Head and neck squamous cell carcinomas (HNSCCs) originate from the mucosal lining of the oral cavity, pharynx, and larynx, making them the most prevalent cancers in this anatomical region [1]. According to the GLOBOCAN cancer estimates ∼890,000 new HNSCC cases were diagnosed in 2018, resulting in ∼450,000 annual deaths, which account for 4.5% of all cancer mortality [2]. The incidence of HNSCC varies globally and is strongly linked to risk factors such as tobacco use and excessive alcohol intake as well as [2] human papillomavirus (HPV) strains, particularly HPV-16 and, to a lesser extent, HPV-18 [3, 4]. Since these high-risk HPV strains are included in FDA-approved vaccines, widespread immunization efforts can potentially prevent HPV-positive HNSCC [1]. Oral cancers related to other causes, particularly smoking are classified as HPV-negative HNSCC [5]. Although some oral precancerous lesions of squamous cell carcinoma, such as leukoplakia (white patches) and erythroplakia (red patches), may progress into invasive disease, most patients are diagnosed at an advanced stage [1]. Since the molecular knowledge behind HNSCC development and progression is still incomplete, searching for novel molecular biomarkers associated with tumor behavior is still relevant.

In HNSCC, cumulative genetic instability drives epithelial stem cells through the hyperplasia, dysplasia, carcinoma in situ and invasive carcinoma progression sequence [1, 6, 7]. This involves e.g., the loss of regions of 9p21 encoding for the tumor suppressor CDKN2A (p16INK4A a CDK4 and CDK6 inhibitor) and ARF (p14, a stabilizer of p53 protein) besides the progressive loss of different parts of the TP53 gene on chromosome 17. Accordingly, several stem cell biomarkers have been detected in HNSCC, including CD44 [8], ALDH1 [9, 10], CD133, SOX2, OCT4, KLF4 and NANOG [11–13].

Extracellular matrix (ECM) and stromal components, have only been recently recognised as important contributors of cancer initiation and progression of malignant tumors [14–20]. SPOCK1 (testican-1), first identified in human seminal plasma [21], is a heparan sulfate/chondroitin sulfate proteoglycan [22]. The SPOCK family members (SPOCK1, SPOCK2, and SPOCK3) are highly conserved ECM proteoglycans with multiple functional domains [23]. When overexpressed, SPOCK may promote cell cycle progression and tumor cell growth by upregulating the PI3K/AKT signaling e.g., in breast, gallbladder, ovarian, colorectal and liver cancers [24–27]. It may also be involved in the elevated invasion and migration e.g., of gastric cancer, glioma, as well as in pancreatic and prostate cancer [17, 19, 28, 29]. Recent studies highlighted the role of SPOCK1 in the activation of cell migration [30, 31] and matrix metalloproteinases (MMPs, e.g., MMP-2), which can support epithelial-to-mesenchymal transition (EMT) in cancer progression [32]. Our group has recently shown in mouse, rat and human liver models that in hepatocarcinogenesis SPOCK1 is upregulated in transformed hepatocytes, with high levels observed in hepatocellular carcinomas and cirrhotic liver, which latter may transform into cancer, but not in the extracellular matrix of the liver [24]. Also, we detected SPOCK1 upregulation in ovarian cancer in relation to poor disease outcome [33].

In the oral cavity of transgenic mice, elevated SPOCK1 expression mediated EMT in a cyclosporin-A induced gingival overgrowth model [34, 35]. Oral mucosa showed an elevated probability for EMT due to increased secretion of MMP9 and reduced TNFα and IL-1β levels. EMT featured by reduced E-Cadherin and emerging N-Cadherin and vimentin production may also be involved in HNSCC progression resulting in increased motility, stromal spreading, stemness and therapy resistance [36, 37]. These observations suggested that SPOCK1 may also influence the basic functions and potentially the malignant transformation and progression of oral epithelial tissues.

In this study we investigated the prognostic role of SPOCK1 expression in HNSCC by using both the in silico mRNA analysis tool, Kaplan-Meier Plotter and the in situ protein expression of different oral cavity locations applying standardized immunohistochemistry. H-Score results were gained in digital sections both by eye-control testing and using a machine-learning-based automated image analysis tool called DensitoQuant.

In the oral mucosa of transgenic mice, SPOCK1-mediated pathways may contribute to drug-induced gingival overgrowth and EMT [34, 35]. Also, SPOCK1 has been recently found in a basement membrane-related gene signature, including, resulting in reduced survival of HNSCC [40]. These data suggested that SPOCK1 may have a role in the biology of human oral epithelia including their tumors e.g., HNSCC. Here we tested SPOCK1 expression in human clinical HNSCC samples, both at the mRNA and protein level and confirmed its role in the adverse clinical outcome of HNSCC. Our results also validated the pathologist-read based protein expression results by machine-learning automated scoring.

First in the KM-Plotter open-source database management software, we found that SPOCK1 mRNA levels were significantly higher in oral cancers than in the normal oral mucosa. Subsequently, the „Pan-Cancer” algorithm of the same software revealed that among 500 HNSCC patients those with elevated SPOCK1 mRNA expression showed significantly worse OS, when patients were divided into two groups at the median antigen expression. This was even more emphasized at the high vs. low neoantigen levels and, obviously, when the best prognostic cut-off was applied. At the protein level, SPOCK1 showed a characteristic granular, peri-/paranuclear cytoplasmic immunostaining in HNSCC cells, consistent with the localization to the Golgi-region. This was in line with the observation of detecting testican-1 protein along with amyloid-β precursor protein in the frontal and temporal cortex of Alzheimer patients [41]. Patients with higher SPOCK1 expression exhibited significantly worse OS in the visually scored cohort with the Cox-regression test of H-Score (p = 0.002), and similarly, when cases were dichotomized at the upper quartile of H-Score, with the log-rank test (p = 0.002). The machine-learning-based H-Score evaluation also supported this observation both with Cox-regression test (p = 0.008), log-rank test (p = 0.044). Splitting SPOCK1 expression mostly at the upper rather than the lower quartile, made our results even more easily reproducible, because strong reaction can be straightforward to recognise and score. Our results imply that SPOCK1 may carry an adverse prognostic information. However, we have to accept the limitations of our project concerning the moderate number of patients collected only in a single department. Since not all of our patients had unequivocal HPV data, we could not correlate SPOCK1 expression with virus infection. Also, whole section vs. TMA validation to more accurately consider tumor heterogeneity, and validation of automated vs. pathologist-read testing, might also be required. Therefore, further validation of our findings is needed using large, multicenter HNSCC cohorts accompanied with complete HPV, p16, smoking and alcohol exposure data.

SPOCK1 has been primarily considered as an extracellular matrix protein secreted by mesenchymal fibroblasts, cancer-associated stromal cells (CAFs) [42]. However, it was found also in the nervous system as a synaptic component [43], as well as in chondrocytes and endothelial cells [44] and in different type of cancers mainly of epithelial origin, as a prognostic factor of adverse outcome [23, 45]. SPOCK1 has also been found in the bloodstream of septic patients [46] and in vitro studies using SKOV3 ovarian cancer cell lines confirmed that it is released into the culture medium to maintain intracellular equilibrium [33]. If this is the case in the human and serum SPOCK1 may indicate malignization, what we have been investigating in several clinical cancers, SPOCK1 blood level could be a promising biomarker in liquid biopsy for tracing of residual tumors. Though some reports link SPOCK1 expression to EMT and one to the thermoregulation of adipocytes [34, 35, 47], the function of SPOCK1 in epithelial cells is largely unknown. SPOCK1 may support tumor cell survival by inhibiting apoptosis and cooperating with TGF-β in promoting EMT [30]. Our group’s recent investigations in mouse and rat models and in clinical samples demonstrated that SPOCK1 (testican-1) is upregulated during hepatocarcinogenesis and in human hepatocellular carcinomas, but without occurring in the extracellular matrix [24, 45].

In the majority of HNSCC the PI3K-AKT pathway is activated, which can promote proliferation, EMT and tumor cell invasion [48]. HNSCC progression has been linked to ongoing dedifferentiation towards a cancer stem cell state accompanied by “partial” or “hybrid” EMT [37], which are responsible for enhanced tumor cell migration and metastatic spreading [49, 50]. This has been underlined by the detection of several stem cell biomarkers, as mentioned also in the introduction, including CD44 [8], ALDH1 [51], CD133, OCT3, OCT4 [52], SOX2, and NANOG [11, 12]. As we are aware, there has been no relevant mechanism of action data on the potential role of SPOCK1 in HNSCC, we can only speculate based on its relevant pathways and data gained in other malignant tumors. The development of an invasive EMT-prone phenotype can probably be mediated through the activation of TGF-β1, EGFR and NOTCH signaling, as well as hypoxia (HIF-1α) [53]. Indeed, the vast majority of HNSCC overexpress EGFR and can be treated using anti-EGFR antibodies and/or tyrosine kinase inhibitors, which however, frequently become resistant due to EGFR mutation [54]. SPOCK1 expression has also been linked also by several studies with the support of EMT phenotype [36, 37]. Further evidence on the negative prognostic role of SPOCK1 came from a study showing that pre-miR-150 can downregulate SPOCK1 to reduce the aggressiveness of HNSCC cells and colon cancer organoids [55, 56]. These findings are in line with ours, showing that elevated SPOCK1 expression may also be a promoter of the advanced HNSCC, may be EMT-prone phenotype with poor prognosis.

While HPV-positive HNSCC can be prevented by vaccination, HPV-negative cases, particularly at advanced stage, require multimodal radio-/chemotherapy besides surgery [1, 5]. Recently, epidermal growth factor receptor (EGFR)-positive cases have been treated using radiotherapy combined with anti-EGFR chimera (mouse/human) antibodies (cetuximab) [46]. Immune checkpoint inhibitors (e.g., pembrolizumab, or nivolumab) have also been approved in cisplatin-refractory recurrent or metastatic HNSCC [30, 47]. Besides its potential prognostic role, SPOCK1 may be offered as a potential target for developing novel treatment options too.

Taken together, these findings support the concept that SPOCK1 can be a surrogate negative prognostic biomarker in HNSCC. Given its reported involvement in supporting an EMT-related tumor progression pathway, SPOCK1 overexpression may indicate a group of oral cancers with aggressive tumor biology. However, our results may require further validation in larger, independent cohorts, ideally stratified by tumor site, HPV/p16 status, and treatment modalities. If reliably detected in the serum, SPOCK1 may represent a promising target for future biomarker-driven, liquid-biopsy-based monitoring of residual tumors. However, further validation and functional experiments are required to prove our pilot findings.