Association of B‐cell lymphoma 2/microRNA‐497 gene expression ratio score with metastasis in patients with colorectal cancer: A propensity‐matched cohort analysis

Abstract Background Deregulated microRNAs (miRs) significantly impact cancer development and progression. Our in silico analysis revealed that miR‐497 and its target gene B‐cell lymphoma‐2 (BCL2) could be related to poor cancer outcomes. Purpose To investigate the BCL2/miRNA‐497 expression ratio in colorectal cancer (CRC) and explore its association with the clinicopathological characteristics and CRC prognosis. Methods Archived samples from 106 CRC patients were enrolled. MiR‐497 and BCL2 gene expressions were detected by Taq‐Man Real‐Time quantitative polymerase chain reaction in propensity‐matched metastatic and nonmetastatic cohorts after elimination of confounder bias. Results B‐cell lymphoma‐2 gene was upregulated in metastatic samples (median = 1.16, 95%CI = 1.09–1.60) compared to nonmetastatic (median = 1.02, 95%CI = 0.89–1.25, p < 0.001). In contrast, lower levels of miR‐495 were detected in specimens with distant metastasis (median = 0.05, 95%CI = 0.04–0.20) than nonmetastatic samples (median = 0.54, 95%CI = 0.47–0.58, p < 0.001). Estimated BCL2/miR‐497 ratio yielded a significant differential expression between the two cohort groups. Higher scores were observed in metastasis group (median = 1.39, 95%CI = 0.9–1.51) than nonmetastatic patients (median = 0.29, 95%CI = 0.19–0.39, p < 0.001). Receiver operating characteristic curve analysis showed BCL2/miR‐497 ratio score to have the highest predictive accuracy for metastasis at presentation. The area under the curve was 0.90 (95%CI = 0.839–0.964, p < 0.001) at cut‐off of >0.525, with high sensitivity 81.1% (95%CI = 68.6%–89.4%) and specificity 92.5% (95%CI = 82.1%–97.0%). Also, the ratio score was negatively correlated with disease‐free survival (r = −0.676, p < 0.001) and overall survival times (r = −0.650, p < 0.001). Kaplan–Meier curves showed lower survival rates in cohorts with high‐score compared to low‐score patients. Conclusion The BCL2/miR497 expression ratio is associated with poor CRC prognosis in terms of metastasis and short survival.


| INTRODUC TI ON
Colorectal cancer (CRC) substantially influences cancer-related death worldwide. 1 Despite the recent advances in CRC management, the associated morbidity and mortality remain high. 2 The last decade has witnessed a massive growth in our understanding of CRC genetic etiopathology. 3 Identifying and highlighting such genetic contribution may help better understand the molecular basis of cancer patient prognosis with potential future targeted therapy. 4 Noncoding RNAs have emerged as central genetic/epigenetic players in several cancers, including CRC. 5,6 The noncoding mi-croRNAs (miRNAs) class has been implicated in CRC tumorigenesis/progression and treatment. 7,8 Indeed, their dysregulation may contribute to poor CRC outcomes, including metastasis and short survival. [9][10][11] Our in silico analysis has revealed the microRNA-497 (miR-497) as one of the most iterated miRNAs in CRC, as will be detailed later on, and the B-cell lymphoma 2 (BCL2) gene as one of its target genes that proved previously to play a central role in the regulation of apoptosis and was implicated in colorectal carcinogenesis, progression, and treatment resistance. 12 Interestingly, previous studies have reported that miR-497 can suppress proliferation and induce apoptosis via the Bcl2-related molecular axis in several tissues and cancers, including neuronal cells, the "human umbilical vein endothelial cells," breast cancer, and multiple myeloma. [13][14][15] Zhu et al. found that miR-497 could decrease the resistance to cisplatin in "human lung cancer cell lines" by targeting BCL2. 16 Also, a recent study by Zheng et al. has proved that miR-497/BCl2 axis could suppress cisplatin resistance in CRC cells. 17 Nevertheless, no previous study demonstrated the impact of BCL2/miR-497 expression ratio score on CRC prognosis and outcome. In this sense, the authors were interested in exploring the association of the BCL2/miR-497 expression profile with the clinic-pathological characteristics and outcomes of CRC patients to help their prognostic stratification and future individualized therapeutic management.  (Table 1). Log fold change and adjusted p-values were identified for each experiment using the Database of Differentially Expressed miRNAs in Human Cancers (dbDEMC v3.0) (https://www.biosi no.org/dbDEM C/index). The average fold change of microRNAs was estimated, the direction of expression across all studies was identified, and the total number of comparisons with significant expression was calculated. MiR-497-5p was selected because it was the most frequently downregulated mi-croRNA across datasets. Functional enrichment analysis and gene targets of miR-497-5p identification in CRC KEGG pathway were identified using the DIANA-miRPath v.3.0 (http://www.micro rna.gr/ miRPa thv3); a "miRNA pathway analysis-based webserver". 18

Conclusion:
The BCL2/miR497 expression ratio is associated with poor CRC prognosis in terms of metastasis and short survival. Samples with incomplete clinical and/or follow-up data, history of receiving any therapeutic modality before resection, secondary CRC as well as samples without available paired noncancer tissue, tiny size tissue specimen available for molecular work, and those with low concentration or the extracted total RNA did not have enough quality to proceed in the downstream real-time qPCR steps, were excluded as showed in Figure 1. The ethical/legal guidelines adopted by the Declaration of Helsinki were followed. The local Medical Research Ethics Committee granted ethical approval for this study, and the patient consent was waived as the enrolled samples in this retrospective study were archived.

| Clinical assessment and follow-up
Patient information was obtained from the medical records. These included patients' demographic data, primary cancer site, pathology

| Propensity scores matching analysis
The survival outcomes of metastatic and nonmetastatic colon cancer patients and the impact of transcriptomic signature of selected markers were compared via a propensity score matching analysis. This analysis was performed to adjust confounder variables using the MatchIt R package. The following covariates were adjusted: age, sex, obesity, tumor site, histopathological diagnosis, pathological grade, tumor size, lymph node metastasis, and lymphovascular invasion. Multivariate logistic regression was applied to create a balancing score as a distant measure for each patient.
Next, metastatic and nonmetastatic cohorts were allocated using a one-to-one nearest neighbor algorithm without caliper adjustment to find pairs of patients that have the closest match in the two study groups. The quality of the matches in the two datasets (N = 53 patients in each group) were evaluated by estimating mean difference and average absolute standardized difference of covariates. 20

| BCL2/miR-497 expression analysis
Total tissue RNA, including miRNAs, was isolated from the CRC samples using miRNeasy FFPE Kit (217504, Qiagen, Hilden, Germany) following the manufacturer's instructions. To ensure DNA-free extracts, each sample was subjected to DNase I treatment (for 2 h at 37°C). RNA concentration/purity and integrity were tested by "NanoDrop ND-1000 spectrophotometer (NanoDrop Tech., Inc.)" and "agarose gel electrophoresis," respectively. Nontemplate and non-RT enzyme negative controls were run with each experiment to exclude amplicon contamination. Then the quantitative Real-Time PCR was carried out in duplicate in "StepOne Real-Time PCR System" (Applied Biosystems) as described in detail previously. 10,11 All the steps of the qRT-PCR were run following the "Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE)" guidelines. 21 The relative expression levels were calculated using the delta-delta CT (cyclic threshold) method. 22

| Statistical analysis
Relative expression levels of microRNA and genes were stratified by metastasis and plotted as box plots. Expression data were nonparametric; therefore, log transformation was employed. The   (Table S1).  Figure S1).

| Baseline characteristics of the study population
The study population included 69 males and 37 females, 53.8% over 55 years old, and 62.3% were obese. Detailed information about clinical characteristics of propensity-matched metastatic and nonmetastatic cohorts is described in      was negatively correlated with disease-free survival (r = −0.676, p < 0.001) and overall survival times (r = −0.650, p < 0.001). Patients with metastasis exhibited lower survival times ( Figure 5A-B). When patients were categorized according to the median ratio score into high-score and low-score groups, Kaplan-Meier curves showed lower survival rates in cohorts with high-score compared to lowscore patients ( Figure 5C-D). A prognostic nomogram to predict metastasis at presentation was generated using the ratio score with demographic characteristics of patients, which showed good agreement with the actual outcome ( Figure 6).

| DISCUSS ION
CRC's tendency to invasion/metastasis is one of the major factors leading to poor prognosis. 23 Identifying new genetic/epigenetic biomarkers associated with CRC metastasis and survival could help improve cancer management. 24 In this work, we explored the association of BCL2, miR-479, and BCL2/miR-479 ratio with poor prognosis in terms of metastasis and short survival in patients with CRC. We found that BCL2 was upregulated in metastatic samples compared to nonmetastatic ones. In contrast, miR-495-5p downregulation was found in specimens with distant metastasis than nonmetastatic samples. The estimated ratio score between BCL2 and miR-497-5p yielded a significant differential expression between the two cohort groups. Also, ROC curve analysis showed BCL2/miR-497 ratio score to have the highest predictive accuracy for metastasis at presentation. Furthermore, the ratio score showed a negative correlation with disease-free survival and overall survival, as well as included in a newly generated prognostic nomogram to predict metastasis, among other parameters.
These results are consistent with previous studies that reported the implication of BCL2 and miR-497 in cancer, including the CRC, 12,[25][26][27] and support that analyzing combined markers is better than an individual molecule in cancer diagnostics and/or prognostication. 28 The pro-survival BCL2 is one of the "anti-apoptotic BCL2 family proteins" implicated in promoting cancer cell proliferation, metastatic spread, and resistance to anticancer therapy. 29

| CON CLUS ION
In summary, our findings in this study suggest the essential role of the BCL2/miR-497 ratio as a prognostic ratio for CRC in terms of association with metastasis and poor survival indices. However, it is worth noting that our study lacks the functional studies that

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest.

AUTH O R CO NTR I B UTI O N
All authors contributed to data analysis, drafting, or revising the article, have agreed on the journal to which the article has been F I G U R E 6 Nomogram for predicting metastasis. (A) The nomogram was constructed based on demographic features of patients and ratio risk score. The outcome measured was metastasis. The logistic regression model was applied. (B) Example for using the nomogram. Assumed having a 45-year-old obese male patient whose tissue microRNA risk score was high at 0.45. Each variable will be scored on its points scale. The scores for all variables are added to obtain the total score "of 123," and a vertical line is drawn from the total points' row to estimate the probability of metastasis "23.7%" submitted, gave final approval of the version to be published, and agree to be accountable for all aspects of the work.