Prediction of response to PD-1 blockade by immune cell profiling.
Abstract
Background: Immune checkpoint inhibitors revolutionized the therapy of several types of advanced cancers, and the gastric cancer is one of them. Expression of the PD-L1 in tumor cells and/or immune cells is believed to be the predictive marker, but the evidence is still very weak. Multi-parameter flow cytometry or mass cytometry using isotope enabled the characterization of immune cells.
Methods: Total 250 pathologic samples from the patients treated with anti-PD-1 antibodies for gastric carcinoma were retrospectively collected. We performed immune cell profiling using sequential immunohistochemistry (IHC) on FFPE tissue array slides. The IHC was processed using DAKO Link 480 and chromogenized with aminoethyl carbazole (AEC). After images were obtained with Aperio AT2, the slides were bleached in 20X saline sodium citrate. The stain-scan-bleach steps were repeated 10 times with multiple antibodies including CD3, CD8, CD4, PD1, PDL1, TIM3, LAG3, FoxP3 and cytokeratin. Images from each stained slides were aligned with CellProfiler program. Average 15,000 cells in each case were analyzed for the staining intensity of each staining and cell density of various immune cells were counted. For the identification of immune cell pattern of responders, t-SNE (t-distributed stochastic neighbor embedding) technique, one of the machine learning algorithm, was applied.
Results: Microsatellite instable (MSI) cases showed the best response on anti-PD-1 antibody drugs. In addition, density of immune cells as well as ratio of the immune cells are the strong predictive markers for PD-1 blockage therapy. MSI samples demonstrated the characteristic pattern of immune cell clusters, and the responders showed similar pattern of immune cell clustering in t-SNE clustering analysis.
Conclusion: The multiplex IHC method is useful for immune profiling, and machine learning algorithms are helpful tools for the analysis of big data from multiplex IHC images.
Tyrosine aminoacyl-tRNA synthetase sensitize breast cancer to the combined chemotherapeutic regimen
Abstract
Although approximately 70% of entire patients are currently receiving the chemotherapy regimen, pathologic complete response (pCR) rate is still low, ranging from 23% to 32.7%. Therefore, the need for a marker predictive of response to a particular cytotoxic regimen, especially before neoadjuvant chemotherapy, is becoming all the more necessary to optimize therapeutic efficacy and to avoid unnecessary complications caused by systemic therapy. An aminoacyl-tRNA synthetase (ARS), also called tRNA-ligase, is an enzyme that attaches the appropriate amino acid onto its tRNA. In humans, the 20 different types of aa-tRNA are made by the 20 different aminoacyl-tRNA synthetases. Certain diseases such as cancer and autoimmune disorders have been correlated to specific mutations of aminoacyl-tRNA synthetases. Here, we performed quantitative proteomics mass spectrometry in twenty paired FFPE biopsy breast cancer samples consist of non-responsive and responsive groups to chemotherapy. To define the best classifier to evaluate the predictive power of signatures, we employed four different machine learning algorithms and performed repeated cross-validation on the training set to classify samples between pCR and non-pCR groups. The candidate proteins selected from the machine-learning algorithms were subsequently validated by immunohistochemistry of 123 cases of independent needle biopsy FFPE samples which obtained before chemotherapy. 10 human breast cancer cell lines enrolled and verified biological functions for protein candidates through molecular biology-driven assays, including RNAi, Glo-cell titer assay, IF, cytometry and the 3D tumor spheroid-based function assays for target validation. The MS analysis of FFPE set yielded 6,069 protein groups. The filtered dataset was subjected to statistically analysis using Student’s t test (p value < 0.05), which resulted in 539 proteins with differential abundances. To identify the most meaningful changes in two conditions, the volcano plot was used. 13 proteins including YARS, TBC1D10C, SMCHD1, WARS, IGKV3-15, HTATIP2, KIAA1522, MAOB, MAPT, BSPRY, CAP2, ABAT and NAT1 were prioritized. We searched for biological process in the Gene Ontology (GO) enrichment analysis in each proteomic cluster. Several immune responses process, apoptotic process, DNA replication process and aminoacylation for protein translation process primarily were represented in group with complete remission. On the other hand, cell adhesion process, cytoskeleton organization process, vesicle organization process and Golgi organization process overrepresented in breast cancer which showed poor responses to the therapy. The machine learning approaches using Random Forest algorithm demonstrated the highest AUC value, 0.978 (sensitivity 1.0 and specificity 0.714) with a combination of 11 proteins including STUB1, PDCD6, YARS, MAOB, PDCD4, NA, FLYWCH2, ABAT, FAM162A and WARS (Figure00, Table 00). For the accuracy prediction, four different algorithms demonstrated evenly high accuracy rates from 0.85 to 0.95 with a combination of STUB1, PDCD6, YARS, MAOB, WARS, RHBDF1 or KIAA1522. the selected seven candidates (KIAA1522, RHBDF1, WARS, YARS, MAOB, STUB1 and PDCD6) evaluated in the subsequent steps of verification using immunohistochemistry (IHC) in 123 patient cohorts. The predictive relevance of individual proteins, YARS and RHBDF1 to distinguish CR from nCR was AUC of 0.605 and 0.630 for all cases and approximately 10.2% higher AUC in luminal breast cancer (AUC=0.749 and 0.717). The overexpression of YARS induced chemotherapeutic sensitivity in hormone receptor positive breast cancer cell lines such as T47D, which was confirmed with multiple molecular biology-driven assays. We confirmed, for the first time, that he ARS-related proteomic marker is critical in chemotherapeutic responses in breast cancer.
Development of Proteomic Multi-Marker Panel Assay for HER2 protein Using Multiple Reaction Monitoring-Mass Spectrometry with Breast Cancer FFPE Tissue Specimens
Abstract
Advances in targeted medications has greatly improved the survival rate of breast cancer patients with molecular marker-positive tumors. Human epidermal growth factor receptor (HER2) is a prominent marker that has been extensively studied in recent years. To date, immunohistochemistry (IHC)-based grading of FFPE tumor tissue slides has remained as the gold-standard for quantifying these proteins. Despite its use as a standard, IHC-based grading is highly subjective as the results are predicated upon a trained individual’s eye rather than numerical quantities. Thus, alternative methods that can account for quantitative levels of molecular markers are gaining popularity – including targeted proteomics using mass spectrometry (MS), which has proven its accuracy and sensitivity. However, technical limitations have impeded the application of MS-based protein quantification to microgram-levels of FFPE tissue. To challenge these difficulties, we performed a proof-of-concept study using multiple reaction monitoring-mass spectrometry (MRM-MS) to investigate its potential for application in pathological examinations. As a result, the MRM-MS method could quantify several peptides of HER2 simultaneously in FFPE tissue slides. The results of multiplexed and targeted proteomics showed high correlation with the conventional IHC readings of experienced pathologists and even differentiated the HER2(++) FISH+ from HER2(++) FISH- with better details. Unlike immunostaining-based approach, MRM-MS has a distinct advantage as it can quantify multiple protein markers in a single run. This study does not only support the plausibility of MRM-MS approaches in FFPE tissue slides, but also the MS-based approaches in precision medicine.
Beijing Institute of Lifeomics, National Center for Protein Sciences –Beijing
Title
In-depth Serum Proteomics Reveals Biomarkers of Psoriasis Severity and Response to Traditional Chinese Medicine
Abstract
Serum and plasma contain abundant biological information that reflect the body’s physiological and pathological conditions and is therefore a valuable sample type for disease biomarkers. However, comprehensive profiling of the serological proteome is challenging due to the wide range of protein concentrations in serum. To address this challenge, we developed a novel in-depth serum proteomics platform capable of analyzing the serum proteome across ~10 orders or magnitude by combining data obtained from Data Independent Acquisition Mass Spectrometry (DIA-MS) and customizable antibody microarrays. Using psoriasis as a proof-of-concept disease model, we screened 50 serum proteomes from healthy controls and psoriasis patients before and after treatment with Traditional Chinese medicine (YinXieLing) with our DIA-MS and antibody microarray platform. We identified 106 differentially-expressed proteins in psoriasis patients involved in psoriasis-relevant biological processes, such as blood coagulation, inflammation, apoptosis and angiogenesis signaling pathways. In addition, unbiased clustering and principle component analysis revealed 58 proteins discriminating healthy volunteers from psoriasis patients and 12 proteins distinguishing responders from non-responders to YinXieLing. To further demonstrate the clinical utility of our platform, we performed correlation analyses between serum proteomes and psoriasis activity and found a positive association between the psoriasis area and severity index (PASI) score with three serum proteins (PI3, CCL22, IL-12B). Taken together, these results demonstrate the clinical utility of our in-depth serum proteomics platform to identify specific diagnostic and predictive biomarkers of psoriasis and other immune-mediated diseases.
Room 404, Nakseongdae R&D Center, 38 Nakseongdae-ro, Gwanak-gu, Seoul, 08790, Korea
Phone: (+82)2–393–8328 E-mail : admin@khupo.org Homepage : http://www.khupo.org
COPYRIGHT ⓒ 2019 KHUPO. ALL RIGHTS RESERVED.
