Intestinal microbiota markers related to immune and nutritional status for prognosis of non-small cell lung cancer and application thereof

Abstract

The application discloses a kind of non-small cell lung cancer prognosis immunity and nutrition state related intestinal microorganism marker and its application, intestinal microorganism marker specifically includes: Eubacterium hallii, Akkermansia, Lachnospira pectinosus, Lachnospira viviparous, Paraclostridium, Haemophilus influenzae.The relative abundance of intestinal flora is used with patient PNI index, and the correlation of the diversity of intestinal flora, intestinal type and specific strain and PNI is established.The relative abundance of 6 specific strains of intestinal flora is used with patient PNI index to construct the evaluation model of nutrition immunity, by using this model, the relative abundance information of 6 specific strains of new individual is input, and the result of nutrition immunity evaluation is output, without invasive examination, can repeatedly detect, and improve the nutrition immunity state of patient by the intervention of specific intestinal flora.This provides a new way for personalized medicine, helps doctor to formulate more accurate treatment strategy.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of biological medicine, and particularly relates to a gut microbial marker related to immune and nutritional status of non-small cell lung cancer prognosis and application thereof. BACKGROUND

[0002] Human intestinal flora and host constitute an interrelated whole, and intestinal microorganisms not only can degrade nutrients in digested food, host vitamins and other nutrients, but also can promote differentiation and maturation of intestinal epithelial cells to activate the intestinal immune system and regulate host energy storage and metabolism, which play an important role in human digestion and absorption, immune response, metabolic activity and the like; therefore, the composition of intestinal microorganisms has an important influence on the immune and nutritional status of individuals. The assessment of the immune and nutritional status of individuals is crucial for health management, disease prevention and improvement of disease prognosis. For example, the prognosis of non-small cell lung cancer patients after surgery is closely related to their nutritional and immune status. However, the current evaluation index of immune and nutrition is inaccurate, imprecise and has a lag; the nutritional index such as PNI for evaluating disease prognosis involves clinical blood sampling and is a invasive detection method, and there is no effective and precise intervention method for improving immune and nutritional deficiency. Therefore, a more effective, simple and intervenable method is needed to evaluate the immune and nutritional status of individuals, so as to better evaluate the risk of affecting the immune system, metabolic health, nutritional absorption and various chronic diseases. SUMMARY

[0003] The purpose of the present application is to provide a gut microbial marker related to immune and nutritional status of non-small cell lung cancer prognosis and application thereof, which can better evaluate the immune and nutritional status of individuals by the abundance of the gut microbial marker, and can play an important role in the treatment and health management of patients, thereby providing a new way for personalized medicine and health management.

[0004] In the first aspect, the present application provides a gut microbial marker related to immune and nutritional status of non-small cell lung cancer prognosis, which adopts the following technical scheme:

[0005] A gut microbial marker related to immune and nutritional status (PNI) of non-small cell lung cancer prognosis, wherein the gut microbial marker comprises Eubacterium hallii, (Eubacterium hallii ), Akkermansia muciniphila, Akkermansia muciniphila ), Ruminococcus callidus, Ruminococcus gnavus ), Parabacteroides distasonii, Clostridium paraputrificum ), Haemophilus influenzae, Haemophilus influenza .

[0006] The results of the present application show that Eubacterium hallii, Akkermansia muciniphila and Ruminococcus flavefaciens are significantly more abundant in the gut of high PNI individuals than in low PNI individuals; while Ruminococcus gnavus is significantly less abundant in the gut of high PNI individuals than in low PNI individuals. While Clostridium subsp. ( Clostridium paraputrificum ) and Haemophilus influenzae ( Haemophilus influenza ) show a trend of being more abundant in low PNI patients than in high PNI patients.

[0007] Preferably, the non-small cell lung cancer prognosis gut microbiota marker further comprises: Eubacterium ventriosum ( Eubacterium ventriosum) , Parvimonas sp. ( Subdoligranulum) , Coprococcus sp. ( Coprococcus) , Odoribacter sp. ( Odoribacter N54 MGS 14) , Ruminococcus sp. ( Ruminococcus A254 MGS 108） , Actinobacteria sp. ( Actinobaculum oral taxon 183 F0552)、 Lactobacillus brevis ( Lactobacillus brevis) , Haemophilus influenzae ( Haemophilus influenza)。

[0008] The results of the present application show that high PNI individuals are enriched in Eubacterium ventriosum, Parvimonas sp., Coprococcus sp., Odoribacter sp., Ruminococcus sp. and Actinobacteria sp.; while low PNI individuals are enriched in Lactobacillus brevis.

[0009] A non-small cell lung cancer prognosis immune and nutritional status (PNI) related gut microbiota marker, said gut microbiota marker shows two enterotypes in cluster analysis: Ruminococcus enterotype Ruminococcus and Bacteroides enterotype Bacteroids .

[0010] According to the experimental results of the present application, high microbial diversity is shown to be enriched in high immune and nutritional status (high PNI) individuals, while low microbial diversity is shown to be enriched in low immune and nutritional status (low PNI) individuals.

[0011] According to the experimental results of the present application, Ruminococcus enterotype Ruminococcus is shown to be enriched in high immune and nutritional status (high PNI) individuals, while Bacteroides enterotype Bacteroids is shown to be enriched in low immune and nutritional status (low PNI) individuals.

[0012] In a second aspect, the present application provides an evaluation model for the immune and nutritional status of non-small cell lung cancer prognosis, which is characterized by the intestinal microorganism markers described above, obtaining the corresponding relationship between the relative abundance of the bacterial species and the PNI, standardizing the relative abundance data of the bacterial species, taking the standardized relative abundance of the bacterial species as the input variable, taking the PNI index as the output variable, and using an Epsilon-Support Vector Regression model to establish a regression model between the relative abundance of the bacterial species and the PNI index.

[0013] In a third aspect, the present application provides a reagent for detecting the non-small cell lung cancer prognosis microorganism markers described above, which comprises a reagent for collecting a biological sample from a subject, a preparation for extracting microorganism marker DNA, and a reagent for detecting microorganism marker DNA and determining its abundance. The biological sample is human feces.

[0014] In a fourth aspect, the present application provides a system for evaluating non-small cell lung cancer prognosis microorganism markers, which comprises the detection results of the intestinal microorganism marker combination and the evaluation model.

[0015] The system further comprises a detection device.

[0016] The detection device comprises a device for sequencing or detecting the abundance information of each intestinal microorganism in the intestinal microorganism combination.

[0017] In a fifth aspect, the present application further provides a probiotic for assisting in improving the immune and nutritional status (PNI) related to non-small cell lung cancer prognosis, which uses the corresponding probiotic to increase the abundance of the bacterial species positively correlated with nutrition and immunity and to reduce the abundance of the bacterial species negatively correlated with nutrition and immunity.

[0018] The present application has the following beneficial effects:

[0019] 1) The present application uses the relative abundance of intestinal flora and the PNI index of patients to establish the correlation between the diversity, intestinal type and specific bacterial species of intestinal flora and PNI.

[0020] 2) The present application uses the relative abundance of six specific bacterial species of intestinal flora and the PNI index of patients to construct an evaluation model for nutrition and immunity. By using this model, the relative abundance information of the six specific bacterial species of a new individual is input, and the result of nutrition and immunity evaluation is output. This model does not require invasive examination, can be repeatedly detected, and can improve the nutritional immunity status of patients through the intervention of specific intestinal flora. This provides a new way for personalized medicine and helps doctors to develop more accurate treatment strategies.

[0021] 3) This invention is applicable not only to patients with non-small cell lung cancer but also to healthy individuals. By analyzing the relationship between microbial composition and immune and nutritional status, this invention helps to identify health risks in patients at an early stage. This enables medical professionals to take intervention measures before the disease worsens, improving the success rate of treatment.

[0022] 4) The application of this invention helps improve the accuracy and reliability of postoperative prognostic assessment, providing a reference for clinical treatment and management decisions. In stage I NSCLC patients, PNI is a strong predictor of overall survival, surpassing TNM in efficacy. The research results of this invention contribute to a deeper understanding of the interrelationships between gut microbiota and immune and nutritional status. This helps advance the fields of gut microbiology and pathophysiology research, providing a new scientific basis for future disease treatment. Attached Figure Description

[0023] Figure 1 shows the predictive results of PNI for the prognosis of surgical treatment in patients with non-small cell lung cancer in Example 1. (A): Results of retrospective evaluation of PNI levels in all clinical trial patients; (B): Results of ROC curves constructed using a logistic regression model for all clinical trial patients; (C): Results of Youden index analysis for all clinical trial patients; (D): Results of retrospective evaluation of PNI levels in patients in Phase I clinical trials; (E): Results of ROC curves constructed using a logistic regression model for patients in Phase I clinical trials; (F): Results of Youden index analysis for patients in Phase I clinical trials.

[0024] Figure 2 Example 1 shows the results of PNI as a predictive indicator for the prognosis of surgical treatment in patients with non-small cell lung cancer. Among them, (G): PNI cutoff value analysis results; (H): Multivariate analysis results of patients in clinical stage I; (I) Prognostic evaluation effect of PNI in patients at different clinical stages.

[0025] Figure 3 Example 2: Differences in gut microbiota composition between the high PNI group and the low PNI group; (A): Venn plot of OTUs shared by the gut microbiota of patients with high and low PNI in non-small cell lung cancer; (B): α-diversity data of the gut microbiota of patients; (C): PLS-DA analysis results of microbiota changes between groups.

[0026] Figure 4 The differences in gut microbiota composition between the high PNI group and the low PNI group in Example 2, where (D): relative abundance data of bacterial 16S rRNA sequences at the phylum level; (E): relative abundance data of bacterial 16S rRNA sequences at the genus level.

[0027] Figure 5, the intestinal microbial composition difference between high PNI group and low PNI group in Example 2, wherein, (F): Cladogram analysis of intestinal bacteria with significant difference in high and low PNI patients; (G): LEfSe analysis of intestinal bacteria with significant difference in high and low PNI patients.

[0028] Figure 6 , specific intestinal types of high PNI group and low PNI group in Example 2; (A): Principal component analysis of the intestinal microbial composition of non-small cell lung cancer patients divides the patients into two groups with good discrimination; (B): Relative abundance clustering determines the optimal clustering number Calinski-Harabasz (CH) Index, showing that the patients are divided into 2 categories; (C) : Cluster analysis shows the relative abundance of Ruminococcus, Bacteroides and Prevotella at the genus level in the two clusters ); (D): The proportion of high and low PNI patients in the two intestinal type clusters.

[0029] Figure 7, qPCR relative abundance of 6 representative intestinal bacteria in non-PNI patients in Example 3. DETAILED DESCRIPTION

[0030] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and examples. Unless otherwise defined, the scientific and technical terms used herein have the meanings commonly understood by one of ordinary skill in the art. The nomenclature and experimental methods used herein are well known and routinely used in the art. The operations used are generally performed using standard techniques according to the product instructions of the instrument consumables manufacturers and the conventional technical requirements and the references provided herein.

[0031] Example 1 to determine the optimal cutoff value of PNI

[0032] In order to determine the optimal cutoff value of PNI in non-small cell lung cancer (NSCLC) patients, first, the PNI levels of patients with different 5-year overall survival rates were retrospectively evaluated. Our analysis included a total of 289 postoperative patients. The formula for calculating PNI is PNI = serum albumin (g / L) + 5 x absolute lymphocyte count (10 9 / L). Overall survival (OS) was defined as the time from surgery to death due to non-small cell lung cancer, and about 79.87% of the patients survived for more than 5 years.

[0033] Figure 1The results of A were retrospectively evaluated for PNI levels in all patients with different 5-year overall survival rates. This analysis included a total of 289 postoperative patients. From the results, the average PNI value for all patients was 48.80 ± 4.20, ranging from a minimum of 40.00 to a maximum of 66.90; the PNI level was significantly higher in patients who survived for 5 years (49.40 ± 4.12) than in patients who died (46.40 ± 3.64) (P < 0.05). To further investigate whether there was a change in the PNI cutoff value in clinical stage I patients, Figure 1 The results of D were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years.

[0034] Figure 1 The results of B were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years. Figure 1 The results of E were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years.

[0035] Figure 1 The results of C were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years. Figure 1 The results of F were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years.

[0036] Figure 2 The results of G were retrospectively evaluated for PNI levels in 154 clinical stage I patients with different 5-year overall survival rates; the scatter plot shows the PNI of patients who survived and the PNI of patients who died, based on the survival of 154 clinical stage I patients for 5 years after surgery. From the results, in clinical stage I patients, the PNI level was significantly higher in patients who survived for 5 years. Figure 2Further multivariate analysis in H showed that PNI had better prognostic evaluation effect than TNM staging in patients with clinical stage I. Figure 2 The results in I showed that PNI had good prognostic evaluation effect in patients with clinical stage I and II. However, PNI did not show good prognostic effect in postoperative NSCLC patients with clinical stage III.

[0037] In summary, these results showed that PNI, as an indicator of the nutritional and immune status of patients, could be a valuable tool for predicting the prognosis of early postoperative non-small cell lung cancer patients.

[0038] Example 2 Differences in intestinal microbiome composition between high PNI and low PNI non-small cell lung cancer patients

[0039] In compliance with relevant ethical requirements, 14 cases of HPNI group non-small cell lung cancer patients and 14 cases of LPNI group non-small cell lung cancer patients were obtained. All specimens were from Xiangya Second Hospital of Central South University. After obtaining fresh fecal specimens from the two groups, the middle part of the feces was immediately scooped out with a sterile cotton swab in a clean bench, placed in a sterile cryopreservation tube, and immediately frozen in liquid nitrogen and stored in a -80°C freezer. The Magen company's fecal microbial DNA extraction kit (HiPure Stool DNA Kits, D3141-02) was used to extract total microbial DNA from the fecal sample according to the instructions, and stored at -20°C for use. In order to detect the quality of the extracted total microbial DNA, the concentration of DNA and the OD 260 / 280 and OD 260 / 230 values were determined using Nanodrop, and 1% agarose gel was prepared, 500 ng of DNA was spotted, and the gel was run at 100 V for 20 min in a DNA electrophoresis tank. Under the blue light, observe whether the DNA is degraded. The Beijing Aovison Gene Technology Co., Ltd. sequencing platform Illumina Miseq / Novaseq 6000 was used to complete the 16S rRNA sequencing of the extracted DNA samples, and the sequencing strategy was PE250 / PE300.

[0040] The data of intestinal microbiome composition were mapped to the reference genome by 16S rRNA sequencing to generate phylogenetic files, and the sequencing data were split by barcode sequence according to different samples by QIIME software. Then the sequencing data were filtered and spliced by Pear software, and the sequences with a length of less than 270 bp were removed after splicing by Vsearch software, and the chimeric sequences were removed by comparison according to the Gold Database database by the uchime method to retain high-quality microbial sequence reads. Then the high-quality sequences were clustered by OTU (Cluster) by using the Vsearch (v2.7.1) software uparse algorithm (the sequence was classified according to the similarity between each other, and a group was an OTU), and the sequence similarity threshold was 97%. The OTU representative sequence was compared with the Silva138 database by using the BLAST algorithm, and the e-value threshold was set to 1e-5, to obtain the species classification information corresponding to each OTU, and the species information of the community was annotated at each level (kingdom, phylum, class, order, family and genus). Finally, the species diversity of the two groups of samples was analyzed and compared by using the R package (RDP classifier), and the significantly different bacteria in the two groups of samples were finally screened and identified.

[0041] In this analysis step, 16s rRNA sequencing analysis uses multiple indicators such as chao1, observed species, PD_whole_tree and shannon index to evaluate the α diversity of the sample and its results can be seen Figures 3-5 .

[0042] The Venn diagram of OTUs shared by the intestinal microbiota of high and low PNI patients in NSCLC patients is shown in Figure 3 A. Figure 3 B is the application of 4 indicators including chao1, observed species, PD whole tree and shannon index to evaluate the α diversity of the intestinal microbiota of patients, and from the results it can be seen that the intestinal microbial diversity of high PNI patients is significantly higher than that of low PNI patients. Figure 3 C is the application of PLS-DA (partial least squares discriminant analysis) to investigate the overall difference of intestinal microbiota between high and low PNI patients, and the results show the PLS-DA analysis of microbial group changes between groups; it shows that there is a significant difference in the composition of intestinal microbiota between high and low PNI patients. Figure 4 D and E are the relative abundance of bacterial 16S rRNA sequences at the phylum level and the genus level, respectively. Figure 5F and G were applied to analyze the differences in the intestinal microorganisms of patients with high and low PNI using Cladogram (phylogenetic tree) and LEfSe (linear discriminant analysis effect size), respectively. The results showed the significant differences in the intestinal microorganisms of patients with high and low PNI analyzed by Cladogram and LEfSe; the species and genera of intestinal microorganisms of patients with PNI higher than 46.2 and lower than 46.2 were screened by LEfSe analysis, and the results are shown in Figure 5 G. The results showed that the microbial composition of the high immune and nutritional status group was rich in A kkermansia、 Eubacterium halli、Eubacterium coprostanoligenes、Eubacterium ventriosum、 Subdoligranulum、Coprococcus、Clostidium methylpentosum DSM 5476、Odoribacter N54 MGS 14、Ruminococcaceae bacterium GD1、Ruminococcus Callidus、Ruminococcus N15 MGS57、Ruminococcus A254 MGS 108、Eubacterium Coprostanoligenes、 Actinobaculum oral taxon 183 F0552、NK4A214、Eubacterium Ventriosum、 Rikenellaceae、Alistipes、Adlercreutzia、Sellimonas、Eggerthellaceae、UCG 002、 Oscillospiraceae、Ruminococcus、Verrucomicrobiae、Verrucomicrobiota; The intestinal microbial composition of the low immune and nutritional status group was rich in Ruminococcus gnavus、Clostridium paraputrifucum、 Lactobacillus brevis, Haemophilus influenzae, Haemophilus, Haemophilus influenzae, Pasteurellaceae, Pasteurellales, Tyzzerella sp Marseille P3062.

[0043] The two groups of samples screened above were clustered by PAM (partitioning around medoids) based on the Jensen-Shannon distance between samples, and the optimal number of clusters was determined by the Calinski-Harabasz (CH) index. The calculation results were visualized by PCoA, and the specific results can be seen in Figure 6 : Principal component analysis (PCA) revealed that lung cancer patients could be classified into 2 obvious clusters, each corresponding to a unique enterotype: Bacteroids (enterotype 1) and Ruminococcus (enterotype 3) (see Figure 6 A). A clustering algorithm based on the relative abundance of each genus was used, and the optimal number of clusters was determined by the Calinski-Harabasz (CH) index as 2 (see Figure 6 B). The main contributors of each enterotype were determined and are shown in Figure 6 C. As shown in Figure 6 D, high PNI patients were mainly associated with Ruminococcus enterotype, while low PNI patients were mainly associated with Bacteroids enterotype.

[0044] Example 3

[0045] According to the results in Example 2, 6 species of bacteria with specific primers at the species level were selected for qPCR verification. According to the representative species in Example 2 with high and low PNI expression differences, 6 representative species were selected: the species with significantly higher abundance in the intestinal tract of patients with high PNI than in patients with low PNI: Eubacterium hallii ( Eubacterium hallii ), Akkermansia muciniphila ( Akkermansia muciniphila ), Ruminococcus callidus(Ruminococcus callidus ) ; and Faecalibacterium prausnitzii (F. prausnitzii), which was significantly more abundant in the gut of low PNI patients than in that of high PNI patients. Ruminococcus gnavus ). While Clostridium subsp. (C. subsp.) and Haemophilus influenzae (H. influenzae) had a trend of higher abundance in low PNI patients than in high PNI patients. Clostridium paraputrificum (Haemophilus influenza

[0046] Firstly, the 16S rDNA gene sequences of the differential species were found in the NCBI database, and the primers for the variable region of bacterial 16S rDNA were designed, as shown in Table 1. The designed primers were checked for bacterial coverage using the TestPrime tool (https: / / www.arb-silva.de). The lower the coverage, the higher the specificity of the primers. Finally, the primers with the lowest coverage were selected for synthesis. In this example, in addition to the differential species primers, a pair of reference primers was set to amplify the universal sequence of microorganisms (Maeda H, et al. Quantitative real-time PCR using TaqMan and SYBR Green for Actinobacillus actinomycetemcomitans, Porphyromonas gingivalis, Prevotella intermedia, tetQ gene and total bacteria. FEMS Immunology & Medical Microbiology. 2003, 39:81-86). The experimental method and steps for detecting the relative abundance of differential microorganisms in fecal samples by qPCR were as follows: first, the optimal annealing temperature of each species primer was determined by gradient annealing temperature PCR; then, the DNA extracted from each sample was used as a template for qPCR with each specific primer of the differential species, and the cycle number (Ct value) of each species corresponding primer was recorded; finally, the relative abundance of each differential species in each sample was calculated. The relative abundance (Ra) of any species i can be expressed as: Ra (i) = (1 / 2)^(Cti-Ctc), where Cti represents the cycle number of species i, and Ctc represents the cycle number of the universal primer. The total volume was 20 μL. The qPCR reaction conditions were as follows: 1) 50°C incubation for 2 min; 2) 95°C pre-denaturation for 2 min; 3) 95°C denaturation for 15 s, 56°C annealing for 15 s, and 72°C extension for 1 min, repeated for 40 cycles; and 4) melting curve analysis, with the default program.

[0047] Table 1

[0048] ​​

[0049] Representative bacterial species showing differential expression were detected by quantitative real-time PCR, and statistical analysis was performed using a t-test, with p-values ​​less than 0.05 considered statistically significant. The results showed that at the bacterial level, *Eubacterium hominis* (…) Eubacterium hallii Akkermansia myxophilus ( Akkermansia muciniphila ), and rumen cocci ( Ruminococcus callidus In patients with high PNI, intestinal abundance was significantly higher than in patients with low PNI (Figure 7), while active rumenococci ( Ruminococcus gnavus In patients with high PNI, gut microbiota abundance was significantly lower than in those with low PNI. These significantly different bacterial species could serve as potential microbial biomarkers for assessing prognostic survival in patients with early-stage non-small cell lung cancer. Figure 7 Clostridium parasitoides ( ). Clostridium paraputrificum ) and Haemophilus influenzae ( Haemophilus influenza The abundance of PNI tended to be higher in patients with low PNI than in patients with high PNI. Figure 7 ).

[0050] A regression model between the relative abundance of the six bacterial species and the PNI (Potential Incidence Nominal Index) was established using Epsilon-Support Vector Regression, and trained for prediction in two independent cohorts of non-small cell lung cancer patients. First, the relative abundance of the six bacterial species in each cohort was standardized, and the mean and standard deviation of each species were recorded. Then, the relative abundance of the corresponding bacterial species in the cohort was standardized to a specific value. On the standardized data, the model used the Gaussian kernel-based support vector regression model Epsilon-Support Vector Regression, where the prediction function was defined as follows:

[0051] Where is the predicted PNI exponent, is a vector composed of the relative values ​​of bacterial species in the training samples, is the Lagrange multiplier of the support vector, and is the Gaussian kernel function. During training, the optimization objective of this Epsilon-Support Vector Regression model is:

[0052]

[0053] in The predicted PNI index. This is a vector composed of the relative values ​​of bacterial species in the training samples. It is the Lagrange multiplier of support vectors. The kernel function is Gaussian. During training, the optimization objective of this Epsilon-SupportVector Regression model is:

[0054]

[0055] Meanwhile, the following constraint conditions are satisfied:

[0056]

[0057] wherein is the control boundary hyperparameter, is the slack variable, and C represents the regularization parameter. The non-small cell lung cancer patient cohort and were trained and predicted using the parameter C = 10, wherein the PNI mean absolute error (MAE) of the training prediction was 2.048, and the PNI mean absolute error (MAE) of the training prediction was 2.918.

[0058] From the above results, it can be concluded that the model performance evaluation was performed on two independent non-small cell lung cancer patient cohorts, and the PNI mean absolute error (MAE) of the training and prediction was obtained, indicating that the model can effectively predict the PNI index based on the relative abundance of 6 microorganisms. By applying the model, the qPCR relative abundance values of 6 bacteria can be input to obtain the PNI index.

[0059] In order to verify the actual application effect of the model, two patients with high and low PNI indexes were selected, and the PNI values of the two patients were calculated according to the method of Example 1, and then the prediction was performed according to the method of Example 3. The specific steps are as follows:

[0060] The PCR and PNI data in the training set were used to train our SVM model;

[0061] The data of the independent test set were input into the trained model to predict the PNI value of the test set. The test results are as follows:

[0062] Low PNI group: the true PNI value is 45.85, and the predicted PNI value is 45.59; high PNI group: the true PNI value is 52.85, and the predicted PNI value is 52.35; wherein the true PNI value is the test value calculated by the method in Example 1, and the predicted PNI value is the test value obtained by the method and model in the present application. These results show that the PNI values tested by our method and the method in Example 1 are very close, indicating that the method of the present application can accurately predict the PNI value of non-small cell lung cancer patients.

Claims

1. A prognostic assessment model for immune and nutritional status in non-small cell lung cancer, characterized in that, Based on gut microbial markers, the correspondence between the relative abundance of bacterial species and the PNI was obtained. The relative abundance data of bacterial species were standardized. The input variable was the standardized relative abundance of bacterial species, and the output variable was the PNI index. The Epsilon-SupportVector Regression model was used to establish a regression model between the relative abundance of bacterial species and the PNI index. The intestinal microbial markers are the following six types: Eubacterium hopterii, Akkermansia, Clavicipitolus rumenococcus, Clostridium parapeptidans, and Haemophilus influenzae. Fecal samples were obtained from 14 patients with non-small cell lung cancer in the HPNI group and 14 patients with non-small cell lung cancer in the LPNI group. Total microbial DNA was extracted from the fecal samples. Gut microbiota composition data was mapped to a reference genome using 16S rRNA sequencing to generate a phylogenetic tree file. The sequencing data was then split into different samples based on barcode sequences using the QIIME software. The sequencing data were filtered and assembled using PEAR software. After assembly, Vsearch software was used to remove sequences shorter than 270 bp. Chimeric sequences were then removed by alignment with the Gold Database using the uchime method, retaining high-quality microbial sequence reads. The Vsearchv 2.7.1 software was used to perform OTU clustering on high-quality sequences using the uparse algorithm. The species diversity of the two groups of samples was analyzed and compared using the R package, and bacteria with significant differences between the two groups were finally screened and identified. The two selected groups of samples were clustered using a centroid-based clustering algorithm based on the Jensen-Shannon distance between the samples. Principal component analysis (PCA) revealed that lung cancer patients were classified into two distinct clusters, each corresponding to a unique intestinal type: Bacteroids Intestinal type 1 and Ruminococcus Intestinal type 3, Six strains with specific primers at the bacterial level were selected for qPCR validation. A regression model between the relative abundance of the above six bacterial species and PNI was established using Epsilon-Support Vector Regression. On normalized data, this model adopted the Gaussian kernel-based support vector regression model Epsilon-Support Vector Regression, where the prediction function is defined as follows: in The predicted PNI index. This is a vector composed of the relative values ​​of bacterial species in the training samples. It is the Lagrange multiplier of support vectors. Given a Gaussian kernel function, the optimization objective of this Epsilon-Support VectorRegression model during training is: The following constraints must be met simultaneously: in, is the control boundary hyperparameter, Ɛ is the relaxation variable, and C represents the regularization parameter. The non-small cell lung cancer patient cohort was trained and predicted using parameter C=10.

Images