Background
Head and neck cancer affects 740,000 new patients each year, with approximately 300,000 deaths worldwide [
1,
2], of which more than 90% are histologically squamous (HNSCC). Despite significant advances in therapy, only 30–50% of patients with locally advanced disease survive more than 5 years, and this has not changed over the past 40 years [
3]. In this context, concurrent chemoradiation therapy (CRT), with or without induction chemotherapy (IC), has emerged as the new paradigm of treatment for patients with locally advanced HNSCC affecting the oropharynx, larynx and hypopharynx [
4]. While the objective of this strategy is to preserve critical functions (airway, speech, swallowing, body image, etc.), a proportion of patients fails treatment and requires salvage surgery. Noteworthy, ineffective pre-treatment results in unwanted delay of curative therapy and in addition, surgery is made challenging in a heavily pre-treated patient and in a radiated field [
5]. It is generally believed that personalized treatment decisions based in non-invasive biomarkers able to stratify patients according to treatment response would help to optimize appropriate patient-specific therapeutic interventions, quality of life and outcomes.
The discovery of extracellular vesicles (EVs) could circumvent these issues and allow analyses of a much ‘cleaner’ plasma compartment that may better reflect the tumor cells or its microenvironment. EVs are nanosize bi-lipid membrane vesicles carrying various cellular components (proteins, lipids, nucleic acids, and sugars) and released by cells into the extracellular space [
6,
7]. The EVs obtained from plasma of cancer patients are enriched in proteins/glycoproteins expressed on cell membranes and/or in the cytosol of the parent tumor cells, and in many immunosuppressive molecules [
8]. The EVs are involved in the transmission of biological signals between different populations of cells [
9].
There are emerging evidences that cancer-derived EVs play a crucial role in the reprogramming of both local microenvironment and distant sites, contributing to tumor capacity for immune evasion, growth, invasion and metastatic spread [
10,
11]. Furthermore, since they are secreted into the intercellular space, EVs can be detected in a variety of biological fluids, including saliva, cerebrospinal fluid, urine and serum/plasma. The potential application of EVs for diagnosis and guiding therapeutics, as well as determining prognosis of pathological conditions has allowed the field of EV-studies to grow steadily in recent years [
12]. Of note, HNSCC are strong EVs producers and the plasma of patients with HNSCC is highly enriched in EVs [
13,
14].
Based on the potential role of tumor-derived EVs as mediators of tumorigenesis, others and we have reasoned that the molecular content of plasma EVs makes them potential biomarker sources [
15‐
17]. Therefore, here we propose the use of antibody arrays to identify and quantify proteins carried by EVs circulating in the plasma of HNSCC patients that do not respond to CRT (NR, Non-Responders), and compared this profile with patients that present a complete response to this treatment (CR, Complete-Responders).
So, in this study, we identified several plasma biomarker candidates present in plasma compartments with the potential to predict beforehand which patients could take benefit from CRT and to guide therapeutic decisions in this context.
Discussion
Over the last years, management and prognosis of patients with HNSCC has improved through multimodality treatment protocols, which includes surgery, radiotherapy and chemotherapy, but long-term survival rate is still around 50% [
3]. Thus, to improve the prognostic of HNSCC patients by selecting the most appropriate curative therapies, this study aimed to identify novel non-invasive candidate biomarkers in plasma EVs that could segregate locally advanced HNSCC patients who respond to chemoradiation therapy from those who do not
.
Different approaches have been used to identify prognostic factors in HNSCC. Many are based on analyses of the primary tumor tissue itself. However, a few studies have focused on plasma for discovery of prognostic biomarkers in HNSCC. Le et al. (2003) [
22] reported that serum Osteopontin levels correlated with tumor hypoxia, and suggested that this protein could be used to identify patients at high risk for tumor recurrence. Similarly, Gehrmann et al. (2014) [
23] showed that serum Hsp70 levels in the HNSCC patients might be useful for tumor detection, and also for monitoring therapeutic response to radiation therapy. In other studies, plasma levels of LCN2/matrix metallopeptidase 9 complex, MMP2, TIMP1, TIMP2 and TIMP3 were correlated to tumor size, lymph node involvement, tumor differentiation and prediction of tumor stage and T status in patients affected with HNSCC [
24‐
26]. However, none of these markers are being used as clinical predictors for outcome.
Antibody arrays are screening assays for the presence and absence of an antigen based on the highly specific recognition between an antibody and its target antigen. Several studies have reported the use of antibody arrays in cancer biomarker discovery studies to discover biomarkers that are potentially valuable for diagnosis, prognosis or treatment response evaluation. Sill et al. (2016) [
27] used this approach to compare gastric cancer cells to adjacent normal tissue and they found 17 proteins with high specificity for gastric adenocarcinoma. Another antibody array study revealed that cyclin D2, cytokeratin 18, cyclin B1, hnRNP m3-m4 and the monophosphorylated ERK were decreased in a doxorubicin resistant breast cancer cell line in comparison to a sensitive one [
28]. Sreekumar et al. (2001) [
29] also used antibody array assays to identify up-regulated apoptotic factors in colon carcinoma treated with ionizing radiation.
Utilization of peripheral blood for the discovery of tumor biomarkers is a very practical approach because it is minimally invasive, inexpensive, highly reproducible, and it could mitigate the intra-temporal heterogeneity-sampling problem [
30]. Zupancic et al. (2014) [
31] analyzed non-depleted plasma samples of healthy volunteers in comparison to glioblastoma multiform (GBM) patients using antibody arrays and identified 11 plasma proteins as biomarker candidates for the diagnosis and prognosis of patients with GBM. Although plasma and serum are reliable sources for cancer biomarker discovery, its usage has several limitations as previously discussed [
20]. One limitation is the presence of high-abundance proteins. Although plasma and serum have more than 10,000 proteins, > 99% of the serum protein mass is dominated by 21 proteins [
32,
33]. Furthermore, depletion of these high abundance proteins to uncover low-abundance biomarkers might introduce artifacts and bias to proteins quantification [
34]. The discovery of extracellular vesicles (EVs), carrying proteins and nucleic acids in various body fluids, such as plasma, is an alternative method to circumvent some of these limitations. In a previous study, we shown evidences that it is possible to isolate at least two distinct populations of EVs in plasma, according to their affinities for CTB or AV ligands, and these EVs might have different origins and molecular contents. Moreover, this method is highly specific for the isolation of phospholipid membrane vesicles with minimal contamination of large non-vesicular biological complexes or high abundant plasma proteins [
20]. Importantly, these methods are able to capture EVs from relatively small plasma volumes and even from freeze-thawed plasma samples, with the potential for use either in large longitudinal studies or routine clinical use.
We therefore performed an antibody array analysis of CTB- and AV-EVs in a pool of plasma samples from HNSCC patients (responders and non-responders to CRT). This analysis revealed that EVs derived from NR patients carry previously reported tumor biomarker candidates. Patel et al. (2014) [
35] showed that inhibition of HSP90 potentiated cell death in HNSCC cells induced by cisplatin and radiotherapy. The increased percentage of circulating CD4+ cells was found to predict response to induction chemotherapy in advanced laryngeal cancer [
36]. Notwithstanding, it is well known that gene repair is one of the most relevant mechanisms of resistance to anti-cancer drugs and two proteins related to DNA repair, PRIM1 and POLB, were detected in both EV subpopulations (CTB and AV) in NR patients. Previously reported tumor biomarker candidates were also detected in the cargo of EVs circulating in the plasma of CR patients. Liu et al. (2014) [
37] proposed that TP73, a protein involved in the signaling pathway of DNA damage, is a potential target for miRNAs to induce cisplatin resistance in ovarian cancer. It was also reported that BRCA2-negative ovarian cancer cells are more sensitive to cisplatin treatment [
38]. The EGFR was found to be overexpressed in 90% of HNSCC patients and its high levels in the tumor tissue has been strongly correlated with worst clinical outcome [
39]. In addition, the loss of STAT1 has been related to the tumorigenesis of several cancer types, and is implicated as a tumor suppressor in esophageal squamous cell carcinoma [
40]. A significant correlation between the ratios of BAX/ BCL2 was reported to segregate radiosensitive versus radio-resistant breast cancer patients [
41]. SNAI1, which is a master regulator of epithelial-to-mesenchymal transition, has shown to be overexpressed in several cancer cells and is often related to prognosis and aggressiveness [
42]. The suppression of IKBKB through miR-16 sensitizes breast cancer cells to paclitaxel treatment [
43].
In the present study, we identified a set of 119 proteins that were specific to the patients that do not respond to CRT. A fraction of proteins detected in the EVs from these patients were associated with Insulin/ MAP kinase and PI3 kinase signaling, and angiogenesis. It is noteworthy that angiogenesis plays a critical role in HNSCC progression and high plasma levels of the angiogenic factors, VEGF and EFGR, have been related to the worst prognosis of HNSCC patients [
44,
45]. On the other hand, 38 different proteins were found specific to the HNSCC patients presenting a complete response to CRT. Among the most significant pathways associated with these proteins, were FAS, p53 and apoptosis cell death signaling, which are critical pathways in HNSCC patients associated with response to chemoradiation therapy [
14,
46‐
48]. These pathway analyses suggest that the content of circulating plasma EVs could have relevant functions in the treatment response of HNSCC patients.
Acknowledgements
DMR-Jr acknowledges the support of the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, 99999.007922/2014-00) and of the São Paulo Research Foundation (FAPESP, 2015/21420-3). ALV and ALC had a Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPq scholarship (300936/2015-0 and 561217/2010-6, respectively). NGI acknowledges the National Medical Research Council of Singapore Clinician Scientist Award (NMRC/CSA-INV/011/2016).