Archives

  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • br E G Receiver operating characteristic ROC curves

    2020-08-18


    (E–G) Receiver operating characteristic (ROC) curves showing sensitivity and specificity of INPP4B for ER status (E), CDK1 for tumor grade (F), and ERBB2 for HER2 status (G).
    Boxes are extended from the 25th to the 75th percentile, with a line at the median. The whiskers extend to the most extreme data point which is no more than 1.5 times the interquartile range (IQR) from the box. The individual points represent outliers or extreme values.
    present study using an iTRAQ-2DLC-MS/MS approach in an attempt to identify metastasis-associated CB-839 in low-grade breast cancer (Bouchal et al., 2015). In that study, we quantified 6% more proteins than in the current study (see also Figure S1); however, there, we were limited by significantly lower sample throughput, only allowing the analysis of pooled and not individ-ual samples in a reasonable time, resulting in inferior statistical power. Compared to the iTRAQ method used earlier by us (Bou-chal et al., 2015) and the Clinical Proteomic Tumor Analysis Con- 
    Figure 4. Independent Validation of INPP4B,
    CDK1, and ERBB2 Association with ER
    Status, Tumor Grade, and HER2 Status
    (D) The Cancer Genome Atlas (TCGA) RNA sequencing dataset of 1,078 patients (see Data S4E for dataset details): transcript levels were significantly different (p < 0.05) depending on ER status (for INPP4B) or HER2 status (for ERBB2); CDK1 was statistically significantly correlated with proliferation marker MKI67 (p < 0.05).
    Boxes are extended from the 25th to the 75th percentile, with a line at the median. The whiskers extend to the most extreme data point which is no more than 1.5 times the interquartile range (IQR) from the box. The individual points represent outliers or extreme values.
    sortium (Mertins et al., 2016), SWATH-MS has a better quantitative accuracy by avoiding the flattening of peptide ra-tios due to the use of the same iTRAQ reporter ions for quantification of co-iso-lated precursors. A recent study using SuperSILAC for the proteomic profiling of 40 breast cancer tissues (Tyanova et al., 2016) identified 10,138 endoge-nous proteins in total, but only a fraction of this number (2,588 proteins) was quantified across all samples (Figure S6). The study found a 19-protein signature discriminative for medium- and high-grade breast cancer subtypes, of which we consistently quantified 14 proteins in our SWATH-MS dataset of 96 patients. The abundance ranks of these 19 pro-teins in the two independent datasets (their 40 patients and our 96 patients)
    were highly similar (Table S1). Compared to the SuperSILAC approach, advantages of SWATH-MS are the lower cost and convenience of the label-free quantification but most importantly the consistent quantification of proteins across large sample sets (Figure S6). One of the gold-standard methods to profile proteins in clinical tissue samples is selected or multiple reaction monitoring (S/MRM). Of 319 breast-cancer-associated proteins quantified by S/MRM by Kennedy and colleagues (Kennedy et al., 2014), our SWATH-MS data cover 305 (96%). Similarly,
    Figure 5. Expression of ERBB2 Protein and Transcript in ER /HER2+ versus ER+/HER2+ Breast Cancer Tissues (A) Intensity of ERBB2 protein in SWATH-MS proteomics data (16 patients of grade 3; Data S1B).
    (B) Immunohistochemistry for ERBB2 in an independent set of patients (78 patients of grade 2+3; Data S1C).
    Boxes are extended from the 25th to the 75th percentile, with a line at the median. The whiskers extend to the most extreme data point which is no more than 1.5 times the interquartile range (IQR) from the box. The individual points represent outliers or extreme values.
    9 of 10 proteins associated with breast cancer biology (repre- correlated with S/MRM. In summary, our SWATH-MS-based sented by 16 of 17 peptides) were quantified by S/MRM in the strategy provided an advantageous combination of sample same set of tumors (Procha´zkova´ et al., 2017) as in our current throughput, quantitative precision (Vowinckel et al., 2013), SWATH-MS dataset with high level of correlation (Spearman and proteome coverage in a large sample set. Applying the correlation coefficients 0.439–0.880 and p values 1.1 3 10 5 to latest technical developments (e.g., ion mobility MS or faster 2.2 3 10 16; Data S5). This comparison well validates our Orbitrap-based instruments) may further improve the quantita- SWATH-MS data using an independent method on individual tive depth of SWATH-MS or similar data independent acquisi- tumor level. A strong correlation between SWATH-MS and tion-based studies. S/MRM was demonstrated already in the first SWATH-MS
    publication (Gillet et al., 2012) and confirmed in other indepen- Biological Relevance of the Key Proteins Selected by the dent studies (Kockmann et al., 2016; Liu et al., 2013; Naka- Decision Tree