Background
Head and neck squamous cell carcinoma (HNSC) is the sixth most common cancer worldwide and is mainly induced by smoking, alcohol consumption, and human papillomavirus [
1]. Currently, clinical treatment modalities for HNSC include surgical resection, targeted therapy, adjuvant chemotherapy, and radiotherapy [
2]. However, due to serious side effects, patients have a limited response rate to these therapeutic measures [
3]. In the USA, the 5-year survival rate of HNSC patients is still relatively low, ranging from 40 to 50% [
4]. To further improve patient outcomes, more precise patient stratified management and individualized treatment approaches are urgently needed.
The PI3K/AKT pathway is one of the most important pathways regulating many cellular processes, and its abnormal activation leads to cell cycle dysregulation, genomic instability, cell differentiation defects, and persistent mitotic signaling in HNSC patients, thereby enhancing the malignancy of tumors [
5]. In terms of the tumor immune microenvironment (TIME), it has been demonstrated that PI3K pathway also regulates many important aspects of immune cell differentiation, development, and function as well as activates numerous immunosuppressive factors, including the expression of immune checkpoints and the infiltration of immunosuppressive cells [
6,
7]. In addition, recent studies have illustrated that disrupting the HER3-PI3K-mTOR oncogenic signaling axis reverses the suppressive TIME, enhancing the immunotherapeutic efficacy of PD-1 inhibitors [
8‐
11]. However, the mutation landscape of the PI3K pathway genes in HNSC and the impact of PI3K pathway mutation on the TIME as well as immunotherapy of HNSC remain unknown.
Following the success of immunotherapy in melanoma, an increasing number of clinical trials have confirmed its remarkable efficacy in specific patients [
12]. Recently, two phase III randomized trials have revealed that PD-1 inhibitors can prolong patient survival and have better efficacy than chemotherapy in second-line platinum-refractory recurrent/metastatic HNSC [
13]. As a classical anti-PD-1 monoclonal antibody, pembrolizumab is superior to cetuximab in combination with chemotherapy in phase III clinical trials and has been approved by the FDA for first-line treatment of recurrent or metastatic HNSC [
14]. Several biomarkers, including PD-1, PD-L1, tumor mutation burden (TMB), IFN-γ signature, and HPV factors, have been identified to evaluate the efficacy of immunotherapy in HNSC patients, but only PD-L1 is widely used in clinical practice, and the overall predictive effect is unsatisfactory [
15]. Given the relatively high cost and immune-related severe adverse events of immunotherapy, there is an urgent need to explore more accurate and practical predictive markers.
In this study, we characterized PI3K pathway mutation landscape in HNSC patients and analyze its prognostic value in immunotherapy and non-immunotherapy cohorts. After stratifying HNSC patients into the mutation and wild groups, we systematically evaluated the relationship between PI3K pathway mutation and immunological characteristics from multiple dimensions including TMB, immune cell infiltration, immunomodulator expression, and immunotherapy response prediction, respectively. In addition, we also explored potentially sensitive drugs in the wild group population that responded poorly to immunotherapy to significantly improve their prognosis. Overall, our study may provide a reference for early clinical identification of immunotherapy-sensitive HNSC patients to receive further individualized therapy.
Discussion
HNSC is the sixth most common cancer worldwide. Currently, treatment modalities for HNSC include surgical resection, targeted therapy, adjuvant chemotherapy, radiotherapy, and immunotherapy [
35,
36]. Mutation status of particular genes could predict therapy outcomes, for example, mutation of the p53 gene is associated with an increased risk of locoregional failure in patients with invasive HNSC who are treated with radiation therapy, PIK3CA mutation predicted worse DFS in a prospective cohort of HPV-associated oropharyngeal squamous cell carcinoma patients treated with definitive chemoradiation, mutation in the epidermal growth factor receptor ligand-binding domain confers increased sensitivity to cetuximab, and TMB and PRKDC mutation are predictive biomarkers for immunotherapy [
18,
37‐
41]. However, the predictive ability of these biomarkers to ICIs response is still limited. We then explored the relationship between PI3K pathway mutation, the most frequently mutated pathway in HNSC, and the immune microenvironment as well as immunotherapeutic efficacy.
In this study, we divided HNSC patients into the mutation and wild groups according to the mutation status of PI3K pathway genes and found that the mutation group had longer OS observably in the immunotherapy cohort. However, the mutation status was not significantly associated with OS and DFS in the non-immunotherapy cohorts. Further studies exhibited that the PI3K pathway mutation group owned significantly higher TMB as well as mutation frequencies of oncogenes and tumor suppressor genes, relatively active immune-related pathways, richer immune cell infiltration, and higher expression levels of immunomodulators. The predicted results of TIDE and SubMap consistently indicated that HNSC patients in the PI3K pathway mutation group were more likely to benefit from immunotherapy. Thus, PI3K pathway mutation status may be a reliable biomarker to predict the efficacy of immunotherapy in HNSC patients.
In HNSC patients, aberrant activation of the PI3K/AKT/MTOR pathway promotes malignant progression and impacts many vital aspects of immune cell development, differentiation, and activation of immunosuppressive factors [
5‐
7,
42]. Moreover, previous studies have demonstrated that the PI3K pathway is the most common mutant oncogenic pathway (30.5%) and predicts a higher mutation rate in HNSC [
43]. We, therefore, investigated the landscape and prognostic value of PI3K pathway mutation in HNSC. The results of all three cohorts consistently indicated that PI3K pathway genes are commonly mutated (38.8%, 28.1%, and 20%; Figs.
1A, B and S
2A). Kaplan-Meier curves and multivariate Cox regression analyses demonstrated that patients who harbored PI3K pathway mutation had significantly prolonged OS in the immunotherapy cohort. Interestingly, there were no statistical differences in OS and DFS between the PI3K pathway mutation and wild groups in the non-immunotherapy cohorts. This suggested that PI3K pathway mutation may herald a better efficacy of immunotherapy in HNSC patients, and it may serve as a potential biomarker to select immunotherapy-sensitive patients for further immunotherapy.
To explore the potential mechanisms that may contribute to the differential efficacy of immunotherapy and prognosis in patients with PI3K pathway mutation and wild groups, we performed an integrated genomic variation analysis between the two groups. Consistent with Lui et al., the mutation group possessed significantly higher mutation rates of tumor suppressor gene CDKN2A and oncogenes TTN, NOTCH1, and MUC16 [
43]. Many studies have confirmed that higher TMB in solid tumors predicts better immunotherapeutic outcomes, and the significantly higher TMB in the PI3K pathway mutation group of the three cohorts also supports its better immunotherapeutic efficacy and prognosis in the immunotherapy cohort [
44‐
46]. In contrast, CNA analysis showed increased copy number amplification and deletion loads at both focal and chromosomal arm levels in the PI3K pathway wild group, suggesting that CNA may play a key role in malignant progression in the wild group. The two malignant biological behaviors, gene mutation, and CNA, predominated in the PI3K pathway mutation and wild groups, respectively, which is consistent with the fact that there were no differences in prognosis between the two groups in the non-immunotherapy cohorts TCGA-HNSC and MD-Anderson [
47‐
49].
Next, we performed GSVA and GSEA enrichment analyses separately to explore the underlying biological mechanisms of the PI3K pathway mutation and wild groups. The results exhibited that the immune-related pathways, including B cell-mediated immunity, MHC protein binding, T cell receptor complex, and positive regulation of T cell proliferation, were significantly enriched in the PI3K pathway mutation group. On the other hand, specific pathways associated with malignant tumor progression such as angiogenesis, positive regulation of mitochondrial fission, and metabolism-related pathways such as fatty acid beta oxidation were relatively activated in the PI3K pathway wild group, indicating that we can develop specific drugs targeting tumor angiogenesis, cell proliferation, and metabolism for patients in the wild group.
Considering the significant enrichment of immune-related biological processes in the PI3K pathway mutation group, we further evaluated the abundance of immune cell infiltration and immunomodulators expression levels between the two groups. The results from xCell and CIBERSORT consistently indicated that killer cells such as memory CD4 T cells and central memory CD8 T cells, which play critical roles in anti-tumor immunity and immunotherapy, were significantly richer in the PI3K pathway mutation group. Similarly, the expression levels of immunomodulators, including co-stimulatory, co-inhibitory, ligand, and receptor molecules, were also significantly higher in the PI3K pathway mutation group. Thus, we have reason to believe that anti-tumor immunity will be more active in the PI3K pathway mutation group after targeting these immunomodulators using ICIs, and reactivated killer immune cells will exert anti-tumor function and improve patient outcomes [
50‐
52].
Our previous findings exhibited that the PI3K pathway mutation group possessed significantly higher levels of TMB, killer immune cell infiltration, and immunomodulators expression, hinting that patients in the mutation group are more sensitive to immunotherapy [
18,
45,
50,
53,
54]. The results from TIDE and SubMap algorithms also consistently demonstrated that patients in the PI3K pathway mutation group were more likely to benefit from immunotherapy. On the other hand, using drug sensitivity data from CTRP and PRISM databases, we identified five potential therapeutic drugs for HNSC patients in the PI3K pathway wild group who did not respond well to immunotherapy. Among them, “BMS-536924” and “linsitinib” are IGF-1R/IR inhibitors that can inhibit glucose metabolism in tumor cells and then exert anti-tumor effects [
55,
56]. This is consistent with the significant enrichment of metabolism-related biological pathways in the PI3K pathway wild group and provides a reference for treating patients with relatively poor prognosis in the wild group.
This study is the first to elaborate the mutation landscape of the PI3K pathway in HNSC and found that PI3K pathway mutation predicts better immunotherapeutic efficacy and significantly prolonged OS in HNSC patients after receiving immunotherapy. Second, we explored the utility and generalizability of our findings and confirmed that PI3K pathway mutation could still accurately predict better immunotherapeutic outcomes in other common tumors such as esophagogastric and colorectal cancers. However, our study also has some shortcomings. For instance, due to the lack of immunotherapy cohorts for HNSC, we collected only 129 samples with complete mutation, immunotherapy, and prognosis information from the MSKCC-2019 cohort, and more multi-center prospective studies are needed to validate our results in the future.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.