The diagnostic and prognostic value of UBE2T in intrahepatic cholangiocarcinoma

Patients

Thirteen IHBD, 23 BilIN-1/2, and 11 BilIN-3 tissues from April 2008 to November 2013 were obtained, and 401 ICC formalin-fixed paraffin-embedded (FFPE) specimens diagnosed at the Eastern Hepatobiliary Surgery Hospital, Shanghai, from July 2000 to December 2008 were included; all specimens were classified into the diagnostic or prognostic group. Two experienced pathologists reexamined the associated hematoxylin and eosin (HE)-stained slides containing FFPE tissues. The inclusion criteria were as follows: (1) pathological diagnosis according to the histological diagnostic criteria of the WHO; (2) no preoperative anti-cancer treatment; (3) partial hepatectomy with curative intent; and (4) the existence of complete data for both the UBE2T H-score and follow-up period. Laboratory tests were conducted on blood samples obtained before surgery. The tumor diameter and surgical margin were tested on the largest tumor by using gross pathologic specimens. Written informed consent was obtained from all patients. The study protocol was approved by the Research Ethics Committee of Eastern Hepatobiliary Surgery Hospital (EHBHKY2015-01-001).

Follow-up

The endpoints of this study were time to recurrence (TTR) and overall survival (OS). TTR was defined from the date of hepatic resection until the detection of tumor recurrence, death or last observation. OS was defined as the interval between surgery and death or last observation. The patients’ follow-up examinations were performed every 3 months during the first year after surgery and every 6 months thereafter. At each visit, the tests for liver functions, alpha fetoprotein (AFP), carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA199) and an abdominal ultrasound were conducted. Contrast-enhanced computed tomography (CT) scanning or magnetic resonance imaging (MRI) was performed once every 3-6 months or when recurrence or metastasis was suspected. Follow-up data were collected until September 2014.

Data mining of the TCGA database

Data on the mRNA expression of UBE2T and associated clinical prognosis were obtained from the data library of The Cancer Genome Atlas (TCGA).

Construction of tissue microarrays and IHC

As previously reported, tissue microarray construction, IHC, and H-score measurements were performed for all the cases (Hirsch et al., 2003; Yu et al., 2019). The HE-stained slides of all patients were reviewed and identified, and then, the typical areas were premarked in the paraffin blocks. The marked areas of each block were punched by tissue cylinders with diameters of 1.5 mm. As in previous studies, the representativeness of a 0.6-mm core in the tissue microarray was equal to that of a larger slide for the optimization of standardized experimental conditions and for assessing focal and heterogeneous expression types (Anagnostou et al., 2010; Jones & Prasad, 2012), so the expression characteristics of a tissue specimen of this size were considered to be representative of the whole tissue slice. Then, the obtained tissues were incorporated into a recipient paraffin block afterwards. After the sections were sliced to 4-µm thickness, they were placed on slides coated with 3-aminopropyltriethoxysilane.

Paraffin sections were deparaffinized in xylene and rehydrated through decreasing concentrations of ethanol (100%, 95%, and 85% for 5 min each). Antigens were unmasked by microwave irradiation for 3 min in pH 6.0 citric buffer and cooled at room temperature for 60 min. Endogenous peroxidase activity was blocked by incubation in 3% H₂O₂/phosphate-buffered saline, and goat serum was used to block nonspecific binding sites. The primary antibody was as follows: rabbit polyclonal antibody to UBE2T (GTX106464; GeneTax, USA; 1:200 dilution). An EnVision Detection kit (GK500705; Gene Tech, Shanghai, China) was employed to visualize UBE2T. Tissue sections were counterstained with hematoxylin for 5 min. Negative control slides without primary antibody were generated for all assays.

The UBE2T-stained slides were scanned on a KFBIO KF-PRO-005-EX digital section scanner (Konfoong Biotech International Co., Ltd., Yuyao, China). Images were analyzed using the HALO 2.0 CytoNuclear Quantification algorithm (Indica Labs), a whole-slide imaging data analysis software program that measures and reports individual cell data that is represented as the percentage of positive cells per mm² tissue. The H-score was used to express the measurement of the HALO software (Fig. S1) (Kargl et al., 2017). ${\displaystyle \mathrm{H\; -\; score}=1\times \left(\text{\%}\mathrm{cells}1+\right)+2\times \left(\text{\%}\mathrm{cells}2+\right)+3\times \left(\text{\%}\mathrm{cells}3+\right).}$

X-tile analysis

The optimal cutoff point used for the survival analysis of the different UBE2T expression groups was calculated with X-tile software version 3.6.1 (Yale University School of Medicine, New Haven, CT, USA) (Camp, Dolled-Filhart & Rimm, 2004). UBE2T expression was represented as the H-score, and X-tile plots were used for optimization of cutoff points based on follow-up data. Statistical significance was evaluated by a standard log-rank method using the cutoff score derived from 401 ICC cases, with the P value obtained from a lookup table.

Statistical analysis

Continuous variables were compared using the Mann–Whitney test or Student’s t-test. The chi-squared test or Fisher’s exact test were used to analyze categorical variables when appropriate. Multiple sets of measurement data were compared by means of a multiple independent samples nonparametric test (Kruskal-Wallis test). Receiver operating characteristic (ROC) area under the curve (AUC) analysis showed the discriminatory power of the putative markers. The TTR and OS rates were calculated by the Kaplan–Meier method and log-rank test. Univariate and multivariate prognostic analyses were conducted using the Cox proportional hazards regression model. A P value less than 0.05 indicated statistical significance. All statistical analyses were performed using SPSS 24.0 software (SPSS, Chicago, IL, USA) and GraphPad Prism 8.0.1 (GraphPad Software, La Jolla, CA, USA).

Expression differences in and prognostic impact of UBE2T on ICC based on the TCGA database

Based on the data obtained from the TCGA database, 9 cases of normal tissue and 36 cases of ICC were analyzed. The UBE2T mRNA expression level in the ICC tissues was obviously higher than that in the normal tissues (P < 0.0001, Fig. 1A). For the 28 ICC patients with follow-up data, when the median UBE2T expression value was used as the cutoff point, the survival analysis showed no difference in OS (P = 0.3679, Fig. 1B).

UBE2T expression profiles of the IHBD, BilIN-1/2, BilIN-3, and ICC and its diagnostic value

The H-scores for 13, 23, 11, and 401 IHBD, BilIN-1/2, BilIN-3, and ICC tissues, respectively, were calculated for UBE2T expression-level comparisons. The immunohistochemical expression characteristics of UBE2T in the IHBD, BilIN-1/2, BilIN-3, and ICC tissues are shown in Fig. 2. The UBE2T expression level of the ICCs was significantly higher than that of the IHBD (P = 0.003), BilIN-1/2 (P < 0.001), and BilIN-3 tissues (P = 0.012) (Fig. 3A). Additionally, the ROC curve revealed that there was a strong discrimination between ICC and the IHBD (AUC = 0.782), BilIN-1/2 (AUC =0.774), and BilIN-3 tissues (AUC = 0.776) (Fig. 3). The baseline characteristics of the three latter groups are shown in Table S1.

Clinicopathological characteristics of the ICC cohort

For the 401 patients with ICC, analysis by X-tile software was performed to assess the best cutoff point for the H-score of UBE2T. Using a standard log-rank method, with P values acquired from a lookup table for TTR and OS, we selected an H-score of <28.96 as the best cutoff point. A total of 288 and 113 patients were divided into low- and high-expression groups, respectively (Fig. S2). The baseline characteristics of the ICC patients are shown in Table 1. The median follow-up time was 42.1 months.

Impact of UBE2T on the clinical outcomes

Kaplan–Meier analysis was conducted to compare the prognoses of ICC patients with low or high expression of UBE2T. The results suggested that the high-expression group had a shorter TTR and OS than the low-expression group. The median TTRs of the high- and low-expression groups were 5.7 months and 8.5 months, respectively. The 1-, 3-, and 5-year TTR rates were 69.3%, 85.6%, and 100%, respectively, in the high-expression group and 56.2%, 72.5%, and 86.2%, respectively, in the low-expression group (P = 0.005, Fig. 4A). The median OS times of the high- and low-expression groups were 18.4 months and 24.2 months, respectively. The corresponding 1-, 3-, and 5-year OS rates were 64.4%, 25.7%, and 1.9%, respectively, in the high-expression group and 73.4%, 40.7%, and 22.2%, respectively, in the low-expression group (P < 0.001, Fig. 4B).

Independent prognostic factors of TTR and OS

Univariable analysis by means of Cox proportional hazards regression revealed that AFP, CA199, tumor size, tumor number, liver cirrhosis, microvascular invasion (MVI), tumor-node-metastasis (TNM) staging, and UBE2T were associated with TTR (all P < 0.05, Table 2) and that AFP, CA199, albumin (ALB), alkaline phosphatase (ALP), tumor size, tumor number, MVI, TNM staging, and UBE2T were related to OS (all P < 0.05, Table 3). Multivariable analysis indicated that multiple tumor nodules, poorer TNM staging, and high UBE2T expression were independent risk factors for TTR (all P < 0.05, Table 2) and that higher CA199, lower ALB, larger tumor size, multiple tumor nodules, presence of MVI, poorer TNM staging, and high UBE2T expression were significant risk factors for OS (all P < 0.05, Table 3).