Purpose: This study aimed to describe the computed tomographic (CT) findings of gastric hepatoid adenocarcinoma (GHA) and determine features distinguishing it from gastric adenocarcinoma (GA).
Methods: Computed tomographic images of pathologically verified GHA (n = 11) and GA (n = 38) were retrospectively reviewed. α-Fetoprotein (AFP), carcinoembryonic antigen (CEA), and CT findings were assessed in our study. Computed tomographic findings included the location, distribution, growth pattern, ulceration, thickness of gastric wall, enhancement pattern, and the ratio of lesion attenuation to aorta CT attenuation. Short diameter of the metastatic lymph nodes, mean short diameter of metastatic lymph nodes, and the ratio of the number of enlarged lymph node on CT to the number of metastatic lymph nodes pathologically were measured and calculated. Data were compared using the χ and Student t tests; significant CT criteria were identified using receiver operating characteristic curve.
Results: α-Fetoprotein, CEA, and CT findings, including the longest short diameter, the mean short diameter, the ratio of the number of enlarged lymph node on CT to the number of metastatic lymph nodes pathologically, the lesion in arterial phase minus portal venous phase, and the lesion/aorta ratio, were statistically significant predictors for the differentiation of GHA from GA (P < 0.05). When only the aforementioned CT findings were used as criteria, the sensitivity and specificity for diagnosing GHA were 82.86% and 90.91%, respectively. When AFP, CEA, and CT findings were used as criteria, sensitivity of 97.14% and specificity of 90.91% were achieved.
Conclusions: Elevated serum AFP level and CT findings could distinguish GHA and GA with a high degree of accuracy.