| Colorectal cancer(CRC)is one of the most common malignant cancers,it is of great significance to develop the signatures for early diagnosis or recurrence of CRCs that could be robustly applied to the clinic.Currently,transcriptome sequencing(RNA-seq)commonly require samples with high-quality RNA,which are named as research-oriented clinical samples,to obtain reliable research results,such as fresh-frozen samples with high tumor purity.For non-research-oriented clinical samples,due to the low-quality RNA caused by degradation or low abundance,the accuracy of gene expression measurements by RNA-seq is very challenging.Therefore,they are generally not recommended to be measured by RNA-seq.However,majority of samples preserved in hospitals or tissue banks worldwide are non-research-oriented samples with complete pathological information and follow-up data.Our previous researches demonstrated that the within-sample relative expression orderings(REO)of gene pairs were insensitive to samples with low-quality RNA.In this study,we measured the expression profiles of 45 stage II CRC non-research-oriented samples which were formalin-fixed paraffin-embedded(FFPE)with about 6 years of follow-up data,demonstrated the feasibility of developing REO-based signatures in non-research-oriented clinical samples,developed two signatures for early diagnosis and recurrence of CRCs,and analyzed transcriptomic characteristics about recurrence and metastasis.It may provide a novel approach for utilizing low-quality samples and help us to understand the mechanism of recurrence and metastasis of CRCs.1.Evaluation of REO-based cancer features in 45 non-research-oriented clinical samplesCompared with the gene expression profiles of 231 stage II fresh-frozen(FF)CRCs from TCGA database,we found that RNA in 45 non-research-oriented samples were severely degraded,and the accuracy of gene expression measurements was significantly challenged,even became zeros.Meanwhile,177,122 and 43,439,977 gene pairs which REO patterns were high stably reversed and significantly reversed in77 stage II CRCs versus 51 normal samples were selected,and defined as REO-based cancer features,respectively.The results of the retention of REO-based cancer features in 45 non-research-oriented samples showed that,compared to 95% of high stably reversed and approximately 60% of significantly reversed gene pairs maintained in FF CRCs,approximately 80% and 55% of gene pairs were maintained in 45 non-research-oriented clinical samples,respectively.The results indicated that majority of REO-based cancer features were insensitive to non-research-oriented clinical samples with low-quality RNA.2.Identification of an REO-based signature for early diagnosis of CRCThe timely and early detection of CRCs,which are usually diagnosed in the middle and advanced stage,is of great clinical significance for the treatment and recovery of patients.The early diagnostic signatures for CRCs,that could be robustly applied into clinic,are urgently needed.Although an REO-based signature with 7gene pairs for early diagnosis of CRCs has been reported,it only accurately identified31.11% of 45 non-research-oriented samples.Using 106 stage I FF CRCs and 51 FF normal samples from TCGA database,we identified a signature with 136 gene pairs for early diagnosis of CRCs.The results showed that the sensitivity and specificity of the signature for 720 CRC samples and 502 non-cancerous samples(including normal samples and inflammatory bowel disease samples)were 98.19% and 99.60%,respectively.Notably,95.56% of 45 non-research-oriented clinical samples were accurately identified.These results indicated that the early diagnostic signature of CRCs with relatively more gene pairs could be robustly applied to non-research-oriented clinical samples with low-quality RNA and it had great value for clinical translational applications.3.Development of an REO-based signature for predicting post-surgery relapse risk of stage II and III CRCAccording to the pathological factors,the accuracy for predicting post-surgery relapse risk of stage II and III CRCs is limited.For example,about 25-30% of stage II CRCs occurred post-surgery relapse.Based on the qualitative features of REO,a signature consisted of 4,500 gene pairs for predicting post-surgery relapse risk of stage II and III CRCs was developed.Compared to that only 25% of non-relapse samples measured in our lab were accurately identified by a previous signature with44 gene pairs,the signature with 4,500 gene pairs could accurately identify 71.42% of21 relapsed samples and 70.83% of 24 non-relapsed samples.And 45non-research-oriented samples could be divided into two groups which were significantly different in disease-free survival intervals.This result further confirmed that the signature with relatively more gene pairs could be robustly applied to non-research-oriented clinical samples with low-quality RNA.4.Analysis of transcriptomic characteristics about recurrence and liver metastasis of CRCDifferentially expressed genes(DEGs)were detected between stage I and IV CRCs,between stage II-III CRCs with high and low risk of post-surgery relapse,and between the relapse and non-relapse samples measured in our lab,respectively.Then,gene function and regulatory relationships of the common DEGs from the above three DEGs lists were analyzed.The results showed that these genes were significantly enriched in immune-related functional pathways,indicated that post-surgery relapse of stage II and III CRCs was closely related to immune dysfunction.Additionally,based on the expression profile data of colorectal tissues(including non-relapse CRCs and normal samples)and liver tissues(including cancer and normal samples),we analyzed the transcriptomic characteristics of colorectal liver metastases.A total of 12 genes that were expressed in liver tissues but unexpressed in colorectal tissues were detected and denoted as "liver-specific genes".We further found that two "liver-specific genes"(ANGPTL3 and CFHR5),were unexpressed in microdissected primary colorectal tumors without metastasis but expressed in both microdissected liver metastases and corresponding primary colorectal tumors(Fisher's exact test,p <0.05).Genes co-expressed with ANGPTL3 and CFHR5 were significantly enriched in metabolism pathways characterizing liver tissues,including “Starch and sucrose metabolism” and “Drug metabolism-cytochrome P450”.This result indicated that,for primary CRC with liver metastasis,both the liver metastases and corresponding primary colorectal tumors may express some liver-specific genes which may help tumor cells adapt the liver micro-environment.And it is important to understand the underlying mechanism of colorectal liver metastasis. |
PDF Full Text Request