Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers, with an overall 5-year survival rate of approximately 8.5% [1]. Majority of the patients present with locally advanced disease or metastases due to aggressive tumor biology which contributes to the poor prognosis. Liver is the most common site of distant metastases in PDAC which has been reported in approximately 80% of patients during the later stages of disease [2]. Additionally, 60–70% of patients who undergo curative surgery are bound to develop metastatic disease within 5 years of resection [3].

MR imaging can provide significantly better diagnostic performance for characterizing focal liver lesions [4]. Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA, Eovist® or Primovist®, Bayer HealthCare) is a liver-specific contrast material that has been found to be useful for the detection of liver metastases in patients with PDAC [5]. The excellent lesion-to-liver contrast on hepatobiliary phase (HBP) images enables improved sensitivity for detection of liver metastases compared with contrast-enhanced CT especially in small lesion < 10 mm (90% in MRI and 62–64% in CT) [5]. This contrast agent is a substrate for specific hepatocyte membrane transporters known as organic anion transporter proteins (OATP; specifically, 1B3 variant). In principle, non-hepatocyte origin hepatic lesions cannot actively uptake gadoxetic acid; hence, they exhibit hypointensity on HBP images. However, it is interesting to note that, in the recent times, studies have shown that the transporter responsible for gadoxetic acid uptake is expressed in various cancer cells [6]. In fact, liver metastases from colorectal cancer often appear hyperintense on HBP images and the respective patients are often associated with poor prognosis [7]. To the best of our knowledge, there is no previously published study that has evaluated the association between imaging findings of liver metastasis from PDAC and patients’ outcome. Therefore, the present study aimed to evaluate the relationship between the patterns of HBP contrast uptake in liver metastases on gadoxetic acid–enhanced magnetic resonance (MR) imaging and overall survival (OS) in patients with PDAC.

Materials and methods

Patients

This retrospective study was approved by our institutional review board, and the need for written informed consent was waived. Between September 2012 and September 2019, 63 consecutive patients with pathologically proven PDAC who underwent gadoxetic acid–enhanced MR imaging and had at least one liver metastasis at presentation of disease were identified. Among them, 6 patients were excluded because of the use of scan parameters different from standard examination (n = 4) and respiratory motion artifact (n = 2). Consequently, the remaining 57 patients (mean ± standard deviation (SD) age, 64.9 ± 9.2 years; age range, 46–92 years) were included in our study. Overall, 30 patients were men (mean ± SD age, 63.8 ± 10.3 years; range, 46–92 years) and 27 were women (mean ± SD age, 66.1 ± 7.8 years; range, 49–81 years). Out of the 57 patients, 44 patients received chemotherapy or chemoradiation therapy as neoadjuvant (n = 12), adjuvant (n = 4), or palliative therapy (n = 28). Neoadjuvant chemoradiation therapy options were 50.4 Gy/28 fractions of external beam radiation and FOLFIRINOX (a regimen composed of folinic acid, fluorouracil, irinotecan, and oxaliplatin) (n = 7), combined gemcitabine and nab-paclitaxel (PTX) (n = 3), or capecitabine (n = 2). Adjuvant chemotherapy option was gemcitabine (n = 4). Palliative chemotherapy options were FOLFIRINOX (n = 19) or combined gemcitabine and PTX (n = 9).

MR imaging parameters

MR imaging of the pancreas was performed using a 1.5-T (Signa HDxt; GE Healthcare and Magnetom Avanto; Siemens) or 3-T MR system (Discovery MR750; GE Healthcare and Magnetom Trio; Siemens) with a phased array receiver coil covering the upper abdomen. The MR imaging protocol consisted of the following sequences: three-dimensional fat-suppressed axial T1-weighted spoiled gradient-echo imaging, in-phase and opposed-phase T1-weighted axial gradient-recalled-echo imaging, respiratory-triggered two-dimensional fat-suppressed axial T2-weighted fast spin echo imaging, and respiratory-triggered two-dimensional axial diffusion-weighted imaging with a single-shot echo-planar sequence.

Dynamic contrast-enhanced images were obtained with 0.025 mmol gadolinium (Gd)/kg body weight for Gd-EOB-DTPA (0.25 mmol Gd/mL; Eovist®, Bayer HealthCare) using T1-weighted fat-suppressed sequences (TR/TE, 3.74/1.82 ms; flip angle, 12°; section thickness, 5 mm; slice gap, 0 mm; field of view, 312 × 400 mm; matrix, 256 × 154; slice number, 44 slices; and acquisition time, 20 s with breath holding): pre-contrast, arterial dominant phase (35–40 s), portal venous phase (60–75 s), late phase (3 min), and HBP (20 min) images. Gadoxetic acid was administered intravenously by automatic power injector at a rate of 1 mL/s followed by 20 mL saline flash with the same rate.

Reference standard of liver metastases

In patients with numerous liver metastases, we counted and evaluated liver metastases up to 10 lesions per each patient. Overall, 344 liver metastases were identified in 57 patients. Among them, 145 lesions were excluded from this study because the tumor size was too small (< 5 mm) to perform image analysis. Pathological confirmation was obtained by CT-guided biopsy (n = 19) or surgical resection/partial hepatectomy (n = 4). In the remaining 34 patients, confirmation of metastases was obtained based on a combination of typical imaging findings, clinical picture, and increase in size of the lesions on follow-up images. The typical imaging findings used for diagnosis of metastases included hyperintensity on T2-weighted imaging, restricted diffusion and hypoenhancement on dynamic early post-gadolinium-enhanced images, and hypointensity on HBP images.

Qualitative image analysis

Qualitative image analysis was performed by two independent readers (Y.N. and T.T.; 8 and 12 years of post-training experience in interpreting abdominal MR images, respectively) on PACS (AGFA IMPAX Viewer). The readers evaluated the HBP images acquired 20 min after injection with contrast agent [8]. The readers were blinded to patients’ clinical information or outcomes. Consequently, the patterns of HBP contrast uptake in liver metastases were evaluated and each nodule was classified as homogeneous or heterogeneous hypointense in consensus. Homogeneity was defined as a signal intensity that was similar to or lower than that of the inferior vena cava, whereas heterogeneity was when the signal intensity was higher than that of the inferior vena cava with areas of internal heterogeneity. During patient-by-patient analysis, patients with only homogeneous hypointense nodules were placed in the homogenous group, whereas patients with only heterogeneous hypointense nodules or both patterns of nodules were placed in the heterogeneous group.

Quantitative image analysis

Quantitative assessment of the liver metastases was performed by quantifying the signal intensities of liver metastasis (SInodule) and liver parenchyma (SIliver) on the HBP images by carefully excluding vessels and bile ducts. The background noise was quantified as the SD of the signal intensity of the liver parenchyma (SDliver), whereas the signal-to-noise ratio (SNR) of the liver metastasis was calculated as a ratio of the liver metastasis signal intensity to the background noise. Furthermore, the tumor-to-liver contrast-to-noise ratio (CNR) was calculated using the following equation: CNR = (SIliver − SInodule) / SDliver. Additionally, the maximum diameter of liver metastasis was measured on HBP axial or coronal images with greatest dimension.

Statistical analysis

Statistical analyses were performed using the MedCalc version 19.1.3 software program for Windows (MedCalc Software). The Mann–Whitney U and Fisher exact tests were conducted to compare patients’ age, sex, presence or absence of chemotherapy, plasma carcinoembryonic antigen (CEA) and carbohydrate antigen (CA) 19-9 levels, number of liver metastases, and presence or absence of extrahepatic metastasis between homogeneous and heterogeneous groups. The Mann–Whitney U test was conducted to compare maximum diameter of liver metastasis, SNR, and CNR between homogeneous and heterogeneous hypointense nodules. Kaplan–Meier analysis and log-rank test were conducted for univariate analysis and Cox proportional hazards regression was conducted for multivariate analysis to evaluate prognostic factors for OS in patients with PDAC and liver metastasis. To assess interobserver variability in the definitions for the homogeneous and heterogeneous groups, κ statistics was used to measure the degree of interobserver agreement. A p value < 0.05 was considered to be statistically significant.

Results

Patient demographics and tumor characteristics

Patient demographics and tumor characteristics are summarized in Tables 1 and 2. The mean follow-up interval for the 57 patients included in this study was 15.3 months (range, 0.1–61.3 months) following the diagnosis of liver metastasis. During the follow-up period, 23 patients died with an estimated 5-year survival rate of 20.1% and the mean survival time of 29.9 months.

Table 1 Background factors between homogeneous and heterogeneous groups
Full size table
Table 2 Tumor characteristics
Full size table

Qualitative assessment of enhancement on hepatobiliary phase images

Majority of the nodules (n = 138, 69%) demonstrate heterogeneous hypointensity, while nearly a third of metastatic nodules were homogeneous (n = 61, 31%). Homogeneous and heterogeneous hypointense nodules were encountered in 18 (32%) and 29 (51%) patients, respectively, and both patterns co-existed in 10 patients (17%). Patients were further classified into homogeneous (18 patients) and heterogeneous (39 patients) groups (Fig. 1).

Fig. 1
figure 1

Flow chart showing included and excluded patients

Full size image

The number of liver metastases was significantly greater in the heterogeneous group than in the homogeneous group (5.5 ± 3.8 vs 2.7 ± 2.3, p = 0.015). Although no significant difference was found in patients’ age, sex, presence or absence of chemotherapy, plasma CEA and CA 19-9 levels, and presence or absence of extrahepatic metastasis (p = 0.13–0.90), tumor markers tended to be higher in the heterogeneous group than in the homogeneous group (Table 1). The sites of extrahepatic metastasis which were confirmed by follow-up imaging included the lung (n = 13), peritoneal carcinomatosis (n = 7), bone (n = 5), lymph node (n = 2), adrenal gland (n = 2), cerebellum (n = 1), and kidney (n = 1).

Quantitative assessment

The maximum diameter of liver metastasis (p = 0.0001) and SNR (p < 0.0001) were significantly greater in the heterogeneous hypointense nodules than in the homogeneous ones. The CNR was significantly greater in the homogeneous hypointense nodules than in the heterogeneous ones (p = 0.0016) (Table 2). For Kaplan–Meier method and log-rank test, patients were classified into two groups using median values of age (64.0 years), maximal diameter of liver metastasis (11.5 mm), and number of liver metastasis (3 lesions). Plasma CEA (5.0 ng/mL) and CA 19-9 (37.0 U/mL) cutoff values were based on our institutional standard.

Patient outcome

Table 3 demonstrates the results of univariate and multivariate analyses of prognostic factors for OS. For Kaplan–Meier method and log-rank test, patients were classified into two groups using median values of age (64.0 years), maximum diameter of liver metastasis (11.5 mm), and number of liver metastasis (3 lesions). Plasma CEA (5.0 ng/mL) and CA 19-9 (37.0 U/mL) cutoff values were based on our institutional standard. Significant prognostic factors for OS in multivariate analysis were as follows: maximum diameter of liver metastasis (hazard ratio (HR): 0.92; p = 0.037) and pattern of HBP contrast uptake (HR: 4.66; p = 0.018). The heterogeneous group had lower OS compared with the homogeneous group (mean OS, 48.5 months vs 23.9 months; p = 0.032) (Figs. 2, 3, and 4).

Table 3 Univariate and multivariate analyses for prediction of overall survival
Full size table
Fig. 2
figure 2

A 73-year-old female with liver metastases. The axial hepatobiliary phase image at 20 min demonstrates the homogeneous hypointense nodule (arrow) with homogeneous hypointensity. The SNR and CNR were 7.0 and 10.2, respectively. The survival duration was 43.2 months

Full size image
Fig. 3
figure 3

A 64-year-old female with liver metastases. The axial hepatobiliary phase image demonstrates the heterogeneous hypointense nodule (arrow) with clustered uptake of hepatobiliary contrast agent in the lesion. The SNR and CNR were 15.0 and 8.4, respectively. The survival duration was 9.4 months

Full size image
Fig. 4
figure 4

Kaplan–Meier overall survival (OS) curves for patients with homogeneous and heterogeneous groups. The heterogeneous group exhibited lower OS rate than the homogeneous group (mean OS, 48.5 months vs 23.9 months; p = 0.032)

Full size image

Discussion

In principle, liver metastases do not exhibit any uptake of hepatobiliary contrast agent on the HBP images and therefore remain hypointense [9]. However, several radiologic studies have described atypical or paradoxical uptake of gadoxetic acid in liver metastases on HBP [9,10,11]. Throughout our clinical experience, we have also observed paradoxical uptake in liver metastases on HBP images in patients with PDAC. To our knowledge, there are limited data on the relationship between the patterns of HBP contrast uptake by liver metastases and patient outcomes. Our study demonstrated that patients with PDAC who present heterogeneous hypointense liver metastasis on HBP images are associated with lower OS rate than patients with homogeneous hypointense nodules.

Gadoxetic acid is a liver-specific MR imaging contrast agent that is taken into hepatocytes by a transporter OATP1B3. Consequently, liver parenchyma and hepatocellular carcinomas or adenomas of hepatocyte origin typically show enhancement on HBP images [12, 13]. Tumors of non-hepatocyte origin, including liver metastases from PDAC, do not express OATP1B3; therefore, they do not exhibit any uptake of gadoxetic acid and appear hypointense on HBP images [14]. However, aberrant OATP1B3 expression has been confirmed in several tumors of non-hepatocyte origin, including colon cancer, breast cancer, and PDAC [15,16,17,18,19]. Heterogeneous hypointense liver metastases from colorectal cancer or breast cancer have been associated with a worse prognosis or therapeutic response to chemotherapy than that observed in homogeneous hypointense metastases [7, 19]. Immunochemical OATP1B3 overexpression has also been correlated to a worse progression-free survival rate than negative OATP1B3 expression [20]. The aberrant OATP1B3 overexpression is associated with decreased apoptosis following chemotherapy and to a significantly more aggressive tumor behavior [15].

Treatment paradigm for PDAC has been changing and early identification of biomarkers for predicting aggressive tumor biology allows dictating treatment strategies. Due to recent effective chemotherapy like FOLFORINOX, patients with locally advanced PDAC and synchronous liver metastasis have a chance to receive subsequent curative surgery [21, 22]. Therefore, in patients with oligometastatic liver disease, identification of pattern of enhancement might allow selection of patients for surgical resection, thereby improving outcome since pancreatic surgeries are not trivial.

Currently, there are no studies that have investigated the relationship between the patterns of HBP contrast uptake in liver metastases on gadoxetic acid–enhanced MR imaging and prognosis in patients with PDAC. However, both colon cancer and PDAC cells have been shown to express aberrant OATP1B3 different from wild-type OATP1B3 expressed in the normal hepatocytes [17]. This aberrant OATP1B3 has only limited transport activity and different subcellular localizations. Although it is necessary to further investigate the biological and clinical significance of the aberrant OATP1B3 expression in PDAC, we believe that OATP1B3 overexpression, which appears as heterogeneous hypointense nodule, is associated with a worse prognosis than negative OATP1B3 expression.

Our study has several limitations. First, we used a retrospective design and performed our study at a single center with a relatively small sample size. Second, pathological evidence was not obtained in all patients. However, there was imaging and clinical confirmation of liver metastases in all patients. Finally, we used various MRI scanners from different manufacturers because of the fact that the study period was relatively long (84 months). Nonetheless, the liver imaging protocol was standardized particularly with respect to the slice thickness and the phase of acquisition. Further prospective clinical studies including the evaluation of immunochemical OATP1B3 expression are needed to validate our results.

In conclusion, patients with heterogeneous hypointense liver metastases exhibited worse prognosis than those presenting homogeneous hypointense nodules. Hepatobiliary contrast uptake pattern in liver metastasis on HBP images was possibly associated with OS in patients with PDAC and liver metastasis.