CN108363907A - A kind of adenocarcinoma of lung personalization prognostic evaluation methods based on multi-gene expression characteristic spectrum - Google Patents

Abstract

The invention discloses a method for evaluating the individualized prognosis of lung adenocarcinoma based on a multi-gene expression profile, which comprises the following steps: obtaining a list of prognostic risk genes and gene weights for lung adenocarcinoma; using tumor tissue transcriptome and survival data of patients with lung adenocarcinoma Construct a prognosis assessment model; calculate the patient's risk score based on the gene expression profile of the tumor tissue of lung adenocarcinoma patients; calculate the patient's annual survival probability based on the patient's risk score. The annual survival probability of lung adenocarcinoma patients obtained by the method of the present invention is highly consistent with the actual annual survival rate (linear correlation R ² =0.994, P value =2.86E-43). It is confirmed that the method has high prediction accuracy, which is highly consistent with the actual survival status. At the same time, for each tumor patient, the present invention can provide the specific survival probability curve of the patient.

Description

A personalized prognostic assessment method for lung adenocarcinoma based on multiple gene expression profiles

technical field

The invention belongs to the fields of biotechnology and medicine, and in particular relates to a personalized prognosis evaluation method for lung adenocarcinoma based on multi-gene expression profiles.

Background technique

Lung adenocarcinoma accounts for about 40% of all lung cancer patients. Lung cancer is the most common cancer worldwide and the leading cause of cancer death in men. The incidence of lung cancer in the female population is second only to breast cancer. Global Burden of Disease (GBD) data shows that in 2016, the number of people suffering from trachea, bronchus or lung cancer in the world exceeded 2.8 million, and the number of patients in China was as high as 1 million. In 2016, the number of deaths from the above-mentioned cancers worldwide was 1.7 million, accounting for 3.12% of the total deaths. In 2016, the number of patients who died in China was 590,000, accounting for 6.11% of the total number of deaths. Statistics show that from 1990 to 2016, the global prevalence and mortality of trachea, bronchus and lung cancer continued to increase, and the prevalence and mortality in China also continued to increase, and the growth trend was relatively consistent with the global growth trend.

The current international tumor staging method is the TNM staging system, which is a malignant tumor classification method proposed by the American Joint Committee on Cancer (AJCC). The National Cancer Institute (NCI) describes the TNM staging as follows: T refers to the size and extent of the main tumor, which is usually called the primary tumor. N refers to the number of nearby lymph nodes with cancer. M refers to whether the cancer has metastasized, that is, spread from the primary tumor to other parts of the body. According to the above indicators, malignant tumors can be roughly divided into stage I, stage II, stage III and stage IV, and the higher the stage, the higher the degree of malignancy of the tumor. The TNM staging system is helpful to the treatment and prognosis evaluation of tumor patients. However, due to the differences in tumorigenesis mechanisms and in vivo microenvironments in different individuals, the survival time of different patients varies greatly, and the TNM staging system cannot well reflect the prognosis of patients. Studies have found that some patients diagnosed with stage I may have a short survival period (1-2 years), while some patients diagnosed with stage IV may have a longer survival period (5 years and above). Therefore, the TNM staging system may be more inclined to describe the average level of a cancer patient population, which is less applicable to individualized diagnosis and treatment. On the other hand, for patients diagnosed as advanced (stage III, stage IV), it will cause certain difficulties for patients and medical staff to choose a treatment plan, resulting in many tumor patients who could have survived for a long time due to excessive medical treatment or medical malpractice. and other patients who should have received appropriate treatment to prolong their survival also led to early death of tumor patients due to abandonment of treatment or improper treatment.

At present, some reports suggest that the use of gene expression profiles can be used to assess the prognosis of cancer patients. However, most of the reports only use single or several genes, can only classify a group, and can only qualitatively divide the individual survival period (such as two indicators of good prognosis and poor prognosis). Therefore, it is necessary to establish a more refined personalized tumor prognosis assessment model to evaluate the patient's survival time and choose an appropriate treatment plan.

Contents of the invention

In view of this, the present invention provides a personalized prognosis assessment method for lung adenocarcinoma based on multi-gene expression profiles, which can accurately predict the annual survival probability of patients.

In order to solve the above technical problems, the present invention discloses a personalized prognosis evaluation method for lung adenocarcinoma based on multi-gene expression profiles,

Include the following steps:

Step 1. Obtain the list of prognostic risk genes and gene weights for lung adenocarcinoma;

Step 2, using the tumor tissue transcriptome and survival data of patients with lung adenocarcinoma to construct a prognosis assessment model;

Step 3, calculating the risk score of the patient according to the gene expression profile of the tumor tissue of the lung adenocarcinoma patient;

Step 4. Calculate the patient's annual survival probability based on the patient's risk score.

Optionally, the acquisition of the lung adenocarcinoma prognosis risk gene list and gene weights in the step 1 is specifically implemented according to the following steps:

Step 1.1. Download the tumor tissue and paracancerous tissue transcriptome data and clinical data of lung adenocarcinoma patients from the Genomic Data Commons Data Portal database, obtain the FPKM value of the tumor tissue gene expression profile of lung adenocarcinoma patients, and perform logarithmic transformation;

Step 1.2, set the total number of samples as m, divide all samples into three groups according to the tertiles of their gene expression values, where the gene expression value is the FPKM value obtained in step 1.1, denoted by V, for the ith The gene is denoted as V _i , and the Cox proportional hazard model is used to calculate the survival risk of the third group compared with the first group, and the hazard ratio HR and P value of each gene are obtained; P value <0.05 is defined as significant, and the screening is significant The survival risk genes of , denoted as n ₁ ; in addition, calculate the correlation between each gene and the patient's survival days, and obtain the correlation coefficient r and P value of each gene; define that P value <0.05 is significant, and the screening is significant The survival-related genes of , denoted as n ₂ ; the intersection of survival risk genes and survival-related genes is defined as the prognosis risk gene, denoted as n, then:

n=n ₁ ∩n ₂ (1)

Step 1.3. Calculate the weight W _i of each gene according to the risk ratio of the i-th gene, and the calculation formula is:

Where i represents the gene number, and HR _i represents the hazard ratio of the i-th gene;

In this way, the weight of each gene is calculated; the final lung adenocarcinoma prognosis risk gene list and gene weight are obtained.

Optionally, the list of prognostic risk genes and gene weights for lung adenocarcinoma are specifically shown in the following table:

Optionally, in step 2, constructing a prognosis assessment model using tumor tissue transcriptome and survival data of patients with lung adenocarcinoma is specifically implemented according to the following steps:

Step 2.1. Define the gene expression value V, and calculate the risk score S _i of the i-th patient according to the expression value and weight of the i-th gene in the j-th sample; the calculation formula is:

Where i represents the gene number, V _ij represents the expression value of the i-th gene in the j-th sample;

Step 2.2, sort all lung adenocarcinoma patient samples according to the risk score from low to high, and use the sliding window model to calculate the average risk score for every 50 samples The calculation formula is:

Where j+49 represents the last 50 samples counted from sample j;

Step 2.3. Use the Weibull distribution to perform curve fitting on the survival data of 50 samples. The probability density function of the Weibull distribution is:

Where k>0 is a shape parameter, and λ>0 is a scale parameter of the distribution.

Step 2.4, calculate for every 50 samples The corresponding k _j and λ _j ; according to experience, k _j is a relatively fixed value, and the mean value is:

Among them, k _j is the shape parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample;

The variation range of the proportional parameter λ _j is relatively large, and the definition of λ _j and The functional relationship is:

Among them, _λj represents the proportional parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample;

Where e is the base of the natural logarithm, α and β are the parameters of the function, and the logarithm of the above formula is obtained:

where logλ _j and is a linear relationship, solved by linear fitting;

According to the average risk score With the fitting curve of Weibull distribution parameter λ _j , the function relationship obtained is:

Will Substituting this function to get the predicted λ _j ′, λ _j ′ is the expected distribution parameter calculated by this function, and calculating the correlation between λ _j and λ _j ′, the correlation coefficient R ² = 0.943, P value = 2.96E-97 .

Optionally, the calculation of the risk score of the patient according to the gene expression profile of the tumor tissue of the lung adenocarcinoma patient in the step 3 is specifically implemented according to the following steps:

Obtain the FPKM value of the gene expression profile of the tumor tissue of patients with lung adenocarcinoma, recorded as: V _i , where i is the gene number; the weight corresponding to the i-th gene is recorded as: W _i , where i is the gene number; patient risk score Recorded as: S; the calculation formula is:

Where i is the gene number and n is the number of genes.

Optionally, the calculation of the patient's annual survival probability based on the patient's risk score in step 4 is specifically implemented in the following steps: the patient's risk score S is brought into the cumulative distribution function of the Weibull distribution to obtain the patient's survival probability function as :

Where t is time, α, β, S, are fixed parameters.

Compared with prior art, the present invention can obtain and comprise following technical effect:

1) Continuous: It can predict the survival probability of tumor patients for continuous time. For example, the survival probability of the patient per month, the survival probability of the patient per year, and the like can be given. However, the current clinical classification method can only give a qualitative judgment.

2) More accurate: Compared with the traditional TNM staging, the personalized prognostic assessment method of lung adenocarcinoma based on multi-gene expression profiles can more accurately reflect the survival status of patients.

3) Personalization: For each tumor patient, the present invention can provide the patient's unique survival probability curve, which is not available in general tumor prognosis assessment models.

Of course, implementing any product of the present invention does not necessarily need to achieve all the technical effects described above at the same time.

Description of drawings

The accompanying drawings described here are used to provide a further understanding of the present invention, and constitute a part of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention, and do not constitute improper limitations to the present invention. In the attached picture:

Fig. 1 is that the average annual survival probability predicted by the present invention compares with the actual annual survival probability;

Fig. 2 is the correlation between TNM tumor staging of the present invention and patient survival time;

Fig. 3 is the fitting curve of average risk score of the present invention and Weibull distribution parameter scale;

Fig. 4 is the fitting residual figure of average risk score of the present invention and Weibull distribution parameter scale;

Fig. 5 is the result of the personalized lung adenocarcinoma prognosis assessment of the present invention.

Detailed ways

The implementation of the present invention will be described in detail below with examples, so as to fully understand and implement the implementation process of how the present invention uses technical means to solve technical problems and achieve technical effects.

The invention discloses a method for evaluating the individualized prognosis of lung adenocarcinoma based on a polygene expression profile, comprising the following steps:

Step 1. Obtain the list of prognostic risk genes and gene weights for lung adenocarcinoma;

Step 1.1, download the tumor tissue and paracancerous tissue transcriptome data and clinical data of lung adenocarcinoma patients from the Genomic Data Commons Data Portal database, and obtain the tumor tissue gene expression profile FPKM ( Fragments P er Kilobase of transcript per Million fragments mapped) values, logarithmic transformation (log2).

Step 1.2, set the total number of samples as m, divide all samples into three groups according to the tertiles of their gene expression values, where the gene expression value is the FPKM value obtained in step 1.1, denoted by V, for the ith The gene was denoted as V _i , and the survival risk of the third group compared with the first group was calculated using the Cox proportional hazards model, and the hazard ratio HR and P value of each gene were obtained. A P value < 0.05 was defined as significant, and significant survival risk genes were screened out, which were recorded as n ₁ . In addition, the correlation between each gene and the patient's survival days was calculated, and the correlation coefficient r and P value of each gene were obtained. A P value < 0.05 was defined as significant, and significant survival-related genes were screened, which were recorded as n ₂ . The intersection of survival risk genes and survival-related genes is defined as the prognosis risk gene, denoted as n, then:

n=n ₁ ∩n ₂ (1)

Step 1.3. Calculate the weight W _i of each gene according to the risk ratio of the i-th gene, and the calculation formula is:

Where i represents the gene number, and HR _i represents the hazard ratio of the i-th gene.

In this way, the weight of each gene was calculated; the final list of lung adenocarcinoma prognostic risk genes and gene weights are shown in Table 1.

Table 1 Gene name and weight

Step 2, using the tumor tissue transcriptome and survival data of patients with lung adenocarcinoma to construct a prognosis assessment model; specifically:

Step 2.1, define the gene expression value V, and calculate the risk score S _j of the i-th patient according to the expression value and weight of the i-th gene in the j-th sample; the calculation formula is:

Where i represents the gene number, j represents the patient number, V _ij represents the expression value of the i-th gene in the j-th sample;

Step 2.2, sort all lung adenocarcinoma patient samples according to the risk score from low to high, and use the sliding window model to calculate the average risk score for every 50 samples The calculation formula is:

where j+49 represents the last 50 samples counted from sample j.

Step 2.3. Use the Weibull distribution to perform curve fitting on the survival data of 50 samples. The probability density function of the Weibull distribution is:

Where k>0 is a shape parameter, and λ>0 is a scale parameter of the distribution.

Step 2.4, calculate for every 50 samples The corresponding k _j and λ _j . According to experience, k _j is a relatively fixed value, and the mean value is:

Among them, k _j is the shape parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample;

The variation range of the proportional parameter λ _j is relatively large, and the definition of λ _j and The functional relationship is:

Among them, _λj represents the proportional parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample;

Where e is the base of the natural logarithm, α and β are the parameters of the function, and the logarithm of the above formula can be obtained:

where logλ _j and It is a linear relationship, which can be solved by linear fitting.

The average risk score is shown in Figure 3 With the fitting curve of Weibull distribution parameter λ _j , the function relationship obtained is:

Will Substituting this function to get the predicted λ _j ′, λ _j ′ is the expected distribution parameter calculated by this function, calculating the correlation between λ _j and λ _j ′, the correlation coefficient R ² = 0.943, P value = 2.96E- 97.

Analysis of the fitted residual plots and QQ plots (Fig. 4) shows that the model reaches significance, i.e. the average risk score A functional relationship with the Weibull distribution parameter _λj is credible.

Step 3. Calculate the risk score of the patient according to the gene expression profile of the tumor tissue of the patient with lung adenocarcinoma; specifically:

Obtain the FPKM value of the gene expression profile of the tumor tissue of patients with lung adenocarcinoma (should include all or most of the genes listed in Table 1), denoted as: V _i (i is the gene number); the i-th gene in Table 1 corresponds to The weight is recorded as: W _i (i is the gene number); the patient risk score is recorded as: S; the calculation formula is:

Where i is the gene number, and n is the number of genes listed in Table 1.

Step 4. Calculate the patient's annual survival probability according to the patient's risk score, specifically:

Substituting the patient's risk score S into the cumulative distribution function of the Weibull distribution gives the patient's survival probability function as:

Where t is time, α, β, S, are fixed parameters.

Figure 5 shows a patient's survival probability curve, where the abscissa in the figure is the number of days, and the ordinate is the survival probability. The patient's annual probability of survival is plotted below the curve. The actual number of days the patient survived is marked in the black box in the upper right corner, and the status (Status) 1 indicates that the patient has died. The red point on the curve (Death) marks the corresponding days and survival probability of the patient when he died. In the figure, the survival probability corresponding to the death of the patient is about 0.22.

In summary, the present invention uses TCGA-LUAD transcriptome and clinical data to perform personalized survival prediction for all patients with lung adenocarcinoma, and uses the cross-validation method to verify the obtained results. The results showed that the annual survival probability of lung adenocarcinoma patients obtained by the individualized prognosis assessment method of lung adenocarcinoma using multi-gene expression profiles was highly consistent with the actual annual survival rate (linear correlation R ² =0.994, P value =2.86E-43 ,figure 1). It is confirmed that the method has high prediction accuracy, which is highly consistent with the actual survival status.

As shown in Figure 2, TNM stage had only a low negative correlation with the survival time of patients with lung adenocarcinoma. Comparing Figure 1 with Figure 2, it can be concluded that the personalized prognosis assessment method for lung adenocarcinoma based on the multi-gene expression profile can more accurately reflect the survival status of patients than the traditional TNM staging.

As shown in Fig. 5, the present invention can predict the survival probability of tumor patients in continuous time. For example, the survival probability of the patient per month, the survival probability of the patient per year, and the like can be given. However, the current clinical classification method can only give a qualitative judgment. For each tumor patient, the present invention can provide the specific survival probability curve of the patient, which is not available in general tumor prognosis assessment models.

The above description shows and describes several preferred embodiments of the invention, but as previously stated, it should be understood that the invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other embodiments. Combinations, modifications and circumstances, and can be modified within the scope of the inventive concept described herein, by the above teachings or by skill or knowledge in the relevant field. However, changes and changes made by those skilled in the art do not depart from the spirit and scope of the invention, and should be within the protection scope of the appended claims of the invention.

Claims (6)

1. A lung adenocarcinoma personalized prognosis assessment method based on polygene expression profile, it is characterized in that, comprises the following steps: Step 1. Obtain the list of prognostic risk genes and gene weights for lung adenocarcinoma; Step 2, using the tumor tissue transcriptome and survival data of patients with lung adenocarcinoma to construct a prognosis assessment model; Step 3, calculating the risk score of the patient according to the gene expression profile of the tumor tissue of the lung adenocarcinoma patient; Step 4. Calculate the patient's annual survival probability based on the patient's risk score. 2. The method for assessing prognosis according to claim 1, characterized in that the acquisition of lung adenocarcinoma prognostic risk gene list and gene weight in said step 1 is specifically implemented according to the following steps: Step 1.1. Download the tumor tissue and paracancerous tissue transcriptome data and clinical data of lung adenocarcinoma patients from the Genomic Data Commons Data Portal database, obtain the FPKM value of the tumor tissue gene expression profile of lung adenocarcinoma patients, and perform logarithmic transformation; Step 1.2, set the total number of samples as m, divide all samples into three groups according to the tertiles of their gene expression values, where the gene expression value is the FPKM value obtained in step 1.1, denoted by V, for the ith The gene is denoted as V _i , and the Cox proportional hazard model is used to calculate the survival risk of the third group compared with the first group, and the hazard ratio HR and P value of each gene are obtained; P value <0.05 is defined as significant, and the screening is significant The survival risk genes of , denoted as n ₁ ; in addition, calculate the correlation between each gene and the patient's survival days, and obtain the correlation coefficient r and P value of each gene; define that P value <0.05 is significant, and the screening is significant The survival-related genes of , denoted as n ₂ ; the intersection of survival risk genes and survival-related genes is defined as the prognosis risk gene, denoted as n, then: n=n ₁ ∩n ₂ (1) Step 1.3. Calculate the weight W _i of each gene according to the risk ratio of the i-th gene, and the calculation formula is: Where i represents the gene number, and HR _i represents the hazard ratio of the i-th gene; In this way, the weight of each gene is calculated; the final lung adenocarcinoma prognosis risk gene list and gene weight are obtained. 3. The prognosis assessment method according to claim 1, wherein the list of prognostic risk genes and gene weights of the lung adenocarcinoma are specifically shown in the following table: 4. The method for evaluating prognosis according to claim 1, characterized in that, in said step 2, the use of tumor tissue transcriptome and survival data of patients with lung adenocarcinoma to construct a prognosis evaluation model is specifically implemented according to the following steps: Step 2.1, define the gene expression value V, and calculate the risk score S _j of the i-th patient according to the expression value and weight of the i-th gene in the j-th sample; the calculation formula is: Where i represents the gene number, V _ij represents the expression value of the i-th gene in the j-th sample; Step 2.2, sort all lung adenocarcinoma patient samples according to the risk score from low to high, and use the sliding window model to calculate the average risk score for every 50 samples The calculation formula is: Where j+49 represents the last 50 samples counted from sample j; Step 2.3. Use the Weibull distribution to perform curve fitting on the survival data of 50 samples. The probability density function of the Weibull distribution is: Where k>0 is a shape parameter, and λ>0 is a scale parameter of the distribution. Step 2.4, calculate for every 50 samples The corresponding k _j and λ _j ; according to experience, k _j is a relatively fixed value, and the mean value is: Among them, k _j is the shape parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample; The variation range of the proportional parameter λ _j is relatively large, and the definition of λ _j and The functional relationship is: Among them, _λj represents the proportional parameter of the Weibull distribution of the survival curve from the jth sample to the j+49th sample; Where e is the base of the natural logarithm, α and β are the parameters of the function, and the logarithm of the above formula is obtained: where logλ _j and is a linear relationship, solved by linear fitting; According to the average risk score With the fitting curve of Weibull distribution parameter λ _j , the function relationship obtained is: Will Substituting this function to get the predicted λ _j ′, λ _j ′ is the expected distribution parameter calculated by this function, and calculating the correlation between λ _j and λ _j ′, the correlation coefficient R ² = 0.943, P value = 2.96E-97 . 5. The prognosis assessment method according to claim 1, characterized in that the calculation of the patient's risk score according to the gene expression profile of the lung adenocarcinoma patient's tumor tissue in said step 3 is specifically implemented according to the following steps: Obtain the FPKM value of the gene expression profile of the tumor tissue of patients with lung adenocarcinoma, recorded as: V _i , where i is the gene number; the weight corresponding to the i-th gene is recorded as: W _i , where i is the gene number; patient risk score Recorded as: S; the calculation formula is: Where i is the gene number, and n is the number of genes listed in Table 1. 6. The prognosis assessment method according to claim 1, characterized in that the calculation of the patient's annual survival probability according to the patient's risk score in said step 4 is specifically implemented according to the following steps: bringing the patient's risk score S into Weibull distribution The cumulative distribution function of gives the patient's survival probability function as: Where t is time, α, β, S, are fixed parameters.