Analyzing the relationship between the cytokine profile of invasive breast carcinoma, its histopathological characteristics and metastasis to regional lymph nodes

Abstract

This study was aimed at analyzing the relations of metastasis to regional lymph nodes (RLNs) with histopathological indicators of invasive breast carcinoma of no special type (IC-NST) and its cytokine profile. Enzyme-linked immunosorbent assays were performed to determine concentrations of IL-2, IL-6, IL-8, IL-10, IL-17, IL-18, IL-1β, IL-1Ra, TNF-α, IFN-γ, G-CSF, GM-CSF, VEGF-A, and MCP-1 in the culture supernatant of IC-NST samples from 48 female patients. Histopathological indicators (degree of tumor cell differentiation, mitoses, and others) and ER, PR, Her2/neu, Ki-67, and CD34 expression levels were determined. By means of three types of neural network models, it was shown that for different parameters of the output layer, different groups of parameters are involved that have predictive value regarding metastasis to RLNs. As a result of multi-dimensional cluster analysis, three clusters were formed with different cytokines profiles of IC-NST. Different correlations between indicators of cytokine production by IC-NST and its histopathological parameters were revealed in groups with different cytokine profiles. It was shown that at simultaneous evaluation of the production of even two cytokines, the importance of which relationship with metastasis was revealed by neural network modeling, can increase the probability of determining the presence of metastasis in the RLNs.

Figure 1

Graphical representation of the results of two-way joining cluster analysis of the data on cytokine production by IC-NST samples and IC-NST histopathological parameters. (a) The horizontal axis of the graph shows the cytokines involved in the classification (according to their production by IC-NST samples, pg/ml), and the vertical axis indicates the encrypted numbers of IC-NST samples from different patients. (b) The graph shows horizontally the histopathological parameters involved in IC-NST classification, and vertically the encrypted numbers of IC–NST samples from different patients. The colors of the intersecting cells indicate that the matrix elements belong to a specific cluster. The color gradient from green to red respectively denotes an indicator’s value below or above the average.

Figure 2

Graphical representation of the results of multi-dimensional cluster analysis of the data of cytokine production by IC-NST samples. The dendrogram is constructed by Ward's method. In the dendrogram, the horizontal axis represents encrypted numbers of IC-NST samples from different patients, and the vertical axis indicates the union distance (Euclidean distances). Clusters I, II and III are formed from relatively similar indicators of different cytokine production (pg/ml) by IC-NST samples and correspond to different profiles of cytokine production by IC-NST samples from different patients.

Figure 3

The 3D Surface Plot (Negative Exponential Smoothing), a 3D distribution surface was constructed, which characterizes the dependence of the number of detected RLNs with metastases on concentrations of various pairs of cytokines (pg/ml) produced by IC-NST samples. (a) The dependence of the number of detected RLNs with metastases on concentrations of TNF-α and GM-CSF produced by IC-NST samples. (b) The dependence of the number of detected RLNs with metastases on concentrations of IL-10 and IL-17 produced by IC-NST samples. (c) The dependence of the number of detected RLNs with metastases on concentrations of IL-18 and VEGF-A produced by IC-NST samples. (d) The dependence of the number of detected RLNs with metastases on concentrations of IL-18 and IFN-γ produced by IC-NST samples.

Figure 4

By means of a 3D surface plot (negative exponential smoothing), a 3D distribution surface was built that characterizes the dependence of the number of detected RLNs with metastases on concentrations of the pair of cytokines (pg/ml) produced by cultured IC-NST samples, and concentrations of different cytokines (pg/ml) and expression of the CD34 differentiation cluster in IC-NST samples. (a) The dependence of the number of detected RLNs with metastases on concentrations of IFN-γ and GM-CSF produced by IC-NST samples. (b) The dependence of the number of detected RLNs with metastases on concentrations of IFN-γ produced by IC-NST samples and indicators of CD34 expression in IC-NST samples. (c) The dependence of the number of detected RLNs with metastases on concentrations of IL-17 produced by IC-NST samples and indicators of CD34 expression in IC-NST samples. (d) The dependence of the number of detected RLNs with metastases on concentrations of IL-18 and indicators of CD34 expression in IC-NST samples.