Recent advances in mass spectrometry (MS) have enabled extensive analysis of cancer proteomes. Here, we employed quantitative proteomics to profile protein expression across 24 breast cancer patient-derived xenograft (PDX) models. Integrated proteogenomic analysis shows positive correlation between expression measurements from transcriptomic and proteomic analyses; further, gene expression-based intrinsic subtypes are largely re-capitulated using non-stromal protein markers. Proteogenomic analysis also validates a number of predicted genomic targets in multiple receptor tyrosine kinases. However, several protein/phosphoprotein events such as overexpression of AKT proteins and ARAF, BRAF, HSP90AB1 phosphosites are not readily explainable by genomic analysis, suggesting that druggable translational and/or post-translational regulatory events may be uniquely diagnosed by MS. Drug treatment experiments targeting HER2 and components of the PI3K pathway supported proteogenomic response predictions in seven xenograft models. Our study demonstrates that MS-based proteomics can identify therapeutic targets and highlights the potential of PDX drug response evaluation to annotate MS-based pathway activities.

Purpose: Response to a complex trastuzumab-based regimen is affected by multiple features of the tumor and its microenvironment. Developing a predictive algorithm is key to optimizing HER2-targeting therapy. Experimental Design: We analyzed 137 pretreatment tumors with mRNA-seq and DNA exome sequencing from CALGB 40601, a neoadjuvant phase III trial of paclitaxel plus trastuzumab with or without lapatinib in stage II to III HER2-positive breast cancer. We adopted an Elastic Net regularized regression approach that controls for covarying features within high-dimensional data. First, we applied 517 known gene expression signatures to develop an Elastic Net model to predict pCR, which we validated on 143 samples from four independent trials. Next, we performed integrative analyses incorporating clinicopathologic information with somatic mutation status, DNA copy number alterations (CNA), and gene signatures. Results: The Elastic Net model using only gene signatures predicted pCR in the validation sets (AUC = 0.76). Integrative analyses showed that models containing gene signatures, clinical features, and DNA information were better pCR predictors than models containing a single data type. Frequently selected variables from the multiplatform models included amplifications of chromosome 6p, TP53 mutation, HER2-enriched subtype, and immune signatures. Variables predicting resistance included Luminal/ER+ features. Conclusions: Models using RNA only, as well as integrated RNA and DNA models, can predict pCR with improved accuracy over clinical variables. Somatic DNA alterations (mutation, CNAs), tumor molecular subtype (HER2E, Luminal), and the microenvironment (immune cells) were independent predictors of response to trastuzumab and paclitaxel-based regimens. This highlights the complexity of predicting response in HER2-positive breast cancer. Clin Cancer Res; 24(21); 5292–304. ©2018 AACR.

Background: High-throughput transcriptomics has matured into a very well established and widely utilised research tool over the last two decades. Clinical datasets generated on a range of different platforms continue to be deposited in public repositories provide an ever-growing, valuable resource for reanalysis. Cost and tissue availability normally preclude processing samples across multiple technologies, making it challenging to directly evaluate performance and whether data from different platforms can be reliably compared or integrated. Methods: This study describes our experiences of nine new and established mRNA profiling techniques including Lexogen QuantSeq, Qiagen QiaSeq, BioSpyder TempO-Seq, Ion AmpliSeq, Nanostring, Affymetrix Clariom S or U133A, Illumina BeadChip and RNA-seq of formalin-fixed paraffin embedded (FFPE) and fresh frozen (FF) sequential patient-matched breast tumour samples. Results: The number of genes represented and reliability varied between the platforms, but overall all methods provided data which were largely comparable. Crucially we found that it is possible to integrate data for combined analyses across FFPE/FF and platforms using established batch correction methods as required to increase cohort sizes. However, some platforms appear to be better suited to FFPE samples, particularly archival material. Conclusions: Overall, we illustrate that technology selection is a balance between required resolution, sample quality, availability and cost.