A novel proteomics-based clinical diagnostics technology identifies heterogeneity in activated signaling pathways in gastric cancers

Abstract

Purpose: The aim of this study was to utilize the proteomics-based Collaborative Enzyme Enhanced Reactive (CEER) immunoassay to investigate protein tyrosine phosphorylations as diagnostic markers in gastric cancers (GCs).

Experimental design: Protein lysates from fresh-frozen 434 advanced stage GCs were analyzed for phosphorylation of HER1, HER2, p95HER2, HER3, cMET, IGF1R and PI3K. The pathway activation patterns were segregated based on the tumor HER2 status. Hierarchical clustering was utilized to determine pathway coactivations in GCs. Prognostic value of pathway activation patterns was determined by correlating disease-free survival times of the various GC subgroups using Kaplan-Meier survival analysis. CEER was also used to determine the presence of tyrosine phosphorylated signaling cascades in circulating tumor cells (CTCs) and ascites tumor cells (ATCs).

Results: Utilizing a novel diagnostics immunoassay, CEER, we demonstrate the presence of p95HER2 and concomitantly activated signaling pathways in GC tumor tissues, CTCs and ATCs isolated from GC patients for the first time. p95HER2 is expressed in ~77% of HER2(+) GCs. Approximately 54% of GCs have an activated HER1, HER2, HER3, cMET or IGF1R and demonstrate a poorer prognosis than those where these receptor tyrosine kinases (RTKs) are not activated. Hierarchical clustering of RTKs reveals co-clustering of phosphorylated HER1:cMET, HER2:HER3 and IGF1R-PI3K. Coactivation of HER1 with cMET renders GCs with a shorter disease-free survival as compared to only cMET activated GCs.

Conclusions: Our study highlights the utility of a novel companion diagnostics technology, CEER that has strong implications for drug development and therapeutic monitoring. CEER is used to provide an increased understanding of activated signaling pathways in advanced GCs that can significantly improve their clinical management through accurate patient selection for targeted therapeutics.

Figure 1. HER2 and p95HER2 expression in gastric cancers.

(A) CU distribution for HER2 expression of GC samples at 0.25 µg lysate. The x-axis represents the IHC/FISH status, and the y-axis represents the CU values from CEER assay as determined from a BT474 standard curve. Separation is illustrated between the two groups with a median of 0 for the IHC/FISH negative population (384 of 434) compared to a median of 11 for the IHC/FISH positive population (50 of 434). One saturated sample, above the limit of quantitation and indicated as ‘Number saturated’ in the corresponding table, is not shown. Boxes represent the interquartile range, with the 75th percentile at the top and the 25th percentile at the bottom. The line in the middle of the box represents the median. Whiskers extend to the highest and lowest value within 1.5 times the interquartile range. P value<.001 was determined by Wilcoxon signed-rank test. (B) CU distribution for p95HER2 in a subset of the tumor samples (58 of 434) at 20 µg lysate. The x-axis represents the IHC status, and the y-axis represents the CU values from CEER assay for p95. Full length HER2 was removed by immuno-depletion prior to the assay. The data points are colored based on the HER2 status by IHC and FISH. As shown, one data point with an IHC of 2 was determined to be positive by FISH analysis. Of the 34 samples determined to be positive for p95 by CEER, 24 (71%) of them were HER2(+) by IHC/FISH. p95HER2 expression could not be determined in one sample which is indicated as ‘Number NA’ in the corresponding table.

Figure 2. Profiling of phosphorylated markers in gastric cancers.

(A) Representative immuno-array images for pathway profiling of indicated signal transduction proteins. Array signal intensity ranges from black/dark blue (low) to red/white (high/saturation). (B) Heat map and hierarchical clustering of the 434 samples based on CU values from CEER assay for phosphorylated markers measured at 10 µ g lysate concentration. Each column represents a marker and each row represents a patient sample. Relative levels of phosphorylation are depicted with a color scale where red represents the highest level of activation and green represents the lowest level. The CU values for each marker (column) were ranked by deciles. Jitter, between 0 and 0.1, was added to each biomarker CU value to create equally sized bins. Row and column dendrogram show the result of the hierarchical clustering calculation.

Figure 3. Disease-free survival differences in gastric cancers based on activated RTK profiling.

(A & B) Disease free survival differences in GC samples of tumor stages II+III. The analysis compared the overall GC sample set (A) or only HER2(−) GC sample set (B) between cohorts with 0 RTK activation vs ≥1 RTK activation. Median survival of the two cohorts in all patients is 32.63 months (≥1 RTK activation) and 76.53 months (0 RTK activation) and in patients with HER2(−) GC is 30.10 months (≥1 RTK activation) and 68.13 months (0 RTK activation). (C & D) Disease free survival differences comparing HER1(−) cMET(+) vs HER1(+) cMET(+) cohorts in HER2(−) (C) or Stage I, HER2(−) (D) GC samples. Median survival of the two cohorts in HER2(−) patients is 46.17 months (HER1(+) cMET(+)) and 82.80 months (HER1(−) cMET(+)). Median survival times in patients with HER2(−) stage I GC are undefined for both cohorts. Sample numbers in each cohort, p-values and hazard ratios are indicated.

Figure 4. Profiling of phosphorylated markers in CTCs and ATCs from gastric cancer patients.

RTK and downstream pathway profiling in ATCs isolated from 3 patients after ligand (EGF, Heregulin, HGF and IGF) stimulation with or without 2 µM inhibitor cocktail (lapatinib and PHA-665,752) or DMSO. Relative CU is defined as the ratio of CUs over baseline (no ligand or drug treatment).