CN116013525A

CN116013525A - Colorectal cancer prognosis model constructed based on iron death characteristics and construction method thereof

Info

Publication number: CN116013525A
Application number: CN202310016828.5A
Authority: CN
Inventors: 邓江; 张艳宇; 赵宁; 吕丽萍; 马平; 张阳阳
Original assignee: Academy of Military Medical Sciences AMMS of PLA
Current assignee: Academy of Military Medical Sciences AMMS of PLA
Priority date: 2023-01-06
Filing date: 2023-01-06
Publication date: 2023-04-25

Abstract

The invention discloses a colorectal cancer prognosis model constructed based on iron death characteristics and a construction method thereof, and the colorectal cancer prognosis model constructed based on the iron death characteristics is characterized in that the model comprises a scoring formula for prognosis evaluation of colon cancer. When the model score of the tester is greater than-0.281, the tester is divided into a high risk group, and the prognosis is poor; when the model score of the tester is less than-0.281, the tester is classified into a low risk group, and the prognosis is good. The invention proposes to replace the expression value of the gene with the gene pair for constructing the model. The gene pair is selected to form a gene pair by two genes, and when the former gene is larger than the latter gene, the gene pair has a value of 1, and when the former gene is smaller than the latter gene, the gene pair has a value of 0. Thus, the construction of the model is independent of specific gene expression values, but only depends on the ordering of the relative expression amounts of genes, thereby overcoming the defects.

Description

Colorectal cancer prognosis model constructed based on iron death characteristics and construction method thereof

Technical Field

The invention relates to the technical field of biology, in particular to a colorectal cancer prognosis model constructed based on iron death characteristics and a construction method thereof.

Background

Colorectal cancer (colorectal cancer, CRC) has seen an increasing trend in recent years in our country in morbidity and mortality. The flow regulation data show that the incidence rate and the death rate of colorectal cancer in China are respectively 3 rd and 5 th in all malignant tumors, wherein the new cases reach 37.6 ten thousand and the death cases reach 19.1 ten thousand. At present, clinical staging of colorectal cancer mainly depends on TMN staging of tumors, and diagnosis and treatment schemes developed according to the staging and treatment schemes comprise operations, radiotherapy and chemotherapy and the like on patients do not obviously change prognosis survival time of the patients. In recent years, along with the continuous popularization of high-throughput sequencing technology, high-throughput sequencing is performed on clinical samples of tumor patients, so that more personalized diagnosis and treatment schemes are possible to be prepared. This relies on a more accurate determination of the prognosis of the patient to make accurate diagnosis and treatment.

Colorectal cancer has a more convenient way to obtain its tumor tissue samples for reasons of the patient's frequent enteroscopy than other tumors, thereby facilitating the development of high throughput sequencing assays. However, the traditional method for discriminating the gene construction model has obvious defects: that is, the gene expression value obtained by each detection is a relative expression value due to the reasons of a sequencing platform, batch difference and the like, and the expression of the gene expression value fluctuates up and down, so that the prognosis score calculated directly according to the gene expression has the defects of insufficient adaptability and difficult wide application in popularization and application.

Disclosure of Invention

The invention aims to provide a colorectal cancer prognosis model constructed based on iron death characteristics and a construction method thereof.

The invention provides a device for prognosis evaluation of colon cancer, which comprises a detection device and a computer-readable carrier, wherein the detection device comprises a substance for detecting the expression quantity of genes to be detected; the computer readable carrier has recorded thereon a scoring formula for prognosis evaluation of colon cancer:

Riskscore＝exp(B)1)

wherein "|" indicates that the gene pair is selected to constitute one gene pair with two genes, and the value of the gene pair is 1 when the previous gene is greater than the expression value of the next gene, and is 0 when the previous gene is less than the expression value of the next gene.

The substance for detecting the expression level of the gene to be detected may be a device for detecting the expression level of the gene to be detected in the prior art, or may be a primer or a primer composition for detecting the expression level of the gene to be detected.

When the model score of the tester is greater than-0.281, the tester is divided into a high risk group, and the prognosis is poor; when the model score of the tester is less than-0.281, the tester is classified into a low risk group, and the prognosis is good.

The computer readable carrier can receive the expression quantity of the gene to be detected by the detection device and automatically calculate a score according to a scoring formula for prognosis evaluation of colon cancer; this process may also be performed manually by an operator.

The invention also provides a colorectal cancer prognosis model constructed based on the iron death characteristics, wherein the model comprises a scoring formula for prognosis evaluation of colon cancer, and the scoring formula is as follows:

Riskscore＝exp(B)1)

The invention also provides a method for constructing the colorectal cancer prognosis model based on the iron death characteristics, which comprises the following steps:

1) Identifying a tumor differential gene of the cancer tissue and the paracancerous tissue;

2) Identifying differentially expressed iron death-related genes;

3) Screening genes existing in the step 1 and the step 2 simultaneously;

4) Performing gene knockout based on the data set, and combining the remaining genes two by two to obtain a plurality of gene pairs;

5) Single factor Cox regression screened pairs of genes, the threshold value for screening was set as: p <0.05;

6) Screening the gene pairs in the step 5) by using a Lasso model method, and then constructing a model by using a multifactor Cox regression method to obtain the colorectal cancer prognosis model constructed based on the iron death characteristics.

The present invention also provides a biomarker composition for prognosis evaluation of colon cancer, characterized in that the biomarker composition comprises at least one gene expressed in CD44, trim 3, ZEB1, LONP1, TFR2, MT1G, CA, SQLE, GPX2, VEGFA, STMN1, CXCL2, TAZ, SLC3A2, DRD4, CDKN1A, HILPDA, FANCD2, BID, PLin4, DRD5, CRYAB, HSD17B11, FADS2, VLDLR, SCD, DUSP1, TXNIP, PRKAA2, DDIT4, CDO1, NQO1, PSAT1, RRM2, CAV1, HELLS, ASNS, CDKN2A, SESN2, GLS2, GDF15, MIOX, AURKA, ACSF, ENPP2, SLC7a11, SLC2A1, HMOX1, ATP6V1G2, RPL8, ABCC1, hand 1, EPAS1, 2, hadx 6, andx B, TSC, ZFP 69D 3, SLC 4, CDO1, npc 3, GPT2, SLC3, GPT2, tsc 3, tsc 2.

Wherein the biomarker composition is the expression level of genes VEGFA, AURKA, CXCL, ABCC1, CDKN1A, SCD, BID, CAV, CRYAB, LPCAT3, HSD17B11, DDIT4, FADS2, CDKN2A, DDIT4, RRM2, EPAS1, HELLS, CDKN2A, ASNS, AURKA, CDKN2A, ENPP2, GDF15, SLC7A5, HMOX1 and TSC22D 3.

The use of a biomarker as described above in the manufacture of a product for assessing prognosis of colon cancer.

Wherein the product is a kit or a detection device.

The invention also provides a product for assessing prognosis of colon cancer, which is characterized in that the product comprises the biomarker.

The present model proposes replacing the expression value of the gene with the gene pair to construct the model. The gene pair is selected to form a gene pair by two genes, and when the former gene is larger than the latter gene, the gene pair has a value of 1, and when the former gene is smaller than the latter gene, the gene pair has a value of 0. Thus, the construction of the model is independent of specific gene expression values, but only depends on the ordering of the relative expression amounts of genes, thereby overcoming the defects.

The method creates tracking group data from The Cancer Genome Atlas (TCGA) database, while testing group data is cross-validated from multiple data sets. The method constructs gene pairs with differential genes of iron death. Iron death is a newly discovered programmed cell death mode depending on iron elements in recent years, and the core mechanism is accumulation of active oxygen of iron elements and lipid, so that cells are subjected to peroxidation and death. Several colorectal cancer studies have found that iron death activation can inhibit the development of tumors, possibly a key element of body tumor immunity. The prognosis model constructed based on the iron death gene can accurately predict the prognosis situation of a tester.

Drawings

FIG. 1 shows 68 genes of a gene pair to be constructed obtained by crossing T-DEGs and FRGs selected in the early stage.

FIG. 2 is a graph of the determination of the optimal diagnostic point for the ROC curve as a prognosis over a year based on Lasso Cox results as-0.281.

FIG. 3 shows the results of TCGA queue model verification: the Survival analysis adopts Kaplan-Meier test, the Survival time adopts comprehensive Survival time (OS), disease-related Survival time (Disease-specific Survival, DSS), disease-free Interval (DFI), and Progression-free Interval (PFI), and P <0.05 (Figure 3).

FIG. 4 shows the Area Under the ROC Curve (AUC) for 1, 2, and 3 years for the time ROC prognostic analysis of scores as 0.771,0.791,0.750 (Figure 4), respectively.

FIG. 5 shows the single-factor and multi-factor Cox independent prognosis analysis of Riskscore with the basal information (age, sex) and tumor Stage (Stage, T Stage, M Stage, N Stage) of the tester, and the age and Riskscore were two independent key factors affecting prognosis (Figure 5).

Fig. 6 is GSE17538 queue model validation results: survival analysis was performed using Kaplan-Meier test, P _OS ＝0.003、P _DSS ＝0.006、P _DFI ＝0.003(Figure 6)。

Fig. 7 shows that the AUC areas for 1, 2, and 3 years, respectively, were 0.655,0.665,0.617 for time ROC prognostic analysis of scores.

Fig. 8 is GSE17538 queue model validation results: survival analysis was performed using Kaplan-Meier test, P _OS ＝0.002、P _DFI <0.001。

Fig. 9 shows that AUC areas for 1, 2, and 3 years, respectively, were 0.625,0.586,0.594 for time ROC prognostic assays of scores.

Detailed Description

The following detailed description of the invention is provided in connection with the accompanying drawings that are presented to illustrate the invention and not to limit the scope thereof. The examples provided below are intended as guidelines for further modifications by one of ordinary skill in the art and are not to be construed as limiting the invention in any way.

The experimental methods in the following examples, unless otherwise specified, are conventional methods, and are carried out according to techniques or conditions described in the literature in the field or according to the product specifications. Materials, reagents and the like used in the examples described below are commercially available unless otherwise specified.

The model of the present invention proposes to replace the expression value of the gene with the gene pair for constructing the model. The gene pair is selected to form a gene pair by two genes, and when the former gene is larger than the latter gene, the gene pair has a value of 1, and when the former gene is smaller than the latter gene, the gene pair has a value of 0. Thus, the construction of the model is independent of specific gene expression values, but only depends on the ordering of the relative expression amounts of genes, thereby overcoming the defects.

The set up of tracking group data in the present invention is derived from the The Cancer Genome Atlas (TCGA) database, while the testing group data is cross-validated by multiple data sets. The method constructs gene pairs with differential genes of iron death. Iron death is a newly discovered programmed cell death mode depending on iron elements in recent years, and the core mechanism is accumulation of active oxygen of iron elements and lipid, so that cells are subjected to peroxidation and death. Several colorectal cancer studies have found that iron death activation can inhibit the development of tumors, possibly a key element of body tumor immunity. The prognosis model constructed based on the iron death gene can accurately predict the prognosis situation of a tester.

Example 1 construction of a model

1. Identification of tumor differential Gene (tumor associated differential expressed genes, T-DEGs) in cancer tissue and paracancerous tissue

Downloading cancer and paracancerous gene sequencing data of colorectal cancer testers from a TCGA database (https:// portal.gdc.cancer/rev/repositisource), obtaining 488 cancer sample data altogether, 42 paracancerous sample data altogether, performing differential analysis by wilcoxon rank sum test through differential analysis, setting the threshold value as P <0.05, |logFC|gtoreq 1, and obtaining 7878 genes altogether;

2. identification of differentially expressed iron death-related genes

Iron death-related genes (ferroptosis associated genes, FRGs) were downloaded from the FerrDb website (http:// www.zhounan.org/ferdbj /), 267 in total;

3. determination of model candidate genes

Intersection of T-DEGs and FRGs screened in the earlier stage is taken, as shown in figure 1, genes of the gene pair to be constructed are obtained, 68 total, CD44, TRIB3, ZEB1, LONP1, TFR2, MT1G, CA, SQLE, GPX2, VEGFA, STMN1, CXCL2, TAZ, SLC3A2, DRD4, CDKN1A, HILPDA, FANCD2, BID, PLIN4, DRD5, CRYAB, HSD17B11, FADS2, VLDLR, SCD, DUSP1, TXNIP, PRKAA2, DDIT4, CDO1, NQO1, PSAT1, RRM2, CAV1, HELLS, ASNS, CDKN2A, SESN2, GLS2, GDF15, MIOX, AURKA, ACSF2, ENPP2, SLC7A11, SLC2A1, HMOX1, ATP6V1G2, RPL8, ABCC1, HAMP, EPAS1, PROM2, PRDX6, ANGPTL7, ZFP69B, TSC D3, SLC1A5, AKR1C2, SLC2A8, 4, CHAC1, SLC 12, SLC7, LPT 3, and GPT 3.

4. Construction of Gene pairs

Model candidate genes are extracted from a training group TCGA queue and 2 verification group queues (the platforms of the 2 verification group queues are Affymetrix Human Genome U Plus 2.0Array, and the platforms of the GSE17538 and the GSE39582 are Affymetrix Human Genome U Plus), and if a certain gene is not detected in the data of the queue, the gene is rejected. Meanwhile, if the expression of the gene pair is lower than 0.2 or higher than 0.8, it is also knocked out. The remaining pairwise combinations were used to construct gene pairs (any two gene pairs, then the next round of screening).

4. Single factor Cox regression screening

Screening the gene pairs constructed in the step three by using single factor Cox regression, wherein the screening threshold is set as follows: p <0.05, a total of 14 gene pairs were finally obtained, as shown in Table 1.

TABLE 1

In the above table, HR is the risk ratio (second column), HR.95L and HR.95H are the range of trusted intervals of HR (calculated from the samples), and the total HR is represented by a 95% confidence interval for his range (third and fourth columns), when the P value in the fifth column is less than 0.05 indicating passing the cox test.

5. Lasso Cox regression screening gene pair and construction model

Screening gene pairs by using a Lasso model method, constructing a model by using a multi-factor Cox regression method, naming the model as a preliminary model, and finally constructing a scoring model as follows:

Riskscore＝exp(B)1)

B＝(VEGFA|AURKA×0.396823529)+(CRYAB|LPCAT3×0.442788086)+

2)

TABLE 2

In the model set, two genes before and after "|" constitute a gene pair, and when the former gene > the expression value of the latter gene, the gene pair has a value of 1, and when the former gene < the expression value of the latter gene, the gene pair has a value of 0.

Lasso Cox results as shown in fig. 2, the optimal diagnostic point for ROC curve was determined to be-0.281 based on Lasso Cox results over a one year prognosis. That is, when the model score of the tester is greater than-0.281, it is classified into a high risk group, and the prognosis is poor; when the model score of the tester is less than-0.281, the tester is classified into a low risk group, and the prognosis is good.

Based on the above model, a device for prognosis evaluation of colon cancer can be prepared, which comprises a detection device comprising a substance for detecting the expression amount of a gene to be tested, and a computer-readable carrier; the computer readable carrier has recorded thereon a scoring formula for prognosis evaluation of colon cancer:

Riskscore＝exp(B)1)

Example 2 verification of model

And (4) classifying all testers of the queue into a high risk group and a low risk group according to the model scores by using a cutoff value of-0.281, and performing survival analysis on the tester queue. For better evaluation of the evaluation model on the prognosis of the testers, the influence of confounding factors is removed, and all the testers with observation time less than 30 days are removed from all the testers. The verification result is as follows:

1. the Riskscore was calculated for 434 colorectal cancer testers data from the tracking group (TCGA (The cancer genome atlas, cancer genomic profile)) according to the scoring model in example 1 of the present invention.

Meanwhile, 434 colorectal cancer testers were subjected to Survival analysis by Kaplan-Meier test, and the Survival time was respectively calculated by using comprehensive Survival time (OS), disease-related Survival time (Disease-specific Survival, DSS), disease-free Survival Interval (DFI), progression-free Survival Interval (PFI), and P <0.05, and the result is shown in fig. 3.

The scores were subjected to a time ROC prognostic analysis with Area Under the ROC Curve (AUC) of 1, 2 and 3 years being 0.771,0.791,0.750, respectively, as shown in fig. 4.

Independent prognosis analysis of Riskscore with basal information of the tester (age, sex) and tumor Stage (Stage, T Stage, M Stage, N Stage) revealed that age and Riskscore are two independent key factors affecting prognosis, as shown in fig. 5.

As can be seen from fig. 3, the OS, DSS, PFI are used as indicators, and there is a significant difference (P < 0.05) between the high risk group and the low risk group; as can be seen from FIG. 4, the area under the ROC curve is more than 0.75, and the survival status of the testee is better judged; as can be seen from FIG. 5, riskscore and age are the best two indicators of prognosis for the predicted test subjects

2. For a pair ofTesting group-1 group: the queue data was derived from 229 colorectal cancer tester data of GSE17538 queue (https:// www.ncbi.nlm.nih.gov/GEO/query/acc.cgiac=gse 17538) of GEO database: after the testers of the queue are calculated according to a Riskscore formula and are grouped according to a cut-off value, all the testers are divided into a high risk group and a low risk group, and the survival conditions of the high risk group and the low risk group are checked according to the real survival data and the clinical data of all the testers, and the results are shown in figures 6-7; wherein, the survival analysis adopts Kaplan-Meier test, the survival condition and time of high and low risk group testers have significant difference, and P is less than 0.05 (P _OS ＝0.003、P _DSS ＝0.006、P _DFI =0.003), as shown in fig. 6; time ROC prognostic analysis was performed on the scores with AUC areas of 0.655,0.665,0.617 for 1, 2 and 3 years, respectively, as shown in fig. 7.

3. For Testing group-2: the queue data is derived from GSE39582 queue (https:// www.ncbi.nlm.nih.gov/GEO/query/acc. Cgiac=GSE 39582) 550 colorectal cancer tester data of the GEO database: after the testers of the queue are calculated according to a Riskscore formula and are grouped according to a cut-off value, all the testers are divided into a high risk group and a low risk group, the survival conditions of the high risk group and the low risk group are checked according to the real survival data and the clinical data of all the testers, and the same experiment is carried out according to the method in the step 1, and the result is shown in figures 8-9; wherein, the survival analysis adopts Kaplan-Meier test, and the survival conditions and time of high and low risk group testers have significant difference, and P is less than 0.05. Wherein, using OS as index, P _OS =0.002, P using DFI as index _DFI <0.001 as shown in fig. 8; time ROC prognostic analysis was performed on scores with AUC areas of 1, 2, and 3 years as follows: 0.625,0.586,0.594, as shown in fig. 9.

The present invention is described in detail above. It will be apparent to those skilled in the art that the present invention can be practiced in a wide range of equivalent parameters, concentrations, and conditions without departing from the spirit and scope of the invention and without undue experimentation. While the invention has been described with respect to specific embodiments, it will be appreciated that the invention may be further modified. In general, this application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. The application of some of the basic features may be done in accordance with the scope of the claims that follow.

Claims

1. A device for prognosis evaluation of colon cancer, the device comprising a detection device comprising a substance for detecting the expression level of a gene to be tested, and a computer-readable carrier; the computer readable carrier has recorded thereon a scoring formula for prognosis evaluation of colon cancer:

Riskscore＝exp(B)1)

2. A colorectal cancer prognosis model constructed based on iron death characteristics, characterized in that the model comprises a scoring formula for prognosis evaluation of colon cancer, the scoring formula being:

Riskscore＝exp(B)1)

3. The method for constructing the colorectal cancer prognosis model based on the iron death characteristics is characterized by comprising the following steps:

2) Identifying differentially expressed iron death-related genes;

3) Screening genes existing in the step 1 and the step 2 simultaneously;

4. A biomarker composition for prognosis evaluation of colon cancer, characterized in that the biomarker composition comprises at least one gene expressed in the genes CD44, trie 3, ZEB1, LONP1, TFR2, MT1G, CA9, SQLE, GPX2, VEGFA, STMN1, CXCL2, TAZ, SLC3A2, DRD4, CDKN1A, HILPDA, FANCD2, BID, PLin4, DRD5, CRYAB, HSD17B11, FADS2, VLDLR, SCD, DUSP1, TXNIP, PRKAA2, DDIT4, CDO1, NQO1, PSAT1, RRM2, CAV1, HELLS, ASNS, CDKN2A, SESN2, GLS2, GDF15, MIOX, AURKA, ACSF2, ENPP2, SLC7a11, SLC2A1, HMOX1, ATP6V1G2, RPL8, ABCC1, HAMP, EPAS1, PROM2, PRDX6, anx 7, ZFP B, TSC D3, SLC 5D 1, NQO1, SLC2, SLC 3C 2, tsc 3, tsc 8, tsc2, tsc 3 and tsc 2.

5. The biomarker composition according to claim 4, wherein the biomarker composition is the expression level of genes VEGFA, AURKA, CXCL, ABCC1, CDKN1A, SCD, BID, CAV, CRYAB, LPCAT3, HSD17B11, DDIT4, FADS2, CDKN2A, DDIT4, RRM2, EPAS1, HELLS, CDKN2A, ASNS, AURKA, CDKN2A, ENPP2, GDF15, SLC7A5, HMOX1 and TSC22D 3.

6. Use of a biomarker according to any of claims 4 to 5, in the manufacture of a product for assessing prognosis of colon cancer.

7. The use according to claim 6, wherein the product is a kit or a detection device.

8. A product for assessing a prognosis of colon cancer, said product comprising a biomarker according to any of claims 4 to 5.