5. Current State and Future Perspectives of RPPA
Of major value for RPPA experimentation is a thorough knowledge regarding protein signaling networks and antibody specificity as this is the key for successful RPPA experimentation. Unfortunately, the majority of antibody vendors leaves antibody validation to customers and provides limited information about suitable controls. Characterizing and handling large numbers of antibodies therefore constitutes a huge workload for RPPA labs, especially since re-testing has to be carried out with the purchase of each new antibody lot. A major concern is that antibody providers might not be able to maintain their quality in the long run as this might be compromised for the sake of improved financial return.
Different printing and signal detection approaches as well as data analysis tools have evolved since the introduction of RPPA as a highly sensitive large-scale platform for targeted proteomics [15]. The RPPA field would definitely benefit from a comparison of different RPPA platforms to evaluate that the different strategies taken for RPPA return the same results. Such efforts are currently ongoing and its publication will be of value for the field. The question of whether scientific findings such as biomarker signatures can be reproduced on different RPPA platforms or even on other technical platforms, e.g., by mass spectrometry, presents also a highly interesting question. Reviewing the current literature returned encouraging examples.
The analysis of hormone receptor-positive breast cancer employed RPPA to compare proteomic signatures of low- and high-grade breast cancer. The identified biomarker signature comprised caveolin-1, NDKA, RPS6, and Ki-67 and could determine the recurrence risk in patients with ER+ breast cancer implying that it could potentially also be applied to predict a need for chemotherapy [7,42]. Moreover, the clinically relevant and biologically insightful observation that caveolin-1 expression of tumor-associated fibroblasts inversely correlates with clinical outcomes in patients with ER+ breast cancer was already made years ago by using IHC [43]. The observation that also stroma proteins might be identified as part of a biomarker signature supports the idea of understanding cancer as a “tumor organ” with stroma cells as integral partners of the tumor that sustain tumor growth. The prognostic relevance of tumor stroma proteins with respect to grading and patient outcome was also identified by a LC-MS/MS-based analysis of breast cancer specimens. Likewise, the mass spectrometry approach identified different sets of proteins and showed that structural proteins of the tumor stroma are highly abundant in low-grade tumors and additionally identified proteins directly or indirectly associated with transforming growth factor β (TGFβ1) signaling as upregulated in the tumor stroma [44].
RPPA-based profiling of breast cancer specimens including all known histological subtypes, including also HER2+ breast cancer and triple negative breast cancer (TNBC) in addition to ER+ breast cancer samples, identified a small subset of breast cancer patients with relatively high levels of pHER2(Y1248) but without overexpression of HER2. This therapeutically relevant finding showed co-expression of HER2 along with either EGFR or HER3, and hints towards a subgroup of breast cancer patients that can potentially benefit from therapies targeting the full EGFR/ERBB module [45].
Hennessy et al. [3] reported six independent groups with distinct proteomic profiles and characteristic overall survival after analyzing 128 breast tumors using 82 antibodies recognizing cancer-relevant proteins and phosphoproteins. Two subgroups showed high level expression of stroma proteins such as collagen VI and caveolin-1, of which collagen VI was also part of a proteomic signature to separate luminal A vs. luminal B tumors. The prognostic relevance of stroma proteins emerged also in the analysis of the TCGA work that reported fibronectin, collagen VI, and caveolin-1 as part of stroma signatures named reactive I and reactive II [46,47]. One of the stroma subclusters comprised exclusively luminal A tumors, the second group consisted of luminal A and luminal B tumor, relying on the transcript-based PAM50 score as reference point. Reactive I and reactive II groups did not differ with respect to tumor cell content, and a supervised analysis did not indicate differences with respect to genetic aberrations, e.g., mutation, DNA copy number or methylation.
These findings illustrate the potential of RPPA for biomarker discovery and place it next to established approaches such as IHC and mass spectrometry [48,49]. A direct comparison of RPPA in terms of assay sensitivity with other targeted immunoassay approaches such as the more recently introduced ABCD-technique [50] might be of interest for the field of immunoassays. In this sense, the high sample capacity and high accuracy of RPPA make this platform promising for the validation and quantification of target proteins identified by mass spectrometry, for example. Ideally, these approaches work hand in hand by exploiting the potential of mass spectrometry for de novo discoveries and the capacity of IHC to identify the subcellular localization of the protein of interest. Moreover, RPPA appears to be extremely valuable to advance personalized medicine where information on the abundance of targetable receptors, for example, is required. Numerous drugs target receptor tyrosine kinases to date, for example, and the majority of these drug targets can be quantified in clinical tissues by RPPA. Thus, using RPPA for tumor profiling seems to be highly promising also for cancer types not as well characterized as breast cancer. This will open up new therapeutic strategies for patients suffering from rare malignancies or metastatic disease who then may soon have new opportunities to benefit from targeted therapies.