Overall, this study identified a prediction score system curated from a six-gene panel, and constructed a predictive model adopting clinicopathological characteristics for pCR rate of NCT in TNBC. Given the distinct response to systemic therapy, potential mechanisms were investigated and the promising signatures were identified.
The heterogeneity of molecular features in TNBC has been long investigated, and it is well recognized that this kind of divergence could result in distinct clinical profiles and survival outcomes [10, 11, 16]. With the consideration of molecular heterogeneities, we identified three distinct subtypes of TNBC with the receipt of NCT through consensus clustering method, and further investigated the genomic difference based on gene expression profiles. Then, enrichment analyses were carried out to comprehensively explore the cellular component, molecular functions, and biological process. Both the up-regulated genes and down-regulated genes were undergone analyses, respectively, of which the results revealed that the increased proactivity of mitochondrion organization, lymph vessel morphogenesis, and focal adhesion could attribute to this kind of molecular heterogeneity. The core functional group of genes was presented in PPI network, suggestive of the significant roles in the biological course.
To facilitate the introduction of a practical prediction model, we conducted analyses on the basis of random forest analysis and Lasso analysis to retrieve the intersected proportion for dimensionally reduction and features selection, which has been widely used for characteristics of cancer diagnosis and therapy [17–19]. After this group of genes identified, the respective prediction values for pCR was successively evaluated by logistic analysis, and the a six-gene panel comprising ATP4B, FBXO22, FCN2, RRP8, SMERK2, TET3 were finally recognized. With the application of mathematic formula, pRS system was quantitatively established with a decent value for pCR rate. To optimize this prediction method, the clinicopathological characteristics were integrated and assessed the predictive values to select the valuable multi-variables, which pRS and clinical stage of tumor size was ultimately determined.
With the selected gene panel and variables, the nomogram which was considered as an intuitive method was constructed to quantitatively assess the predicted pCR rate in TNBC [20]. Results from validation analyses was suggestive of the well performance of this model and the rationale of broad application in clinical practice. Several have managed to establish prediction models for pCR of NCT in TNBC which was estimated to perform well in clinical practice [21–23]. However, most of them basically focused on the characteristics from imaging or laboratory indexes. This nomogram took full consideration of the molecular heterogeneities detected from transcriptome information and clinicopathological characteristics to provide practical prediction of pCR rate for TNBC patients planned to NCT. Besides, we also discussed the potential mechanisms of benefit and non-benefit phonotypes in order to provide promising evidence for practice. Biological processes including apoptosis, hypoxia, mTORC1 signaling and myogenesis, and oncogenic features of EGFR and RAF were in proactivity to attribute to an inferior response. In fact, the potential triggers of distinct response to NCT remain undetermined, considering, these findings to some extent enlighten the quest for improvement of efficacy, which were in accordance with the previous studies [24–27]. Current trials have exerted efforts and assess the efficacy of this kind of targeted therapies in TNBC [28, 29], however, these results were controversial and remained to be updated through randomized controlled trials with a large sample cohort.
Indeed, there were some inevitable limitations of this study. Firstly, this was a retrospective analysis with the adoption of the identified datasets from publicly databases, and the heterogeneities among populations from different cohorts could not be removed. Secondly, a few characteristics, such as the Ki67 index and therapeutic protocols, could not be taken into considerations due to the lack of records in publicly available database, which could potentially weaken the prediction power of this model. Thirdly, the cutoff value of pRS was determined as the median which was practical yet less precise, which was necessary to be validated and optimized. Last, both experimental and clinical research are supposed to conducted and validate these findings obtained this study.