Introduction

Neuregulin 1 (NRG1) is a member of the epidermal growth factor (EGF) family and functions as a signaling protein that regulates intercellular interactions. It is predominantly expressed in various tissues, including the brain, heart, and mammary glands, and plays a crucial role in tissue development and maturation1,2. As a ligand of the EGF family, NRG1 binds to ERBB family receptor tyrosine kinases, including ERBB2, ERBB3, and ERBB4, to activate downstream signaling pathways involved in biological processes3,4. When NRG1 undergoes fusion to form a novel fusion gene, it expresses a fusion protein on the cell membrane that retains the active EGF-like domain of NRG1, preserving its physiological activity. Consequently, NRG1 fusions lead to the persistent activation of related signaling pathways, resulting in uncontrolled cell proliferation and tumorigenesis5,6,7,8.

Historically, numerous pharmaceutical agents have been investigated as potential treatments for patients with NRG1 fusions. Notably, the drug zenocutuzumab, which targets NRG1 fusions, recently received accelerated approval from the FDA, marking a significant advance as the first targeted therapy for patients with NRG1 fusions.

NRG1 fusions represent important molecular events that are implicated in the tumorigenesis of various cancers. However, their occurrence is rare, with a frequency of less than 1% across tumors. The highest frequency of NRG1 fusions is observed in lung cancer. Nevertheless, they are also detected in other cancers. NRG1 fusions involve multiple fusion partner genes, with CD74, ATP1B1, and SDC4 being the most common5. In lung cancer specifically, several NRG1 fusion partners, including CD74, ATP1B1, SLC3A2, SDC4, RBPMS, and WRN, have been identified9,10,11. Other NRG1 fusion partners, including POMK, APP, CDH6, ATP1B1, and CLU, have also been found in various solid tumors12,13,14.

This study aimed to comprehensively investigate the frequency, structural characteristics, co-occurring mutations, and prognostic impact of NRG1 fusions. Using next-generation sequencing (NGS), we conducted molecular profiling of 25,203 solid tumor samples from Chinese patients and analyzed data from 49 patients with NRG1 fusions, obtaining valuable clinical insights and identifying therapeutic opportunities for patients harboring NRG1 fusions in solid tumors.

Results

The frequency of NRG1 fusions across solid tumor types

We collected next-generation sequencing (NGS) data from tumor samples from 25,203 patients across 15 types of solid tumors (Table 1). Among these patients, 49 patients with NRG1 fusions were identified, resulting in an overall frequency of 0.2%. Lung cancer (49.4%) was the most prevalent tumor type among all patients included in this study, followed by colorectal cancer (11.1%), liver cancer (8.5%), gastric cancer (6.0%), pancreatic cancer (3.7%), sarcoma (2.6%), breast cancer (2.5%), glioma (2.1%), head and neck cancer (1.7%), cervical cancer (1.6%), esophageal cancer (1.6%), cholangiocarcinoma (1.1%), prostate cancer (0.6%), medulloblastoma (0.1%) and other tumors. In the total cohort, 57.7% of patients were male, and 21% were classified as stage IV at diagnosis.

Table 1 Baseline characteristics of patients with NRG1 fusion and non-fusion
Full size table

A total of 36 NRG1 fusion events were identified in 12,458 lung cancer patients, reflecting a frequency of 0.29% within this group (Fig. 1a). Notably, the frequency of NRG1 fusion was greater in several other tumor types than in lung cancer, with values of 0.65% in prostate cancer and 0.47% in breast cancer. In addition, the observed frequencies of NRG1 fusions in other tumor types were 0.25% in esophageal cancer, 0.17% in colorectal cancer, 0.13% in gastric cancer, 0.11% in pancreatic cancer, and 0.05% in liver cancer (Fig. 1a). We did not capture NRG1 fusion events in sarcoma and glioma. Compared with male patients, female patients were more likely to present with NRG1 fusions (p = 0.003); however, there were more male patients overall in the cohort.

Fig. 1: Frequencies of NRG1 fusion and fusion partners.
figure 1

a Frequencies of NRG1 fusion in different tumor types. b Frequencies of fusion partners identified in the study.

Full size image

The genomic landscape of NRG1 fusions

NRG1 consists of a total of 13 exons, with a functional EGF-like domain in exons 6 and 7. Exon 8 encodes a transmembrane domain. Gene fusions can occur at both the 5′ end and the 3′ end of NRG1. In the present study, we identified a total of 36 fusion partners associated with NRG1 (Fig. 1b). The most frequent partner was CD74 (29.3%), followed by NRG1-IT3 (5.2%), B3GNT6 (3.4%), IRG2BP1 (3.4%), LINC01288 (3.4%), and PTPRB (3.4%) (Fig. 1b). Among the detected partners, 16 fused to the 3′ end of NRG1, and 11 of these fusions disrupted the EGF-like domain (Fig. 2a). In contrast, 22 partners fused to the 5′ end of NRG1, and only three of these fusions compromised the integrity of the EGF-like domain (Fig. 2b). Most of the identified fusion partners were located on the same chromosome as NRG1, namely, chromosome 8. Other fusion partner locations include chromosomes 1, 2, 5, 10, 11, 12, 17, 19, and 22 (Fig. 2c).

Fig. 2: Schematic diagram of NRG1 fusions.
figure 2

a NRG1 type I fusion, 5′ NRG1 joined to 3′ partners. b NRG1 type II fusion, with 5′ partners joined to 3′ NRG1. c locations of the fusion partners on the chromosome. The blue and orange lines indicate NRG1 type I fusions and type II fusions, respectively.

Full size image

Co-occurrence of gene mutations and NRG1 fusions

The genomic alterations that co-occurred with NRG1 fusion were analyzed across all the samples. The most frequently mutated genes included TP53 (63%), CD74 (35%), MYC (29%), CDKN2A (22%), and EGFR (22%) (Fig. 3a). Patients with NRG1 fusions presented significantly higher mutation frequencies of CD74 (p < 0.001), FGFR1 (p = 0.028), and RET (p = 0.007) than did patients without NRG1 fusions (Fig. 3b). In the context of lung cancer, significantly more gene mutation events, including mutations in CD74 (p < 0.001), EGFR (p = 0.019), MYC (p = 0.045), GNAS (p = 0.025), RET (p = 0.019), TET2 (p = 0.044), CIITA (p = 0.009), KIF5B (p = 0.003), VEGFB (p = 0.01), and ZNF217 (p = 0.017), occurred in patients with NRG1 fusions (Fig. 3c).

Fig. 3: Mutational landscape of patients with NRG1 fusions.
figure 3

a Heatmap of the comutations of 49 patients with NRG1 fusions. b Comparison of mutation frequency between patients with NRG1-fusion and NRG1-nonfusion solid tumors. c Comparison of mutation frequency between NRG1 fusion and NRG1-nonfusion lung cancer patients. The genes highlighted in orange signify a significant mutation frequency between fusion and non-fusion. Chi-square test or Fisher’s exact test, P < 0.05.

Full size image

Tumor biomarkers in patients with NRG1 fusions

We further compared the differences in pancancer biomarkers between patients with and without NRG1 fusion. Our analysis revealed no significant difference in TMB between the two groups (p = 0.17, Fig. 4a). The variation in TMB across both groups suggests that additional factors beyond NRG1 fusion status may influence mutational profiles. Notably, according to some studies, more than 80% of microsatellite instability-high (MSI-H) tumors are classified as TMB-high (TMB-H); however, only 18.3% of TMB-H tumors are MSI-H15. Nevertheless, we observed a significant difference in the frequency of MSI-H tumors between the two groups. Patients with NRG1 fusion presented a markedly lower proportion of MSI-H than did those without this fusion (p = 0.042, Fig. 4b). Specifically, all patients with NRG1 fusions were found to be microsatellite stable (MSS). These findings suggest that patients with NRG1 fusions may have a poor response to immune monotherapy. Gastrointestinal tumors, for which MSI status is frequently utilized clinically as a guide for treatment, were analyzed independently. However, due to the limited number of NRG1-fusion-positive cases in this subset, no statistically significant associations with MSI status could be established (p = 1, Fig. 4c).

Fig. 4: Immunotherapy-related biomarkers in patients with NRG1 fusion.
figure 4

a comparison of TMB between fusion and non-fusion. The centre of the box is represented by the median value. Wilcoxon test, P < 0.05. Comparison of MSI status between fusion and non-fusion in pan-cancer (b) and gastrointestinal tumors (c). Chi-square test or Fisher’s exact test, P < 0.05.

Full size image

NRG1 fusions as potential therapeutic targets

The NRG1 fusion drug zenocutuzumab has been approved by the FDA, offering a new treatment option for patients. To explore additional potential therapeutic targets and facilitate comprehensive treatment decisions, we analyzed actionable mutation profiles in patients with NRG1 fusions across lung cancer, colorectal cancer, and breast cancer.

In the lung cancer cohort, nearly half of the patients (15/36) with NRG1 fusion lung cancer had targets for approved/potential drugs that could be treated with other targeted agents (Fig. 5a). Specifically, seven patients with EGFR mutations had the potential to benefit from more than a dozen approved EGFR-targeting drugs, such as gefitinib and osimertinib. The EGFR mutations identified include exon 19 deletions (19del, n = 4), exon 20 insertions (20ins, n = 2), and L858R (n = 1). One patient with ERBB2 amplification could be treated with ERBB2-targeted therapies. Additionally, approved targeted agents for RET fusions were available for two patients with RET fusions, and patients with MET amplification were potentially suitable for Met-targeting drugs such as sevotinib and britinib. The presence of pathogenic mutations in SMARCA4 and ARID1A opens further avenues for targeted therapies, particularly given the growing interest in drugs that inhibit these pathways.

Fig. 5: Therapeutic or potential targets in patients with NRG1 fusion.
figure 5

Exploration of therapeutic or potential targets in NRG1 fusion patients with lung cancer (a), colorectal cancer (b), and breast cancer (c). The numerical values displayed on the heatmap represent the level of the gene mutation annotated in the OncoKB database. d Kaplan‒Meier curve showing overall survival between NRG1-fusion patients and nonfusion patients in the TCGA cohort.

Full size image

The treatment strategy for patients with colorectal cancer can be determined by RAS/RAF mutation status (Fig. 5b). Notably, we identified actionable ERBB2 mutations (n = 1) indicating potential benefit from targeted inhibitors. Furthermore, one patient exhibited TMB-H status, suggesting possible eligibility for immune checkpoint inhibitor therapy. Within the cohort, three patients diagnosed with breast cancer were found to have NRG1 fusions. However, the number of approved or potential drug targets in breast cancer patients is limited, and only 1 patient with ERBB2 amplification had a relevant target (Fig. 5c). These results indicate the potential to further expand the population benefiting from ERBB2 inhibitors.

Prognosis of patients with NRG1 fusion

Preliminary studies have indicated a disappointing prognosis for patients diagnosed with NRG1 fusion tumors16,17. We conducted a survival analysis comparing patients with and without NRG1 fusion, utilizing data extracted from the TCGA database. The findings revealed no significant difference in overall survival across the pan-cancer cohort (Fig. 5d). However, owing to the limited number of patients with NRG1 fusions across each cancer type, our study did not investigate the impact of NRG1 fusions on prognosis within specific cancer types.

Discussion

To demonstrate the landscape of NRG1 fusion, we retrospectively analyzed data from more than 25,000 samples. In this study, NRG1 fusions were observed in lung cancer, colorectal cancer, breast cancer, gastric cancer, liver cancer, esophageal cancer, pancreatic cancer, prostate cancer, and medulloblastoma. The frequencies of NRG1 fusion in prostate cancer, breast cancer, lung cancer, and esophageal cancer were greater than the overall frequency of 0.2%. In addition, the frequencies of NRG1 fusion in colorectal cancer, gastric cancer, pancreatic cancer, and liver cancer were lower than the overall frequency. Unexpectedly, no NRG1 fusion events were identified in sarcoma or glioma, although the sample numbers of both tumor types were greater than 0.2%. Interestingly, multiple types of gene fusions have been reported to be abundant in sarcoma18,19,20,21. The scarcity of NRG1 fusions in sarcoma indicates that the development of NRG1 fusion-targeted therapy for sarcoma may not be high-priority.

The positions of most breakpoints of NRG1 fusions ranged from intron 1 to intron 5. Whether the EGF-like domain is disrupted depends on which end of NRG1 is retained after fusion. The typical fusions usually contain the 3′ end of NRG111,12. In this study, we detected several cases that contained the 5′ end of NRG1. Regardless of which end of NRG1 the partners fused into, the functions of the fusions remain to be characterized. In this study, CD74 was identified as the most frequent fusion partner of NRG1, which was consistent with previous findings. CD74 encodes a multifunctional protein involved in diverse biological processes22,23,24. As a surface receptor, CD74 can activate pathways involved in the proliferation and survival of tumors23,25. As a high-affinity surface receptor of MIF, CD74 participates in the regulation of antigen presentation in the immune response26,27,28.

To elucidate the potential functions of these fusion events in oncology, we analyzed the co-occurrence of gene mutations in patients with NRG1 fusions. TP53 mutation occurred in most NRG1 fusion cases. However, the difference in the frequency of TP53 mutation between patients with and without NRG1 fusion was not significant. Regardless of the tumor type, the mutation frequencies of CD74, FGFR1, and RET were significantly greater in patients with NRG1 fusion than in patients without NRG1 fusion. The co-occurrence of NRG1 fusion and oncogenes, such as FGFR1 and RET, might provide a new option for physicians in making treatment decisions. Alterations in more genes, including CD74, EGFR, MYC, GNAS, RET, TET2, CIITA, KIF5B, VEGFB, and ZNF217, were found to cooccur with NRG1 fusions in lung cancer patients.

Prior studies have documented that low TMB is associated with reduced efficacy of immunotherapy29,30. The efficacy of immunotherapy relies largely on the ability of immune cells to recognize cancer cell-specific antigens. Consequently, cancer cells with a greater number of genetic mutations are theoretically more likely to be recognized by immune cells because of producing a greater number of neoantigens30,31,32. A large multicenter clinical trial, eNRGy1, demonstrated that patients with lung cancer who exhibit an NRG1 fusion with low PD-L1 expression (28%) and low TMB demonstrate a reduced response to immunotherapy17. The findings from our analysis reveal important insights into the tumor biology associated with NRG1 fusions. The lack of a significant increase in TMB in the NRG1 fusion cohort suggests that these tumors may not follow the typical mutational patterns observed in other genomic alterations that result in a greater mutational burden. MSI-H and NRG1 fusion are two distinct markers in tumors that may coexist in some cases and may affect patient response to therapy when coexisting33. The presence of MSI-H in non-NRG1 fusion patients suggests the potential for distinct oncogenic mechanisms in NRG1 fusion patients versus nonfusion patients, which may influence treatment response. These findings emphasize the need for further research on more effective targeted therapies for NRG1 fusion patients.

Patients with NRG1 fusions usually respond poorly to standard chemotherapy, immunochemotherapy, or ICIs (such as PD-1/L1 monoclonal antibodies)34,35. A range of pharmaceutical agents are currently available, with a primary focus on targeted therapy drugs and monoclonal antibodies. One of these medications is a small-molecule tyrosine kinase inhibitor (TKI) that targets the EGFR/HER2 pathway, while the other is a large-molecule monoclonal antibody that binds to HER2/HER336,37,38. Recently, zenocutuzumab was approved for the treatment of adult patients with advanced unresectable or metastatic non-small cell lung or pancreatic cancer who are positive for NRG1 fusion and who have experienced disease progression despite other treatments. This is the first targeted drug approved by the FDA for the treatment of patients with NRG1 fusion-carrying non-small cell lung cancer or pancreatic cancer38,39.

Our results revealed multiple potential drug targets in NRG1 fusion patients with lung cancer, colorectal cancer, and breast cancer. According to the findings of previous studies, the occurrence of NRG1 fusion in solid tumor patients has been demonstrated to be associated with a favorable response to the clinical implementation of Erb family inhibitor treatment, which has been shown to result in improved prognoses and survival outcomes36,40,41. The emergence of RTK fusions is one of the mechanisms of EGFR-TKI drug resistance. Dual inhibition of EGFR-RTK is safe and effective in patients with targetable RTK fusions after the progression of EGFR-TKIs42. In our study, patients with colorectal cancer or breast cancer showed the potential for response to drugs targeting ERBB2. Drugs targeting the ERBB2/3 kinase domain can inhibit the continued activation of the ERBB2/3 pathway via NRG1 gene fusion, thereby inhibiting cancer progression. In addition to the previously mentioned TKI drugs, a variety of monoclonal antibodies, including trastuzumab and pertuzumab, have been shown to inhibit dimerization and promote antibody-dependent cellular toxicity by binding to the ERBB2 extracellular domain. Other therapeutic agents in this class include seribantumab, lumretuzumab, elgemtumab, GSK2849330, KTN3379, and AV-203, which target ERBB3. Moreover, ADC drugs, such as patritumab and zenocutuzumab, have demonstrated therapeutic potential in clinical trials35,43,44. Concurrently, the presence of KRAS mutations in colorectal cancer patients suggests the potential benefit of targeted drugs. This highlights the need for personalized treatment strategies based on individual genetic profiles. Tumor genetic testing facilitates the identification of novel drug targets for patients with NRG1 fusions.

However, the limitation of this study is that only DNA sequencing was used to characterize the map, and the lack of multiomics joint verification may lead to incomplete NRG1 fusion detection. A combination of DNA and RNA testing may be necessary to obtain more comprehensive information and further refine the research findings. Understanding the broader context of NRG1 fusions within the landscape of oncogenic drivers may ultimately enhance personalized treatment strategies and improve patient outcomes. Additionally, the absence of detailed treatment history limits our ability to assess therapy-induced changes in the mutational landscape.

In conclusion, our findings indicate that tumors harboring NRG1 fusions have distinct genetic and molecular characteristics. Understanding these differences is critical for developing personalized treatment strategies and improving the clinical prognosis of patients with these specific genetic alterations.

Methods

Patient cohort

From November 2021 to September 2022, a total of 25,203 patients with 15 cancer types who underwent NGS were enrolled in this retrospective study. Targeted sequencing using either a 539-gene or 551-gene panel to identify a comprehensive genomic profile was performed by Simcere Diagnostics Co., Ltd. (Nanjing, China), which is certified by the Clinical Laboratory Improvement Amendments (CLIA), College of American Pathologists (CAP), and ISO15189. The clinical data collected included age, sex, tumor stage, and tumor type. This study was carried out in accordance with the Declaration of Helsinki. Approval was obtained from the Ethics Committee of Liuzhou Worker’s Hospital and The First Hospital of China Medical University. The Ethics Committee granted a waiver of informed consent based on review and determination that this research meets the following requirements: (i) this research is a observational study; (ii) the personal information of the subjects is strictly confidential; (iii) the waiver will not adversely affect the rights and welfare of the subjects.

Library preparation and NGS sequencing

DNA was extracted from patient samples using the DNeasy Tissue Kit. The KAPA Library Preparation Kit was utilized for library construction, while the Invitrogen Qubit 4.0 was used to evaluate the library concentration. Sequencing was performed with an average depth of 1500x using an Illumina NextSeq 550 or NovaSeq 6000 system.

Variant calling and bioinformatics analysis

Fastp (v.2·20·0)45 was used to trim adapters and filter low-quality reads. BWA-mem (v.0·7·17)46 was then used to align the cleaned paired-end reads to the human reference genome (hg19). Single nucleotide variants (SNVs) and insertion‒deletion variants (indels) were identified using the tools VarDict (v.1·5·7)47 and InterVar48. Copy number variants (CNVs) were detected using CNVkit (dx1·1), and FACTERA (v1·4·4) was utilized to call fusion events. High-level amplification was defined as ≥ 6 copies49. A total of 334 homopolymer repeat loci were selected to determine microsatellite instability (MSI) status, with the cutoff for MSI-high (MSI-H) defined as 0.15. The tumor mutational burden (TMB) was calculated by summing all the SNVs and indels in the coding region of the targeted genes, with TMB-high (TMB-H) defined as ≥ 10 mut/Mb.

Therapeutic target analysis

OncoKB (v4.24), a precision oncology knowledge database, was used to identify therapeutic or potential targets50,51. Level 1 indicates an FDA-recognized biomarker for predicting drug response. Level 2 includes standard-of-care biomarkers that are predictive of the response to an FDA-approved drug. Level 3 reflects compelling clinical evidence or investigational biomarkers. Level 4 indicates compelling biological evidence.

Statistical analysis

All the data were analyzed using R version 4.3.2. The genomic landscape was visualized using the R package ComplexHeatmap (v2.18.0). The chromosome circos plot was generated with the R package RCircos (v1.2.2). Clinical and genomic data from The Cancer Genome Atlas (TCGA) were retrieved from cBioPortal. The Kaplan‒Meier curves were analyzed with the log-rank test using the R packages Survival (v3.5.7) and Survminer (v0.4.9). Differences in TMB between groups were assessed using the Wilcoxon test, whereas categorical variables were analyzed using either the chi-square test or Fisher’s exact test. P < 0.05 was considered to indicate statistical significance.