(a–e) SUNE1 cells stably overexpressed the vector or HOPX (n=8 per group) were injected into the foot pads of mice to construct inguinal lymph node metastasis models. (a) Representative images of primary foot pad tumour and metastatic inguinal lymph node. (b) Representative images (× 200) of microscopic primary tumours in foot pads stained with haematoxylin and eosin (H&E). Scale bar, 100 μm. (c) Representative images and quantification of the average volumes of the inguinal lymph nodes. Mean±s.d.; *P<0.01 compared with vector; Student’s t-tests. (d) Immunohistochemical staining for pan-cytokeratin-positive tumour cells in inguinal lymph nodes (× 40 and × 200). Scale bar, 100 μm. (e) Metastatic ratios of inguinal lymph nodes; Chi-square test was used for statistical analysis. (f–h) SUNE1 cells with vector or HOPX overexpression group (n=5 per group) were injected into the tail veins of mice to construct a lung colonization model. (f) Representative images of macroscopic tumour colonization and growth in lung tissues and the number of tumour nodules in mice xenograft. Mean±s.d.; *P<0.01 compared with vector; Student’s t-tests. (g) Representative images (× 100 and × 400) and quantification of the average number of microscopic tumour nodes in lungs stained with H&E. Scale bar, 100 μm; mean±s.d.; *P<0.01 compared with vector; Student’s t-tests. (h) Immunohistochemical staining for HOPX and SNAIL expression in the lungs of mice xenograft (× 100 and × 400). Scale bar, 100 μm. (i) Relative HOPX and SNAIL expression values were determined via real-time RT–PCR in the NPC tissues (n=24). Statistical analysis was performed using the Pearson’s coefficient test. Each mouse sample was considered as an independent experiment; three technological replications were repeated in each sample.