Thyroid carcinoma (THCA) is the fifth most prevalent malignancy affecting women (Zhang et al., 2019b) and a major cause of annual endocrine malignancies death (Aboelnaga and Ahmed., 2015). Over the past decades, the incidence of THCA has been escalating worldwide (Cabanillas et al., 2016). THCA is generally classified into four pathological subtypes: papillary thyroid cancer (PTC, 80–85%), follicular thyroid cancer (FTC, 10–15%), medullary thyroid carcinoma (MTC, less than 2%) and anaplastic thyroid carcinoma (ATC, less than 2%) (Laha et al., 2020). Although most THCA is well-differentiated PTC with a 10-years survival rate of over 95%, some variants of PTC may demonstrate increased aggressive behavior, particularly in older patients, contributing to significant mortality (Nath and Erickson., 2018). Therefore, surgical resection and radioiodine (RAI) therapy were considered standard treatments for most THCA patients. Nonetheless, survival rates for advanced thyroid cancer patients remain low. For these reasons, novel therapeutic strategies, such as immunotherapies and targeted molecular therapy, are under investigation for treating advanced or metastatic THCA patients.

In recent years, immune checkpoint inhibitor (ICI) has achieved remarkable progress in treating breast cancer (Santa-Maria and Nanda., 2018), lung cancer (Anagnostou et al., 2017), melanoma (Amaria et al., 2018; Barrios et al., 2020) and squamous cell carcinoma of the head and neck (HNSCC) (Ferris et al., 2016; Ferris et al., 2018). It has been shown that numerous immune cells and their mediators exist in the tumor microenvironment (TME) of THCA with various interactions between them (Varricchi et al., 2019). In addition, it was demonstrated that increased frequency of regulatory and PD-1⁺ T cells was associated with the recurrent or aggressive PTC (French et al., 2012), and that the frequency of PD-1⁺ T cells was higher in patients with extranodal invasion than in those without lymph nodes metastases, indicating that it may be associated with the THCA prognosis. Furthermore, Gunda and his colleagues have recently revealed that anti-PD-1/PD-L1 therapy could beneficially modulate the immune microenvironment in orthotopic ATC murine model while simultaneously enhancing the efficacy of lenvatinib, a multi-targeted tyrosine kinase inhibitor (Gunda et al., 2019). The efficacy of ICIs has also yielded encouraging results in clinical trials, manifesting potent antitumor effects and improved tolerability in advanced FTC patients (Mehnert et al., 2019). However, the expensive cost of ICIs, which averages around $150,000 per year (Oiseth and Aziz., 2017), and their unpredictable efficacies (Giannone et al., 2020), prelude their widespread clinical use. Considering these factors, screening for appropriate molecular markers to improve treatment precision is highly demanding.

Tumor mutational burden (TMB), or the number of nonsynonymous mutations in a genomic region of somatic cells, is a biomarker employed to predict immunotherapy efficacy in various cancers (Samstein et al., 2019). As is widely known, tumor-infiltrating immune cells (TICs), an essential component of TME, are critical for in tumor initiation and progression (Bissell and Hines., 2011). Furthermore, gene locus mutations are observed in many histological subtypes of THCA, such as anaplastic lymphoma kinase (ALK), neurotrophic receptor tyrosine kinase 1 (NTRK1) genes rearrangements, BRAF and GTPase RAS family genes (Bos, 1989; Nikiforov and Nikiforova., 2011; Cancer Genome Atlas Research Network, 2014; Arndt et al., 2018; Varricchi et al., 2019). As high-throughput sequencing advances, large-scale acquisition of relevant cancer genomic data has become possible. However, few studies have conducted in-depth investigations to evaluate immunotherapy efficacy and prognosis in THCA.

Herein, based on Cancer Genome Atlas (TCGA) and ImmPort databases, we screened immune-related differentially expressed genes (IRDEGs), explored their functional enrichment, and constructed a prognostic prediction model. Additionally, we further investigated the association of TMB with immune infiltration and prognosis in THCA.

THCA is the fifth most prevalent cancer in the United States, with an annual incidence increase of ∼5% (Varricchi et al., 2019). Although most THCA patients have a favorable prognosis, approximately 15–20% of DTC patients, most ATC patients display resistance to standard treatment methods, such as RAI. Sorafenib and lenvatinib, two recently approved multikinase inhibitors (MKIs), have demonstrated encouraging outcomes in progressive, RAI-refractory DTC. However, adverse consequences have been identified, restricting their use (Cabanillas and Habra., 2016; Krajewska et al., 2017). Notably, the development of ICIs, such as anti–CTLA-4 and anti–PD-1 agents, have revolutionized THCA treatment (Antonelli et al., 2018; Varricchi et al., 2019). However, not all patients benefit from them due to individual variances. In the current study, we constructed a prognostic prediction model by screening IRDEGs to predict immunotherapy efficacy and survival outcome of THCA.

In the present study, we first identified 294 IRDEGs between THCA and normal samples and then explored potential functional enrichment pathways of these IRDEGs via KEGG and GO functional enrichment analyses. We discovered that the IRDEGs were primally enriched in some immune-related pathways, suggesting that IRDEGs might play a critical role in alteration of THCA immune microenvironment. Subsequently, three hub IRDEGs associated with THCA prognosis were identified, including PPBP, SEMA6B and GCGR. PPBP, alternatively known as C-X-C chemokine ligand 7, stimulates various cellular processes, such as DNA synthesis, glucose metabolism, intracellular cAMP accumulation and fibrinogen activator synthesis. Some studies have indicated that PPBP and its encoded proteins might be linked to the progression of Wilms tumor (Guo et al., 2017) and gastric cancer (Yamamoto et al., 2019). SEMA6B a member of the semaphorin family, is mainly involved in developing peripheral and central nervous systems (Andermatt et al., 2014). SEMA6B has been demonstrated to have a critical role in the prognosis of various tumors, such as gliomas (Sun et al., 2019), breast cancer (Murad et al., 2006) and testicular cancer (Ji et al., 2020). A recent study has shown that SEMA6B promoted occurrence and development of THCA via regulating Notch signaling pathway (Lv et al., 2021). GCGR, a member of G protein-coupled receptor family, is critical in regulating glucose homeostasis. Previous studies indicated that GCGR aberrant expression might lead to adverse survival in endometrial stromal sarcoma (Davidson et al., 2013), renal cell carcinoma (Schrödter et al., 2016), and non-small cell lung cancer (NSCLC) (Li et al., 2020). In addition, it has been revealed that GCGR overexpression results in poor survival of PTC by activating epithelial-mesenchymal transition (EMT) and P38/ERK pathway (Jiang et al., 2020). Although we speculated that these genes might be a potential therapeutic target and/or prognostic biomarker for treating THCA patients, this hypothesis still requires additional validation in future studies.

Next, we firstly built a THCA prognostic model using the three screened hub genes (PPBP, SEMA6B and GCGR), and validated the reliability of the prognostic model in THCA. The results in the combined, test, and training sets revealed that THCA patients in the high-RS group exhibited poor survival outcomes than those in the low-RS group, and AUC values were over 0.6, implying that the constructed prognostic model accurately predicted OS in THCA. Besides, we found that older patients in the more advanced stage had significantly greater RS levels than younger ones in the earlier stage, consistent with the conclusion drawn by Ibrahimpasic et al. (Ibrahimpasic et al., 2019).

TMB, as a novel biomarker, has been recently known to forecast the clinical efficacy of immunotherapy in many cancers (Samstein et al., 2019), since it is closely associated with tumor immune infiltration and microenvironment alteration (Kang et al., 2020). Zhou et al. (Zhou et al., 2021) discovered that patients in the low-TMB group exhibited a worse survival outcome and immune response than those in the high-TMB group in. However, few studies focused on the predictive value of TMB for immunotherapy in THCA patients. In this current study we revealed that THCA patients in the high-RS group possessed higher TMB than those in the low-RS group. Herein, we speculated that higher TMB might be correlated with a worse prognosis in THCA, consistent with the conclusion drawn by Zhang et al. (Zhang et al., 2019a) in their work on clear cell renal cell carcinoma. Besides considering the important roles of TICs in TME on prognosis of numerous malignancies (Zhang et al., 2003; Tran Janco et al., 2015), we further assessed the distribution of 22 TIC subtypes in THCA samples from high- and low-RS groups and the relationship with survival outcome. The analysis results revealed that patients in the high-RS group chiefly possessed a higher level of monocytes, M2 macrophages and activated dendritic cells (DCs). In contrast, those in the low-RS group exhibited higher proportions of M1 macrophages and resting DCs, manifesting that tumor-associated macrophages were associated with tumor progression (Jackaman et al., 2017). Nevertheless, the underlying mechanism remains yet unclear. Similarly, Travers et al. (Travers et al., 2019) found that increasing tumor-killing M1 macrophages and decreasing M2 macrophages in TME contributed to reduced TMB and improved survival in mice with ovarian cancer. Additionally, a study indicated a strong correlation between DCs and advanced patients with PTC (Bergdorf et al., 2019), which might explain why THCA samples in the high-RS group exhibited advanced pathological stages. Besides, it has been reported that tumor-infiltrating memory B cells are linked to superior clinical outcomes in breast cancer (Garaud et al., 2019). In contrast, in the present study, we revealed that a high percent of memory B cells was highly correlated with poor survival in THCA.

Immune checkpoint molecules, such as PD-L1, PD-1 and CTLA4, have been demonstrated to connect with the efficacy of immunotherapy (Kalbasi and Ribas., 2020) Hence, we further investigated these biomarkers, including PD-L1, PD-1 and CTLA4, expression levels of THCA patients between in the high- and low-RS group. Results showed that PD-L1 expression was significantly upregulated in the low-RS group compared to the high-RS group, suggesting that this prognostic model might have ability to determine THCA patients’ response to immunotherapy. Based on these findings, we supposed that high TMB and low PD-L1 expression in THCA patients might respond poorly to immunotherapy.

In conclusion, this study constructed and validated a THCA prognostic prediction model based on TCGA database, displaying good predictability and reliability for THCA prognosis. To our knowledge, our group was the first to screen out three potential therapeutic target genes and elucidate the association of TICs with RS and OS in THCA. Additionally, we figured out that THCA patients in the high-RS group had high TMB and low PD-L1 expression, establishing a baseline and reference for predicting THCA immunotherapy efficacy in clinical trials. However, future investigations require relevant basic experiments and large sample clinical trials.