br Statistical analysis br Statistical

2.9. Statistical analysis
Statistical analysis was performed using SPSS Statistics 21 (IBM, Armonk, NY, United States),Graphpad Prism 5 software (La Jolla, CA, United States) and MedCalc for Windows, version 15.0 (Ostend, Belgium). Wilcoxon t-test (non-parametric) or paired t-test (parametric) was used to calculate the difference in o-Phenanthroline between tumor and adjacent normal tissues. Mann-Whitney U test (non-parametric) or unpaired t-test (parametric) was use to compare sub-groups within same group and to calculate difference between the expression of tumor and normal samples in TCGA datasets. The continuous data were re-presented as Mean ± SEM. To detect and exclude any potential out-liers, we considered the following points (1) either the expression va-
Clinicopathological parameters were assessed using Fisher's exact test. The diagnostic potential of the genes was evaluated by employing Receiver Operating Characteristic (ROC) curve. Kaplan–Meier survival curves and log-rank test were used for survival analysis in TCGA data. Cox proportional hazards regression model based univariate and mul-tivariate analyses were conducted to predict prognosis. Correlations were carried out using Pearson's or Spearman's coefficient. All p-values were two-sided and were considered significant when < 0.05.
3. Results
3.1. Patient characteristics
The characteristics of the patients enrolled in the study (n = 37) are presented in Table 2. There were 28 males and 9 females with the median age of 55 years. All cases were adenocarcinomas out of which 33 were well/moderately differentiated (89.1%) whereas three cases of mucinous adenocarcinoma (8.2%) and one case of papillary carcinoma (2.7%) was present. Further, patients were characterized on the basis of TNM staging. There were 9 patients (24.32%) with tumor reaching to muscularis propria (pT2) and 28 patients (75.68%) with locally ad-vanced tumors invading subserosa and serosa (pT3 + pT4). Lymph node metastasis was observed in 12 patients (32.43%) and only one  BBA - Molecular and Cell Biology of Lipids 1864 (2019) 1314–1327
Table 2
Demographic and clinical characteristics of CRC patients (local cohort).
Clinical parameters Patient cohort (n = 37)
Sex
Male 28 Female 9 Age
Left-sided colon 16 Right-sided colon 14 Rectum 7 Histological subtypes
Adenocarcinomas 33 Mucinous Adenocarcinoma 3 Papillary Adenocarcinoma 1 Pathology TNM stage
I 7 II 18 III 11 IV 1 Pathology T Stage
Well/Moderately differentiated 34 Poorly differentiated 3 Preoperative Serum CEA (ng/ml)
Clinicopathological features of TCGA cohort are summarized in Table 3.
3.2. Histopathological examination of the tumor and adjacent normal tissues
Frozen sections could be carried out in the 10 cases of tumor tissue and adjoining normal mucosa. Oil Red O staining in these frozen sec-tions to investigate cellular lipid content in the tumor and adjacent normal tissues exhibited increased intracytoplasmic lipid accumulation in the tumor cells whereas sections of the adjacent normal mucosa did not show any visible lipid droplets (Fig. 1).
3.3. Accumulation of cholesterol in tumor tissues
Table 3
Clinicopathological features of TCGA cohort and their association with HMGCR, NR1H3 and NR1H2.
Characteristics TCGA cohort(n = 375)
HMGCR expression p-Value NR1H3 expression p-Value NR1H2 expression
p-Value
Gender
Age
Tumor site
Pathology T stage
Pathology N stage
Pathology M stage
Pathology TNM Stage
Bold p-values indicate statistically significant correlations.
cholesterol levels. The mean cholesterol levels were 29.95 ± 3.36 μg/ mg protein and 22.64 ± 1.23 μg/mg protein in females and males, respectively (Fig. 2B). No notable association was found between cho-lesterol level and other clinical parameters.