Between January 2011 and December 2012, a total of 276 patients with pathologically proven gastric cancers and results of HER2 expression status, were selected from our pathology database. Among them, we included patients who fulfilled the following criteria: 1) patients who had locally advanced (T4b) or metastatic gastric cancer (M1) on histopathology or imaging studies, 2) patients who had CT images with optimal gastric distension, 3) patients whose primary gastric lesions were detectable on axial scans of CT images, and 4) patients whose results of HER2 expression were not questionable. We only included patients with locally advanced or metastatic gastric cancer (unresectable gastric cancer) because the HER2 status is critically important in these patients, and since chemotherapy, including targeted therapy, can be used in patients with unresectable gastric cancer. We excluded 95 patients as they had resectable gastric cancers (n = 83), CT images with suboptimal gastric distention (primary cancer cannot be clearly differentiated from adjacent gastric wall [n = 11]), and had primary gastric lesions that were not detectable (n = 1). There were no patients in whom the results of HER2 status were questionable. Thus, finally 181 patients with locally advanced (n = 37) and/or metastatic (n = 172) gastric cancers were included in our study population (Fig. 1). There were 138 men and 43 women, and their mean age was 60.6 ± 13.3 years (age range, 22–95 years). Patients were divided into two groups according to their HER2 expression status: HER2-positive and negative groups. For determining the HER2 positivity, IHC for HER2 overexpression and fluorescence in-situ hybridization (FISH) for gene amplification were used (15). In IHC for HER2 overexpression, anti-HER2 antibody (A0485, DAKO, Carpinteria, CA, USA) was applied for 10 minutes, and antibody binding was detected using an avidin-biotinperoxidase complex (Universal Elite ABC Kit, Vectastain, Burlingame, CA, USA). HER2 immunostaining was scored using the instructions in the DAKO HercepTest kit, as follows: score 0, no membrane staining or membrane staining in < 10% of tumor cells; score 1+, faint/barely perceptible partial membrane staining in ≥ 10% of tumor cells; score 2+, weak to moderate staining of the entire membrane in ≥ 10% of tumor cells; and score 3+, strong staining of the entire membrane in ≥ 10% of tumor cells. Scores of 0 and 1+ were considered negative for HER2 overexpression, and a score of 3+ was considered positive. For patients with score 2+, FISH test using dual-color Vysis kits (Vysis, Downers Grove, IL, USA) were used to clarify the HER2-status (15). A HER2:CEP17 (centromeric probe 17) ratio of ≥ 2 was defined as positive for HER2 amplification (11).

The clinical and histologic features were reviewed by one author (having 5 years of experience) using electronic medical records (EMR) of our hospital. For each patient, the type of treatment for both primary and metastatic tumors, and palliative anti-cancer treatment regimen including cytotoxic chemotherapeutic agents and HER2-directed therapy, were collected. Any surgery undergone was also analyzed. Clinical outcomes included the current status of the patient (dead or alive) and survival days after the initial diagnosis of metastatic gastric cancers. Overall followup survival data was completed by reviewing our hospital EMR, as well as contacting the Resident Service Division of the Ministry of Public Administration and Security. The endpoints of this study were either demise of the patient, or status on October 6, 2015. For histologic analysis, location of the tumor and results of HER2 expression were recorded.

This retrospective study used various CT scanners. All patients with locally advanced or metastatic gastric cancers underwent MDCT with one of 4, 8, 16, 64, 128, or 320 detector-rows, either at our institution (n = 123) or at an outside hospital (n = 58). The parameters used for these MDCTs were as follows: detector collimation, 0.63–1 mm; tube voltage, 120 kVp; tube current, 150–250 mAs; pitch, 0.89–1.35; rotation time, 0.5–0.75 seconds; slice thickness, 2.5–5 mm; and reconstruction interval, 2–5 mm. Among the 181 patients, 108 patients (59.7%) underwent stomach protocol CT with gastric distension, with ingestion of either one pack of an effervescent agent (n = 103) or > 1 liter of water (n = 5). The remaining 73 patients underwent dynamic-enhanced CT after ingestion of 500 mL of water (12). In most patients (n = 138, 76.2%), 3D coronal and/or sagittal multiplanar reconstruction images as well as transverse images were reconstructed, and were used for CT image analysis.

For the contrast-enhanced CT study, 1.5–1.6 mL/kg of a 350 or 370 mgI/mL iodinated contrast agent was injected at a rate of 3–5 mL/sec, using an automatic power injector. CT images were obtained as follows: single portal venous phase (36 patients), dual phases that included the arterial and portal phase (37 patients), and triple phases that included arterial, portal, and delayed phases (104 patients). Dynamic-enhanced images were obtained by scanning the images at 13–17, 60–75 secconds, and 2–3 minutes after attenuation of the descending thoracic aorta reached 100 Hounsfield units, using the bolus tracking technique for arterial, portal, and delayed phases, respectively (12). The other dynamic CT protocol used in 3 patients was the pancreatobiliary protocol including early arterial, late arterial, and portal phases. CT scan with portal and delayed phases was performed for 1 patient.

Morphologic CT features of the primary gastric neoplasms, as well as their enhancement patterns, were assessed by two radiologists (with 5 and 19 years of experience, respectively) in a consensus manner, on a picture archiving and communication system workstation monitor. Both radiologists were aware that the study population had locally advanced or metastatic gastric cancers, but were blinded to their histologic subtype, the degree of differentiation, and IHC results including HER2 expression status.

The analyzed CT findings for the primary gastric lesion were as follows: multiplicity, Bormann type, maximum tumor thickness, tumor margin, clinical tumor node metastasis (cTNM) staging, homogeneity, degree of enhancement on each phase, and dynamic pattern of enhancement (12). Multiplicity was assessed into three categories: single, multiple, or diffuse. Bormann type was classified as: polypoid, type I; ulcerofungating with a sharp raised margin, type II; ulcerated with a poorly defined infiltrative margin, type III; and infiltrative, type IV. The maximum tumor thickness was measured on the axial CT scan using an electronic caliper. Tumor margin was defined as either a well-defined or an ill-defined margin. cTNM staging was assessed similar to prior studies (1617). The CT criteria for T staging are summarized in Table 1. The degree of enhancement was evaluated as high, iso-, or low attenuation on arterial, portal, and delay phases, referencing the enhancement of the normal gastric mucosa. Thereafter, the dynamic pattern of enhancement was assessed as follows: persistently low attenuation (low/low), progressive enhancement pattern (low/iso, low/high, or iso/high), wash-in/wash-out pattern (iso/low, high/iso, or high/low), and persistently high attenuation (high/high) (12). The presence of vascular invasion was also recorded, when the tubular vascular thrombi were noted around the primary gastric lesion.

For evaluating the lymph nodes (LNs), the location and number of metastatic LNs were recorded. We defined the location of nodal stations according to the Japanese classification of gastric cancer (18). LN metastasis was considered to be present if there were any LNs ≥ 8 mm in the short-axis diameter, if a cluster of ≥ 3 peri-lesional nodes was identified regardless of their size, or if the LNs contained necrotic foci (1617). The presence of necrosis was also evaluated in cases of metastatic LNs. The shortest diameter of the biggest metastatic LN was also measured. Finally, the degree of enhancement of the biggest metastatic LN was assessed in comparison with the enhancement of the back muscle.

The location of distant metastasis was classified into distant LN, liver, peritoneum, lung, bone, adrenal gland, tumor thrombi, and others. We also assessed the presence of infrarenal metastatic LNs, and periureteral or peribiliary metastasis. Distant metastasis was considered to be present in 171 patients, was pathologically confirmed in 64 patients, and proven based on images including follow-up studies in 107 patients. The presence or absence of peritoneal seeding was also assessed. Peritoneal seeding was confirmed by histology (n = 44), or based on a combination of CT findings and follow-up studies (n = 46), determined by the consensus of both radiologists.

Hepatic metastasis was analyzed in greater detail by the same two radiologists. The two reviewers assessed the metastatic lesions either as hyper- or hypo-attenuated, in comparison to the adjacent normal parenchyma, and the highest enhancing peripheral portion was considered for qualitative interpretation. In addition, the dynamic enhancement pattern of hepatic metastasis was also assessed in the same manner as the primary lesion. Margin, presence or absence of necrosis, and capsular retraction of hepatic metastasis, were also evaluated.

All statistical analyses were performed using SPSS 21.0 for Windows (SPSS Inc., Chicago, IL, USA). The prevalence of each CT finding between HER2-positive and negative groups were compared using univariate statistical tests such as Fisher's exact test or the Student t test. After identifying significant CT features using univariate analysis, binary logistic regression analyses with the backward projection method were performed to determine the most significant differential CT features. Survival curves were calculated using the Kaplan-Meier method, and the difference was analyzed with the log-rank test. For all statistical analyses, a p value < 0.05 was considered to be statistically significant.

Table 2 summarizes the clinical and histological findings in 181 patients. There were no significant differences in age and sex between the HER2-positive and negative groups. Histopathologically, 37 patients with HER2-negative (24.8%) and 6 patients with HER2-positive (18.8%) were diagnosed after surgery, such as palliative operation. In the remaining 138 patients, endoscopic biopsy was performed to prove gastric cancers. Of the 43 patients who underwent surgery, pathologic TNM staging was used for analysis, while cTNM staging was utilized for patients who did not undergo surgery. Among the 43 patients, gastrectomy with/without LN dissection was performed in 22 patients. In the HER2-negative group, all 149 patients underwent cytotoxic chemotherapy, while combination therapy with cytotoxic agents and trastuzumab was administered in 62.5% (20/32) of the HER2-positive patients. The difference was statistically significant (p < 0.001). In addition, HER2-negative cancers showed a tendency to have more advanced T stage (T4a [68.5%, 102/149] or T4b [23.5%, 35/149]). To the contrary, lower T stages (T2 [9.4%, 3/32] or T3 [12.5%, 4/32]) were more common in HER2-positive cancers (p = 0.015). N and M stage, and morphology of the tumor, did not show any significant difference between the two groups.

Figure 2 shows the survival curves of HER2-positive and negative cancers according to the primary systemic therapy. Median overall survival was significantly longer in HER2-positive cancer patients treated with trastuzumab (13.7 months; 95% confidence interval [CI], 8.5–18.9) than in HER2-positive (5.0 months; 95% CI, 0.4–9.5; p = 0.019) and HER2-negative cancer patients treated with chemotherapy alone (9.6 months; 95% CI, 7.8–11.4; p = 0.035). However, there were no significant differences in survival between HER2-positive and negative cancers treated with chemotherapy alone (p = 0.205).

CT findings of primary tumors and metastatic LNs with results of univariate statistical analysis are presented in Tables 3 and 4. HER2-negative cancers (mean thickness, 1.95 cm) were slightly thicker than HER2-positive cancers (mean thickness, 1.66 cm), but the difference was not statistically significant (p = 0.089). In terms of the degree of primary tumor enhancement, HER2-positive cancers (62.5%, 20/32) more frequently showed hyperattenuation on the portal phase, compared with HER2-negative cancers (30.9%, 46/149) (p = 0.003) (Figs. 3, 4). As for the other phases, the difference in the degree of enhancement was not statistically significant. Furthermore, the pattern of dynamic enhancement did not show a significant difference (Table 3).

In terms of LN metastases, HER2-positive cancers represented slightly larger metastatic LNs when compared to HER2-negative cancers (2.66 cm vs. 2.34 cm, respectively), but the difference was not statistically significant (p = 0.253). Moreover, all other variables (i.e., size, shape, homogeneity, presence or absence of necrosis, and degree of enhancement of LN metastasis) were also not statistically different between the two groups (p > 0.05) (Table 4).

Hepatic metastases were more frequently found in HER2-positive cancers (20/32 [62.5%] vs. 48/149 [32.2%], p = 0.001), while peritoneal seeding was more often found in HER2-negative cancers (80/149 [53.7%] vs. 10/32 [31.3%], p = 0.021) (Fig. 3). Periureteral or peribiliary metastasis was exclusively noted in HER2-negative cancers (13/149 [8.7%] vs. 0/32 [0%]), but the difference was not statistically significant (p = 0.129) (Fig. 4). CT findings of hepatic metastases were not significantly different between the two groups, for all analyzed items (p > 0.05) (Table 5).

Results of binary regression analysis are presented in Table 6. Hyperattenuation of the primary tumor (odds ratio [OR], 4.68; p < 0.001) and the presence of hepatic metastasis (OR, 4.43; p = 0.001) were significant differentiators of HER2-positive cancers from HER2-negative cancers. Representative examples are presented in Figures 3 and 4.