Come on Monday to the CCR spirit one morning

What I share with you today is an article recently published in Clinical cancer research, with an impact factor of 10.
107.

Cancer cell-intrinsic and immunological phenotypes determine clinical outcomes in basal-like breast cancer Breast cancer (BLBC) is an aggressive molecular subtype of breast cancer that lacks targeted therapy.

There is currently no clinically available test for risk-stratified treatment of BLBC patients.

So in this article, the author conducted a genome sequencing study using paired design and verified it in five independent cohorts.

The authors performed transcriptome sequencing on matched recurrent and non-recurrent BLBC tumors to identify prognostic phenotypes.

The author mainly applied variable selection algorithm to screen prognostic-related genes, and through leave-one-out cross-validation, trained a random forest classifier based on the first 21 genes (BRAVO-DX).

It was verified in five independent BLBC cohorts, and the robustness of the biomarkers was proved by PCR.

Alright~ Let's take a look at the specific research that the author has done~ The results show 1.
Discover and verify the characteristics of the cohort.
The analyzed cases are 20-69-year-old women with invasive breast cancer, from 1408 cases I- A total of 949 patients were identified as eligible for this analysis among patients with stage III TNBC (ER- / PR- / HER2-).

After analyzing the patient's comprehensive information, the author finally selected 67 relapsed and 67 non-relapsed BLBC patients, and performed RNA-seq analysis.

The public expression data of five independent cohorts of TNBC patients were used for verification (Figure 1).

Figure 1.
Study design and description of the basal-like cohort 2.
Transcriptome analysis of the found concentrated formalin-fixed paraffin-embedded tissue.
Breast cancer tissue collected at the time of direct surgery or diagnosis of the primary cancer was obtained from the treated hospital Formalin-fixed paraffin-embedded (FFPE) blocks before treatment.

RNA was extracted from 134 patients and used to characterize the molecular and phenotypic characteristics of recurrent and non-recurrent breast tumors.

The authors first compared the samples with the TCGA breast cancer cohort analyzed using poly(A)+ RNA-seq, including basal-like, luminal, HER2 amplified, and normal-like tumors.

Basal-like TCGA carcinoma was found, and 90.
8% of tumors were classified as PAM50 basal-like intrinsic subtype.

Next, the authors verified that the recurrence status was not affected by technical covariates, and found that there was no correlation between tumor purity and recurrence.

In addition, the levels of ER (ESR1), PR (PGR) or HER2 (ERBB2) in the two samples did not increase.

3.
The relationship between tumor molecular characteristics and recurrence In this part, the author first identified the differentially expressed genes (DE) of recurrent and non-recurrent tumors, and then performed a functional enrichment analysis of important DE genes (Figure 2).

A total of 560 differentially expressed genes were identified.

Among them, 239 genes are associated with a good prognosis, and 321 genes are associated with a poor prognosis.

Enrichment analysis of risk genes found that they were significantly enriched in epithelial markers and several cell differentiation characteristics.

Survival genes are significantly enriched in immune cell markers and immune-related markers.

The authors next assessed whether the recurrence is related to different molecular or histochemical subtypes.

According to the expression levels of the top 100 risk-related genes, patients were clustered into three categories, and their risk of recurrence varied greatly.

Pathological examination and evaluation found that the low-risk group had higher lymphocyte infiltration, while the high-risk group had low immune infiltration, and there was a significant difference in expression between the high-risk and low-risk groups.

These results indicate that the risk of BLBC recurrence is significantly related to the transcriptome phenotype that integrates the inherent expression patterns of cancer cells and immune cells.

Figure 2.
Transcriptional characteristics of recurrent and non-recurrent BLBC tumors.
4.
The correlation between immune genome and recurrence-free survival.
Based on the above results, the author discusses the immunity between recurrent and non-recurrent BLBC tumors in more detail in this section Learn the difference.

The author first used the CIBERSORT method to characterize the tumor microenvironment according to cell composition, and found that tumors grouped according to the risk of recurrence differ greatly in the total level of immune infiltration and composition characteristics (Figure 3).

The author also compared the levels of macrophages, T cells and B cells between the three groups, and found that the ratio of these three cells in low-risk tumors was significantly higher than the other two groups.

As the degree of risk increases, the proportion of regulatory T cells (Treg) increases.

However, natural killer (NK) cells showed the opposite trend.

The authors also observed a significant correlation between the abundance of tumor-infiltrating T and B lymphocytes and the degree of clonal expansion (lower Gini index).

Figure 3.
The correlation between BLBC recurrence risk and immunogenomics 5.
Development and validation of robust models for recurrence-free survival in BLBC and TNBC.
As the phenotype and prognosis of BLBC tumors are clearly stratified, the authors developed prognostic genes and classification algorithms here (Figure 4).

The author first carried out strict variable selection and cross-validation.

Among them, risk genes and survival genes are highly correlated, so they cannot be used as independent prognostic markers.

The author uses a variety of variable selection algorithms to screen characteristic genes.

The ranking is based on the non-redundancy of genes and the expected effect of classification.

By training a random forest (RF) classifier, LOOCV is used to increase the number of genes, and finally the top 21 genes are selected as BRAVO-DX for further evaluation.

It was found that BRAVO-DX represents a larger set of independent genes with related expression patterns.

In order to assess whether this set of genes covers the expression patterns of cancer cells internal and immune cells, the authors correlated the expression of the BRAVO-DX gene with the predicted immune cell abundance.

The results showed that about half of the genes were negatively related to the expected immune cell abundance, rather than immune cell markers, thus indicating that they are internally expressed by tumor cells.

The author uses the previously proposed cost-sensitive learning to train a classifier with the required classification features.

Under the best sampling rate, the classifier has high sensitivity in identifying recurrence cases.

And the BRAVO-DX score can reliably identify groups of patients with significant differences in recurrence-free survival.

In order to further prove the clinical applicability of BRAVO-DX, the author continued to test its prognostic performance in an independent validation cohort, and defined a subset of BRAVO-DX, BRAVO-IMMUNE, which contains 12 tumor immune phenotype-related genes.

It is also defined that the smallest subset of BRAVO-DX contains only the three most significant risk and survival genes (IKZF3, AIM2 and ELF3).

Figure 4.
Development and characterization of prognostic genetic markers.
Next, the authors selected three breast cancer cohorts to evaluate the accuracy of the classifier (Figure 5).

The results showed that the overall survival rates of BRAVO-DX and BRAVO-IMMUNE were higher in all validated cohorts, and the three independent genes IKZF3, AIM2 and ELF3 were also significantly associated with recurrence-free survival.

Figure 5.
Verification of BRAVO-DX gene expression markers In order to test the application of BRAVO markers in the clinical field, the author used TaqMan-based real-time PCR technology, using a portion of tumor samples to verify the selected biomarkers (Figure 6).

It was found that the quantification of selected genes by real-time PCR was highly consistent with RNA-seq, and the TaqMan PCR assay well preserved the differences in the expression of risk genes and survival genes between relapsed and non-relapsed cases.

In order to test the accuracy of qPCR results in predicting recurrence, the Ct value was converted to logFPKM using the formula derived from linear regression and applied to the random forest model trained from RNA-seq data.

It turns out that the prediction results of the two platforms are highly consistent.

Figure 6.
Use TaqMan real-time PCR assay to verify the BRAVO-DX logo.
That’s all for this article.
The results of this study show that the phenotypic characteristics of BLBCs and their microenvironment are related to relapse-free survival, and prove that The utility of intrinsic and extrinsic phenotypes as independent prognostic biomarkers in BLBC.

Welcome to follow Shengxinren transcriptome | methylation | resequencing | single cell | m6A | multiomics cytoscape | limma | WGCNA | water bear legend | linux electrophoresis | PCR | a brief history of sequencing | karyotype | NIPT | basic experimental genes | 2019-nCoV | Enrichment Analysis | Joint Analysis | Microenvironmental Plague Pursuit | Summary of Ideas | Scholars | Scientific Research | Withdrawal | PhD Reading | Work