Correlation between Periostin Expression and Pro-Angiogenic Factors in Non-Small-Cell Lung Carcinoma

Abstract

The role of periostin (POSTN) in remodeling the microenvironment surrounding solid tumors and its effect on the tumor cells in non-small-cell lung carcinoma (NSCLC) have not yet been fully understood. The aim of this study was to determine the relationship between POSTN expression (in tumor cells [NSCLC cells] and the tumor stroma) and pro-angiogenic factors (CD31, CD34, CD105, and VEGF-A) and microvascular density (MVD) in NSCLC. In addition, these associations were analyzed in individual histological subtypes of NSCLC (SCC, AC, and LCC) and their correlations with clinicopathological factors and prognosis were examined. Immunohistochemistry using tissue microarrays (TMAs) was used to assess the expression of POSTN (in tumor cells and cancer-associated fibroblasts [CAFs]) and the pro-angiogenic factors. A significant positive correlation was found between the expression of POSTN (in cancer cells/CAFs) and the expression of the analyzed pro-angiogenic factors (CD31, CD34, CD105, and VEGF-A) and MVD in the entire population of patients with NSCLC and individual histological subtypes (AC, SCC). In addition, this study found that POSTN expression (in tumor cells/CAFs) increased with tumor size (pT), histopathological grade (G), and lymph-node involvement (pN). In addition, a high expression of POSTN (in tumor cells and CAFs) was associated with shorter survival among patients with NSCLC. In conclusion, a high expression of POSTN (in cancer cells and CAFs) may be crucial for angiogenesis and NSCLC progression and can constitute an independent prognostic factor for NSCLC.

Keywords: angiogenesis; cancer; cancer-associated fibroblasts (CAFs); endothelial cells (ECs); extracellular matrix (ECM); invasion; non-small-cell lung carcinoma (NSCLC); periostin (POSTN); pro-angiogenic factors; tumor microenvironment.

Figure 1

Comparison of the expression of POSTN (in cells (A) and in stroma (B)) and the expression of pro-angiogenic markers (VEGF-A (C); CD31 (D); CD34 (E); CD105 (F)) between non-malignant lung tissue and cancer cells. The significance of the differences was determined using the Mann–Whitney U test. The images in (G–K) show punches from tissue microarrays illustrating an example of the immunohistochemical (IHC) reaction for each individual protein, as described above. Error bars represent the standard deviation (SD). *** p < 0.001.

Figure 2

Bar charts showing the expression of POSTN in cells (A–C) and in stroma (D–F) across all cancer types with regard to the T-stage (A,D) and N-stage (B,E) of the TNM classification as well as the tumor stage (C–F). The significance of the differences was determined using the Kruskal–Wallis ANOVA test and the differences between individual groups were evaluated using the appropriate post hoc test. Error bars represent the standard deviation (SD). * p < 0.05; ** p < 0.01; *** p < 0.001.

Figure 3

Bar charts showing the expression of POSTN in cells (A–D) and in stroma (E–H), across all cancer types (A,E), in adenocarcinoma (B,F), and in squamous-cell carcinoma (C,G), with regard to the tumor grade. The significance of the differences was determined using the Kruskal–Wallis ANOVA test (p < 0.001 in bar plots (A–C,E–G)), and differences between individual groups were evaluated with the appropriate post hoc test. The results obtained for large-cell carcinoma were statistically insignificant (D,H). The images in (I–Q) show representative punches demonstrating the IHC reaction for POSTN in adenocarcinoma (I–K), squamous-cell carcinoma (L–N), and large-cell carcinoma (O–Q) with respect to the tumor grade. Error bars represent the standard deviation (SD). *** p < 0.001.

Figure 4

Bar plot showing the only statistically significant difference between males and females (Mann–Whitney U test). Error bars represent the standard deviation (SD). * p < 0.05.

Figure 5

Correlation plots comparing POSTN expression in cells to POSTN expression in stroma, across all cancer types (A), in adenocarcinoma (B), in squamous-cell carcinoma (C), and in large-cell carcinoma (D). The plots include p-values and R-values (calculated with Spearman’s correlation test) and the number of cases, which can be seen in the lower right corner of each plot.

Figure 6

Correlation plots comparing the quantities of CD31-positive (A), CD34-positive (B), and CD105-positive (C) vessels quantified via the Chalkley and Weidner methods. The plots include p-values and R-values (calculated with Spearman’s correlation test) and the number of cases, which can be seen in the lower right corner of each plot.

Figure 7

Correlation plots comparing the expression of POSTN in stroma with the expression of pro-angiogenic markers, as quantified via the Weidner (A–C) or Chalkley (D–F) method. The pro-angiogenic factors are CD31 (A,D), CD34 (B,E), and CD105 (C,F). The plots include p-values and R-values (calculated with Spearman’s correlation test) and the number of cases, which can be seen in the lower right corner of each plot.

Figure 8

Correlation plots comparing the expression of POSTN in cells (A–C) and in stroma (D–F) with the expression of pro-angiogenic markers, as quantified via the Chalkley method, assessed in adenocarcinoma. The pro-angiogenic factors are CD31 (A,D), CD34 (B,E), and CD105 (C,F). The plots include p-values and R-values (calculated with Spearman’s correlation test) and the number of cases, which can be seen in the lower right corner of each plot. The images in (G–I) show punches from tissue microarrays demonstrating an example of the immunohistochemical (IHC) reaction for each individual protein, as previously described. The images in (J–L) show representative magnified regions of the microarray punches.

Figure 9

Correlation plots comparing the expression of POSTN in cells (A–C) and in stroma (D–F) with the expression of pro-angiogenic markers, as quantified via the Chalkley method, assessed in squamous-cell carcinoma. The pro-angiogenic factors are CD31 (A,D), CD34 (B,E), and CD105 (C,F). The plots include p-values and R-values (calculated with Spearman’s correlation test) and the number of cases, which can be seen in the lower right corner of each plot. The images in (G–I) show punches from tissue microarrays demonstrating an example of the immunohistochemical (IHC) staining for each individual protein, as previously described. The images in (J–L) show representative magnified regions of the microarray punches.

Figure 10

Correlation plots comparing the expression of POSTN in stroma (A–D) with the expression of VEGF-A, as assessed with the semi-quantitative IRS method, across all cancer types (A), in adenocarcinoma (B), in squamous-cell carcinoma (C), and in large-cell carcinoma (D). The images in (E–G) show punches from tissue microarrays demonstrating an example of the immunohistochemical (IHC) staining for each individual protein, as previously described. The images in (H–J) show representative magnified regions of the microarray punches.

Figure 11

Kaplan–Meier plots of survival for patients expressing POSTN in their cells (A), POSTN in their stroma (B), and VEGF-A (C) at levels below or above the median. The graphs in (D,E) show the plots of survival for patients with regard to the quartile classifications of the levels of POSTN expression in their cells (D) or stroma (E). Quartiles are shown in ascending order, with I representing the first 25% of results, II representing 26–50%, III representing 51–75%, and IV representing 76% and above. The significance of the differences was determined with the log-rank test. The p-value and the number of cases are given in the lower right corner of each plot.

Figure 12

Kaplan–Meier plots of survival for patients with POSTN in their cells (A,B) or stroma (C,D) with regard to the tumor type (adenocarcinoma (A,C) or squamous-cell carcinoma (B,D)). The significance of the differences was determined using the log-rank test. The p-value and the number of cases are given in the lower right corner of each plot.

Figure 13

Receiver operating characteristic (ROC) curve demonstrating the sensitivity and specificity of the logistic regression model for predicting patient survival over a five-year period, as detailed in Table 5. This curve illustrates the impact of various cut-off values on the model’s sensitivity and specificity. The area under the curve (AUC) can be regarded as a comprehensive indicator of the model’s overall performance. The values in the lower right quadrant represent the model’s performance when the cut-off is set to 50% of the survival probability.