High PROX1 expression in gastric cancer predicts better survival

Background PROX1 is a transcription factor involved in the development of various organs. It has also an important function in colorectal cancer progression. The aim of this study was to investigate the prognostic role of PROX1 expression in gastric cancer. Methods We evaluated PROX1 expression in gastric cancer by immunohistochemistry of tumor-tissue microarrays including tumor specimens from 283 patients who underwent surgery at Helsinki University Hospital. We investigated the association of PROX1 expression with clinicopathologic variables and patient survival. Results Cytoplasmic PROX1 reactivity was high in 56 (20.5%) and low in 217 (79.5%) cases. Low PROX1 immunostaining associated with diffuse cancer type (p = 0.002). In subgroup analysis, PROX1 was a significant marker of better prognosis in patients aged under 66 (p = 0.007), in those with intestinal cancer (p = 0.025), among men (p = 0.019), and in tumors of less than 5 cm diameter (p = 0.030). Patients with high PROX1 expression had a cancer-specific 5-year survival of 65.6% (95% CI 52.7–78.5), compared to 37.1% (95% CI 30.2–44.0) for those with low expression (p = 0.004, log-rank test). This result remained significant in multivariable analysis (HR = 0.56; 95% CI 0.35–0.90; p = 0.017). Conclusion In gastric cancer, high cytoplasmic PROX1 expression is an independent marker of better prognosis.


Introduction
Gastric cancer is a highly malignant tumor and is globally one of the major causes of cancerrelated death [1]. Its incidence in the last decades has decreased, but its 5-year survival rate, despite curative surgery is, in the Western world, still less than 30% [2]. Besides the established prognostic factors such as TNM-stage together with the possibility of curative surgery, additional prognostic information may come from tumor markers.
The transcription factor PROX1 is a key regulatory protein in the development of various organs. It is involved in cell-fate determination and progenitor-cell regulation in the central nervous system, eye, liver, pancreas, lymphatic system, and the heart. It also plays a role in cancer development. Previous studies have indicated that PROX1 can have either tumor-suppressive or oncogenic properties. Various levels of PROX1 protein occur in tumor cells, but the role of PROX1 in the proliferation, migration, and invasion of cancer cells is unclear [3].
In gastric cancer, PROX1 may promote tumor progression by inducing cancer-cell proliferation and lymphangiogenesis [16]. PROX1 expression when, appearing in gastric cancer tissue, may serve as a potential prognostic factor and treatment target. Moreover, tumorigenesis and tumor progression is also linked to dysregulation of microRNAs. Recently Zhang et al. [17] found that miR-489 was significantly downregulated in human gastric cancer tissues and cell lines, and PROX1 was a direct miR-489 target.
The aim of this study was to determine the prognostic value of immunohistochemical staining of PROX1 in gastric cancer.

Tissue samples and immunohistochemistry
We collected formalin-fixed and paraffin-embedded surgical tissue samples from the archives of the Department of Pathology, and de-identified and analyzed the tissue samples anonymously. All histological slides were re-evaluated by an experienced pathologist who defined and marked areas representing the highest grade of the individual tumor. Four 1.0-mm cores from each tumor block, two from the invasive front, and two from the center of the tumor, were sampled and embedded in a new paraffin block by a semi-automatic tissue microarrayer (Tissue Arrayer 1, Beecher Instruments Inc., Silver Spring, MD, USA) as previously described [18]. Sections of 4 μm were cut and processed for immunohistochemistry.
Sections were fixed on slides and dried for 12 to 24 hours at 37˚C, then they were deparaffinized in xylene and rehydrated through gradually decreasing concentrations of ethanol to distilled water. For antigen retrieval, sections were treated in a PreTreatment module (Lab Vision Corp., Fremont, CA, USA) in Tris-HCl (pH 8.5) and in Tris-EDTA (pH 9) buffer for 20 minutes at 98˚C. Sections were stained in an Autostainer 480 (Lab Vision). Tissues were incubated with anti-human Prox1 Antibody (R&D Systems, Inc., Minneapolis, MN, USA; diluted to 1:1800 = 18 μg/ml) overnight at room temperature. For detection ImmPRESS HRP Polymer Detection Kit, Peroxidase, anti-goat IgG (Vector Laboratories, Burlingame, CA, USA) was used. Colon cancer tissue served as a positive control for staining.

Scoring of immunoreactivity
PROX1 immunoreactivity was successfully scored in 273 tumors. We scored cytoplasmic PROX1 intensity in cancer cells as 0 to 3. Negative immunoreactivity was scored as 0, weak positivity 1, moderate positivity 2, and strong positivity 3. All samples were scored independently by two researchers (A.L. and J.H.) without knowledge of clinical status and outcome data. Samples with discordant scores were re-evaluated until consensus was reached. The highest score of the four cores served for further analysis.

Statistical analyses
Associations between PROX1 positivity and clinicopathological variables were assessed by the chi-square test or Fisher's exact test. Disease-specific survival was calculated from date of surgery to death from gastric cancer.
Survival curves were constructed according to the Kaplan-Meier method and compared with the log-rank test. For univariable and multivariable survival analysis, the Cox proportional hazard model had the following covariates entered: age, gender, stage, TNM classification, Lauréns classification, tumor size, and PROX1 expression. Stage, TNM classification, and tumor size were categorical covariates. A p-value of <0.05 was consider statistically significant. All statistical analyses were done with IBM SPSS Statistics version 22.0 (IBM Corporation, Armonk, NY, USA).

Immunohistochemistry and association of PROX1 with clinicopathological variables
Of 273 cases, the cytoplasmic PROX1 reactivity was weakly positive in 99 (36.3%), moderately positive in 39 (14.3%), strongly positive in 17 (6.2%), and negative in 118 (43.2%). In the final analysis, negative and weakly positive was regarded as low expression, and moderately and strongly positive as high expression. Representative images of immunostainings are presented in Fig 1. PROX1 staining occurred mainly in the cytoplasm, and in some strongly stained samples we noted also nuclear immunopositivity. Low PROX1 immunostaining associated with diffuse cancer type (p = 0.002; Table 1). No other clinicopathological variables associated with PROX1.

Survival analyses
In gastric cancer-specific survival analysis according to the Kaplan-Meier method patients with strong and the moderate immunostaining have clearly better survival than patients with weak or negative immunoexpression (p = 0.033, log-rank test; Fig 2A). In the final analysis, we combined strong and moderate immunoexpression to represent high expression and weak and negative immunoexpression to represent low expression. Gastric cancer patients with high PROX1 expression had a cancer-specific 5-year survival of 65.6% (95% CI 52.7-78.5), compared to 37.1% (95% CI 30.2-44.0) for patients with low expression (p = 0.004, log-rank test; HR = 0.52 (95% CI 0.33-0.82), Fig 2B, Table 2).
In subgroup analysis, PROX1 was a significant marker of better prognosis in patients aged under 66 (p = 0.007), among men (p = 0.019), and among patients with tumors of less than 5 cm in diameter (p = 0.030). PROX1 was a significant marker of better prognosis in the subgroup of intestinal cancer type (p = 0.025; Fig 3A) whereas it did not show a prognostic significance among patients with diffuse cancer type (p = 0.290; Fig 3B). In other subgroups studied PROX1 did not serve as a prognostic marker.
PROX1 expression remained significant in multivariable analysis (HR = 0.56; 95% CI 0.35-0.90; p = 0.017; Table 3). Other independent prognostic factors in multivariable analysis were advanced age and distant metastasis, and tumor size greater than 5 cm.

Discussion
Here we show that cancer-specific 5-year survival was significantly better for gastric cancer patients with PROX1-positive tumors. In multivariable analysis, high PROX1 immunopositivity served as an independent marker of better prognosis. Among patients with intestinal cancer, those with high PROX1 immunostaining had superior survival.
PROX1 was expressed evenly throughout the cytoplasm, and in some strongly stained samples we noted also nuclear positivity. Earlier studies have reported both cytoplasmic and nuclear staining patterns [16,19]. The cytoplasmic PROX1 expression by immunohistochemistry correlates with PROX1 mRNA amplification [19]. High mRNA levels correlate also with regional lymph node metastasis but not with distant metastasis. Depending on the cancer, the staining pattern varies. In colonic [4] and hepatocellular carcinomas [6] and in gliomas [15], the PROX1 staining is mainly nuclear, whereas in pancreatic cancer, mainly cytoplasmic staining is observed [9]. The role of cytoplasmic PROX1 expression still remains unknown. One hypothesis is that PROX1 is enriched and activated in the cytoplasm of the cell before being translocated to the nucleus to become functionally active [4].
In gastric cancer, Park et al. [16] studied PROX1 by silencing its expression in cell lines by small interfering RNA against PROX1. Cell proliferation was inhibited by PROX1 knockdown, suggesting that PROX1 may play a role in regulating cell fate by reducing apoptosis as well as by promoting proliferation of gastric cancer cells. They also investigated the prognostic role of PROX1 in 327 gastric cancer patients by immunohistochemistry, finding that the survival of patients with PROX1-positive tumors was significantly worse than for those with PROX1-negative tumors. Unexpectedly, our result contradicted this. Reasons for the difference between their results and the present ones may be several. That 2017 Korean study did not include patients with metastasis. Our chosen antibody was anti-human Prox1 Antibody (R&D Systems  Inc.), whereas they used a polyclonal rabbit anti-human PROX1 (Santa Cruz Biotechnology). They apparently stained whole tissue sections, whereas we utilized the TMA technique. As we scored cytoplasmic intensity in hot-spot tumor areas, they multiplied the stained area by the intensity score, and attributed a score more than or equal to 6 out of 9 for positive grouping.
The effects of PROX1 expression on patient prognosis need to be further studied to determine its value as a prognostic tumor marker. MicroRNAs (miRNAs) are involved in a variety of biological processes, and their dysregulation is linked to tumorigenesis and tumor progression. Earlier studies suggest an important role in tumor biology for miR-489. Zhang et al. [17] studied the expression levels of miR-489 in gastric cancer tissues and cell lines to investigate its role in the regulation of aggressive behavior of gastric cancer cells. They demonstrated that miR-489 is downregulated in human gastric cancer tissues and cell lines. Interestingly, miR-489 transfection reduces PROX1 protein in gastric cancer cells, and PROX1 silencing seems to inhibit proliferation. Further, PROX1 overexpression reverses miR-489-mediated suppression of proliferation and invasion of gastric cancer cells. In gastric cancer tissues, this study found a negative correlation between miR-489 and PROX1 protein expression. Zhang et al. suggest that PROX1 is a direct target of miR-489 in gastric cancer cells. As PROX1 depletion suppresses proliferation, one would expect that low PROX1 expression would correlate with better survival. In contrast, we noted the opposite result, because gastric cancer patients with high PROX1 expression had a better survival than did those with low or no expression.
Petrova et al. [5] in 2008 demonstrated similar findings in colon cancer. According to this study, PROX1 facilitates tumor progression by affecting cell adhesion and polarity. As a downstream effector of TCF/β-catenin signaling in colorectal cancer, loss of PROX1 does not directly affect cell proliferation, but it rather shifts the transcriptional phenotype of colon cancer cells towards more slowly growing adherent cells. Further, overexpression of PROX1 potentiates intestinal adenoma development, suggesting its important role in formation and growth of in situ lesions, which may progress to carcinoma. In addition to this oncogenic role of PROX1, it may in gastric cancer play several roles, as we found that high PROX1 expression was associated with a better prognosis.
The strength of this study is its large gastric cancer patient cohort with a long and reliable clinical follow-up and survival data. By the TMA method, only small areas of each tumor are evaluated, not whole-tissue sections. On the other hand, the TMA technique allows analysis of large patient cohorts with a homogenous staining method. In conclusion, our study shows that PROX1 expression serves as a marker of favorable prognosis in gastric cancer.