Enhancement parameters of contrast-enhanced computed tomography for pancreatic ductal adenocarcinoma: Correlation with pathologic grading

Woorim Seo, Young Chul Kim, Seon Jeong Min, Sang Min Lee

Abstract

Key words: Computed tomography; Pancreatic ductal carcinoma; Diagnostic imaging; Clinical pathology; Neoplasm grading; Prognosis

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDA) is a malignancy with a high mortality rate and short survival time[1]. Several important prognostic factors including tumor size, lymph node status, pathological grading and differentiation of the tumor influence survival in patients with PDA[2]. The pathological grade of adenocarcinoma is associated with the intratumor microvessel density (MVD)[3]. The process of neoangiogenesis is mediated by tumor angiogenic factors. Adenocarcinomas that develop in various organs tend to have a characteristic neovascularization pattern[4-6].

Computed tomography (CT) is an imaging modality used for evaluating tumors. The degree of CT enhancement is thought to be dependent upon the increase or decrease of intratumor MVD[7]. Some reports have described the relationship between CT enhancement, tumor angiogenesis, and the pathological grade of PDA[8]. It was reported that the degree of CT enhancement was directly proportional to the pathological grade of lung cancer but inversely proportional to that of PDA[8,9]. However, to date, few quantitative studies have compared CT enhancement parameters and the pathologic grade of PDA[10,11]. Therefore, the aim of this study was to investigate the relationship between various CT enhancement parameters and the pathologic grade of PDA.

MATERIALS AND METHODS

Patients

This retrospective study was approved by the institutional review board and the requirement for informed consent was waived. We conducted a computerized search of electronic medical records for patients with PDA. Forty-eight patients underwent surgery following CT examination from October 2012 to June 2017. We excluded 6 patients because they did not undergo arterial and pancreatic phase CT. A total of 42 patients were enrolled in our study. Forty-two patients with PDA (head and uncinate process: 30, body: 11, tail: 1) were treated using Whipple’s procedure (n= 6), pylorus preserving pancreaticoduodenectomy (n= 28), and distal pancreatectomy (n= 8).

CT examination

All CT images were obtained with two 128-channel multi-detector scanners (Siemens SOMATOM Definition AS and Flash, Siemens Healthcare, Erlangen, Germany). The CT parameters were as follows: Slice thickness, 3-5 mm; field of view (FOV), 50 cm × 50 cm; matrix, 512 × 512; beam collimation, 128 mm × 0.625 mm; beam pitch, 0.7; gantry rotation time, 0.5 s; tube voltage, 100-120 kV; and automated dose modulation with a maximum allowable tube current set at 200 mA.

Each patient received 120-150 mL of iohexol 300 (300 mg iodine) (Bonorex 300; Central Medical Service, Seoul, South Korea). An automatic power injector operating at an injection rate of 3.5 mL/s was used. The arterial, pancreatic, and late phase images were obtained with delays of 40 s, 65 s, and 105 s, respectively, after the injection of the contrast agent.

Imaging analysis

Two radiologists, blinded to the clinical data, performed consensual analysis of the axial CT images on a picture archiving and communication system (PACS; G3, Infinitt Healthcare, Seoul, South Korea). The CT attenuation values [Hounsfield Unit (HU)] of the tumor were measured by drawing circular regions of interest (ROIs) on the arterial and pancreatic phases. The attenuation values of the tumor were analyzed in the arterial (VTarterial) and pancreatic phases (VTpancreatic) and expressed in HU; visible necrosis, adjacent pancreatic parenchyma, and large vessels were excluded[12]. The same ROIs were reproduced at the aorta of the corresponding images, which measured the ROI of the tumor, the attenuation value of the aorta in the arterial phase (VAarterial), and the attenuation value of the aorta in the pancreatic phase (VApancreatic).

The enhancement parameters,i.e., the tumor-to-aorta enhancement ratios of the arterial (TARarterial) and pancreatic phases (TARpancreatic) were the division of VTarterialto VAarterialand VTpancreaticto VApancreatic, respectively.TARarterial=VTarterial/VAarterial;TARpancreatic=VTpancreatic/VApancreatic.

The tumor-to-aortic enhancement fraction (TAF) represents the ratio of difference between the attenuation of the tumor on arterial and parenchymal images to the difference between the attenuation of the aorta on arterial image and pancreatic images. The difference in tumor enhancement between the arterial and pancreatic phases (DT) was calculated by subtracting VTarterialfrom VTpancreatic[13]. The difference in aortic washout between the arterial and pancreatic phases (DA) was calculated by subtracting VApancreaticfrom VAarterial(Figure 1A and 1B). Thereafter, TAF was calculated by dividing DT by DA. Three equations can be summarized as follows:DT=VTpancreatic-VTarterial;DA=VAarterial-VApancreatic;TAF=DT/DA= (VTpancreatic-VTarterial)/(VAarterial-VApancreatic).

Statistical analysis

Statistical analyses were performed with SPSS software (SPSS Statistics for Windows, version 20.0; IBM Corp, Armonk, NY, United States). The Kruskal-Wallis analysis of variance (ANOVA) and the Mann-WhitneyUtest were used to evaluate differences among the three groups,i.e., poorly, moderately, and well-differentiated pancreatic tumors[14]. The patients’ age, sex, tumor size, lesion location, TARarterial, TARpancreatic, and TAF were compared. Moreover, receiver operating characteristic (ROC) analysis was used to compare the diagnostic performance of TARarterial, TARpancreatic, and TAF for predicting the pathologic grade of PDA.P< 0.05 was considered statistically significant.

RESULTS

The study included 23 men and 19 women with a mean age of 62.43 years (SD: 11.42; range: 34–85 years). The 42 lesions investigated in our study were located in the pancreatic head and uncinate process (n= 30), body and neck (n= 11), and tail (n= 1).

Figure 1 Axial computed tomography images on arterial phase (A) and pancreatic phase (B) of a 68-year-old woman, who was classified in the poorly differentiated group. A: Computed tomography (CT) attenuation values [Hounsfield Unit (HU)] of the tumor in the arterial phase (VTarterial ) and the aorta in the arterial phase (VAarterial) were measured by drawing two separate circular regions of interest in the aorta and pancreas on arterial phase image; B: CT HU of the tumor in a pancreatic phase (VTpancreatic) and the aorta in the pancreatic phase (VApancreatic) were measured in the aorta and pancreas on the pancreatic phase 65 s after the injection of the contrast agent.

A total of 42 PDAs were categorized into three groups: (1) The well-differentiated group (WD) (n= 13); (2) The moderately differentiated group (MD) (n= 21); and (3) The poorly differentiated group (PD) (n= 8). The size of the lesions ranged from 1.5 to 7 cm for WD lesions, 1.8 to 5.0 cm for MD lesions, and 2.2 to 13.0 cm for PD lesions, respectively. There were no significant differences in size and location of the lesion among the WD, MD, and PD groups (P≤ 0.076). Patient characteristics are summarized in Table 1.

There were no significant differences in values of VTarterial, VTpancreatic, VAarterial, VApancreatic, DT, and DA among the three groups (Table 2). Moreover, there was no significant difference in the value of TARarterialamong the WD (mean: 0.26, 95%CI: 0.1903-0.3340), MD (mean: 0.27; 95%CI: 0.2284-0.3122), and PD groups (mean: 0.19; 95%CI: 0.1295-0.2465). There was no significant difference in the value of TARpancreaticamong the WD (mean: 0.45, 95%CI: 0.3493 to 0.5435), MD (mean: 0.48; 95%CI: 0.3988-0.5557), and PD groups (mean: 0.37; 95%CI: 0.2724-0.4759) (P< 0.0001) (Table 3).

The value of TAF was statistically different among the three groups; WD (median and mean, 0.19 and 0.28, respectively; 95%CI: 0.1370-0.4289), MD (median and mean, 0.17 and 0.19, respectively; 95%CI: 0.1377-0.2468), and PD (median and mean, 0.10 and 0.10, respectively; 95%CI: 0.02948-0.1675) (P< 0.05).

The diagnostic performances of TARarterial, TARpancreatic, and TAF for the prediction of the pathological grade of PDA are shown in Table 2. The diagnostic performance of TAF (Az = 0.692-0.757) was higher than that of TARarterial(Az = 0.509-0.71) and TARpancreatic(Az = 0.512-0.654) for predicting the pathological grade of PDA, although the difference was not statistically significant (P> 0.093).

DISCUSSION

The pathological tumor grade is an important prognostic factor of survival in patients with PDA[2]. PDA has unique characteristics and different CT enhancement patterns (such as lung and renal cancers)[8,9,12,15,16], based on the proportion of MVD, degree of fibrosis, and residual normal pancreatic tissue.

There were no significant differences in VTarterial, and VTpancreaticamong the three groups in our study. Several researchers[8,10]have studied the correlation between CT enhancement parameters and the histological findings of pancreatic adenocarcinomas. Wanget al[8]reported that the pathological grade showed a good correlation with VTpancreaticand MVD. In contrast, Hattoriet al[10]’s study on pancreatic ductal cancer reported that VTarterialand VTpancreaticwere negatively correlated with the degree of fibrosis. VTarterialshowed a significant correlation with vascular endothelial growth factor and MVD but VTpancreaticwas not correlated with MVD. Hattoriet al[10]reported that TARarterialwas positively correlated with MVD and negatively correlated with the extent of fibrosis. However, our findings demonstrate that there were no significant differences in TARarterialand TARpancreaticamong the three groups.

There were no significant differences in the values of DT and DA among the WD, MD, and PD groups. Aortic enhancement curves showed a decreased slope from the arterial to the pancreatic phases, after the arterial phase and tumor enhancement curves showed an increased slope. However, the degree of aortic enhancement is influenced by the dose of the contrast media, rate of injection, appropriate timing ofcontrast-enhanced imaging, heart rate and cardiac output of the patient, weight, and age[17]. The renal cancer study divided tumor enhancement with aortic enhancement to correct these intrinsic factors[15].

Table 1 Distribution of patient characteristics, lesion size and location

Table 2 Diagnostic performance of computed tomography parameters for the prediction of pathological grading

Finally, the TAF values were statistically different among the WD (median and mean, 0.19 and 0.28, respectively; 95%CI: 0.1370-0.4289), MD (median and mean, 0.17 and 0.19, respectively; 95%CI: 0.1377-0.2468), and PD (median and mean, 0.10 and 0.10, respectively; 95%CI: 0.02948-0.1675) (P< 0.05). Perfusion imaging demonstrates blood flow in the target organ using single-photon emission computed tomography, CT, and magnetic resonance imaging. Perfusion CT can identify vascularity and fibrosis in the diseased pancreas[18,19]. Various perfusion CT parameters can be generated by postprocessing the CT data. Perfusion CT has a smaller FOV, requires additional radiation exposure, and processing time. However, TAF may be obtained with conventional pancreatic CT, without additional radiation exposure and processing time, and is more useful for practical staging than perfusion CT parameters.

There are several limitations to our study. First, we drew two similar ROIs at the aorta and tumor for minimizing intraobserver variation: Two radiologists consensually reviewed PDA lesions in the arterial and pancreatic phases. Therefore, we could not ascertain the inter or intraobserver variations. Second, this study had an inherent bias owing to its retrospective design. Third, our sample size was small, which made it difficult to obtain statistically significant data. Finally, there was no statistically significant difference, but the number of patients with MD PDAs was greater than patients with PD and WD PDAs. Therefore, prospective studies with large populations are needed in the future, to overcome these limitations.

Table 3 Differences in computed tomography parameters among the well-differentiated, moderately differentiated, and poorly differentiated group groups

ARTICLE HIGHLIGHTS

Research conclusions

TAF was statistically different among the three pathologic grade groups. So, the TAF might be correlated with histological finding of PDA. Therefore, calculation of TAF using conventional CT might be useful in predicting the pathologic grade of PDA.

Research perspectives

The conventional CT has been useful modality for diagnosis of PDA. In our study, we suggest the CT enhancement parameter, TAF, could be used as a value for predicting pathologic grade of PDA. The pathologic grade is related to prognosis of PDA, then we can use conventional CT not only for diagnosis, but also for predicting pathologic grade and prognosis of PDA. Also, TAF may be obtained with conventional pancreatic CT, without additional radiation exposure and processing time, and is more useful for practical staging than perfusion CT parameters.