Oncogene amplification of CCND1 via extrachromosomal DNA impacts overall survival in head and neck squamous cell carcinoma

Abstract

Background

Extrachromosomal DNA (ecDNA) drives oncogene amplification and transcriptional rewiring in many cancers. For head‑and‑neck squamous cell carcinoma (HNSCC), the prevalence of ecDNA has been reported on the top across various cancer types. However, the clinical significance of ecDNA remain incompletely defined. This study aimed to (1) systematically evaluate ecDNA frequency and architecture in HNSCC, (2) examine its associations with clinical characteristics, and (3) identify driver oncogenes amplified by ecDNA, if any, that might impact patient prognosis.

Methods

We analyzed primary tumor data from 153 HNSCC cases in TCGA. ecDNAs were detected using the AmpliconArchitect pipeline on TCGA whole-genome sequencing. HPV status was retrieved from the TCGA HNSC landmark study. Differences in the frequency of ecDNA based on anatomical subsite and sex were assessed using chi-square tests. Patient age was compared using independent samples t-test. We compiled a panel of 38 candidate oncogenes from TCGA HNSC landmark paper and published ecDNA studies, then evaluated how often these genes were amplified on ecDNA in our cohort. The association of each ecDNA-amplified gene with overall survival (OS) was estimated using Kaplan-Meier curves, and differences in survival were analyzed using the log-rank test.

Results

ecDNA was present in 68.0% (104/153) of tumors. The most common ecDNA structures were linear (47.7%), followed by break‑fusion‑break (BFB, 30.1%), heavily rearranged (26.1%), and circular (25.5%). No significant differences were observed by anatomical site or sex. Patients with ecDNA were younger (median 57.7 vs. 62.0 years, p = 0.038). Among the 38 oncogenes examined, amplification of a cluster of 11q13 genes including FADD, CTTN, and CCND1 via ecDNA occurred most frequently, followed by EGFR, BIRC2, WHSC1L1 and YAP1. Overall, ecDNA status alone did not significantly affect OS (log‑rank p = 0.442). However, patients harboring CCND1‑ecDNA had markedly poorer OS (log‑rank p = 1.5 × 10⁻⁴), underscoring the adverse prognostic impact of CCND1 amplification through ecDNA. Notably, CCND1 ecDNA events are mutually exclusive with HPV-positive tumors, suggesting a distinct oncogenic pathway in HPV-negative HNSCC.

Conclusions

ecDNA is common in HNSCC and is associated with a younger age of onset. Although ecDNA presence itself does not independently predict overall survival, ecDNA‑mediated amplification of CCND1 is strongly linked to worse outcomes. These findings highlight the amplification of CCND1 and other 11q13 genes via ecDNA as a key oncogenic mechanism in HNSCC and suggest that it could serve as a prognostic biomarker and potential therapeutic target.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-05199-3.

Introduction

Head and neck squamous cell carcinoma (HNSCC) represents a diverse group of malignancies that arise from the mucosal epithelium of the oral cavity, pharynx, and larynx. Even with today’s multimodal treatments involving surgery, cytotoxic chemotherapy, and radiation, HNSCC still imposes a heavy health burden globally and is characterized by high rates of morbidity and mortality [1, 2]. A key challenge in improving cancer outcomes is our limited knowledge of the mechanisms that drive tumor growth, recurrence, and treatment resistance. Traditional analyses have largely focused only on mutations within the chromosomes, but recent evidence suggests that structural instability in the genome is actually a critical factor in how cancer progresses clinically [3].

One particularly potent mechanism of this instability is the presence of extrachromosomal DNA (ecDNA). Unlike the linear DNA found in chromosomes, most ecDNA, especially the cancer-relevant ones, exists as large, circular particles that lack centromeres [4]. This structural difference causes them to segregate randomly during cell division [5]. As a result, tumors can rapidly accumulate high numbers of oncogene copies and develop significant genetic variety within a single tumor mass. Furthermore, the chromatin in ecDNA is highly accessible, which allows for massive transcriptional rewiring and the overexpression of oncogenes [6].

HNSCC is notable for harboring one of the highest frequencies of ecDNA among solid tumors [6, 7]. Despite this distinction, the specific genomic landscape remains poorly characterized, particularly regarding the structural topology and oncogenic cargo found in these patients. Consequently, the clinical relevance of these extrachromosomal elements is largely unknown, and their impact on patient survival has not been definitively established.

To address this gap, we designed this study to (1) systematically evaluate the frequency and architecture of ecDNA in HNSCC, (2) examine how it relates to clinical characteristics, and (3) identify specific driver oncogenes amplified by ecDNA that may independently impact patient prognosis. By clarifying these mechanisms, we hope to provide new insights into the biology of HNSCC and identify potential biomarkers for better risk stratification.

Methods

ecDNA identification and classification

We analyzed primary tumor data from 153 HNSCC cases in TCGA. The tool AmpliconArchitect [8] was used to infer the architecture of amplicons. Briefly, amplicons are defined as genomic segments greater than 10 kb with copy numbers exceeding four copies. AmpliconArchitect uses aligned Whole-Genome Sequencing (WGS) data and user-defined seed intervals as input, constructs a breakpoint graph from identified intervals and rearrangement junctions, and reconstructs the underlying amplicon structures. Based on graph topology, amplicons are classified into four categories: circular, linear, break-fusion-bridge (BFB), and heavily rearranged. Circular paths within the breakpoint graph are indicative of ecDNA-like structures. Representative AmpliconArchitect plots illustrating these major structural classes are provided in Supplementary Fig. 1.

HPV status determination

HPV status was retrieved from the definitive TCGA HNSC landscape study [9]. This classification was established through the rigorous mapping of RNA-Seq reads to viral genomes, which provided a robust molecular definition of viral infection concordant with genomic and sequencing data.

Clinical characteristics relevance

Associations between ecDNA frequency and clinical variables were assessed using Chi-square tests for categorical factors (sex, anatomic site, pathological stage). Continuous variables (age) were compared between groups using Student’s t-test, based on differences in mean values. Survival distributions were visualized with Kaplan-Meier curves and analyzed by the log-rank test, with hazard ratios estimated via Cox proportional hazards regression.

Characterization of ecDNA-mediated oncogene amplification

We curated a targeted panel of 38 candidate driver genes to characterize the functional impact of extrachromosomal amplification. This set was selected based on two criteria: genes established as canonical drivers with high amplification frequency in HNSCC, and those known to be preferentially amplified via ecDNA in pan-cancer studies. To strictly define ecDNA-mediated amplification, we mapped these loci to the hg19 reference genome. A candidate gene was classified as being amplified by ecDNA only if its genomic coordinates were fully encapsulated within a reconstructed ecDNA circular topology. We then evaluated the status of these specific drivers to determine their association with patient survival and clinical outcomes.

Differentially expressed genes analysis

To characterize the specific transcriptional perturbations driven by CCND1 amplification on ecDNA (ecCCND1), we compared the global gene expression profiles of ecCCND1-positive tumors against negative controls. Statistical comparisons were conducted using two-tailed independent-samples t-tests. To account for multiple hypothesis testing, raw p-values were adjusted using the Benjamini-Hochberg False Discovery Rate (FDR) method. We defined significantly differentially expressed genes (DEGs) based on a dual threshold: FDR < 0.25 and a minimum 2-fold difference in expression levels between the groups.

Gene set enrichment analysis (GSEA)

To elucidate the downstream transcriptional consequences of CCND1 amplification on ecDNA, we performed GSEA to detect transcriptomic differences in cancer signaling pathways between ecCCND1-positive and negative tumors. Genes were preranked according to their t-statistic derived from the differential expression analysis. Using the Broad Institute’s GSEA software (v2.2.1), we interrogated the MSigDB Hallmark gene sets (v2025.1) to focus on distinct cellular processes [10, 11]. Pathways were considered significantly enriched if they met the threshold of a nominal p < 0.05.

Results

Prevalence and clinical feature association of ecDNA in HNSCC

As depicted in Figs. 1A and 68% of the 153 HNSCC cases contained identifiable focal amplifications. The most common architectures were linear (47.7%), followed by break-fusion-bridge (BFB, 30.1%), heavily rearranged (26.1%), and circular forms (25.5%). Patients with ecDNA were significantly younger (median 57.7 vs. 62.0 years, t-test p = 0.038, Fig. 1B). However, no association was observed between ecDNA presence and sex, tumor site, or pathological stage (Table 1).

Fig. 1.

Characterization of ecDNA prevalence, clinical association, and oncogene content in HNSCC. A Prevalence of distinct amplicon structures (Circular, HR, BFB, Linear) and overall ecDNA frequency across HNSCC tumors. B Comparison of patient age at diagnosis between ecDNA-negative and ecDNA-positive groups (p = 0.038). C Frequency of oncogenes amplified by ecDNA. Gene names highlighted in bold indicate localization to the 11q13 chromosomal region (CTTN, FADD, CCND1)

Table 1.

Demographic and clinical characteristics of the study cohort

		Tumors with ecDNA	Tumors without ecDNA
		N = 104	N = 49	P values
Age		57.7 ± 13.1	62 ± 11	0.038†
Gender
	M	25 (24%)	17 (34.7%)	0.235‡
	F	79 (76%)	32 (65.3%)
Site
	Hypopharynx	2 (1.9%)	0 (0%)	0.383‡
	Larynx	17 (16.3%)	4 (8.2%)
	Oral Cavity	68 (65.4%)	35 (71.4%)
	Oropharynx	17 (16.3%)	10 (20.4%)
Stage
	I	1 (1%)	3 (6.1%)	0.305‡
	II	22 (21.2%)	11 (22.4%)
	III	22 (21.2%)	9 (18.4%)
	IV	59 (56.7%)	26 (53.1%)
HPV status
	Positive	22 (22%)	13 (28.9%)	0.492‡
	Negative	78 (78%)	32 (71.1%)
Median overall survival (year)		2.12	2.34	0.442§
	%death	50%	41%

Patients were stratified into ecDNA-positive (N = 104) and ecDNA-negative (N = 49) groups. Values are expressed as mean ± SD or n (%). Statistical comparisons were performed using the t-test (†), Chi-square test (‡), or Log-rank test (§). Bold text indicates statistical significance (P < 0.05)

The 11q13 amplicon dominates the ecDNA landscape in HNSCC

A primary mechanism by which ecDNA drives tumor initiation and progression is through the massive amplification of driver oncogenes. To determine the specific oncogenic cargo carried by ecDNA in HNSCC, we interrogated a panel of 38 established driver genes derived from the landmark TCGA study [9]. Our analysis revealed a striking predominance of the 11q13 locus among the amplified targets. Specifically, genes located within this amplicon, including FADD, CTTN, and CCND1, were the most frequently detected on ecDNA and were observed in 39, 39, and 37 cases respectively. Beyond this cluster, EGFR was the next most common driver (13 cases), followed by BIRC2 (9 cases), WHSC1L1 (9 cases), and YAP1 (9 cases) (Fig. 1C).

CCND1 amplification by ecDNA (ecCCND1) significantly related with worse overall survival

To assess clinical relevance, we analyzed survival outcomes relative to ecDNA status. Comparisons between the total ecDNA-positive and ecDNA-negative tumors revealed similar survival outcome (p = 0.442, log-rank test). While tumors driven by circular ecDNA structures showed a tendency toward more aggressive behavior, this difference remained statistically non-significant (p = 0.084, log-rank test, Fig. 2A). However, at the gene level, distinct associations emerged. Although amplifications of neighboring 11q13 drivers like ecFADD and ecCTTN were significantly linked to shorter survival (p = 0.00057, log-rank test), the association was even stronger for ecCCND1. This specific amplification of CCND1 served as the most robust predictor of mortality (Hazard Ratio = 2.5, log-rank p = 0.00015, Fig. 2B).

Fig. 2.

Prognostic impact, molecular profile, and viral exclusivity of CCND1-containing ecDNA (ecCCND1). A, B Kaplan-Meier analysis of overall survival stratified by (A) circular ecDNA presence (p = 0.084) and B ecCCND1 status. Patients with ecCCND1 exhibit significantly reduced survival (p = 0.00015, HR = 2.5). C Scatter plot of CCND1 copy number versus log2 expression, showing a significant positive correlation (Spearman ρ = 0.75, p = 8e-29), colored by ecCCND1 status (blue, no ecCCND1; red, with ecCCND1). D Volcano plot of differentially expressed genes in ecCCND1-positive tumors, highlighting key upregulated 11q13 genes. E Venn diagram illustrating significant mutual exclusivity between HPV-positive status and ecCCND1 presence (p = 0.0012)

CCND1 expression is higher in ecDNA+ tumors containing CCND1 than tumors without ecDNA

We next sought to confirm that the observed genomic amplification results in a corresponding upregulation of gene expression. Analysis of matched transcriptomic profiles revealed a clear dosage-dependent effect: tumors harboring ecCCND1 exhibited significantly higher CCND1 transcript levels compared to ecCCND1-negative tumors (Figure S2B, p < 1e − 10). Consistently, tumors with ecCCND1 showed both higher CCND1 copy number and expression (Fig. 2C), with a strong positive correlation between copy number and transcript levels (Spearman ρ = 0.75, p = 8e-29). Together, these findings identify ecCCND1 as a major driver of transcriptional heterogeneity in this population.

Genes over- and under-expressed in CCND1 containing ecDNA+ tumors

We applied independent t-tests to delineate the transcriptional signature specific to ecCCND1-positive tumors. This analysis revealed a distinct overexpression profile heavily enriched for genes co-located within the 11q13 amplicon. Fifteen genes were significantly upregulated, including CCND1 itself, alongside key neighboring oncogenes such as PPFIA1, FADD, ORAOV1, CTTN, ANO1, and FGF19. Other upregulated targets included MRPL21, TPCN2, and IGHMBP2. Notably, 11 of the 15 significantly upregulated genes co-located within the 11q13 region. Conversely, the analysis identified significant downregulation of specific markers, most notably the carcinoembryonic antigen-related cell adhesion molecules CEACAM5 and CEACAM7. (Fig. 2D)

Pathways up- and down-regulated in CCND1 containing ecDNA+ tumors

To elucidate the functional landscape driven by ecCCND1, we performed Gene Set Enrichment Analysis (GSEA). This analysis revealed a profound rewiring of cellular states, with 14 Hallmark pathways significantly upregulated in ecCCND1-positive tumors (Fig. 3 green bars). The enriched phenotype was characterized by heightened proliferative signaling (Myc targets, E2F targets, G2M checkpoint) and metabolic reprogramming (Glycolysis, Oxidative Phosphorylation). Additionally, pathways associated with tumor aggressiveness, such as Epithelial-Mesenchymal Transition (EMT), Angiogenesis, and Hypoxia, were markedly induced. Conversely, we observed a significant downregulation of seven gene sets (Fig. 3 brown bars), predominantly those involved in immune surveillance. These included the suppression of Interferon-gamma, IL-2/STAT5, and IL-6/JAK/STAT3 signaling, as well as genes typically downregulated by KRAS activation (Fig. 3, Supplementary Table S1A & S1B, Supplementary Figure S1).

Fig. 3.

Functional enrichment analysis of ecCCND1-associated pathways. Bar chart displaying Gene Set Enrichment Analysis (GSEA) results for Hallmark gene sets. Bars represent GSEA prerank nominal p-values. Green bars indicate pathways significantly upregulated in ecCCND1-positive tumors (e.g., G2M checkpoint, E2F Targets, MYC Targets), while orange bars represent pathways upregulated in the negative group, predominantly involving immune signaling (e.g., Interferon Gamma Response, IL6-JAK-STAT3)

CCND1 ecDNA events are mutually exclusive with HPV-positive tumors

Given that HPV and CCND1 are recognized as independent primary drivers in HNSCC, we examined the relationship between viral status and ecCCND1 amplification. Our analysis revealed a profound mutual exclusivity between these two factors (Fig. 2E). Of the 37 HPV-positive tumors in the cohort, only one concurrently harbored ecCCND1. Conversely, 34 of the 35 ecCCND1-positive tumors were HPV-negative. This segregation was highly statistically significant (p = 0.0012, Chi-square test), confirming that ecCCND1 is a distinct driver characteristic of the HPV-negative disease subtype.

ecCCND1 is an independent prognostic factor in multivariate analysis

To evaluate whether ecCCND1 is independently associated with clinical outcomes, we performed multivariate Cox proportional hazards analysis adjusting for age, sex, tumor stage, and HPV status. In this model, ecCCND1 remained significantly associated with worse survival (HR = 1.93, p = 0.0166). Tumor stage (HR = 1.45, p = 0.0114) and HPV positivity (HR = 0.49, p = 0.0497) were also significant predictors, whereas age and sex were not. These results indicate that the prognostic impact of ecCCND1 is independent of established clinical and molecular factors. (Supplementary Table S2)

Discussion

The molecular pathogenesis of HNSCC is traditionally understood as a disruption of key functional pathways regulating cellular proliferation, epithelial differentiation, and cell survival [2]. In addition, evidence suggests that circadian regulator genes contribute to transcriptional heterogeneity and clinical outcomes in HNSCC, highlighting an additional layer of regulatory complexity in tumor biology [12]. While focal amplifications of oncogenes are recognized drivers in this landscape, our findings suggest that the topology of these amplifications is as critical as the copy number itself. In this study, we identify that the amplification of CCND1 in HNSCC is frequently mediated by extrachromosomal DNA. This finding places HNSCC alongside other aggressive malignancies where ecDNA serves as a primary vehicle for oncogene hyper-amplification, such as glioblastoma (EGFR) and neuroblastoma (MYC) [3].

In this study, we adopted a broader operational definition of extrachromosomal DNA based on structural classification by the AmpliconArchitect algorithm. While ecDNA is often strictly defined as circular DNA in prior literature, extrachromosomal amplifications can span a structural continuum. Accordingly, we included all AmpliconArchitect-defined extrachromosomal amplicons in our analysis, while considering circular structures as high-confidence ecDNA, consistent with established definitions.

At the gene level, ecCCND1-positive tumors display a distinctive transcriptional pattern characterized by coordinated amplification and overexpression of multiple genes within the 11q13 locus. In addition to CCND1, genes such as PPFIA1, FADD, ORAOV1, CTTN, ANO1, and FGF19 were consistently overexpressed, a pattern that is broadly consistent with prior reports suggesting that the 11q13 amplicon may function as a clustered oncogenic unit in HNSCC [13, 14]. Although our analysis does not establish causality, the observed co-expression pattern aligns with previously described roles of these genes in tumor biology. For example, amplification of FADD has been associated with both its canonical pro-apoptotic adaptor function and additional roles in cell-cycle regulation, NF-κB signaling, and tumor progression across several malignancies [15]. Similarly, ANO1, a calcium-activated chloride channel, has been reported to contribute to cell migration and chemotherapy resistance in head and neck cancers [16, 17]. The concurrent upregulation of CTTN, which participates in cytoskeletal remodeling and regulation of Arp2/3-mediated actin dynamics, may further be compatible with enhanced invasive potential described in previous studies [18, 19]. In contrast, the reduced expression of CEACAM5 and CEACAM7, genes implicated in epithelial adhesion and differentiation, could be consistent with a shift toward a less differentiated and more motile cellular phenotype. Together, these transcriptional patterns suggest that ecCCND1-positive tumors may be associated with a broader regulatory program involving the 11q13 locus that has previously been linked to aggressive tumor behavior [20].

The relatively limited number of significant differentially expressed genes is likely due, in part, to limited statistical power from the small number of ecCCND1-positive tumors. In addition, the marked biological heterogeneity of HNSCC further reduces the ability to detect gene-level differences after multiple-testing correction. The transcriptional effects of ecCCND1 may also be more localized to specific pathways rather than reflected as widespread genome-wide changes.

Pathway analysis further reinforces this aggressive biological program. ecCCND1-positive tumors are characterized by the upregulation of aggressive hallmark pathways, including Myc targets, E2F targets, and the G2M checkpoint. Together, these changes are consistent with the central role of Cyclin D1 in enforcing CDK4/6-mediated G1/S transition and cell-cycle progression in HNSCC [21]. The concomitant induction of glycolysis and oxidative phosphorylation reflects metabolic reprogramming characteristic of rapidly proliferating head and neck tumors and has been associated with treatment resistance and poor clinical outcome [22, 23]. In parallel, strong enrichment of epithelial–mesenchymal transition (EMT), angiogenesis, and hypoxia signatures is consistent with known transcriptomic hallmarks of invasion, neo-vascularization, and radio-/chemoresistance in HNSCC [24]. Together, these pathway alterations indicate that ecCCND1 does not merely increase CCND1 dosage but coordinates a broader rewiring of cell-cycle, metabolic, and microenvironmental programs that collectively underlie the highly aggressive clinical phenotype.​.

The localization of CCND1 on ecDNA confers a unique structural advantage driving tumor aggression, distinguishing this mechanism from the functional constraints of linear chromosomal amplification. Circular ecDNA lacks centromeres, allowing for unequal segregation during mitosis and the rapid accumulation of massive copy numbers within specific tumor subpopulations [6, 25]. Furthermore, recent chromatin topology studies have demonstrated that ecDNA exhibits a highly accessible, open chromatin structure, allowing for ultra-long-range interactions between oncogenes and enhancers [26, 27]. This explains our observation that ecCCND1 tumors exhibit significantly higher mRNA transcript levels than tumors with linear amplifications. The “dosage effect” driven by this accessible circular architecture, together with co-amplified 11q13 oncogenes and the proliferative and EMT pathway activation described above, likely fuels the extreme proliferative capacity characteristic of these tumors [14, 28].​ Of note, the 11q13 CCND1 amplicon’s topology (ecDNA or integrated HSR) is a critical but often overlooked variable in HNSCC functional studies. For example, while established lines like FaDu and Detroit 562 are known to harbor these amplification events [29], ecDNA in these cancer cell lines can exhibit significant inter-isolate variability. Over prolonged periods of in vitro passaging without stringent selective pressure, ecDNAs can be lost or may reintegrate into linear chromosomes, forming homogeneously staining regions (HSRs) [30]. When cultured cells spontaneously convert ecDNA to HSRs, the epigenetic landscape flattens, copy number becomes fixed, and the unequal segregation necessary to model adaptive resistance is lost [30]. This structural heterogeneity also hampers downstream validation: CRISPR targeting of high‑copy loci frequently induces cytotoxicity, and multi‑omic integration requires temporally matched structural data. Therefore, rigorous longitudinal monitoring of amplicon topology is essential to ensure biologically faithful epigenetic and functional genomic analyses in HNSCC cell lines.

The ecCCND1 amplification is particularly relevant when considering the clinical dichotomy of HNSCC. Our data reveals a strict mutual exclusivity between ecCCND1 amplification and HPV infection, reinforcing the “two-disease” model of head and neck cancer. HPV-positive tumors are driven by viral oncoproteins (E6 and E7) that inactivate p53 and Rb, thereby bypassing cell cycle checkpoints [31]. In contrast, our findings suggest that HPV-negative HNSCC relies on catastrophic structural genomic instability to achieve a similar cell-cycle dominance. By amplifying CCND1 on ecDNA and concomitantly upregulating Myc/E2F-related and metabolic pathways, these tumors directly saturate the cell with Cyclin D1, forcing the G1/S transition via CDK4/6 activation and simultaneously enhancing EMT, hypoxia, and angiogenesis programs that are tightly linked to aggressive behavior in HPV-negative disease [32].​.

Consequently, the intersection of HPV-negative status and ecCCND1 amplification defines an ultra-high-risk subset of patients. Clinically, this suggests that current TNM staging may be insufficient for these individuals. The unique biological properties of ecDNA position it as a superior prognostic marker. Pan-cancer analyses consistently show that ecDNA presence correlates with increasing tumor stage [33] and significantly shorter overall survival [6]. Crucially, this stark clinical association persists even when controlling for specific tissue types [34]. Therefore, patients harboring ecCCND1 should likely be categorized into a high-risk stratification group regardless of their traditional pathological stage.

Finally, identifying ecCCND1 as a structural driver opens new therapeutic avenues. Since these tumors are molecularly addicted to the Cyclin D1-CDK4/6 axis and exhibit coordinated activation of Myc/E2F, metabolic, and EMT programs, they may represent the ideal candidates for targeted CDK4/6 inhibitors, a strategy that has previously shown mixed results in unselected HNSCC populations [35, 36]. Moving forward, therapeutic strategies must evolve to counteract not just the protein product of the oncogene, but also the structural instability and coordinated pathway rewiring that sustains it.​ Emerging anti-ecDNA therapies offer a complementary approach by exploiting the unique vulnerabilities of extrachromosomal inheritance. For instance, the high replication stress and asymmetric segregation inherent to acentric ecDNA create specific dependencies on DNA damage response pathways, exacerbating this stress via CHK1 or ATR inhibitors could induce synthetic lethality specifically in ecDNA-bearing cells [37].

Recent evidence highlights ecDNAs as powerful regulatory hubs in HNSCC, particularly in HPV-positive oropharyngeal models. These hybrid human-HPV ecDNAs physically cluster into nuclear hubs, utilizing open chromatin and high copy numbers to facilitate opportunistic enhancer hijacking. Through strong cis-interactions, shared enhancers drive cooperative hyper-transcription of both viral (E6/E7) and co-amplified somatic oncogenes (e.g., EGFR, MYC) [38–41]. Importantly, this strong dependence on ecDNA-driven enhancer networks creates druggable vulnerabilities [38–40]. Disrupting the protein machinery that sustains these hubs, either via targeted CRISPR interference or BET bromodomain inhibitors, selectively collapses the oncogenic transcriptional programs in ecDNA-positive tumors. Furthermore, the massive transcriptional burden of these cancers sensitizes them to broader strategies, including agents that target transcription-replication conflicts (e.g., CHK1 inhibitors) [42] or actionable transcriptional nodes like CDK7 [43]. Taken together, these observations raise the possibility that effective therapeutic strategies in ecDNA-positive HNSCC may require not only inhibition of downstream oncogenic signaling pathways, but also disruption of the structural and regulatory features that enable ecDNA-driven amplification. However, further functional and clinical studies will be necessary to clarify the extent to which these approaches translate into clinical benefit.

In this study, HPV status was determined based on tumor RNA sequencing data. Notably, emerging liquid biopsy approaches, such as circulating tumor HPV DNA (ctHPVDNA) sequencing, have demonstrated high diagnostic accuracy and the ability to capture additional prognostic features (e.g., viral integration and mutational profiles), highlighting the potential of non-invasive methods for more comprehensive HPV characterization in HNSCC [44].

Conclusions

This study suggests that ecDNA-mediated amplification of CCND1 may represent an important structural and transcriptional feature associated with a high-risk, HPV-negative subset of HNSCC. Tumors harboring ecCCND1 showed coordinated overexpression of multiple oncogenes within the 11q13 amplicon and exhibited a transcriptional profile characterized by elevated cell-cycle activity, metabolic pathway enrichment, and gene signatures related to epithelial–mesenchymal transition, hypoxia, and angiogenesis, together with relative suppression of several immune signaling pathways. While these observations do not establish causality, the overall expression pattern is consistent with biological programs previously linked to aggressive tumor behavior and adverse clinical outcomes. In this context, ecCCND1 status may represent a potentially informative molecular feature for future risk stratification when considered alongside established clinical parameters such as TNM staging. Collectively, these data position ecDNA-mediated CCND1 amplification as a key driver and prognostic biomarker in HPV-negative HNSCC. In addition, the observed transcriptional programs raise the possibility that therapeutic strategies targeting both Cyclin D1–CDK4/6 signaling and mechanisms underlying ecDNA-mediated oncogene amplification could be relevant for this subgroup, although further functional and clinical studies will be required to clarify these implications.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

CJ.W. conceptualized the study, designed the methodology, collected and analyzed the data, wrote the original draft, reviewed and edited the manuscript, and administered the project. P.T. conceptualized the study, designed the methodology, collected and analyzed the data, wrote the original draft, reviewed and edited the manuscript.

Funding

This study was supported by grant from the Taipei Medical University [TMU113-AE1-B34] and from the National Science and Technology Council, Taiwan [NSTC114-2320-B-038-077-MY3].

Data availability

Publicly available datasets were analyzed in this study. Clinical information and RNA sequencing-based gene expression data were obtained from the NCI Genomic Data Commons (GDC) Data Portal ( [https://portal.gdc.cancer.gov](https:/portal.gdc.cancer.gov) ) under the TCGA-HNSC cohort. Frequencies and ranges of tumor ecDNAs were obtained from Supplementary Table 1 of a previously published TCGA pan-cancer study ( [https://static-content.springer.com/esm/art%3A10.1038%2Fs41588-020-0678-2/MediaObjects/41588_2020_678_MOESM2_ESM.xlsx](https:/static-content.springer.com/esm/art%3A10.1038%2Fs41588-020-0678-2/MediaObjects/41588_2020_678_MOESM2_ESM.xlsx) ) [6]. HPV status was retrieved from Supplementary Data 1.2 of the TCGA-HNSC landmark study ( [https://static-content.springer.com/esm/art%3A10.1038%2Fnature14129/MediaObjects/41586_2015_BFnature14129_MOESM116_ESM.zip](https:/static-content.springer.com/esm/art%3A10.1038%2Fnature14129/MediaObjects/41586_2015_BFnature14129_MOESM116_ESM.zip) ) [9].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.