FormalPara Key Summary Points

The incidence of keratinocyte carcinomas—basal cell carcinoma (BCC) and squamous cell carcinoma (SCC)—is on the rise, with the associated number of deaths surpassing those caused by melanoma, according to recent reports.

Biomarker profiling of skin tumors plays an important role in deeper disease understanding, as well as in guiding clinical diagnosis and patient management.

Biomarkers have influenced the development or repurposing of targeted therapies, such as the Hedgehog pathway inhibitors or the immune checkpoint blockade, now used for patients with advanced BCC and SCC.

In this work, we review the known and emerging biomarkers of BCC and SCC relevant for aspects of patient management, including prevention, risk assessments, tumor diagnosis, and therapy selection

Introduction

Skin cancer is the most common cancer type in the USA, with over five million annually treated cases and one in five Americans predicted to develop the disease by the age of 70 [1]. Skin cancer can be classified as melanoma or non-melanoma (NMSC), the latter including basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (SCC), among others [2]. Melanoma represents 1% of skin cancers, but has historically claimed the majority of skin cancer-related deaths, which has made it a major focus of diagnostic and drug development efforts in the realm of skin malignancies [3, 4]. By comparison, BCC and SCC have potential for improved clinical outcomes; however, given their high incidences, a significant number of these cancers fall into the aggressive category, capable of local invasion, recurrence, and metastasis; in fact, new studies report increasing NMSC incidence rates in recent years and number of deaths surpassing those caused by melanoma [5, 6]. The costs to treat skin cancer increased five times faster than those of other cancers between 2002 and 2018, climbing from 3.6 to 8.9 billion dollars; more than 50% of these expenditures are related to BCC and SCC [7,8,9]. In addition, these estimates are likely to be low because of common exclusion of non-melanoma skin cancers from national cancer registries [3, 10].

Development of BCC and SCC is impacted by environmental, behavioral, and genetic risk factors, most prominent of which is the environmental exposure to ultraviolet radiation (UVR) [11,12,13,14,15,16] (Fig. 1). Correlations between chronic UVR exposure and NMSC suggest that changes in sun-related behaviors may be able to prevent a significant portion of cutaneous carcinomas; however, the implementation of preventative measures has been hampered by low public compliance and conflicting evidence on efficacy [17,18,19]. The US Preventive Services Task Force (USPSTF) cites insufficient evidence to recommend for or against skin cancer screening of the general population [20, 21], while data from countries like Germany and Australia show that routine skin cancer screening can be both beneficial and cost-effective [22, 23]. One of the challenges in screening for NMSC is inconsistent diagnostic accuracy upon primary examination; clinical presentations of NMSC are diverse, and may carry considerable phenotypic similarities to benign lesions [24, 25]. Nonetheless, early diagnosis is key for the successful treatment of patients with skin cancer, especially those with aggressive disease, underlying the need for the development of additional skin investigation tools and validation of new biomarkers. Skin biomarker profiling plays an important role in deeper disease understanding, as well as in guiding clinical diagnosis and patient management, prompting the use of both invasive and non-invasive approaches to evaluate specific biomarkers; common examples include surgical biopsy with histology, dermatoscopy, and confocal microscopy [26, 27]. An emerging non-invasive approach that holds promise for broadening the skin tumor profiling is the collection of stratum corneum via tape stripping [28]. Epidermal cells from the outermost skin layer are removed with the tapes and can be used for the investigation of molecular biomarkers, allowing for both initial assessments and follow-up insights within the same skin area. Cost reductions in next-generation sequencing (NGS) and technical advances in the field of multiomics allow access to unprecedented amounts of information on patients’ molecular profiles, accelerating the development of skin molecular biomarkers, evidenced by the availability of gene panel tests for the profiling of melanoma and SCC [4, 29].

Fig. 1
figure 1

Risk factors in cutaneous keratinocyte carcinomas. a Development of BCC and SCC is influenced by environmental, behavioral, and genetic factors, most prominent of which is the exposure to ultraviolet radiation (UVR). Individuals with Fitzpatrick skin types I/II, a family or personal history of skin cancer, or those affected by immunosuppression have a higher risk of developing skin cancer. b Exposure to risk factors may result in different clinical manifestations, such as actinic keratoses (benign lesions which can progress to SCC) or directly promote the development of SCC and BCC. Image created with BioRender.com

Full size image

Methods

PubMed and ClinicalTrials.gov databases were independently searched by the authors for key terms including “biomarkers in BCC”, “biomarkers in SCC”, “genomic biomarkers in NMSC”, “proteomic biomarkers in NMSC”, “therapy of BCC”, “therapy of cSCC”. Original and review articles were analyzed by the authors, selecting those that described known or putative molecular and histologic biomarkers related to risk assessments, molecular or histologic profiling of the tumors, therapy selection, and patient follow-up.

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Biomarkers in BCC

Prevention and Risk Assessment

UVR

BCC arises upon malignant transformation of the cells in the basal epidermal layer and is the most common global malignancy; in the USA, its incidence continues to increase by 4–8% annually [30]. Risk of developing BCC is higher for individuals with Fitzpatrick skin types I/II or family history of skin cancer, as well as those affected by genetic disorders such as Gorlin syndrome and xeroderma pigmentosum [31, 32]. The most prominent environmental risk factor is the exposure to ultraviolet radiation (UVR). UV photons absorbed by DNA molecules induce the formation of DNA lesions, most common of which are cyclobutane pyrimidine dimers and 6–4 photoproducts [33]. These result in high proportion of C > T and CC > TT substitutions, which are the hallmarks of the UV mutational DNA signature [34]. Frequent mutations of certain areas of the DNA double helix (UVR mutational hotspots) were found within P53, NOTCH1, and NOTCH2 genes, in both lesional and normal sun-exposed skin, indicating a positive clonal selection [35, 36]. While mutations in P53 are unlikely to predict the evolution of an existing BCC, their accumulation in normal skin may signal an increased risk of skin cancer development [35,36,37]. Genomic biomarkers of cancer promoting UVR damage, together with phenotypic and clinical data, could be useful to stratify higher-risk patients who would benefit from routine skin cancer screening and preventative interventions [38].

Tumor Diagnosis and Molecular Profiling

Major histological subtypes of BCC are superficial, nodular, micronodular, and morpheaform [39]. Nodular and superficial are the less aggressive subtypes; however, tumors may have mixed subtype patterns, including those with aggressive attributes. The clinical presentation of BCC is variable and can resemble benign skin dermatoses, potentially delaying the medical evaluation and complicating the diagnosis. A rare and aggressive type of BCC, the basosquamous carcinoma (BSC), displays a relatively benign clinical appearance with dermoscopic patterns characteristic of both BCC and SCC, creating considerable controversy in its classification [40]. In contrast to the clinical presentation, BSC is characterized by an aggressive subclinical extension, high rate of recurrence (12–51% for surgical excision, 4% for Mohs), a high rate of metastasis (5–10%), and potentially a higher incidence of secondary skin tumors in the affected patients [41,42,43]. Punch biopsy and examination of histologic biomarkers is the gold standard for diagnosing and subtyping BCC, leading to a high number of invasive interventions [25]. Several studies have evaluated the initial diagnostic accuracy by comparing punch biopsies to excisional specimens of BCC; these studies estimate that 11–26% of punch biopsies exclude features of aggressive cancer subtypes [44, 45]. The ambiguity in diagnosing aggressive forms of BCC fuels the interest in identifying genomic biomarkers of histological subtypes, as these would complement the current approaches in early diagnosis and patient management.

Hedgehog Pathway

Activation of the Hedgehog (HH) signaling pathway has been shown to play a critical role in the oncogenesis of BCC [46, 47]. In brief, pathway activation is initiated by the cell-surface protein, Smoothened (SMO), which is inhibited by the transmembrane receptor Patched (PTCH1); binding of the HH ligand to PTCH1 prevents the inhibition of SMO, which migrates to the primary cilium and regulates the transcription of GLI transcription factors to activate signaling. Loss-of-function mutations in the PTCH gene or gain-of-function mutations in SMO can lead to constitutive activation of the HH pathway. Germline mutations in the PTCH gene were initially detected in patients with Gorlin’s syndrome [48]. Subsequently, approximately 85% of sporadic BCCs were also found to contain mutations in PTCH1 and other genes related to the HH pathway, such as SMO and SUFU [46, 49]. While it seems that the activation of the hedgehog signal transduction pathway may be a necessary event in the development of BCC, mutations of the pathway are frequent in BCCs of different size, histology, or recurrence status [46, 49, 50]. Biomarkers in the HH pathway may help distinguish BCC from benign lesions, but the development of additional markers is needed to account for non-HH drivers or modifiers. Furthermore, given the ambiguity in diagnosing aggressive forms of BCC, identification of genomic biomarkers related to histological subtypes would be highly desirable.

Hippo-YAP Pathway

Yes-associated protein (YAP) is a co-transcriptional activator with a role in regulation of proliferation in normal skin [51]. Upon Hippo signaling activation, YAP is phosphorylated and transported to the cytoplasm where it is bound by 14-3-3σ and can no longer induce the expression of target genes [52]. In primary mouse keratinocytes, YAP activation was shown to accelerate proliferation, suppress differentiation, and inhibit apoptosis [51]. Bonilla et al. carried out a genetic profiling of 293 BCCs and reported recurrent loss-of-function mutations in PTPN14 (23%) and LATS1 (8%) that can promote transcriptional activation of YAP1 [49]. Recurrent mutations were also detected in genes so far unrelated to BCC, such as the serine/threonine protein phosphatase PPP6C, mutated in 15% of the BCCs analyzed [49]. PPP6C promotes phosphorylation-dependent activation of LATS1 and inactivation of YAP [53]. Besides its role in the regulation of the Hippo pathway, PPP6C is also involved in the regulation of the cell cycle, DNA damage repair, and cutaneous tumorigenesis [54]. The overall mutational burden of BCC was found to be exceptionally high (65 mutations/Mb) [49], which may complicate the detection of significant mutations in driver genes. Several candidate genes were found to be frequently mutated in BCC, however without reaching statistical significance. These include ARID1A, CASP8, KRAS, NRAS, and RAC1, among others [49, 55].

MYCN

MYC family of transcriptional activators is involved in multiple cellular functions such as DNA repair, proliferation, and apoptosis [56]. Missense mutations in MYCN have been identified in 30% of BCCs [49], most of these mapping to the MYC box 1 domain, involved in the interaction with FBXW7 tumor suppressor [49]. FBXW7 is a component of an ubiquitin ligase complex that promotes degradation of N-MYC [57]. Most commonly identified N-MYC amino acid substitutions in BCC (T58A, P59L, P60L, and P63L) were shown to impair binding to FBXW7, resulting in increased N-MYC protein levels [49]. Interestingly, deleterious mutations and loss-of-heterozygosity (LOH) events in the FBXW7 gene occur in 5% and 8% of BCC samples, respectively, indicating that increased N-MYC stability may be advantageous for the development of this cancer [49].

TERT Promoter

Telomere repeats cap the ends of chromosomes and are critical to genome integrity; their length is maintained through a tightly regulated process, involving the activity of an essential reverse transcriptase, telomerase [58]. TERT gene encodes the catalytic subunit of telomerase, and its promoter region is considered the most important regulatory element for telomerase expression. Mutations in TERT promoter occur at high frequency in numerous cancers and have been correlated with increased expression, longer telomeres, and poor patient outcomes [59]. TERT promoter mutations are frequent in skin cancers, including BCC, suggesting that the increased expression of telomerase might play an important role in their pathogenesis. Common C > T and CC > TT substitutions reveal the UV signature underlying most of the TERT promoter mutations in cutaneous malignancies [60,61,62].

DPH3-OXNAD1 Promoter

Frequent BCC UV signature mutations in the bidirectional promoter region of DPH3-OXNAD1 were reported in one study (57/137, 42%) [63]. Mutations occurred in or adjacent to the ETS/TCF transcription factor binding motif, near the transcription start site, but without detecting changes in the expression levels of DPH3 or OXNAD1, their functional significance still needs to be determined.

KNSTRN

Kinetochore-localized ASTRIN/SPAG5 binding protein (KNSTRN) encodes a kinetochore-associated protein in the mitotic spindle, crucial for correct chromosomal segregation. Recurrent point mutations in KNSTRN that disrupt chromatid cohesion and promote chromosomal instability have been identified in melanoma, SCC, and more recently BCC [64, 65]. Comparison between 18 advanced (inoperable and > 3 cm in size) and 30 early stage (< 2 cm in size) BCCs suggests that KNSTRN mutations may be more common in advanced or aggressive disease, although this remains to be more firmly established by evaluating a larger patient cohort [64].

Non-coding RNAs

MicroRNAs (miRNA) are small non-coding RNAs involved in a variety of biological processes through their roles in the regulation of gene expression [66]. Aberrant activity of miRNAs has been liked to various human diseases, including cancer [67]. Most miRNAs are transcribed into primary miRNAs (pri-miRNAs) and processed into precursor miRNAs (pre-miRNAs) and mature miRNAs by the microprocessor complex [68]. After being incorporated into the RNA-induced silencing complexes (RISC), mature miRNA will interact with their target mRNAs to suppress expression [69]. Compared to healthy controls, BCCs were shown to have altered expression profiles of the microprocessor complex, RISC, and different miRNAs [70,71,72]. miR-203, proposed to act as a tumor suppressor in BCC, was shown to be downregulated in the tumors following the activation of HH and EGFR/MEK/ERK/c-JUN signaling [73]. A study profiling global miRNA expression in nodular and infiltrative BCCs found six miRNAs that showed significantly different expression between the two tumor subtypes [74]. In this cohort, miR-183 was found to have no overlap between nodular and infiltrative tumors, suggesting that distinct miRNA profiles could be one of the variables underlying the differences in the aggressiveness of BCC subtypes [74]. In addition to miRNAs, preliminary studies have shown that other types of regulatory non-coding RNAs, such as circular and LINC (long non-coding) RNAs display different expression patterns in BCCs compared to non-lesional skin [75, 76].

Therapy Selection and Follow-up

Treatment primarily consists of surgical approaches, although some patients can undergo topical, radiation or systemic therapy. Classification of BCC lesions as low or high-risk shapes the initial treatment decisions, and these are made after consideration of tumor size, location, immune status of the patient, recurrent lesions, and histological subtyping [77]. While small BCCs have a low metastatic potential (0.5%), recent data suggests that the risk of metastasis and death can be as high as 6.5% for tumors larger than 2 cm, especially those located on the head or neck [78]. In the cases of BSC, a small tumor size is not indicative of a low-risk lesion. Standardized formal guidelines for the management of BSC are not yet established, and given the high rates of recurrence, metastasis, and secondary tumors, the ideal management of patients with BSCs would include a full surgical removal of the tumor through Mohs surgery, as well as a more frequent follow-up schedule compared to other types of BCC. Low-risk superficial BCCs can be treated with topical 5-fluorouracil (5-FU) and imiquimod 5% creams [79, 80], although without a confirmatory biopsy there is limited histological assurance over complete tumor clearance. Localized skin reactions are common and may include erythema, pain, bleeding, and changes in pigmentation. Another treatment option for low-risk BCCs is cryosurgery, consisting of a controlled application of liquid nitrogen to the visible tumor [81]. While the procedure is fast, hypertrophic scars or permanent pigment alterations may develop in the treated area. Radiation therapy is used as a primary option or an adjuvant treatment for BCC when surgery is contraindicated [82]. For high-risk and recurrent BCCs, Mohs surgery is the gold standard [39].

Systemic therapy with HH pathway inhibitors (HHI) can be used to treat patients with metastatic BCC, or those with advanced/recurrent disease who are not candidates for surgery. Two HHIs targeting SMO are approved by the US Food and Drug Administration (FDA): vismodegib and sonidegib. Reported objective response rates are 30–45%, although some patients can achieve a complete response [83, 84]. Long-term treatment presents challenges related to tolerability, as well as resistance; treatment discontinuation related to adverse events has been reported in up to 55% of patients; the most common events included nausea, muscle spasm, alopecia, weight loss, diarrhea, and fatigue [85,86,87]. Resistance to treatment can be conferred by the activation of non-canonical HH pathway, acquisition of additional SMO mutations, or the activation of alternative pathways [88,89,90]. As an adjacent approach to first-line therapy, HHI are being evaluated for neoadjuvant use, with the primary goal of downstaging tumors and managing surgical procedures [91]. For patients that are resistant or intolerant to HHI therapy, immune therapy is typically recommended as the second-line choice. In a phase II trial involving 84 patients with advanced BCC, immunotherapy with cemiplimab, a human monoclonal antibody directed against programmed death 1 (PD-1) receptor, showed an acceptable toxicity profile and objective responses in 31% of the patients, leading to its approval for locally advanced and metastatic BCCs that are not candidates for HHI or those that failed on primary HHI treatment [92]. So far, no clear correlation has been established between the response to systemic therapy and factors such as age, number of sites affected by BCC, tumor subtype, or prior treatment with radiation.

Biomarkers in SCC

Prevention and Risk Assessment

UVR

SCC contributes to 20% of all skin cancers, its frequency second only to BCC; metastasis occurs in approximately 5% of cases, accounting for 75% of NMSC and 25% of all skin cancer-related deaths [24]. Similarly to BCC, genetic and environmental risk factors can drive the development of SCC, with the leading risk factor being the exposure to UVR [36]. Recurrent UVR signature mutations in P53 and NOTCH genes have been described as early events and likely drivers of SCC pathogenesis; however, these mutations can exist in phenotypically normal skin for decades before the onset of the disease [36, 93], creating opportunities for population stratification through screening.

Actinic Keratosis

Actinic keratoses (AKs), common SCC precursor lesions, appear as irregularly shaped and rough papules on sun-damaged skin, resulting from the proliferation of atypical epidermal keratinocytes. AKs usually follow one of three clinical outcomes: spontaneous regression, persistence, or evolution into SCC [94]. AKs have been classified as grade I–III lesions, based on the extent of epidermal involvement with the basal keratinocyte atypia; however, it has been shown that grade evolution is not necessary for progression to SCC and that all types of AKs should be regarded as potentially invasive and appropriately treated [95, 96]. From a genetic standpoint, AK and SCC exist on a progressive disease spectrum and display similar genetic alterations, making the clinical and histological differentiation between the two conditions challenging. A study comparing photo-protected skin, photo-exposed skin, AK, and SCC showed progressive abnormal gene expression levels along this spectrum [97]. This study outlined an 89-gene classifier distinguishing between normal and transformed skin, but it could not separate AK and SCC, in line with their close genetic relationship [97].

Tumor Diagnosis and Molecular Profiling

Biopsy and histological analyses are the commonly used tools in diagnosing SCC, enabling the distinction between in situ and invasive forms, tumor differentiation, and subtypes. SCC subtypes include common, spindle cell, clear cell, and bowenoid variants [98]. Bowen’s disease (BD), or SCC in situ, histologically shows full-thickness epidermal dysplasia with a complete loss of normal maturation; BD is predicted to progress to invasive SCC in 3% and 10% of cases, a third of which are reported to be metastatic [99]. De novo SCCs, which are not derived from a precursor AK or BD lesion, have increased risk of recurrence and metastasis, with an overall poor prognosis [100].

Chromosomal Alterations in SCC

Changes in chromosomal number, insertions, deletions, and translocations are common in many cancers, including SCC, which displays extensive chromosomal instability [101]. LOH studies have shown recurrent regions of loss and gain in SCC, including loss of 3p (65%), 9p (75%), 2q, 8p and gains on 3q, 8q, 9q, and 11q [101]. The extent of chromosomal instability may correlate with the tumor differentiation status, as significantly fewer changes have been identified in well-compared, poorly differentiated SCCs [101]. Tumor suppressor locus CDKN2A on chromosome 9p21 encodes the p16(INK4a) and p14(ARF) genes that function as cell cycle regulatory proteins in the p53 and RB pathways; these are susceptibility genes for a number of cancers, including melanoma and the squamous carcinomas of the head and neck [102]. Inactivation of the CDKN2A locus by both genetic and epigenetic mechanisms are frequently reported in SCC [101, 103], in line with common LOH events at the 9p region. Within the same region, a frequent microdeletion at 9p23 was reported, found within the gene encoding protein tyrosine phosphatase delta (PTPRD). SCC tumors carrying the PTPRD microdeletion were more likely to be poorly differentiated and associated with a higher metastatic risk [101]. PTPRD may be a candidate tumor suppressor gene in SCC; however, given its chromosomal localization, further functional studies are needed to explore this hypothesis.

Histone Methyltransferases

KMT2C and KMT2D encode histone methyltransferases that regulate gene expression through the targeted modification of histone H3. Frequent inactivating mutations in both genes were reported in aggressive SCCs and associated with poor prognosis [104]. KMT2C mutations had a significant positive correlation with bone invasion, shorter recurrence-free survival and shorter overall survival in patients with SCC [105]. Significantly higher rates of mutations in KMT2D were reported for metastatic (62%) relative to non-metastatic SCCs (31%) [106].

CTCF Insulator Binding Sites

A study by Mueller et al. profiled mutational patterns in 15 SCC metastases and found an uneven mutational load in certain regions of the genome, especially insulator elements [107]. The main human insulator, CTCF, had a high prevalence of mutations in its consensus DNA binding motif. CTCF simultaneously binds to multiple DNA sites, thereby approximating distant chromatin regions and forming 3-dimensional DNA loops termed topologically associated domains (TADs). Metastatic SCCs were found to have a high mutation density in the CTCF binding motif, affecting 422 TADs encompassing a total of 1979 genes, and including 101 genes previously identified as tumor modulators [107]. Additional analyses are needed to determine the functional impact of these mutations in SCC.

Protein Biomarker Panel for SCC Progression

Sun et al. used immunohistochemistry to compare the protein profiles of 42 noncancerous, 34 precancerous lesions, and 51 SCC samples. The SCC specimens were further distinguished as well- (18), moderate- (13) and poorly- (20) differentiated subgroups. A total of 10 selected proteins were profiled, including keratins (CK10, CK17), cell–cell adhesion factors (CD44, EZR, E-cadherin, and β-catenin), chaperones (Hsp75 and Hsp90-α), transcription regulator EXOSC10, and mitochondrial redox protein SOD2 [108]. Keratins, cell adhesion proteins, and EXOSC10 showed progressive changes in the intensity of immunohistochemical stains between the three test groups, correlating with disease progression, but not SCC differentiation. SOD2 was virtually absent from non-SCC tissue, but showed a significant increase in accumulation in SCC, and a positive correlation with the tumor differentiation status (rs = 0.423, P < 0.005). The protein levels of Hsp75 and Hsp90-α significantly increased along the transformation spectrum from noncancerous tissue to SCC. The strength of Hsp90-α staining in the SCC group was associated with differentiation status (rs = 0.389, P = 0.005), with higher protein expression seen in poorly differentiated samples.

Epidermal Homeostasis

Disruption of skin homeostasis is a risk factor for the development of SCC, possibly by creating a permissive environment for tumor initiation and progression. Epidermolysis bullosa (EB) is a heterogeneous group of inherited skin disorders characterized by dysfunction of structural components of the cutaneous basement membrane resulting in chronic skin trauma [109]. Patients suffering from EB, and in particular those with severe recessive dystrophic EB (RDEB), are at a high risk of developing SCC [110, 111]. RDEB is caused by mutations in the COL7A1 gene, resulting in the deficient expression of its protein product, the type VII collagen (COL7). RDEB-SCCs have a high propensity of relapse and metastasis; however, early diagnosis remains challenging because of the difficulty in differentiating the tumors from benign epidermal hyperplasia [111]. Combination of delayed diagnosis and the exceptionally aggressive features of RDEB-SCC contribute to the poor prognosis in these patients; indeed, the US National EB registry reports the cumulative risk of death from SCC in patients with RDEB to be 57.2% by age 35 and 87.3% by age 45 [112]. Structural proteins or enzymes involved in would healing and skin homeostasis, including matrix metalloproteinases (MMPs), have been considered as biomarkers or regulators of SCC carcinogenesis [113]. MMPs are a large family of proteolytic enzymes able to degrade almost all components of the extracellular matrix and basement membranes. Progressively increased accumulation of various MMP family members was reported between AKs, BD, non-EB SCCs, and RDEB-SCCs, indicating a possible role for these enzymes in SCC progression and aggressiveness [114].

S100A7

S100A7 (psoriasin) is a calcium-modulated protein, originally identified in psoriatic keratinocytes [115, 116]. Expression of S100A7 is upregulated in many types of squamous cell carcinomas, including lung, oral, bladder, and skin, where it may play a role in carcinogenesis and metastasis [117,118,119]. Several studies report increased expression levels of S100A7 in highly differentiated SCCs, compared to weak or absent expression in less differentiated tumors, suggesting a possible association with the balance of proliferation and differentiation in SCC [119, 120]. Further suggesting the involvement of S100A7 in differentiation is its location within the epidermal differentiation gene cluster on chromosome 1q21; indeed, the upregulation of S100A7 was found to be paralleled by the upregulation of squamous differentiation markers, including keratin-4, keratin-13, TG-1, and involucrin [121, 122].

The Complement System

Activation of complement C3 is a part of the innate immune system involved in the inflammatory response initiation, resulting in the formation of membrane attack complex and lysis of the target cell; several complement components and inhibitors have been described as regulators of the tumor microenvironment and potential biomarkers of SCC progression [123]. Overexpression of complement factor H isoforms (CFH and FHL-1) was shown in SCC relative to normal keratinocytes, as well as following progression from AK to invasive SCC, resulting in promotion of cancer proliferation, migration, and association with poor prognosis [123]. Complement factor B, C1r, and C1s were shown to be upregulated by SCC cells in culture and by SCC tumors in vivo, regulating proliferation and migration of SCC cells and promoting growth of SCC xenografts; conversely, knockdown of C1r and C1s was shown to inhibit activation of the extracellular signal-related kinase (ERK)1/2 and phosphoinositide 3-kinase (PI3K) signaling pathways, promote the apoptosis of SCC cells and suppress vascularization and growth of SCC xenografts in vivo [124, 125], suggesting their potential use as disease biomarkers, with a role in viability, apoptosis resistance, and invasion.

Therapy Selection and Follow-up

Noninvasive SCCs are successfully treated with techniques such as curettage, cryosurgery, and photodynamic therapy. High-risk features of SCC include tumor size, depth of invasion, and poor differentiation [126]. Surgery with safety margins is the treatment of choice in patients with primary SCC; Mohs surgery is indicated in patients with high-risk tumors [127]. Radiation therapy is a viable treatment for small tumors, or for patients who are not eligible for surgery [128].

Systemic treatments available for patients with SCC include targeted therapy to the epidermal growth factor receptor (EGFR), chemotherapy, and anti-PD-1 immunotherapy [129, 130]. Two monoclonal antibodies targeting EGFR, cetuximab and panitumumab, were evaluated as first-line treatment in patients with advanced or metastatic disease. Objective response rates were between 28% and 53%, with mostly acceptable safety profiles [131, 132]. An oral anti-EGFR agent, gefitinib, showed an overall response rate of 16% and 3.8 months of median progression-free survival a phase II study including 40 patients with advanced SCC [133]. Systemic chemotherapies are associated with significant toxicity and their use in advanced SCC is indicated when other therapeutic approaches are limited or exhausted, with systemic immune therapy becoming the new standard of care for advanced and metastatic SCC [134]. Expression of PD-L1 has been observed in 35–70% of advanced SCCs and in 58–100% of metastatic SCCs, and could potentially be used as a biomarker in predicting response to checkpoint inhibitors [135]. Intralesional administration of cemiplimab in patients with recurrent resectable SCC (or BCC) is being evaluated in early phase clinical trials [136]. Systemic therapy with cemiplimab is indicated for the treatment of locally advanced or metastatic SCC not amenable to surgery or radiotherapy, with objective response rates 44–50%, including patients achieving a complete response [137,138,139].

Discussion and Conclusions

Non-melanoma skin cancer is the most diagnosed cancer in the USA and represents a substantial public health issue. Environmental exposure to UVR is the key environmental risk factor for the development of skin cancer; however, the long lag between the initial UVR-induced skin damage and the onset of skin cancer decreases the perception of UVR danger and hinders preventative efforts. Management of NMSC relies on profiling of multiple biomarkers, including histologic and molecular types; in the latter group, biomolecules such as DNA, RNA, and proteins have been explored in the context of NMSC and some of these markers have influenced the development or repurposing of targeted therapies, such as the HH inhibitors or the immune checkpoint blockade, now used for patients with advanced BCC and SCC. Despite the clear clinical benefit of HHIs, both intrinsic and acquired resistance remain consistent challenges in the treatment of the target population [83, 84]. The use of combination therapy or therapy switching between HHIs and other treatment modalities are potentially effective strategies to reduce resistance, as shown in several smaller studies [140,141,142]. Besides the approved HHIs, vismodegib and sonidegib, other HHIs are under investigation in advanced BCC, including patidegib, itraconazole, and arsenic trioxide. An ongoing phase III trial is investigating patidegib for the reduction of disease burden in Gorlin syndrome [143] and in a phase 2 trial of persistently developing high frequency BCC in non-Gorlin patients [144]. Itraconazole is a systemic antifungal found to antagonize the HH pathway activation by binding to SMO at a site distinct from the other HHIs [145]. Arsenic trioxide destabilizes GLI2 to inhibit transcription of HH target genes and has shown activity in combination with itraconazole in BCC refractory to approved HHIs. The activation of alternative signaling pathways presenting cross talk with HH, notably the WNT pathway, has been suggested as another mechanism of resistance to HHI [90]. WNT activation involves a complex cascade triggered by a variety of signals, the outcomes of which are context dependent; inhibitors of the pathway, such as those of porcupine and β-catenin, are currently under clinical investigation in patients with advanced solid malignancies, including non-melanoma skin cancers [146, 147].

Although the combination approaches to HHI resistance may be effective in some patients, further identification of biomarkers of resistance or response is needed, as these hold the promise of improving the therapeutic efficacy of HHIs by personalizing treatment. The in-depth profiling of tumor biomarkers, aided by the widening use of omics technologies and high-throughput data analysis tools, will propel the promise of personalized treatment into the reality of clinical practice. Use of non-invasive methods for skin evaluation and sampling, such as tape stripping, will likely play a key role in enabling the widespread reach into the patient communities and the effective collection of the needed numbers of longitudinal samples required for biomarker discovery and validation. A list of molecular biomarkers cited in this work is summarized in Table 1. Relevant combinations of new molecular and clinical biomarkers are expected to improve diagnosis, staging of skin tumors, and prediction of therapeutic responses, covering critical steps in structuring the patient management strategy. Further research into NMSC biomarkers is expected to continue to transform patient care and improve outcomes.

Table 1 Molecular biomarkers in keratinocyte carcinomas
Full size table