Introduction

Locoregionally advanced head and neck squamous cell carcinoma (HNSCC) is associated with poor 5-year survival of only 40% for non-viral mediated disease, and substantial functional morbidity with combined multimodality therapy1,2,3,4. Due to the poor success of systemic cytotoxic chemotherapy in treating recurrent/metastatic (R/M) HNSCC, the recent clinical focus has shifted to immunotherapy with antibodies targeting T cell inhibitory receptors that function as immune checkpoints, such as programmed death 1 (PD-1). Nonetheless, PD-1 inhibitors were reported to unleash anti-tumor immunity and achieve durable clinical responses in only 15–20% of treated patients in front-line recurrent/metastatic (R/M) setting5,6,7,8, with 4-year overall survival rate of about 15% overall and up to 22% among cases with a PD-L1 combined positive score (CPS) of 20 or greater5,6,7,8,9. Such disparity in treatment benefits between patients, paired with the absence of any standard effective therapies that target immunotherapy resistance, necessitates the identification of predictive biomarkers to better inform clinicians’ therapeutic decisions3. While it was reported that higher PD-L1 expression, assessed by the CPS, is predictive of a favorable response to immune checkpoint inhibitors (ICI), CPS remains an imperfect biomarker, with the majority of patients ultimately develo** therapeutic resistance. Therefore, identification of improved biomarkers of response and resistance is of great importance, as biomarker-directed therapeutics after progression on immunotherapy (such as HRAS) suggest that determining specific mechanisms driving treatment resistance in HNSCC is key to therapeutic development10,11. Furthermore, little is known regarding the mechanisms of resistance to ICIs in head and neck cancer, further limiting the ability to predict non-responders and to investigate novel therapeutics targeting mechanisms of resistance to ICIs in R/M HNSCC12,13.

A high prevalence of alterations in the MYC oncogene is well documented in various solid malignancies, along with its association with aggressive disease and poor clinical outcomes14,15,16. While recurrent MYC gain-of-function mutations were found in certain human lymphomas17,18,19, in HNSCC, amplification appears to be the predominant genetic aberration that occurs in the MYC gene20, with an estimated prevalence of 12% in the HNSCC cohort of The Cancer Genome Atlas (TCGA), whereas mutations occur in only a small subset (1.2%) of patients21. MYC amplification is known to have broad influence on the transcriptome, driving upregulation of various mitogenic and survival signaling pathways associated with cell growth, proliferative, anti-apoptotic, and metabolic processes in ovarian, breast, and lung cancers among others22,23,24,25. Furthermore, MYC plays an important role in suppressing the host anti-tumor immune response, through various mechanisms involving modifications to the tumor microenvironment via inhibitory cytokines (e.g., TGFβ and immunomodulatory molecules such as PD-L1, CD47, and MHC I)26,27,28,29. While the tumorigenic effects of MYC are well known, the molecular mechanisms by which the mutation or amplification of this gene may confer treatment resistance in HNSCC remains under-investigated30. A recent case report of a patient with recurrent/metastatic (R/M) HNSCC highlighted a differential response to nivolumab in metastatic lesions secondary to the acquisition of MYC amplification. The MYC amplified lesion was resistant to the ICI while all other lesions, devoid of MYC amplification, responded to treatment, suggesting MYC amplification may play a role in ICI resistance in R/M HNSCC31. Supporting this suggestion, another study showed that MYC amplification regulates PD-L1 expression and is involved in a decreased response to ICI therapy in esophageal squamous cell carcinoma32.

Here we present a unique case of a patient with R/M HNSCC who, despite initial response to nivolumab-based therapy, developed rapidly progressive, metastatic disease after the acquisition of a MYC amplification. We conducted transcriptomic analysis of this patient’s HNSCC to identify the potential molecular processes through which MYC amplification may confer resistance to immunotherapy. In light of these findings, we searched our in-house genomic sequencing platform and identified seven additional patients with MYC amplified HNSCC along with 48 MYC wild-type matched controls, to further elucidate the broader clinicopathologic characteristics of MYC-driven disease. Finally, we performed gene expression and pathway analysis using the transcriptomic data obtained from TCGA to further explore key molecular networks associated with MYC amplification in HNSCC. Collectively, this work seeks to highlight MYC amplification as a potential mechanism of treatment resistance and suggest its use as a predictive biomarker and a potential therapeutic target in the treatment of R/M HNSCC24,33,34.

Results

Case presentation

A 58-year-old male with an 80-pack-year tobacco history but quit at diagnosis presented in July 2020 noting a right-sided neck mass (Fig. 1). A CT soft tissue neck with contrast demonstrated abnormally enlarged bilateral lymph nodes with a heterogenous appearance and a soft tissue prominence of the anterior hypopharynx at the level of the glottis. A subsequent ultrasound-guided fine-needle aspiration of a right cervical node demonstrated scant metastatic squamous cell carcinoma involving fibrous and lymphoid tissue, at which time the patient was referred to our institutional multidisciplinary head and neck cancer team. Upon endoscopic evaluation, a 3 cm epiglottic pedunculated lesion was noted. A biopsy of the epiglottic lesion was performed and demonstrated squamous cell carcinoma. Immunohistochemical staining for p16 was negative. Subsequently, a PET/CT scan was conducted, which showed metastatic cervical lymph nodes, most prominent at bilateral level 2, with mild hypermetabolic activity of a lesion along the anterior commissure of the glottis and was staged as (cT1N2cM0, Stage IVa, American Joint Committee on Cancer 8th edition). A baseline Oncoplus molecular analysis was retrospectively conducted on the primary epiglottic specimen, showing CDKN2A loss, TP53 loss, BAP1 rearrangement, and KDM6A mutation. The specimen was found to be microsatellite stable (MSS) and had a tumor mutation burden (TMB) of 18.0 mutations/mb. PD-L1 immunohistochemical staining was negative (CPS < 1%).

Fig. 1: Course of disease.
figure 1

Timeline depicting the course of disease and treatment, during which the patient acquired a MYC amplification that was not detected on baseline genetic testing and subsequently developed highly aggressive, immunotherapy-resistant disease. Numbered images at the bottom correspond to the indexes indicated across the course of the disease depicted above. BL: bilateral, EBUS-FNA: endobronchial ultrasound-guided fine-needle aspiration.

Full size image

The patient was enrolled in a phase II clinical trial for locally advanced, HPV-negative HNSCC evaluating nivolumab-based chemoimmunotherapy followed by response-stratified locoregional therapy (NCT03944915). He initiated induction chemoimmunotherapy with three cycles of nivolumab 360 mg day 1, paclitaxel 100 mg/m2 days 1/8/15, and carboplatin AUC 5 day 1 of 21-day cycle followed by imaging demonstrating 42% tumor shrinkage per RECIST 1.1 and was subsequently treated per protocol with chemoradiotherapy (CRT) consisting of cisplatin 100 mg/m2 q21 days with daily fractionated radiation therapy (RT) at 2 Gy per fraction to total of 70 Gy over 35 fractions to gross disease and bilateral neck. Following completion of CRT he initiated adjuvant nivolumab per protocol. Three months after completion of CRT, the patient underwent PET demonstrating a complete metabolic response. MRI demonstrated no residual cervical lymph node metastases or discernible measurable laryngeal tumor consistent with a complete response by RECIST 1.1 criteria, and direct laryngoscopy showed a complete clinical response. However, during adjuvant nivolumab, six months after completion of chemoradiation, the patient underwent CT imaging demonstrating new mediastinal and hilar lymphadenopathy. A subsequent PET scan showed 18F-fluorodeoxyglucose (FDG) avidity of the mediastinal and bilateral hilar adenopathy along with a left adrenal nodule. Endobronchial ultrasound and fine-needle aspiration of a mediastinal lymph node demonstrated p16 negative, metastatic SCC in September 2021. Repeat Oncoplus molecular analysis was performed on the mediastinal lymph node and revealed an acquired MYC amplification in addition to the genetic aberrations detected in the baseline analysis (e.g., CDKN2A and TP53 loss, BAP1 rearrangement as well as KDM6A mutation) supporting a clonal relationship between primary and progressive disease. Of note, this biopsy was also found to be PD-L1 negative on immunohistochemical staining.

The patient started a chemotherapy regimen consisting of carboplatin (AUC 5 day 1), 5-fluorouracil (5-FU, 1000 mg/m2/d days 1–4), and cetuximab (400 mg/m2 loading followed by 250 mg/m2 weekly of 21-day cycle). Repeat imaging demonstrated progressive disease with interval increase in both the intrathoracic lymphadenopathy (LAD) and adrenal lesion as well as new vertebral metastases in T10 and T12 after four cycles of chemotherapy/cetuximab. The patient received palliative XRT to the right hip, however, his performance status rapidly worsened and he ultimately died from progressive disease five months after starting chemotherapy/cetuximab for recurrent/metastatic HNSCC. Given the observation of an acquired MYC amplification upon rapidly progressive disease while receiving anti-PD-1 therapy following an initial partial response, we proceeded to characterize clinical-pathologic features of MYC amplified HNSCC from our internal dataset, performed RNA-Seq of the paired samples with and without MYC amplification from our case study, and analyzed transcriptomic data of MYC amplified cases obtained from TCGA-HNSC dataset.

Clinical-pathologic characterization of recurrent MYC amplified HNSCC

To further interrogate the clinical and molecular profile of tumors with acquired MYC amplification, we performed a retrospective medical records review of patients with HNSCC treated at our institution between 2018 and 2021. We identified 8 cases (including the index patient) who were positive for MYC amplification based on the OncoPlus assay at the time of recurrence, and 48 matched control cases bearing a wild-type MYC (Supplementary Data 1). MYC amplified cases demonstrated a median age of 61, 25% were p16 positive oropharynx, all cases (100%) had the local disease at the time of diagnosis and initially received chemoradiotherapy, while the majority (63%) ultimately received immunotherapy during their treatment course. MYC amplification was significantly associated with enrichment for the laryngeal primary site (Supplementary Table 1) and showed numerical trend toward elevated frequency of TP53 and CDKN2A genetic aberrations, similar to previous reports (Supplementary Table 2)35,36,37. While median age, gender, treatment modality, p16 status (a surrogate biomarker for HPV-positivity), as well as tumor stage at diagnosis, were not different between the two groups, MYC amplified patients showed non-significant trend toward higher rates of develo** recurrent and/or metastatic disease following primary therapy, with 100% of cases among MYC amplified compared to 72.9% among the wild-type MYC counterparts (p = 0.08; Supplementary Table 1). Furthermore, although the overall survival was not statistically significantly different with this small sample size available for analysis, median survival of MYC amplified patients was 40.6 months, compared to 49.1 months across the MYC wild-type individuals (Supplementary Table 1)36.

RNA-Seq analysis of paired patient samples with acquired MYC amplification

Given that transcriptional changes associated with the acquisition of MYC amplification in setting of immunotherapeutic resistance is poorly understood in HNSCC, we next performed RNA-Seq analysis of the tumor specimens collected from the patient described in the case report at baseline and upon rapid progression of R/M disease while receiving nivolumab (Supplementary Data 2). A purely descriptive comparative transcriptomic analysis of paired samples showed that genes known to be modulated by MYC were upregulated at disease progression, such as WNT/\(\beta\)-catenin pathway agonists, genes central to the glycolytic pathway (e.g., encoding phosphofructokinase, hexokinase II and enolase), fatty acid metabolism (e.g., acetyl-CoA carboxylase, fatty acid synthase, and ATP-citrate lyase), regulators of nucleotide (e.g., ribonucleotide reductase and ectonucleoside triphosphate diphosphohydrolase) and protein (e.g., eukaryotic translation and elongation factors) synthesis networks, as well as molecules that play role in cell proliferation and survival (Fig. 2). Conversely, WNT antagonists, tumor suppressors known to negatively regulate cell cycle, apoptosis and cell adhesion were downregulated in the MYC amplified disease. Notably, genes involved in host anti-tumor immune response such as immune activation regulators, class I human leukocyte antigens (HLAs), STAT 1/2, and chemokines associated with recruitment of immunosuppressive cells were almost universally downregulated in the MYC amplified tumor compared to baseline. While these descriptive observations may provide a snapshot of the transcriptomic changes associated with acquired MYC amplification, thousands of genes (nearly 15% of transcriptome) are predicted to be direct targets of MYC, highlighting the complexity of the MYC-regulated signaling network38,39,40.

Fig. 2: RNA-Seq analysis of rapidly progressive, metastatic disease after the acquisition of a MYC amplification.
figure 2

RNA sequencing was performed on the matched tumor specimens collected at baseline and upon progression. The reads were aligned against the human reference genome GRCh38/hg38 and gene counts were calculated. The figure indicates fold change between expression levels of the selected genes in tumor biopsy collected upon disease progression (acquired MYC amplification) and baseline specimen (wild-type MYC). Genes with fold change value ± 1.5 were included and grouped by their major roles in WNT signaling, metabolism, survival, and proliferation signaling networks.

Full size image

Acquisition of MYC amplification is associated with broad transcriptomic changes

As we could not generate statistically meaningful conclusions based on the analysis of a single case sequencing, we next performed transcriptomic evaluation using samples obtained from the TCGA-HNSC dataset (Supplementary Data 3). Notably, tumors carrying amplified MYC (n = 59) were enriched for laryngeal disease, TP53 and CDKN2A mutations (Supplementary Table 3, Supplementary Fig. 1), and exhibited significant upregulation of numerous driver genes known to be associated with carcinogenesis in several types of cancer, including HNSCC (e.g. ELAVL2, CXCL5, FGFR4, CCNP, STC2, POU5F1B and SYT12)41,42,43,44,45,46,47,48,49,50. Significantly down-regulated genes contained known tumor suppressors and regulators of immune response (MYD88, NAT1, DDB2, PPP2R5C, LPAR6, CNP, CASP1, CXCL10, and CEACAM1)51,52,53,54,55,56,57,58,91,92. To study the dysregulation of cellular processes between MYC amplified and MYC wild-type cases, iPANDA algorithm67 was applied using the Reactome pathways database93. iPANDA calculates the activation or inhibition score for each pathway by combining precalculated gene coexpression data with gene importance factors based on the degree of differential gene expression and pathway topology decomposition. Significantly dysregulated pathways with iPANDA score > 0.01 or < −0.01 were considered as activated and inhibited, respectively (Supplementary Data 6). P-values for the iPANDA pathway activation scores were obtained using weighted Fisher’s combined probability test. Detailed description and statistical credibility of the iPANDA score was previously published67. Unsupervised complete-linkage clustering was performed following Farthest Point algorithm and Euclidean metric94. Gene set enrichment analysis (GSEA) was performed with the GSEAPY python package95 using two collections of gene sets obtained from Enrichr library96 - MSigDB_Hallmark_2020 and KEGG_2021_Human (Supplementary Data 5).

Statistical analysis

Survival analysis was prepared in PandaOmics using the Kaplan–MeierFitter function from the lifelines Python package (two-sided log-rank test). The differences in clinical and disease-specific characteristics i(e.g. age, gender, primary site, HPV/p16 status, TNM stage, gene mutation frequency, etc.) between MYC amplified and wild-type cases (in the internal cohort and TCGA-HNSC dataset) were calculated by two-sample t-tests and chi-square tests of independence using GraphPad Prism software. The significance level was defined as 0.05.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.