3. Discussion
In this study, 55 TSCC/BOTSCC biopsies and corresponding normal samples were analyzed for the expression of 155 cancer and immune related proteins, and differences in protein expression between tumor and normal tissue as well as between HPV-positive and HPV-negative TSCC/BOTSCC, were identified. Furthermore, proteins potentially related to clinical outcome, for HPV-positive and HPV-negative tumors, respectively, were identified. One of these proteins, VEGFA, was validated and was shown to be related to disease-free survival for HPV-positive TSCC/BOTSCC.
The proteins included in the Olink panels utilized in the present study, had been selected as directly or indirectly cancer related and involved in processes such as angiogenesis, cell-cell signaling, cell-cycle control, tumor-immunity, chemotaxis, apoptosis, and cell killing (
www.olink.com/products). Therefore, it was not surprising that many differences in protein expression between tumor and normal tissue were found. That fewer (78/155) proteins differed in expression between HPV-negative tumors as compared to HPV-positive tumors (111/155) vs. their corresponding normal samples was likely a combined effect of a smaller cohort (i.e., lower statistical power) and/or a less active immune defense in the former.
The most upregulated proteins in the tumor tissue were mainly chemokines and cytokines, indicating an increased activity of the immune response, as especially shown for chemokines CXCL9, 10, 11, 13, and CCL3, 4 and 20, and cytokines IL-6 and 8. A recent analysis of 20 different cytokines and chemokines (some also included here) in culture supernatants harvested from HNSCC tumor tissue-derived cell suspensions, showed, in accordance with the present study, high tumor related levels of CXCL9, CXCL10, and CCL20 and decreased levels of CCL5 [
23]. In parallel to this study, higher expression of CXCL9 and CXCL10 was noted in HPV-positive tumors than in HPV-negative tumors. Furthermore, in a study on melanoma a 12-chemokine signature that was related to immune infiltrates included high expression of CXCL9, 10, 11, 13, and CCL 3 and 4, similarly to TSCC/BOTSCC in the present study [
24].
Proteins with a distinct lower expression in HPV-positive TSCC/BOTSCC vs. normal samples included kallikreins KLK13, HK8, and HK11. Kallikreins have mostly been investigated as serum markers in prostate cancer, but also in some other cancer types e.g., lung cancer where HK8 has been found to be related to favorable outcome and shown to suppress tumor invasiveness in lung cancer, while little is known about their expression in HNSCC and TSCC/BOTSCC [
25].
Proteins with higher expression in HPV-positive vs. HPV-negative TSCC/BOTSCC were mainly immune related. Several were surface proteins expressed on immune cells, e.g., CD8A, and PD-1, or on cancer cells and affecting immune activity, e.g., PD-L1, FASL, and notably PD-1 and its ligand PD-L1 were similarly upregulated. A higher CD8A expression was expected, since increased numbers of CD8+, FoxP3+, and CD4+ TILs in HPV-positive vs. HPV-negative OPSCC have been noted earlier, where in particular a higher number of CD8+TILs is linked to a better clinical outcome, irrespective of the HPV status of the tumor [
19,
21,
22].
PD-1 and PD-L1 exhibited a higher expression in HPV-positive when compared to HPV-negative tumors in this study. Interaction of PD-1, expressed on the surface of activated T-cells, B-cells, and macrophages, and PD-L1, expressed on immune and cancer cells can suppress the activity of CD8+ T-cell mediated immune response, and have therefore received attention for immunotherapy [
26,
27]. Recent studies investigating PD-L1 and PD-1 expression in OPSCC gave partly contradictory results, with two studies showing higher PD-L1 mRNA levels in HPV-positive cancer, while the third study reported the opposite result [
19,
23,
28]. Notably, since PD-L1 is expressed on both cancer and tumor infiltrating immune cells, this study, similarly to global mRNA analysis, does not differentiate the expression between these cell types.
In parallel with the higher immune infiltration in HPV-positive than in HPV-negative tumors, some cytokines and chemokines e.g., IL12, CCL4, CCL20, CXCL10, CXCL17 had higher expression in the former, with IL12, notably being an important regulator of T-cell and NK-cells cytotoxicity, and its receptor, IL12R1-β1, was also upregulated in the HPV-positive tumors.
Finally, Mucin-16 (also known as MUC-16), a transmembrane protein that functions as a barrier to bacterial infections, protects cancer cells from being killed by immune cells and contains the CA125 peptide [
29], demonstrated a nearly 5-fold difference between HPV-positive and HPV-negative tumors. Mucin-16/CA125 has especially been studied in ovarian cancer where it is used both as a diagnostic and predictive marker and to evaluate the response to therapy in this tumor type [
29]. The reason behind the increased expression in HPV-positive vs. HPV-negative TSCC/BOTSCC is yet unclear and needs further investigation.
Few other studies have analyzed a large number of proteins simultaneously in OPSCC in relation to tumor HPV status and/or clinical outcome. Proteomic profiling of OPSCC by mass spectrometry disclosed e.g., enrichment of E2F1 and E2F4 in HPV-positive OPSCC, while reverse-phase protein array profiling revealed differences in e.g., P13K/AKT/mTOR and receptor kinase pathways [
30,
31].
Here, most HPV-positive tumors separated into clusters, mainly due to differences in the expression of immune related proteins on the cell surface of tumor infiltrating immune cells, e.g., PD-1, FasL, NCR1, and KLRD1 or soluble cytokines/chemokines e.g., IL12 and CCL4 (
Figure 1B). High expression of FasL, PD-1, and KLRD1 indicates a high infiltration of several parts of the immune defense with PD-1, mainly expressed on T-, B-cells and macrophages, and NCR1 and KLRD1 mainly expressed on NK-cells [
32,
33]. Notably, CD8A expression did not differ between cluster II and III indicating no major differences in numbers of CD8+ T-cells between these clusters, in contrast to that observed between HPV-positive and HPV-negative tumors. Similarly, several of the chemokines differing in the expression between HPV-positive and HPV-negative tumors do not differ between these clusters either, e.g., CXCL9 and 10.
Global mRNA expression between HPV-positive and HPV-negative OPSCC has also shown differences in proteins not included in the present study e.g., CDKN2A, NF-KB, and STAT3, making comparisons difficult [
34,
35,
36]. Notably, in one study HPV-positive OPSCC was split into two subtypes, a Classical subtype (CL)-HPV and an Inflamed/mesenchymal subtype (IMS)-HPV, where the latter was characterized by higher expression of e.g., immune response genes, such as CD8A, ICOS, LAG3, and HLA-DRA, related to CD8 T-cell infiltration [
36]. Here, out of these, only CD8A was analyzed and it did not present differences between the two major clusters although HPV-positive tumors in a third cluster (I), dominated by HPV-negative tumors, presented a low amount of CD8A expression.
In this study, some of the proteins with the most pronounced potential correlation to clinical outcome in HPV-positive TSCC/BOTSCC were at least partly related to angiogenesis e.g., DLL1, ESM1 (endocan), VEGFA, CYR61, and PlGF. Higher VEGFA expression in the tumor cells was confirmed by IHC to be associated with poorer DSF. High VEGFA expression, especially in combination with high EGFR expression has earlier been linked to local recurrence in TSCC in one study, and likewise in oral and laryngeal cancer, but not in a third study, including HPV-positive and HPV-negative OPSCC [
37,
38,
39,
40]. Notably, VEGFA induces the expression of ESM1, which is a mediator of the angiogenic effect of VEGFA [
41]. Moreover, ESM1 has been related to shorter survival in e.g., breast, liver, and nasopharyngeal cancer, but has not been investigated in OPSCC [
42,
43].
Angiogenesis is closely linked to hypoxia, and VEGFA together with CA9, are also markers for hypoxia, both being regulated by HIF1α and are involved in the induction of members of the Notch-pathway where DLL1 functions as a ligand for Notch1 [
44]. Also, PlGF, which is a ligand for VEGFA, is induced by hypoxia [
45].
Noteworthy, the IGF receptor IGF1R also demonstrated a higher expression in tumors that recurred. IGF1R is potentially targetable for therapy and can inhibit the effect of anti-EGFR therapy by acting on the same downstream pathway [
46,
47]. Since current TSCC and BOTSCC therapy often includes EGFR inhibitors, there is a potential need to combine such treatment with anti-IGF1R therapy. IGF1R has also previously been linked to poor prognosis in HPV-negative OPSCC [
48].
There is a need for new targets for therapy of TSCC and BOTSCC irrespective of HPV-status. Recently, earlier treatment that entailed conventional radiotherapy and/or surgery has been intensified by increased radiotherapy, chemoradiotherapy, and/or EGFR-inhibitors. Intensified treatment leaves the patients with more side effects, but has not improved survival for patients with HPV-positive cancer in Stockholm, Sweden [
49]. Treatment for recurrent TSCC and BOTSCC is a real challenge and the results are poor. Proteins related to recurrence in the present study, such as e.g., IGF1R and TNFRSF19/Troy, may possibly be utilized as such targets.
Protein expression was also analyzed in relation to tumor T-stage and major differences related to T-stage were found for some proteins in HPV-negative tumors, but not for HPV-positive tumors. Although T-stage has been shown to be related to clinical outcome [
49] the proteins with the strongest relation to T-stage for HPV-negative tumors, MMP7, CSCL17, and CEACAM5, were not found to be related to survival. There are limitations in the present study. Only 59 tumors were included, and when HPV-positive and especially HPV-negative tumors were analyzed separately or compared, random correlations might have been obtained. In addition, both the treatment and T-stage of the tumors included in the cohort were heterogeneous, as presented in
Table 1. For proteins related to clinical outcome these issues may have had an effect on the results obtained, especially given the few events. Thus, the result presented here must be interpreted with caution.
In summary, when comparing protein expression in TSCC/BOTSCC with normal tissue, the most prominent differences were found for chemo-and cytokines. Between HPV-positive vs. HPV-negative tumors, most of the differences were detected in the immune related proteins as well as cyto-and chemokines. Some proteins were tentatively related to clinical outcome. For HPV-positive tumors, such proteins were mostly related to hypoxia and angiogenesis, and some may be potentially targetable and need to be further evaluated.