Clinical Significance of Extrachromosomal DNA in FGFR2-Amplified Gastric Cancer: Therapeutic Implications From Clinical Experience

Introduction

The fibroblast growth factor receptor (FGFR) family, consisting of FGFR1 to FGFR4, belongs to the receptor tyrosine kinase superfamily. FGFR2 amplification is detected in approximately 5% of patients with advanced gastric cancer (AGC) and is recognized as a potential therapeutic target.¹ Although FGFR inhibitors have shown clinical efficacy in these patients, the resistance mechanisms involved remain poorly understood.^2-5

Understanding the molecular basis of FGFR2 amplification has become increasingly important for optimizing therapeutic strategies. Extrachromosomal DNA (ecDNA) has recently emerged as a critical mechanism driving oncogene amplification and therapeutic resistance across various cancers. ecDNA, characterized as circular DNA elements existing outside chromosomes, exhibits unique properties such as uncontrolled replication and non-Mendelian inheritance patterns; these features contribute to elevated oncogene copy number (CN) and intratumoral heterogeneity, ultimately leading to treatment resistance.^6,7

The clinical relevance of ecDNA in AGC is substantial. ecDNA-based amplification in AGC occurs in approximately 37.9% of patients and correlates with poorer prognosis than focal amplification.⁸ Furthermore, a whole-genome sequencing of 170 gastric cancer genomes suggest that FGFR2 amplification is associated with ecDNA.⁹ However, to our knowledge, there have been no clinical reports to date investigating the therapeutic effects and resistance of FGFR-targeted therapy in patients with FGFR2 amplification on ecDNA.

Herein, we present a case of an AGC patient with FGFR2 amplification on ecDNA and discuss its potential role in therapeutic response and resistance. Additionally, to elucidate the prevalence and clinical significance of this phenomenon, we used fluorescence in situ hybridization (FISH) to determine the ecDNA status of patients with FGFR2-amplified AGC who were enrolled in the MONSTAR-SCREEN-2 study (UMIN000043899), a multicenter biomarker development initiative that used artificial intelligence–driven multiomics analysis for a variety of solid tumors.¹⁰

The clinical information presented in this report was collected throughout the patient's course of care. The patient provided informed consent for the use and publication of their health information and medical imaging.

Case

A 48-year-old man was diagnosed with AGC after staging laparoscopy that revealed positive peritoneal cytology. The patient received systemic chemotherapy with S-1 plus oxaliplatin and intraperitoneal chemotherapy with paclitaxel for 5 months at the previous hospital, followed by total gastrectomy. The patient developed para-aortic lymph node (PALN) metastases 2 months after surgery and was treated with systemic chemotherapy using S-1 plus paclitaxel. After transferring to our institution, computed tomography (CT) was performed, revealing obvious disease progression with further enlargement and increase of the PALN, multiple bone metastases, and lung lymphangitis. The patient received a combination of capecitabine, oxaliplatin, plus nivolumab while undergoing next-generation sequencing (NGS) using Caris MI Profile (Caris Life Sciences, Phoenix, AZ) for tumor tissue resected at surgery and Caris ASSURE (Caris Life Sciences) for plasma samples. Before the NGS results were available, his condition deteriorated with worsening of the lymphangitis, prompting the initiation of irinotecan. However, 2 days later, the patient was hospitalized because of respiratory distress and the onset of disseminated intravascular coagulation (DIC). NGS revealed FGFR2 amplification (CN, 264.9) along with pathogenic short variants in CDH1, SMAD4, and TP53 in the tumor tissue. Liquid genotyping identified FGFR2 amplification (plasma CN [pCN], 16.6) along with the pathogenic short variants of CDH1 and FGFR2. Histologic evaluation with immunohistochemical staining (IHC) for FGFR2 demonstrated cytoplasmic staining with 10% IHC 3+ and 5% IHC 2+ without membrane positivity (Figs 1A and 1B). On the basis of the NGS findings, off-label treatment with pemigatinib, a selective FGFR1-3 inhibitor, was initiated. Although the patient required platelet and fresh-frozen plasma transfusions for 3 days, the DIC resolved, and the respiratory condition improved. Resistance to irinotecan could not be fully assessed because of its short duration of use; therefore, pemigatinib was combined with irinotecan (initiated 2 days pre-pemigatinib) after approval from the institutional review board. The patient's condition gradually improved, and he was discharged 3 weeks after admission. Liquid NGS revealed a decrease in the detected FGFR2 pCN (Table 1). Positron emission tomography-CT (PET-CT) 7 weeks later from the initiation of pemigatinib showed a marked reduction of [¹⁸F] fluorodeoxyglucose uptake in multiple bone metastases, accompanied by decreased tumor marker levels (Fig 1C).

FIG 1.

Clinical and pathologic findings of the case. (A) Hematoxylin-and-eosin–stained biopsy specimen of the primary tumor. (B) IHC analysis of FGFR2 showing 10% IHC 3+ and 5% IHC 2+ in the cytoplasm without membrane positivity. (C) Clinical course depicting changes in tumor markers (CEA and CA 19-9) and PET-CT scans before and after pemigatinib initiation at our hospital. The arrows indicate the timing of liquid biopsy sampling at −6, 2, 8, and 13 weeks, with reference to the pemigatinib initiation. CA 19-9, carbohydrate antigen 19-9; CAPOX, capecitabine plus oxaliplatin; CEA, carcinoembryonic antigen; FGFR, fibroblast growth factor receptor; IHC, immunohistochemistry; IRI, irinotecan; NGS, next-generation sequencing; nivo, nivolumab; PET-CT, positron emission tomography-computed tomography.

TABLE 1.

Genomic Dynamics in Gene Alterations at Each Time Point

Specimen	Pre-Pemigatinib (6 weeks before pemigatinib)		PR Post-Pemigatinib (2 weeks after peimigatinib)		PD Post-Pemigatinib (8 weeks after pemigatinib)		PD Post-Erdafitinib (13 weeks after pemigatinib)
Liquid	Caris Assure		Caris Assure		Guardant360		Guardant360
	Amplification	pCN	Amplification	pCN	Amplification	pCN	Amplification	pCN
	FGFR2	16.6	FGFR2	5.2	FGFR2	110.1	FGFR2	34.3
	Mutation	VAF	Mutation	VAF	MYC	45.4	MYC	11.2
	CDH1, L214P	46	CDH1, L214P	0.6	KRAS	3.0	KRAS	ND
	FGFR2, C382R	0.1	CDH1, W409*	0.6	Mutation	VAF	Mutation	VAF
					FGFR2, L617V	8.4	FGFR2, L617V	7.4
					FGFR2, E565A	7.1	FGFR2, E565A	11.6
					FGFR2, V564L	2.3	FGFR2, V564L	6.6
					FGFR2, N549H	0.8	FGFR2, N549H	1.1
					FGFR2, N549K	0.7	FGFR2, N549K	0.8
					FGFR2, V564F	0.2	FGFR2, V564F	0.2
					Fusion		FGFR2, V564I	0.3
					FGFR2-PRNCR1	0.08	FGFR2, E565G	1.3
							FGFR2, N549D	0.7
							FGFR2, E565K	0.6
							TP53, E339fs	12.5
							Fusion
							FGFR2-PRNCR1	0.05
Tissuea	MI Profile
	Amplification	CN
	FGFR2	264
	Mutation	VAF
	CDH1, L214P	58
	SMAD4, E330K	44
	TP53, E339fs	46

Abbreviations: CN, copy number; FGFR, fibroblast growth factor receptor; ND, not detected; PD, progressive disease; pCN, plasma CN; PR, partial response; VAF, variant allele frequency.

^a Tumor tissue resected at surgery.

However, disease progression occurred 2 months after pemigatinib initiation, with worsening lower back pain attributed to progressive bone metastases, confirmed by PET-CT at 9 weeks after pemigatinib initiation (Fig 1C). Plasma NGS using Guardant360 (Guardant Health, Redwood City, CA) revealed a marked increase in FGFR2 pCN (110) and the emergence of MYC amplification (pCN 45.4), KRAS amplification (pCN 3.0), various FGFR2 mutations (L617V, N549H, E565A, V564L, N549K, and V564F), and an FGFR2-PRNCR1 fusion. Progressive DIC necessitated continuous platelet and plasma transfusions. Subsequent off-label use of other FGFR inhibitor therapy with erdafitinib followed by futibatinib was ineffective, resulting in disease progression and the patient's death. The final liquid NGS analysis showed an expanded array of gene alterations, including increased FGFR2 mutations (Figs 2A and 2B; Table 1).

FIG 2.

Genomic dynamics in liquid NGS analysis. (A) The pCN variations of FGFR2, MYC, and KRAS monitored by liquid NGS using Caris Assure (first two time points) and Guardant360 (subsequent time points) were correlated with the treatment response. (B) Fish plot visualization of clonal evolution depicting the emergence of multiple acquired FGFR2 mutations during disease progression. The time points are indicated relative to pemigatinib initiation. FGFR, fibroblast growth factor receptor; IRI, irinotecan; NGS, next-generation sequencing; pCN, plasma copy number; VAF, variant allele frequency.

ecDNA Evaluation in AGC Patients With FGFR2 Amplification

We assessed AGC patients with FGFR2 amplification enrolled in the MONSTAR-SCREEN-2 study in our institution. This study was conducted in accordance with the Declaration of Helsinki and Ethical Guidelines for Medical and Biological Research Involving Human Subjects, and approved by the Institutional Review Board of the National Cancer Center, Japan. Written informed consent was obtained from all participants. Of 207 patients with AGC enrolled at our institution from May 2021 to March 2024, 34 were excluded because of insufficient tissue samples or failed analysis. Tissue NGS detected FGFR2 amplification in 2.9% (5/173) of the patients, including Patient 1 presented in this report. FISH analysis was performed in all five patients, with repeat testing performed in Patients 2 and 3. The analysis revealed scattered extrachromosomal FGFR2 signals, consistent with ecDNA, in four patients (Patients 1-4). By contrast, Patient 5 exhibited patterns consistent with homogeneously staining region (Fig 3).

FIG 3.

FISH in AGC patients with FGFR2 amplification enrolled in the MONSTAR-SCREEN-2 study. The yellow arrowheads indicate scattered extrachromosomal FGFR2 signals suggesting ecDNA in patients 1-4. The white arrows in patient 5 indicate HSRs. AGC, advanced gastric cancer; ecDNA, extrachromosomal DNA; FGFR, fibroblast growth factor receptor; FISH, fluorescence in situ hybridization; HSR, homogeneously staining region.

Discussion

We present a patient with an extraordinarily high CN of FGFR2 amplifications driven by ecDNA, treated with FGFR-targeted therapies. The patient received concurrent pemigatinib and irinotecan, on the basis of the limited exposure to previous irinotecan and the preclinical evidence that combining cytotoxic chemotherapy with FGFR-targeted tyrosine kinase inhibitors enhances antitumor effects.¹¹ However, the duration of response was short, and disease progression was marked by the emergence of numerous genetic alterations.

A high CN is a reported characteristic of ecDNA, with amplified oncogenes on ecDNA showing significantly higher CNs (mean CN ≥50) compared with those on intrachromosomal DNA.¹² Previous studies demonstrated tumor responses to FGFR-targeted therapy in patients with FGFR2 amplification, especially in those with a high CN, although the duration of response was generally short.^2,13 These findings may suggest ecDNA involvement. However, in our analysis of the MONSTAR-SCREEN2 cohort, ecDNA was also detected in patients with low CN, indicating that high CN is neither necessary nor sufficient for the definitive diagnosis of ecDNA. Another characteristic of ecDNA is early therapeutic resistance through unique mechanisms such as increased genetic heterogeneity, rapid genome changes driven by super-enhancers, and amplification of resistance genes, leading to a poor prognosis.^7,14,15The rapid emergence and diversity of acquired FGFR2 mutations and a fusion observed in our patient would be highly unusual if confined to chromosomal DNA, given conventional mutation rates and selective pressures.^16,17 These alterations are thought to occur on ecDNA, accumulating unrepaired genetic alterations during its rapid amplification and replication processes. We previously reported an AGC patient with markedly high EGFR amplification in ctDNA who acquired multiple EGFR mutations and MET amplification after treatment with cetuximab.¹⁸ The clinical course of this patient closely resembled that of our present patient. Therefore, FISH analysis was performed in this AGC patient with EGFR amplification, revealing the presence of ecDNA (Appendix Fig A1).

Additionally, we investigated the association between FGFR-amplified AGC and ecDNA. A previous study using whole-genome sequencing analysis of gastric cancer revealed association between FGFR2 amplification and self-joining amplicons, structural variant clusters characteristic of ecDNA.⁹ In fact, four of the five patients with FGFR2-amplified AGC in our cohort of MONSTAR-SCREEN-2 also exhibited ecDNA. Although the cancer types were different, this association may underlie the limited efficacy of FGFR inhibitors in FGFR2-amplified AGC compared with biliary tract cancer with FGFR2 fusions. In a phase II study of futibatinib in FGFR-amplified AGC, the overall response rate (ORR) and progression-free survival (PFS) were 18% and 2.9 months, whereas in another study on biliary tract cancer with FGFR fusion, ORR and PFS were 42% and 9.0 months, respectively.^19,20

These findings highlight the need to overcome ecDNA-driven resistance. A study from Memorial Sloan Kettering Cancer Center showed improved survival in AGC patients with ERBB2 ecDNA treated with HER2-targeted therapy plus first-line fluoropyrimidine and oxaliplatin.²¹ Although high CN may confound results, DNA damage from multiple cytotoxic agents appears effective against ecDNA-positive tumors. Irinotecan, a topoisomerase I inhibitor, may have potential efficacy against ecDNA; however, we did not observe such an effect in our patient, suggesting combination chemotherapy may be required. Recent evidence shows that ecDNA causes transcription-replication conflict, replication stress, and DNA damage, activating the S-phase checkpoint. In preclinical AGC models with FGFR2 amplification on ecDNA, the checkpoint kinase 1 (CHK1) inhibitor BBI-2779 suppressed tumor growth and prevented resistance to the FGFR inhibitor infigratinib.²² Additionally, the phase I/II POTENTIATE trial evaluated BBI-355, a selective CHK1 inhibitor, as monotherapy and in combination with futibatinib or erlotinib in metastatic solid tumors with FGFR or EGFR amplifications driven by ecDNA (ClinicalTrials.gov identifier: NCT05827614). However, these trial arms were discontinued because of tolerability issues, including hematologic toxicity and a narrow therapeutic index with continuous dosing regimens. Given the expected synergistic effect and nonoverlapping toxicity profiles, a novel combination of BBI-355 with BBI-825, a selective ribonucleotide reductase inhibitor, is being developed, with this combination arm of the POTENTIATE trial planned to be initiated in the second half of 2025.²³

Important caveats of our study include the absence of standardized ecDNA diagnostic criteria and the inability to reliably identify ecDNA using IHC. Although we explored IHC as a more accessible diagnostic approach, FGFR2 often exhibits nonspecific cytoplasmic staining as an artifact, even in cases without FGFR2 amplification, which complicates interpretation. Our analysis suggests that IHC alone cannot definitively identify ecDNA-based amplification. These findings emphasize that the development of more specific diagnostic approaches for ecDNA detection remains a critical area for future research.

In conclusion, this case report provides valuable insights into FGFR2-amplified AGC driven by ecDNA. The distinct patterns of response and resistance highlight the need for novel therapeutic strategies. Targeting the primary oncogenic driver and the underlying mechanisms of ecDNA-mediated heterogeneity is critical for achieving more durable treatment outcomes.

APPENDIX

FIG A1.

FISH analysis of an AGC patient with EGFR amplification. This patient exhibited a markedly high EGFR plasma copy number in ctDNA, as previously reported by our institution. After treatment with cetuximab, multiple EGFR mutations and MET amplification were acquired (Nakamura et al¹⁸). FISH analysis revealed a scattered EGFR staining pattern, indicating the presence of ecDNA. AGC, advanced gastric cancer; ecDNA, extrachromosomal DNA; FISH, fluorescence in situ hybridization.

AUTHOR CONTRIBUTIONS

Conception and design: Yu Aoki, Tadayoshi Hashimoto, Nobuyuki Takahashi, Saori Mishima, Yoshiaki Nakamura, Izuma Nakayama, Nobuyuki Nakamura, Naoya Sakamoto, Genichiro Ishii, Yasuhito Arai, Tatsuhiro Shibata, Kohei Shitara

Financial support: Kohei Shitara

Administrative support: Tadayoshi Hashimoto, Nobuyuki Nakamura, Kohei Shitara

Provision of study materials or patients: All authors

Collection and assembly of data: Yu Aoki, Tadayoshi Hashimoto, Hideaki Bando, Nobuyuki Nakamura, Naoya Sakamoto, Takao Fujisawa, Takayuki Yoshino, Kohei Shitara

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).