P1177 - Bioinformatic Analysis of Prognostic Value of LOXL1-4 in Colorectal Cancer

Chenyu Sun, MD, MSc¹, John Pocholo W. Tuason, MD¹, Na Hyun Kim, MD¹, Chandur Bhan, MD¹, Nikitha Manem, MD¹, Sudha Misra, MD¹, Mohammed R. Gandam, MBBS, MD¹, Yue Chen, ², Shaodi Ma, ^3
1AMITA Health Saint Joseph Hospital, Chicago, IL; ²School of the First Clinical Medicine, Anhui Medical University, Hefei, Anhui, China; ³School of Public Health, Anhui Medical University, Hefei, Anhui, China

Introduction: Colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancer-related death. Previous studies found increased expression of some members of Lysyl oxidase–like (LOXL) genes in colorectal cancer. Therefore, this study was conducted.

Methods: The expression levels of LOXLs in different cancers were first explored by using Oncomine database. Analysis of different expression levels between tumor and normal tissues was conducted via data obtained from The Cancer Genome Atlas (TCGA) dataset. Log2 transformed FPKM (fragments per kilobase exon-model per million reads mapped) were used. The expression analysis was performed by using the Wilcoxon rank-sum test and visualized by ‘ggplot2’ package of R software. To further explore their expression levels of different histologic types, the difference in transcriptional levels was assessed via UALCAN online tool. The best discriminate cut-off point between the high and low expression groups was assessed by the ROC curve and area under the curve (AUC) values for LOXLs. Log2 transformed FPKM was used. R software by using pROC and ggplot2 packages were used.p < 0.05 was considered as statistically significant.

Results: The expression levels of 4 LOXLs family members were analyzed in 20 different types of human cancer and compared with normal individuals with the Oncomine database. The mRNA expression levels of LOXL1 and LOXL 2 were significantly elevated in the colorectal cancer tissues. Based on data obtained from TCGA and analysis result via ULCAN, the expression level of LOXL1 and LOXL 2 was significantly higher in colorectal cancer tissues than normal tissues, while LOXL4 expressed significantly lower in the colorectal cancer tissues. However, subgroup analysis indicated that in rectal adenocarcinoma, there is no difference in LOXL1 expression levels between tumor and normal tissues. The expression levels of LOXL1, LOXL 2 and LOXL4 were significantly increased in mucinous adenocarcinoma compared with adenocarcinoma. Further analysis of ROC curve shown that AUC of LOXL 2 was 0.908 (0.877-0.939) for CRC, with a positive prognostic value ((PPV) of 0.992.

Discussion: LOXL1 and LOXL2 are overexpressing in CRC. LOXL1, LOLX2, and LOXL4 may have the potential for differentiating mucinous adenocarcinoma from adenocarcinoma. LOXL2 can be a good candidate for diagnosing CRC.


Figure: A) The expression levels of LOXL1-4 in 20 different types of human cancer on the Oncomine database; B) The expression levels of LOXL1-4 in the colorectal cancer tissues and corresponding normal tissues; C) The expression levels of LOXL1-4 in the colon adenocarcinoma tissues and corresponding normal tissues based on analysis via ULCAN; D) Different expression levels of LOXL1-4 in the mucinous adenocarcinoma versus adenocarcinoma tissues based on analysis via ULCAN; E) ROC curve of LOXL2

	Comparison	P value
LOXL1	Adenocarcinoma-vs-Mucinous-adenocarcinoma	4.30E-03
LOXL2	Adenocarcinoma-vs-Mucinous-adenocarcinoma	6.80E-03
LOXL3	Adenocarcinoma-vs-Mucinous-adenocarcinoma	5.48E-02
LOXL4	Adenocarcinoma-vs-Mucinous-adenocarcinoma	2.13E-02