Prognostic significance of TRIM28 expression in patients with breast carcinoma

2.1 Patients and tissue samples

One hundred and fourteen tissue specimens were randomly collected from the modified radical mastectomy for BC in the Affiliated Hospital of Nantong University from January 2013 to December 2014. And the corresponding specimens of adjacent tissues about 5 cm from the cancer tissue border were obtained as controls. One hundred and fourteen pairs of tissues were immediately cryopreserved in liquid nitrogen at −80°C after operation. Histopathologic specimens were all confirmed as a single primary BC by pathologists. No antitumor treatment such as surgery, chemotherapy, or radiotherapy was applied before surgery. Patients involved were all female without serious basic diseases before surgery, aged between 35 and 81 years (the median age was 52 years). Patients were followed up in our outpatient clinic or by telephone surveys, and the follow-up time was from the date of the operation to the date 5 years later or when the patient died. The last follow-up time was December 2019. The study was approved by the Ethics Committee of the Affiliated Hospital of Nantong University, and all selected patients signed an informed consent.

2.2 Pathological parameters of patients

The pathological parameters of patients, such as tumor size, vascular or nerve invasion condition, lymph node condition, histologic grade, TNM stage, estrogen receptor (ER), progesterone receptor (PR), Ki67, human epidermal growth factor receptor 2 (HER-2) expression levels, and molecular subtype, were recorded. The histologic grade of BC was divided into I, II, and III by the Pathology department according to the degree of differentiation from high to low [14]. TNM stage was based on the eighth edition of the breast cancer AJCC staging system [15]. The expression levels of relevant biomarkers, including ER, PR, HER-2, and Ki67, were all detected using IHC or in situ hybridization (ISH) by pathologists [16]. Ki67 expression greater than 20% was defined as positive [17]. According to the diagnosis and treatment guidelines of breast cancer in 2019 Chinese Society of Clinical Oncology (CSCO), breast cancer was divided into four kinds of molecular subtypes: Luminal A, Luminal B, HER-2-positive, and triple negative. ER, PR, and HER-2 expressions were all negative in triple-negative BC. Detailed pathological parameters are shown in Table 1.

2.3 Quantitative real-time polymerase chain reaction assay

Approximately 80 mg of each tissue was taken, shredded, and placed in a homogenizer, and 1 mL of TRIzol reagent (Invitrogen, USA) was added for homogenization, followed by RNA extraction according to the manufacturer’s instructions. The extracted total RNA was used as a template and added with the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA) according to the instruction. Then the reverse transcription was performed on the C1000 gene amplifier (BioRad, USA) to synthesize complementary DNA (cDNA). The synthesized cDNA was used as a template for real-time PCR (qPCR) and analyzed in a Roche Cobas 480 automatic fluorescent PCR analyzer (Roche Molecular Systems, Switzerland). 18S was used as an internal reference, and the relative expression level of TRIM28 was calculated with 2^−△△CT value. The sequence of primers used is as follows (Sangon Biological Engineering Company, Shanghai): the forward of TRIM28: TGTTTCCACCTGGACTGTCA, the reverse of TRIM28: CCAGCAGTACACGCTCACAT; the forward of 18S: GTAACCCGTTGAACCCCATT, the reverse of 18S: CCATCCAATCGGTAGTAGCG.

2.4 Immunohistochemical staining analysis

Formalin-fixed paraffin-embedded tissues were taken for serial section at 5 µm thickness by LEICA RM2035 microtome (Leica, Germany). All sections were deparaffinized in xylene and rehydrated with gradient ethanol, followed by rinsing in phosphate-buffered saline (PBS) buffer. Slides were heated in the boiled EDTA antigen retrieval buffer (PH 9.0) for 18 min and then cooled to the room temperature. A PAP Pen was used to drawn a circle at the tissue margin. And 100 µL of 3% H₂O₂ was added for blocking endogenous peroxidase. After incubation at room temperature for 15 min, PBS was used for rinsing slides again. The rabbit antihuman TRIM28 polyclonal antibody (1:100, Abcam, UK) was added to cover tissues, and then incubated with PBS for 16 h at 4°C. The goat anti-rabbit IgG labeled with Biotin (DAKO, Denmark) was added to tissues and incubated at room temperature for 20 min. Slides were put into DAB color developing solution (1:20) (Invitrogen, USA), and the color development degree was observed under BX51 microscope (Olympus, Japan). After that, hematoxylin counterstaining and hydrochloric acid ethanol differentiation were performed for 10 and 3 s, respectively. Having been dehydrated with gradient ethanol, slides were placed in xylene. Last, Neutral resin was used to seal tissue sections. PBS was utilized as a negative control of the primary antibody.

2.5 Immunohistochemical scoring

The signal of TRIM28 was mainly expressed in the nucleus. And brown-yellow particles being emerging in the nucleus, it was judged as positive. The expression level of TRIM28 was evaluated by the percentage and staining intensity of positive cells. Randomly select three high-magnification microscope fields, calculate the percentage of positive cells in each field, and then calculate the average number, which was the required percentage of positive cells. The percentage scores of positive cells were as follows:1 (0–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%). The positive cell staining intensity scores were as follows: 0 (negative), 1 (weak positive), 2 (medium positive), and 3 (strong positive). The percentage of positive cells multiplied by the staining intensity score was the total immunohistochemical score. The mean value (5.66) of immunohistochemical score was used as cutoff. Those with a score ≥6 were defined as positive.

2.6 Statistical analysis

SPSS 25.0 and GraphPad Prism 8.0 software were used for statistical analysis. All quantitative data were expressed by χ ¯ ± s , and analysis of variance and t test were used for comparison between groups. All categorical data were expressed by frequencies and rates; χ ² test was used for comparison between groups. The prognosis of BC patients was evaluated by univariate and multivariate Cox proportional hazards regression models. Kaplan-Meier method was used for survival curve analysis, and differences were compared by Log-Rank test. By using the ROC curve analysis, the area under the curve (AUC) was estimated to assess the diagnostic role of TRIM28 in BC cases. The Optimal Cutoff Value, sensitivity, and specificity were obtained by the calculation of the highest Youden index (the sum of sensitivity and specificity minus 1). The corresponding figure of the ROC curve was plotted by the software of SPSS 25.0. P value less than 0.05 was defined as statistically significant.

3.1 Correlation between TRIM28 expression and clinical features of BC

A total of 114 breast cancer patients were enrolled in this study, in order to analyze the relationship between TRIM28 expression level and clinicopathological parameters, including age, tumor size, vascular or nerve invasion condition, lymph node condition, histologic grade, TNM stage, ER, PR, Ki67, HER-2 expressions, and molecular subtype. According to the result of IHC, patients were divided into TRIM28-positive group and TRIM28-negative group. Highly expressed TRIM28 in BC was associated with lymph node metastasis (P = 0.013), higher TNM stage (P = 0.013), and worse molecular subtype (P = 0.014). Meanwhile, TRIM28 expression was not significantly related to age, tumor size, vascular or nerve invasion condition, histologic grade, ER, PR, HER-2, and Ki67 expressions (Table 1).

3.2 TRIM28 overexpression in BC

In RT-qPCR, the expression of TRIM28 mRNA was detected in 114 BC and paired adjacent tissues. The results demonstrated that TRIM28 mRNA was significantly increased in BC tissues than adjacent tissues (P < 0.001, Figure 1a). And the overexpression of TRIM28 in cancer tissues was significantly correlated with advanced TNM stages (P < 0.001, Figure 1b). However, there was no significant difference in TRIM28 expression between different molecular subtypes (P = 0.105).

Figure 1

The analysis of TRIM28 expression by PCR. (a) TRIM28 expression in BC tissues was significantly higher than that in adjacent normal tissues (P < 0.001). (b) TRIM28 expression in different TNM stages (P < 0.001). ANT, adjacent normal tissue; T, tumor.

To explore the expression of TRIM28 in different TNM stages, IHC was carried out for further analysis. The results showed that TRIM28 was mainly located in the nucleus. There were 65 cases (57.0%) with positive expression of TRIM28 in BC tissues, and 21 cases (18.4%) with positive expression of TRIM28 in adjacent tissues. The difference was statistically significant (P < 0.001, Figure 2a). The immunohistochemical score of BC tissues was significantly higher than that of adjacent cancer tissues (P < 0.001, Figure 2c). Of all 114 samples, 18 (15.8%) were triple-negative breast cancer, whose immunohistochemical score was significantly higher than that of non-triple-negative breast cancer (P = 0.01, Figure 2d), suggesting that the expression level of TRIM28 was associated with molecular subtypes in BC. Additionally, among 114 BC tissues, there were 13 cases at stage 0 (11.4%), 37 cases at I (32.5%), 45 cases at II (39.5%), and 19 cases at III (16.7%). The number of TRIM28 overexpression in all cases were 4 (30.8%) at stage 0, 18 (48.6%) at I, 27 (60.0%) at II, and 16 (84.2%) at III. TRIM28 expression gradually increased with TNM stages. Most tissues at advanced stages had stronger staining (Figure 2b). Some representative immunohistochemical images were shown below (Figure 3a–d).

Figure 2

The IHC staining of TRIM28 expression in tissues. (a) The comparison of the percentage of positive TRIM28 staining in tissues between tumors and ANTs (P < 0.001). (b) The distribution of positive TRIM28 staining at different TNM stages (P = 0.014). (c) The IHC score of TRIM28 in cancer and adjacent tissues (P < 0.001). (d) The IHC score of TRIM28 in triple-negative breast cancer and non-triple-negative breast cancer (P = 0.01). ANT: adjacent normal tissue; T: tumor; TNBC: triple-negative breast cancer; NTNBC: non-triple-negative breast cancer.

Figure 3

Representative IHC staining of TRIM28 in BC tissues and corresponding tissues adjacent to cancer (IHC ×200). a1–b2: TRIM28 expression in two groups of BC and adjacent tissues, a1 and a2 were in the first group, b1 and b2 were in the second group, a1 and b1 were cases of breast cancer tissues, a2 and b2 were cases of adjacent tissues. c1 and c2: c1 was a case of TRIM28 expression in triple-negative breast cancer, and c2 was a case of TRIM28 expression in non-triple-negative breast cancer. d1–d4: TRIM28 expression in different TNM stages, d1 was at stage 0, d2 was at stage I, d3 was at stage II, d4 was at stage III.

3.3 Prognostic significance of TRIM28 in BC patients

In this study, the relationship between TRIM28 expression and BC prognosis was also evaluated. Kaplan-Meier method was performed to analyze the association between overall survival (OS), progression-free survival (PFS), and TRIM28 expression in BC patients. As shown in Figure 4, TRIM28-positive group (n = 65) showed shorter OS time (Figure 4a, P = 0.001) and PFS time (Figure 4b, P < 0.001) than those of the negative group (n = 49).

Figure 4

Kaplan-Meier survival analysis of high TRIM28 expression (n = 49) and low TRIM28 expression (n = 65) in BC patients. (a) The relationship between TRIM28 expression and overall survival rate (P = 0.001); (b) The relationship between TRIM28 expression and progression-free survival rate (P < 0.001).

Furthermore, Cox proportional hazards regression model was used to assess the prognostic value of TRIM28 for BC. Because of their role as potential risk factors for poor prognosis, TRIM28 expression, age, tumor size, vascular or nerve invasion condition, lymph node condition, histologic grade, TNM stage, ER, PR, HER-2, Ki67 expression, and molecular subtype were selected as risk variables. In univariate and multivariate analysis, molecular subtype and TRIM28 expression were independent risk factors regarding OS (Table 2, HR = 0.105, 95% CI = 0.017–0.665, P = 0.017; HR = 3.061, 95% CI = 1.008–9.297, P = 0.048). TRIM28 and HER-2 expression, as well as molecular subtype, were independent risk factors for PFS (Table 3, HR = 3.719, 95% CI = 1.406–9.838, P = 0.008; HR = 3.136, 95% CI = 1.039–9.466, P = 0.043; HR = 0.179, 95% CI = 0.037–0.879, P = 0.034).

Finally, in order to explore the diagnostic usefulness of TRIM28 in BC, IHC scores in 114 cases of breast cancer tissues and the corresponding adjacent normal tissues were included in the study. From the ROC curve analysis, it can be proved that the expression level of TRIM28 can distinguish breast cancer tissues from normal ones. The AUC was 0.832, with the sensitivity of 0.921, the specificity of 0.675, and the 95% confidence interval of 0.777–0.886 (Figure 5).

Figure 5

ROC curve analysis to detect the predictive value of TRIM28 in BC patients.