. In all, 43 patients with lung lesions were enrolled in the study.

For pre-experimental part of the study: 15 of 43 patients (8 males and 7 females, mean age 64 years, age range 51–78 years) undergoing DE-CT examinations were evaluated to determine the scan time of DE-CT in prospective part of the study.

For prospective part of the study, the remaining 28 patients (19 males and 9 females, mean age 60 years, age range 28–78 years) undergoing DI-PCT and DE-CT examinations were enrolled in the final analysis. The DI-PCT and DE-CT scanning protocols were performed as mentioned below. The DI-PCT and DE-CT can be completed in 1 examination, DE-CT scan was initiated with a delay of 4.5 seconds after DI-PCT.

The radiation dose of the dynamic CT was calculated from the dose-length product (DLP) listed in the exposure summary sheet generated by the CT equipment and multiplied by the correlation coefficient K of 0.014.[

DI-PCT imaging technique

Before the examination, all patients underwent breathing exercise training to ensure that they could hold their breath during the entire perfusion procedure. Superficial and regular abdominal breathing was also permitted in patients who were unable to persist for the entire period of perfusion examination. A 20-gauge intravenous catheter was placed in the antecubital vein to inject the contrast agent.

An initial unenhanced chest CT scan was acquired with a section thickness of 5 mm to localize the tumor and select the volume on which perfusion scans had to be performed. A total of 48 mL of the contrast agent with an iodine concentration of 370 mg I/mL (Iopromide, Bayer Schering, Berlin, Germany) was injected at an injection rate of 6 mL/s. Fifteen consecutive dynamic volume scans with z-axis coverage of 16 cm were performed with a delay of 2 seconds after the start of the injection without table movement.

The dynamic volume scanning parameters were as follows: 100-kv tube voltage, 50-mA tube current, 0.5-second gantry rotation time speed, 0.5-mm slice thickness, 30 seconds total breath-hold time. The first volume scan, acquired before the contrast reaching the heart, served as a baseline, and the other volumes were perfusion scans. After scanning, the scan data were processed with adaptive iterative dose reduction and reconstructed to 320 images per volume with 0.5-mm slice thickness, 0.5-mm slice interval, and a total of 4800 images for the entire perfusion data set.

Determination of DE-CT initial scanning time

Based on Gordic method of the determination of DE-CT initial scanning time in hepatocellular carcinoma [ and the enhancement curve of aorta and lung cancer, DE-CT scan is initiated during the peak enhancement of the lung cancer.

The DE-CT acquisition was started along with contrast agent injection, the purpose of which is to obtain at least 1 unenhanced phase for scanning. Each patient was scanned for 11 axial acquisition phases, with 3-second intervals for 1 to 8 phases, 4.5-second intervals for 8 to 9 phases, 5.0-second intervals for 9 to 11 phases, and the total scanning time was 48.7 seconds. The parameters of DE-CT scanning were as follows: gantry rotation, 350 ms; slice thickness, 0.5 mm; single acquisition, 1.2 seconds; tube voltage pair was set at 80 and 135 kVp, both with real-time adjustable variable tube current. Each of the acquisitions covered the same craniocaudal width of 160 mm. Following all the acquisitions, DE-CT images were reconstructed with a section thickness of 0.5 mm and intervals of 0.5 mm for evaluation.

In the postprocessing, to derive the peak enhancement of aorta and the pulmonary artery, the arterial input curve of contrast agent concentration was required by drawing a region of interest (ROI 25 mm2) the aorta and the pulmonary artery at the level of carina of trachea. The ROI was drawn freehand around the maximum peripheral boundary of the visible tumor on axial imaging. Care was taken to exclude the surrounding lung tissue and normal vascular structures, and also the necrotic tissue in the tumor. Then, the time-density curves (TDCs) of the aorta, pulmonary artery, and the lesion were automatically generated.

Analysis of the individual and the average time-attenuation indicated that peak enhancement of the aorta was at 28.50 seconds (range 17–38 seconds), pulmonary artery at 19.85 seconds (range 11–27 seconds), and the lesions at 34.5 seconds (range 20–41 seconds). Based on the results above, the optimal DE-CT acquisition was set at the moment of a delay of 34.5 seconds after contrast agent injection, with a duration of 4.5 seconds after DI-PCT scanning (Fig. 1).

Figure 1

The time-attenuation curves (TDCs) of 1 patient's DE-CT scan shows curves of the lesion (blue), the aortic (red), and the pulmonary artery (green). DI-PCT and DE-CT scan protocols were marked in the diagram. DI-PCT scan was initiated with a delay of 2 seconds after the start of CA injection, and the optimal DE-CT acquisition was set 4.5 seconds after DI-PCT scan. CA = contrast agent.

DE-CT imaging technique

The DE-CT scan was initiated with a delay of 34.5 seconds after the start of contrast agent injection. The CT protocol was performed with the following parameters: 0.35-second gantry rotation speed, 0.5-mm slice thickness, 1 acquisition of 1.2 seconds; tube voltage pair was set at 80 and 135 kVp, both with real-time adjustable variable tube current. Each of the acquisitions covered the same craniocaudal width of 160 mm. The scan raw data were reconstructed with a section thickness of 0.5 mm and intervals of 0.5 mm for postprocessing.

Data postprocessing

The data processing and analysis of the DI-PCT data were performed using perfusion software available on postprocessing system (Body Perfusion, dual-input maximum slope analysis, Toshiba Medical Systems, Otawara, Japan). First of all, volume registration was performed, to correct for motion between the dynamic volumes and create a registered volume series used for subsequent perfusion analysis. After that, circular ROIs (25 mm2) were manually drawn in the descending aorta, pulmonary artery at the level of carina of trachea, to generate the TDCs representing the bronchial artery flow (BF) input function and pulmonary artery flow (PF) input function. Another circular ROI (25 mm2) was placed in the left atrium and the time of left atrium TDC was used to differentiate bronchial circulation and pulmonary circulation. Freehand ROIs were delineated along the maximum contours of the visible tumor on axial images, avoiding necrosis and other normal anatomy tissue, then TDC of the contrast medium's first-pass attenuation for lung cancer was automatically generated by the system (Fig. 2). The threshold of perfusion analysis range was set from 0 to 150 HU to restrict the perfusion analysis to soft tissue only and to remove the effects of bone and lung parenchyma. Finally, color-coded maps of AF, PF, and PI were automatically generated by the software with matrix of 512 × 512. ROIs were placed in the largest solid part of the lesion on axial, sagittal and coronal slices in the maps to measure the AF, PF, and PI values of the whole tumor.

Figure 2

Time-density curves (TDCs) of PF (blue curve), AF (red curve), and left atrium (orange curve) and lung tumor (green, right axis). The vertical orange dotted line indicates the left atrial peak time, which is centered between the PF and AF peaks, representing the boundary between the pulmonary and systemic circulation. During the DI-PCT, 2 peak enhancements were demonstrated from the TDCs of pulmonary circulation and systemic circulation, respectively. The latter one with greater slope demonstrated predominance in this case.

The DE-CT data postprocessing were analyzed using software package available in the Toshiba workstation (DE Raw Data Analysis, Toshiba Medical Systems, Otawara, Japan). After DE-CT data were loaded, the software automatically generated 2 × 2 groups of images, including 2 single-energy images, a mixed-energy image, and a density-energy curve. A freehand ROI was placed encompassing the solid part of the lesion in the same way with DI-PCT analysis. The CT values (HU) of the lesion were measured in the single energy of 40 KeV image and 70 KeV image, respectively. The difference (λ = 40 KeV CT value 70 KeV CT value) and the slope (λHU = λ/[40 − 70]) served as estimates of iodine uptake at the lesion. Finally, a color-coded iodine concentration profile was generated by the software.[

Data statistical analysis

Data analysis was performed using statistical software (SPSS 19.0). The reproducibility of DI-PCT and DE-CT derived parameters between different observers was evaluated with intraclass correlation coefficients (ICCs). ICC values smaller than 0.4 indicate poor reproducibility, those ranging from 0.4 to 0.75 indicate fair reproducibility, and those above 0.75 indicate good reproducibility. A single sample Kolmogorov-Smirnov test was performed on AF, PF, PI, λ, and λHU to determine whether they were normal distribution. Then the Pearson correlation test was performed between data with normal distribution and Spearman correlation test was performed between data not according without normal distribution. P values less than .05 were considered to indicate a significant difference.

IRB statement

The institutional review board approved this study, and informed consent was obtained from all patients enrolled in the study.

Results

All subjects were pathologically proved as lung cancers (including 16 squamous cell carcinomas, 8 adenocarcinomas, and 4 small cell lung cancers) by surgery or CT-guided biopsy histology. All the 28 patients successfully completed the examination. No severe adverse events occurred.

Interobserver reproducibility in DI-PCT (PF, AF, PI) and DE-CT (λ, λHU) were relatively good to excellent (ICCInter = 0.8726–0.9255, ICCInter = 0.8179–0.8842, ICCInter = 0.8881–0.9177, ICCInter = 0.9820–0.9970, ICCInter = 0.9780–0.9971, respectively).

The DI-PCT perfusion images and the DE-CT iodine profiles are visualized using color-coded maps and matched with the original images (Fig. 3). The measurements of variable parameters are listed in Table 1. The results of correlation coefficient between DI-PCT and DE-CT parameters are demonstrated in Table 2. There was a significant positive correlation between λ and AF, and PF, with a correlation coefficient of 0.589 (P = .001; Fig. 4) and 0.383 (P = .044; Fig. 5). Significant positive correlation was also found between λHU and AF, and PF (r = 0.564, P < .002; r = 0.388, P = .041, respectively) (Figs. 6 and 7). No significant correlation was found between PI and λ, and λHU (P = .346, P = .354, respectively).

Figure 3

A 63-year-old male patient showed a right upper lobe adenocarcinoma. (A) Iodine map of the lesion showed iodine distribution of the lesion: solid portion with significant iodine uptake. (B) Coronal reconstruction showed the nodule located in the right upper chest. (C) Curves of CT numbers of the aorta (red line), pulmonary artery (dark blue), nodule (light blue), and lymph node (yellow). (D) DE-CT showed the perfusion of the lesion.

Table 1

Measurements of parameters derived from different scan modes.

Table 2

Results of correlation analysis for parameters between first-pass dual-input perfusion CT and dual-energy CT.

Figure 4

Correlation scatter diagram of AF-λ indicated significant positive correlation, with the correlation coefficient being 0.589 (P < .01).

Figure 5

Correlation scatter diagram of PF-λ indicated significant positive correlation, and the correlation coefficient being 0.383 (P < .05).

Figure 6

Correlation scatter diagram of AF-slope (λHU) indicated significant positive correlation, with the correlation coefficient being 0.564 (P < .01).

Figure 7

Correlation scatter diagram of PF-slope (λHU) indicated significant positive correlation, with the correlation coefficient being 0.388 (P < .05).

The average DLP of DI-PCT was 386.9 mGy·cm, with an effective radiation dose of 5.4 mSv, whereas the average DLP of DE-CT was 193.0 mGy·cm, with a radiation dose of 2.7 mSv.

Discussion

The results of this study showed good correlations between iodine uptake with DE-CT imaging and perfusion parameters from DI-PCT in lung cancer patients; meanwhile, the exposure dose of DE-CT was significantly lower than that of DI-PCT. In a rabbit experiment of liver tumors, Zhang et al[ demonstrated a positive correlation between the iodine maps of the DE-CT acquired during the arterial and the portal venous phases of enhancement and the blood volume determined at perfusion CT. In another study, Fuld et al[ demonstrated that the pulmonary perfusion blood volume determined at DE-CT could serve as a surrogate for true CT-based measures of the regional pulmonary parenchymal perfusion. By performing DE-CT 15 seconds after contrast agent injection, their study showed good correlation between perfusion CT-based measures of regional pulmonary parenchymal perfusion and DE-CT-based measurements of pulmonary blood volume.

In patients with pancreatic tumors, or liver tumors, or kidney cancer, all the results showed good correlations between parameters acquired from DE-CT and perfusion CT.[ In a study of cervical lymph node lesions, Tawfik et al[ found significant differences in iodine content and iodine overlay acquired from DE-CT between normal, inflammatory lymph nodes, and metastatic squamous cell carcinoma cervical lymph nodes. Moreover, Thaiss et al[ reported that DE-CT and perfusion CT showed good agreement in terms of diagnosis and evaluation of therapy response. Although multiple studies have shown that there was a good agreement between DE-CT and contrast enhanced or perfusion CT scanning,[ only a few of studies on the correlation between parameters from DE-CT and DI-PCT in lung cancer patients have been reported.

Traditional contrast-enhanced CT served as the main imaging modality for the diagnosis of lung cancer and has limitations in tumor staging and follow-up. Although the 2 circulations in lung cancer have been found for nearly half a century, the traditional contrast-enhanced CT only demonstrates the dominant circulation. In lung cancer, usually bronchial artery blood supply is more predominant than the pulmonary artery. This feature brings difficulties to differentiate small lesions.[ With the development of DI-PCT perfusion imaging, tumor hemodynamic parameters derived from perfusion measurements were used to identify lung cancer, which makes CT more suitable for diagnosing, estimating prognosis, and evaluating therapy effect in lung cancer.

Although DI-PCT has many advantages over traditional enhanced CT in qualitative diagnosis, prognosis evaluation, and therapy response evaluation, it still has some limitations. Firstly, DI-PCT requires specialized hardware, with a greater width of the detector to cover the entire lesion.[ Secondly, duration of DI-PCT examination is longer than traditional enhanced CT. In the breath-holding scanning series, patients need to hold breath for a longer time of over 20 seconds, which is difficult for patients with poor respiratory function. Thirdly, DI-PCT requires multiple repeated scans over the same field of view, which significantly increases patients’ radiation dose. In addition, there are discrepancies in the different kinetic modeling approaches among vendors and incompatibility between software versions.[

With the benefit of faster scanning speed and lower exposure dose, DE-CT can reduce scanning time and decrease incidences of radiation adverse events. In addition, analysis of the dual-energy dataset enables the precise iodine enhancement separation and quantification of iodine in a single ROI, owing to the 3-material decomposition algorithm (fat, soft tissue, and iodine). Virtual noncontrast or iodine-enhanced images can also be created.[ Several previous studies have confirmed the reliability and accuracy of this determination,[ which allows the patient to obtain both noncontrast images and enhanced images in 1 examination, and reduces radiation dose.

The results of our study in lung cancer patients showed that the best mean time point for initiating DE-CT scan was 34.5 seconds after contrast agent injection. In a previous study, the scan of DE-CT was initiated 35 to 40 seconds after injection of contrast agent in scans of the lung with multi-row spiral CT.[ In studies of pancreatic cancer, DE-CT was initiated about 32 seconds after contrast agent injection, and the parameters obtained from DE-CT have good correlations with those of perfusion CT (r = 0.77).[ The initiating time in our study was similar to those in previous reports, which ensures the accuracy and practicality of parameters analysis. Moreover, during 1 contrast agent injection, patients in our study could take examinations both DI-PCT and DE-CT, which reduced the contrast agent injection. An important issue in DI-PCT is the radiation dose. In our study, the radiation dose of DE-CT was much lower than that of DI-PCT.

The following limitations of this study must be addressed. First, the sample size of patient groups was small and the histopathology of cancer was limited to primary malignancy tumors, which could lead to selection bias. Second, the scanning protocol might not have been ideal, and the scanning time of DE-CT was based on typical cases, which may lead to overestimation of performance. In addition, compared with other studies, more subjects are needed for further trials. For example, whether DI-PCT is reliable in benign and malignant tumors,[ and in lymph node evaluation, the correlation between DI-PCT, DE-CT, and gold standard FDG-PET/CT is controversial.[

Conclusions

In conclusion, the correlations of iodine uptake and perfusion parameters in lung cancers between dual-energy CT imaging and first-pass dual-input perfusion CT are significant. Meanwhile, DE-CT needs lower radiation dose and shorter scanning time than DI-PCT. In clinical studies, DE-CT can be used to evaluate the hemodynamic characteristics of lung cancers.