NRG1 fusion-driven tumors: biology, detection and the therapeutic role of afatinib and other ErbB-targeting agents

Abstract

Oncogenic gene fusions are hybrid genes that result from structural DNA rearrangements, leading to deregulated activity. Fusions involving the neuregulin-1 gene (NRG1) result in ErbB-mediated pathway activation and therefore present a rational candidate for targeted treatment. The most frequently reported NRG1 fusion is CD74-NRG1, which most commonly occurs in patients with invasive mucinous adenocarcinomas (IMA) of the lung, although several other NRG1 fusion partners have been identified in patients with lung cancer, including ATP1B1, SDC4, and RBPMS. NRG1 fusions are also present in patients with other solid tumors, such as pancreatic ductal adenocarcinoma (PDAC). In general, NRG1 fusions are rare across different types of cancer, with a reported incidence of <1%, with the notable exception of IMA, which represents ≈2‒10% of lung adenocarcinomas and has a reported incidence of ≈10‒30% for NRG1 fusions. A substantial proportion (≈20%) of NRG1 fusion-positive non-small-cell lung cancer (NSCLC) cases are non-mucinous adenocarcinomas. ErbB-targeted treatments, such as afatinib, a pan-ErbB tyrosine kinase inhibitor, are potential therapeutic strategies to address unmet treatment needs in patients harboring NRG1 fusions.

INTRODUCTION

Gene fusions are hybrid genes that result from structural DNA rearrangements including translocations and insertions, transcription read-through, or splicing.¹ Oncogenic gene fusions result in deregulated activity, and are the driver events responsible for initiation and maintenance of various types of cancer.^1–3

For some cancers driven by gene fusions, the successful development of targeted therapies has improved clinical outcomes. For example, tyrosine kinase inhibitors (TKI) such as imatinib, dasatinib and nilotinib have markedly improved outcomes in patients with chronic myeloid leukemia (CML), which is driven by the BCR-ABL1 tyrosine kinase, a fusion protein created by a chromosomal translocation.^2,4

The last few years have seen the identification of a number of activating gene fusions and the demonstration of efficacy and subsequent approval of a number of TKIs targeting these driver fusions. A prime example is the anaplastic lymphoma kinase (ALK) protein; ALK TKIs have been shown to be highly effective in ALK-rearranged lung adenocarcinomas,³ leading to dramatic improvement of survival rate.^5,6 ALK fusions are also found in other tumors, albeit at a lower frequency than in non-small-cell lung cancer (NSCLC).⁷ Other targetable fusions have been identified across a number of tumor types such as the rearrangements of ROS1 in NSCLC,⁸ RET in thyroid malignancies and in NSCLC,^9,10 and the three neurotrophic receptor tyrosine kinase genes (NTRK1, NTRK2, and NTRK3) that can undergo chromosomal rearrangements leading to functional gene fusions.^11–14

In this context, fusions involving the neuregulin-1 gene (NRG1) were first identified by transcriptome sequencing of lung adenocarcinomas, which revealed fusion of CD74 to NRG1.^15,16 Subsequently, many other NRG1 fusion partners have been identified.^17,18 NRG1 fusions are present at a low incidence in multiple tumor types but are enriched in invasive mucinous adenocarcinomas (IMA) of the lung.^18–22

NRG1 fusions result in aberrant expression of the Epidermal Growth Factor (EGF)-like domain of neuregulin-1 (NRG1) on the cell surface, which serves as a ligand for ErbB3 (HER3) and induces the formation of heterodimers, most frequently ErbB2-ErbB3, but also with EGFR (ErbB1) and ErbB4. This leads to pathologic activation of the phosphoinositide 3-kinase-protein kinase B (PI3K-AKT), mitogen-activated protein kinase (MAPK), and other signaling pathways, resulting in abnormal cell proliferation.¹⁶ NRG1 fusions and the subsequent ErbB-mediated constitutive pathway activation therefore represent a rational therapeutic target.

The aim of this narrative review is to provide information on the biology and detection of NRG1 fusions, their role in the development of certain types of cancer, and potential intervention modalities that target the ErbB pathway, including afatinib, as it has been used most frequently in a clinical context, for the treatment of NRG1 fusion-driven tumors. To identify relevant published data on NRG1 and NRG1 gene fusions, we conducted literature searches of PubMed using “Neuroregulin-1” and “afatinib” and “gene fusion” and “NRG1 fusion”. We also searched Google Scholar, as well as conference proceedings from recent major congresses, including ASCO, ESMO, AACR, and WCLC. Searches were updated on March 16^th, 2020.

Biology of NRG1 and NRG1 fusions

Neuregulins and their normal function

The ErbB family of receptor tyrosine kinases is composed of four members (EGFR, ErbB2 [HER2], ErbB3, and ErbB4).¹⁶ There are 13 extracellular ligands for the ErbB family members, which can be divided into those that bind EGFR and ErbB4, and those that bind ErbB3 and ErbB4.¹⁶ This latter group is known as the neuregulins and comprises four members (NRG1–4).²³ NRG1 and NRG2 bind to ErbB3, and NRG1‒4 bind to ErbB4. Upon ligand binding, these receptors form homo- or heterodimers, which results in phosphorylation of the intrinsic kinase domain, and activation of the PI3K-AKT, MAPK, and other pathways.¹⁶

The NRG1 gene is located on chromosome 8 (8p12 region) and encodes NRG1.²⁴ The NRG1 gene is translated to generate six different proteins (I–VI) and at least 31 isoforms.^25–28 NRG1 proteins are structurally related to EGF and contain an EGF-like motif that binds and activates ErbB3 and ErbB4.^29–34 Most NRG1 isoforms are synthesized as membrane-anchored precursors with their EGF domain positioned outside the cell. Proteolytic processing leads to the release of mature, diffusible NRG1, which can activate ErbB3 and ErbB4 on distant cells.³⁵ An exception is NRG1 type III isoform, which passes through the membrane twice and retains its EGF domain tethered to the membrane after proteolytic processing;³⁶ this restricts type III NRG to signaling to adjacent cells.¹⁶

NRG1 binding to ErbB3 can induce heterodimerization, preferentially with the ErbB2 receptor. This allows ErbB2 to phosphorylate the kinase-defective ErbB3 and leads to activation of downstream pathways including the PI3K-AKT and MAPK pathways (Figure 1).^34,37,38 NRG1 also has the capacity to bind ErbB4, similarly leading to dimerization as either a homodimer or a heterodimer with ErbB2/ErbB3, and subsequent activation.^24,39–41 Different ErbB homo- and heterodimers activate different pathways to different degrees;⁴² this might explain differences in response to targeted therapies observed in patients.

Figure 1: Mechanism of action figure.

Figure 1A. NRG1-fusion proteins drive tumorigenic signaling pathways (adapted from Fernandez-Cuesta and Thomas, 2015¹⁶, Ferrer et al 2018,⁷⁸ Gille et al 1996,⁷⁹ Populo et al 2012,⁸⁰ Sakatomoto et al 2009,⁸¹ Shi et al 2007,⁸² Yang et al 1995,⁸³ Manning and Cantley, 2007⁸⁴) and common NRG1-fusions (redrawn from Jonna et al 2019⁴⁸).

A. Wild type NRG1 may be cleaved by proteases to release soluble NRG1 containing the EGF-like region. Soluble NRG1 diffuses to activate distant ErbB3 or ErbB4 receptor subunits (paracrine signaling). Activated receptors dimerize, forming ErbB3- or ErbB4-containing homo- or hetero ErbB dimers, preferentially involving ErbB2.

B. NRG1 fusions by losing, altering or misplacing cleavage sites may become poorer substrates for proteases, e.g. ADAM17, resulting in excess NRG1 tethered to the cell surface.

C. NRG1-fusion proteins hold the EGF-like domain close to the cell surface, causing uncontrolled juxtacrine signaling. Aberrant NRG1 signaling drives cell proliferation and avoidance of apoptosis via PI3K and RAS dependent signaling and activation of downstream signalling molecules, including RALBP-1, CREB, ELK-1, 4EBP, S6K and NFκB.

D. NRG1-fusion proteins may also drive autocrine signaling via the same PI3K/Ras dependent pathways.

Figure 1B. Common NRG1-fusions (redrawn from Jonna et al 2019⁴⁸). Schematics of the wild-type NRG1 gene (green border) and the most common NRG1-fusions (red border, from 21,858 tumours, of more than 26 different types⁴⁸). The CD74-NRG1 fusions shown are CD74 exon 6; CD74 (exon 8)-NRG1 fusions were also reported. The dotted line indicates the point at which NRG1 and the fusion partner are joined. The EGF-like region is shown in purple.

The role of NRG1 fusions in cancer

Discovery and incidence of NRG1 fusions

The first NRG1 fusion was reported in 2014 by Fernandez-Cuesta and colleagues.¹⁵ This team identified a fusion of CD74 to NRG1 while screening a cohort of 25 EGFR and KRAS wild-type lung adenocarcinoma specimens, using a novel computational data analysis strategy. The NRG1 isoform involved in the fusion belonged to type III and contained a β-type EGF domain, which binds with greater affinity to receptors than the α-type.³⁷ Subsequently, they identified an additional four cases of CD74–NRG1 while screening a cohort of 102 pan-negative adenocarcinomas. All five cases were in never-smoker females with IMA of the lung.

Subsequently, CD74–NRG1 has emerged as the most frequently reported NRG1 fusion,^24,43 and most CD74–NRG1 fusions have been reported in patients with lung IMA.^{15,17,19,44–48}

Other NRG1 fusion partners identified in patients with lung cancer include SLC3A2,^17,19,48 SDC4,^43,45,48 RBPMS,^45,48,49 WRN,⁴⁵ VAMP2,⁵⁰ ATP1B1,^18,21,48ROCK1,^43,48 RALGAPA1,⁵¹ TNC,⁴⁸ MDK,⁴⁸ DIP2B,⁴⁸ MRPL13,⁴⁸ DPYSL2,⁴⁸ PARP8,⁴⁸ and ITGB1.⁴⁹ In an analysis of 21,858 tumor specimens that underwent anchored multiplex PCR for targeted RNA sequencing, the prevalence of NRG1 fusions was 0.2%. Of these, CD74 was the most common partner (29%), followed by ATP1B1 (10%), SDC4 (7%), and RBPMS (5%).⁴⁸

Several other NRG1-fusion partners have also been identified across various solid tumor types,^43,48 including POMK (colorectal cancer),^41,52 APP (pancreatic ductal adenocarcinoma [PDAC]),^21,53 CDH6 (PDAC),²¹ ATP1B1 (cholangiocarcinoma and PDAC),^18,21,53 and CLU (ovarian cancer).⁵⁴

Proposed pro-tumoral mechanism of NRG1 fusions

Characterization of the CD74–NRG1 fusion protein provided insight into the pro-tumoral effects of this aberration. Type III NRG1 proteins have cytosolic N-termini and membrane-tethered EGF-like domains. In CD74–NRG1 fusions, the CD74 portion is predicted to replace the transmembrane domain of NRG1 III-β3, resulting in a membrane-tethered EGF-like domain. Fernandez-Cuesta and colleagues demonstrated that the EGF-like domain of the fusion proteins was overexpressed on the cell surface.¹⁵ Maintenance of the extracellular EGF domain at the cellular surface in CD74–NRG1 fusions promotes activation of the ErbB/PI3K/NF-κB pathway. Subsequently, ErbB3 expression was noted at high levels, and the proteins were phosphorylated, in fusion-positive cases.¹⁵ In addition, high ErbB3 expression was significantly associated with NRG1 fusions in a cohort of patients with lung cancer.²² It was hypothesized that CD74–NRG1 signals through induction of ErbB2–ErbB3 heterodimers that activate the PI3K-AKT pathway, as has been shown for wild-type NRG1.¹⁵ This was confirmed by transduction of lung cancer cell lines with viruses expressing the CD74–NRG1 fusion. Transduced cells showed increased levels of phosphorylated ErbB2 and ErbB3, AKT and S6K compared with the empty vector control. The presence of phosphorylated ErbB3 and AKT both depended on the presence of the EGF-like domain of CD74–NRG1 in the fusion. The AKT pathway was also stimulated in Ba/F3 cells expressing normal levels of ErbB2 and ErbB3 when they were co-cultured with NIH-3T3 cells expressing CD74–NRG1.¹⁵

Other groups have confirmed that NRG1 fusions promote signaling through ErbB2/ErbB3 heterodimers.^17,19,43,50 Signaling through EGFR was less important than through ErbB2 and ErbB3.⁴³ The role of ErbB4 remains unclear. It has also been consistently reported in different cancer types, and with different NRG1 fusions, that ErbB2/ErbB3 heterodimers activate signaling through both the PI3K-AKT pathway and the MAPK pathway, with phosphorylation of PI3K, AKT and mTOR being observed, as well as ERK.^{17,19,43,45,50}

In another study, the CD74–NRG1 fusion was shown to confer cancer stem cell (CSC)-like properties, including sphericalization, on immature, progenitor-like cells.⁵⁵ CSCs are a subgroup of cancer cells that have the ability to self-renew and differentiate into other types of mature cells.²⁴ Murayama and colleagues found that expression of CD74–NRG1 induced signaling through ErbB2-ErbB3 heterodimers, which activated the PI3K-AKT pathway, leading to activation of NF-κB and insulin-like growth factor 2 (IGF2), leading to sphericalization of the cells. The CD74–NRG1 fusion not only induced sphere-forming ability in vitro, but also enhanced the tumor-initiating ability in vivo, suggesting that the fusion gene could be involved in tumor development by inducing CSC properties. The CD74–NRG1 fusion did not stimulate rapid growth of CSCs; however, CSCs are thought to be resistant to stressful conditions, hinting at a potential resistance to conventional chemotherapeutics in tumors expressing this fusion.⁵⁵

Others have observed the induction of cancer-like properties in cells expressing NRG1 fusions. Shin et al. showed that an SLC3A2–NRG1 fusion induced cancer cell migration via increased phosphorylation of FAK and Src.¹⁷ They also found that ErbB2/ErbB3 heterodimers were activated by a soluble factor in a juxtacrine and/or autocrine manner. Dhanasekaran and colleagues found that overexpression of CD74–NRG1 induced epithelial to mesenchymal transition (EMT) as shown by increased VIM and SNAIL protein expression and decreased CDH1 levels. Microarray analysis identified overexpression of EMT markers, SRC and ErbB pathways, and under-expression of cell adhesion markers, supporting an oncogenic phenotype.⁴⁵

It has been suggested that the fusion partner may influence the localization of NRG1 to the plasma membrane,⁴⁸ causing it to mimic wild-type NRG1 type III. The NRG1 fusion partners SLC3A2, VAMP2, RBPMS, WNR and SDC4 were predicted to retain the membrane-bound EGF-like domain of the wild-type NRG1 III-β3 form, suggesting they would retain the oncogenic potential of CD74–NRG1.²² Jonna and colleagues identified a small proportion of fusions where exon 2 of NRG1 was spliced to upstream non-coding exons of the fusion partner; this could result in an N-terminal truncated version of NRG1, currently of unknown clinical significance.⁴⁸

Most NRG1 fusions result in the expression of chimeric proteins (e.g. CD74–NRG1), but a WRN–NRG1 gene fusion was found to result in the overexpression of full-length NRG1 regulated by the WRN gene promoter.⁴⁵ Fusion of the NRG1 gene downstream (3ʹ) of a 5ʹ partner brings the EGF-like domain-containing exons of this gene under the control of a foreign transcription regulator.

A number of studies have shown that the NRG III-β isoform is involved in the fusion.^15,21 Isoforms of NRG1 with the β isoform of the EGF-like domain have higher affinity for the receptors than those with the α isoform.³⁷ It could be hypothesized that the combination of overexpression and the higher affinity EGF-like domain result in greater pro-tumoral potential.

Detection of NRG1 fusions

Until recently, fusion proteins and genes in solid tumors were detected primarily using immunohistochemistry and fluorescence in situ hybridization (FISH) techniques; important advances in gene fusion detection include DNA next-generation sequencing (NGS) and the introduction of targeted gene fusion panels on RNA (Table 1).^1,56

Table 1:

The advantages and disadvantages of different assays available for the detection of NRG1 and other gene fusions

Type of assay	Input	Advantages	Disadvantages
FISH56,75	DNA	Widely available and used routinely for other fusions; laboratories are familiar with this technique Faster than NGS techniques	Labor intensive and difficult to interpret (requires diagnostic expertise) Lacks scalability for high-volume multitarget testing Does not provide fusion partner breakpoint precision
Anchored multiplex PCR for targeted RNA sequencing (AMP)75	RNA	Clinical sensitivity and specificity better than standard FISH assays Unlike other PCR methods, AMP enables enrichment of a target region with knowledge of only one of its ends and may be used to assess sequence read complexity based on random start sites	Quality of archived samples may be compromised Costly
Next-generation sequencing of DNA (MSK-IMPACT)43,56	DNA	Next-generation sequencing generally requires a low quantity of tumor material (low DNA inputs)	Results may not be as reliable as targeted RNA sequencing Deep sequencing is required to detect NRG1
Targeted RNA sequencing (MSK Solid Fusion Assay, Archer FusionPlex)43,56	RNA	May be more reliable than next-generation sequencing of DNA Not impacted by large intronic regions Requires low RNA inputs	Long-term storage of samples not recommended (quality of sample may be compromised)
IHC22,75	Proteins (antigens)	Widely available Faster than other techniques Detection of phosphorylated ErbB3 by IHC may potentially be a useful screening method to select for NRG1-positive patients (good correlation with FISH results)	Relies on availability of good-quality antibodies and on qualitative scoring Does not provide fusion partner breakpoint precision

FISH = fluorescent in situ hybridization; IHC = immunohistochemistry; NGS = next generation sequencing; PCR = polymerase chain reaction

In a large analysis of 17,485 solid tumors, detection of NRG1 fusions was undertaken using DNA NGS (MSK-IMPACT) for all tumors and targeted RNA sequencing (MSK-Solid Fusion Assay, Archer FusionPlex) for certain tumours.⁴³ Targeted RNA sequencing detected NRG1 fusions in several KRAS wild-type IMAs and one pancreatic adenocarcinoma, which were not previously detected by next-generation DNA sequencing alone.⁴³ These findings suggest that NRG1 fusions may be missed due to the complexity and diversity of fusion architecture and the large intronic regions characteristic of NRG1. Overall, RNA sequencing may be more reliable than DNA NGS at detecting NRG1 fusions. The unidirectional targeted approach of RNA sequencing allows the detection of unknown fusion partners, which will be particularly useful in light of the variability of NRG1 fusion partners.⁴³ Also of interest are findings from a whole-exome sequencing and RNA-seq analysis of Japanese patients with lung adenocarcinoma who were negative for the driver oncogenes defined by conventional laboratory testing, which identified CD74-NRG2α as an actionable novel fusion.⁵⁷

Prevalence and characteristics of patients with NRG1 fusions

In general, NRG1 fusions are rare across different types of cancer, typically occurring in <1% in most reported series (Figure 2).^{15,19,43,48,58} For example, using prospective DNA NGS, Drilon and colleagues detected NRG1 rearrangements in 3 of 2,079 patients (0.14%) with lung adenocarcinomas, 1 of 791 patients (0.13%) with pancreatic adenocarcinoma, and 1 of 2,703 patients (0.04%) with breast carcinoma.⁴³

Figure 2: Incidence of NRG1 fusions in cancer.

CRC = colorectal cancer; IMA = invasive mucinous adenocarcinoma; KRAS WT = Kirsten rat sarcoma viral oncogene homolog wild-type; NSCLC = non-small cell lung cancer; RCC = renal cell carcinoma

Note, for pancreatic cancer, where reported (Jonna et al. 2019⁴⁸), all cases were KRAS WT

Subsequent analysis of RNA-seq data for 8,984 tumors from The Cancer Genome Atlas identified NRG1 rearrangements across various tumor histologies, as follows: breast (AKAP13-NRG1: 0.01%); head and neck cancer (THBS1-NRG1 and PDE7A-NRG1: 0.49%); lung adenocarcinoma (SDC4-NRG1: 0.22%); renal clear cell carcinoma (PCM1-NRG1: 0.19%); squamous cell lung cancer (THAP7-NRG1 and SMAD4-NRG1: 0.21%); ovarian cancer (RAB3IL1-NRG1: 0.24%); pancreatic cancer (ATP1B1-NRG1: 0.55%); uterine carcinosarcoma (NRG1-PMEPA1: 1.75%), and prostate cancer (NRG1-STMN2: 0.3%).⁴³

Jonna and colleagues⁴⁸ reported on the incidence of NRG1 fusions in 21,858 solid tumor specimens that underwent RNA sequencing, and found that 0.2% harbored an NRG1 fusion. Although the greatest incidence of NRG1 fusions was in NSCLC (25 of 41 cases in total), this represented only 0.3% of NSCLC cases tested. Other NRG1-postive tumor types included gallbladder cancer, renal cell carcinoma, bladder cancer, ovarian cancer, pancreatic cancer, breast cancer, neuroendocrine tumor, sarcoma, and colorectal cancer. As previously noted, NRG1 fusions do appear to be more common in IMA and have been reported in 5 of 72 IMA cases (7%) using anchored multiplex PCR to perform NGS,²⁰ 6 of 90 IMA cases (7%) using whole-transcriptome sequencing,¹⁹ 16 of 59 IMA cases (27%) using targeted sequencing and RT-PCR,¹⁷ and 16 of 51 IMA cases (31%) using FISH.²² In Drilon and colleagues’ study,⁴³ in patients with KRAS wild-type IMA, NRG1 fusions were detected in 4 of 36 patients (11%) using targeted RNA sequencing.

KRAS mutations are the most frequent oncogenic driver mutations in IMA, occurring in 28‒87% of cases.⁵⁹ It was thought that NRG1 fusions were mutually exclusive with KRAS mutations.^19–21 However, a growing number of studies have shown this may not be the case. One series of 16 IMA patients with NRG1 fusions identified KRAS codon 12 mutations in nine cases (56%).¹⁷ Trombetta et al. found concurrent NRG1 rearrangements with KRAS mutations in six of 18 cases (33%).²² With respect to PDAC, it appears that almost all patients harboring NRG1 fusions are KRAS wild-type, which provides a useful screening strategy for identifying PDAC patients with NRG1 fusions.^21,53,60

Initial reports of NRG1 fusions in patients with IMA indicated that most of the patients were female and never smokers,¹⁹ and much of the work was done in an East Asian population.¹⁵ However, a recent evaluation of a cohort of 85 Caucasian lung cancer patients found NRG1 rearrangements in 31% of patients with IMA and 3% of those with non-IMA.²² All of the IMA cases also lacked EGFR mutations and ALK rearrangements.²² These findings indicate that NRG1 fusions are important drivers of IMA in both Asian and Caucasian populations. Moreover, recent results from the international NRG1 registry indicate a substantial proportion (≈20%) of NRG1-positive NSCLC are non-mucinous adenocarcinomas, and there were also cases of non-adenocarcinoma NSCLC.⁶¹

Prognostic significance of NRG1 fusions

The prognostic value of NRG1 fusions in IMA is unclear. As noted earlier, NRG1 fusions may lead to CSC formation;⁵⁵ and CSCs are known to be related to tumor recurrence and metastasis, chemo-resistance and poor prognosis.²⁴

In an evaluation of survival in a series of 59 patients with IMA, patients harboring tumors with NRG1 fusions had a significantly shorter overall survival (OS) than those without NRG1 fusions (median 51.9 months vs. not reached; p=0.019). This was also true for patients with stage I disease (median 48.1 months vs not reached; p=0.009).¹⁷ In contrast, Shim and colleagues reported that OS was no worse in patients with mucinous compared with non-mucinous lung adenocarcinoma, and that there was no difference in prognosis for patients with gene fusions.²⁰

Potential treatments targeting the NRG1/ErbB pathway

Preclinical evidence of ErbB inhibition activity in NRG1-positive models

As NRG1 proteins are ligands of ErbB receptors,^21,62 ErbB-targeted treatments, such as monoclonal antibodies GSK2849330 and zenocutuzumab (MCLA-128) and small molecules afatinib, tarloxotinib, lapatinib, erlotinib and neratinib, are of particular interest. For example, afatinib downregulates ErbB signaling by binding to the kinase domains of EGFR, ErbB2, and ErbB4 and irreversibly inhibiting tyrosine kinase autophosphorylation.⁶³ In a preclinical model in which EFM-19 cells were exposed to conditioned media from human lung cancer cells expressing exogenous CD74-NRG1 fusion protein, afatinib and lapatinib suppressed ErbB2, ErbB3, and ERK phosphorylation, indicating downregulation of ErbB signaling.¹⁹ Anti-tumor activity and downstream ErbB signaling inhibition was also demonstrated with the anti-ErbB3 monoclonal antibody GSK2849330, and to a lesser extent with afatinib, in an NRG1-rearranged patient-derived xenograft model.⁴³ Preclinical studies have also demonstrated efficacy with tarloxotinib, a hypoxia-activated pan-ErbB kinase inhibitor in NRG1 fusion-positive cell lines and patient-derived xenografts.⁶⁴ Initial anti-tumor activity has also been demonstrated with zenocutuzumab (MCLA-128).⁶⁵ In cell lines with aberrant NRG1-expression, exposure to GSK2849330, afatinib, erlotinib, and neratinib significantly inhibited growth and suppressed ErbB2, ErbB3, and AKT phosphorylation.⁴³

Studies using NRG1 fusions from patient cases have shown that afatinib reduced phosphorylation of downstream pathways when cells were treated with it.^19,21 Although, in another study, afatinib was not effective at preventing signaling in cells carrying NRG1 fusions; whereas, an anti-ErbB3 monoclonal antibody, GSK2849330, was.⁴³

As part of the transcriptomic analysis of 5 patients with NRG1 fusions (1 lung cancer, 1 cholangiocarcinoma and 3 PDAC), 4 had extremely high levels of RNA expression ranging between the 93^rd and 100^th percentile of expression compared to NRG1 levels seen across cancers in the TCGA dataset.^18,53 Moderately high RNA expression levels were also observed in ErbB2 and ErbB3 for all 5 patients, and low to normal levels were seen in EGFR and ErbB.

Clinical evidence of ErbB inhibition activity in NRG1-positive patients

Clinical evidence of response to ErbB inhibition in patients with NRG1 fusion-positive tumors has been seen with several different agents (Table 2). To date, the most frequently reported agent showing responses in this setting has been afatinib, although reports of its efficacy in NRG1 fusion-positive tumors is currently limited to case reports in patients with IMA, PDAC, lung adenocarcinoma, ovarian cancer and cholangiocarcinoma (Table 2). Partial responses were achieved for up to 10 months in patients with IMA, and up to 5.5 months in patients with PDAC (Table 2). As well as the case reports discussed here, a global, multicenter registry has been set up to identify patients with NRG1 fusion-positive NSCLCs.⁶¹ Eighty patients have been included on the registry so far, including 12 treated with afatinib. Four of the 12 patients had an objective response [partial response (PR) or stable disease (SD)] to afatinib treatment, with a median progression-free survival of 3.5 months (range 0.6–16.5 months).⁶¹

Table 2:

Reported cases of ErbB-targeted treatment in patients with NRG1 fusion-positive tumors

Patient	Treatment	Tumor type	NRG1 fusion partner	Best response	Duration of response (months)	Reference
1	Afatiniba	Lung adenocarcinoma	SLC3A2	PR	12	Gay et al. 201776
2	Afatiniba	Lung adenocarcinoma	SDC4	PR	12	Jones et al. 201718
3	Afatiniba	IMA	CD74	PR	10	Gay et al. 201776
4	Afatiniba	IMA	CD74	PR	6.5	Cheema et al. 201777
5	Afatiniba	IMA	CD74	SD	3	Drilon et al. 201843
6	Afatiniba	IMA	CD74	PD	-	Drilon et al. 201843
7	Afatiniba	IMA	SDC4	PD	-	Drilon et al. 201843
8	Afatiniba (dose reduced to 30 mg/day due to toxicity)	PDAC	ATP1B1	PR	5.5	Jones et al. 201953
9	Afatiniba 30 mg/day	PDAC	APP	PR	5 (ongoing)	Jones et al. 201953
10	Afatiniba (dose reduced due to cutaneous AEs)	PDAC	ATP1B1	PR	3	Heining et al. 201821
11	Afatiniba monotherapy followed by trastuzumab and pertuzumab	Ovarian cancer	CLU	SD	>36	Murumagi et al. 201954
12	Afatiniba (dose reduced to 30 mg/day due to rash/diarrhea)	Cholangiocarcinoma	ATP1B1	PR	8	Jones et al. 201718
13	GSK2849330b 30 mg/kg IV q2wk	IMA	CD74	PR	19	Drilon et al. 201843
14	Lumretuzumabc 800 mg IV q2wk plus erlotinib	IMA	SLC3A2	SD	4	Han et al. 201766
15	Lumretuzumabc 800 mg IV q2wk plus erlotinib	IMA	SLC3A2	SD	4	Han et al. 201766
16	Zenocutuzumabd (MCLA-128) 750 mg IV q2wk	PDAC	ATP1B1	PR	7+ (ongoing)	Schram et al. 201965
17	Zenocutuzumabd (MCLA-128) 750 mg IV q2wk	PDAC	ATP1B1	SD	7+ (ongoing)	Schram et al. 201965
18	Zenocutuzumabd (MCLA-128) 750 mg IV q2wk	NSCLC	CD74	PR	4.5+ (ongoing)	Schram et al. 201965
19	Pertuzumab and erlotinib	PDAC	SARAF, CDH6	PR	3	Heining et al. 201821

^aOral afatinib, 40 mg/day unless stated otherwise;

^bGSK2849330 (anti-ErbB3 monoclonal antibody) has undergone a phase 1 trial (NCT01966445) but does not appear to be undergoing further development (clinicaltrials.gov);

^cLumretuzumab is an anti-ErbB3 monoclonal antibody;

^dZenocutuzumab (MCLA-128) is a bispecific humanized monoclonal antibody.

AEs = adverse events; APP = amyloid beta precursor protein; ATP1B1 = ATPase Na+/K+ Transporting Subunit Beta 1; CDH6 = Cadherin6; CD74 = CD74 molecule (Major Histocompatibility Complex, Class II Invariant Chain); CLU = clusterin; IMA = invasive mucinous adenocarcinoma; IV = intravenous; NSCLC = non-small-cell lung cancer; PD = progressive disease; PDAC = pancreatic ductal adenocarcinoma; PR = partial response; q2wk = every 2 weeks; SD = stable disease; SARAF = Store-Operated Calcium Entry Associated Regulatory Factor; SDC4 = syndecan 4; SLC3A2 = Solute Carrier Family 3 Member 2

By contrast, some case reports have shown primary resistance to afatinib in patients with NRG1 fusions. Drilon and colleagues investigated afatinib treatment in four patients with NRG1-rearranged lung cancers, and found progressive disease (PD) shortly after starting treatment. Three of the four patients had CD74–NRG1 fusions and one had an SDC4–NRG1 fusion. One of the patients had previously been treated with the anti-ErbB3 molecule GSK2849330 (no longer in development), and had developed PD after a PR lasting 19 months.⁴³

These response discrepancies highlight the need for a global data and treatment sharing platform to gather as much data as possible on these uncommon drivers. In this context, a global multicenter registry of thoracic oncologists was launched to improve our understanding of clinicopathologic characteristics and response to treatment of patients with NRG1 fusion-driven NSCLC.⁶¹

Other case reports or basket studies in patients with NRG1 fusion-driven solid tumors include the anti-ErbB3 agent lumretuzumab (two patients with SD for 4 months)⁶⁶ and the bispecific humanized monoclonal antibody zenocutuzumab (MCLA-128) (three patients with ongoing responses: 7 months PR, 4.5 months PR, and 7 months SD; currently in clinical phase 1/2)^65,67 (Table 2). Other potential targeted treatments against NRG1 fusions include the ErbB2 inhibitors lapatinib, trastuzumab, and pertuzumab, the pan-HER inhibitor tarloxotinib, as well as the ErbB3 inhibitors patritumab (U3–1287; AMG-888), seribantumab (MM-121), AV-203, 9F7-F11, and LJM716. However, only preclinical or phase 1 (pharmacokinetic/pharmacodynamic) data are currently available for these agents.^68–74

Association of NRG1 fusions with resistance to ALK inhibitors

Activation of NRG1 fusion-driven pathways may also act as a mechanism of resistance following treatment with certain TKIs, including ALK inhibitors. A novel NRG1 fusion (RALGAPA1–NRG1) was detected in an ALK+ patient who had progressed on alectinib. Cells engineered to express this fusion were resistant to the ALK inhibitor crizotinib, but sensitive to afatinib.⁵¹ In addition, increased phosphorylated ErbB3 was observed in these engineered cells. The induction of the NRG1 fusion in these cells led to the loss of phosphorylation of SHP2, which is known to be a signaling adaptor for ALK. Analysis of samples from the patient found that signaling via both the ALK and NRG1 fusions was functional. Further analysis of earlier tumor specimens from this patient showed that the RALGAPA1–NRG1 fusion had coexisted with the ALK fusion, and so was not a de novo mechanism of resistance but may have contributed to the short response to crizotinib.

Key points/conclusions

NRG1 fusions are present in multiple cancer types, and in a high proportion of lung IMA. Although these gene fusions are relatively uncommon in most cancer types, they are detectable and targetable vulnerabilities, and recent NRG1 registry data indicate that a substantial proportion (≈20%) of NRG1 fusion-positive NSCLC cases are non-mucinous adenocarcinomas. Other NRG1-positive tumor types include pancreatic cancer (e.g. KRAS wild-type PDAC), gallbladder cancer, renal cell carcinoma, bladder cancer, ovarian cancer, breast cancer, neuroendocrine tumor, sarcoma, and colorectal cancer.

Development of new drugs for rare diseases is challenging, but evaluation of drugs already approved for other indications is a pragmatic option. In the case of NRG1 fusion-driven tumors, existing ErbB-targeted treatments, have potential as targeted therapies in tumors harboring NRG1 fusions. Responses have been reported with afatinib in both IMA and gastrointestinal tumors harboring NRG1 fusions. Future studies to evaluate the efficacy of various ErbB-targeted treatments in a larger number of patients with NRG1 fusion-driven tumors may be best suited to a tumor agnostic real-world analysis or basket trials where feasible.

This approach is ongoing in two prospective studies investigating already licensed drugs in patients with potentially actionable genetic alterations. The Drug Rediscovery Protocol trial (DRUP; NCT02925234) is ongoing in the Netherlands and will enroll around 400 participants with any of more than ten targetable tumor profiles. Similarly, the Targeted Agent and Profiling Utilization Registry study (TAPUR; NCT02693535) is ongoing in the US and will enroll over 3000 patients with any of 13 targetable tumor profiles. Other prospective studies exploring novel agents are also underway, including a phase 2 trial evaluating the clinical activity of tarloxotinib in patients with NSCLC harboring an EGFR Exon 20 insertion or HER2-activating mutation and other advanced solid tumors with NRG1/ERBB family gene fusions (NCT03805841), and a phase 1/2 trial of zenocutuzumab (MCLA-128) in patients with solid tumors harboring an NRG1 fusion (NCT02912949).

As the implementation of more broad spectrum genomic and transcriptomic sequencing approaches becomes more common, the number of niche opportunities for targeting cancers expands. The growing number of targetable cancer-driving fusions is an example of the changing landscape of cancer therapy. While tumors with NRG1 fusions do not represent a substantial proportion of cancers overall, the emerging evidence supports the use of existing anticancer agents effectively repurposed to target these cancer drivers.

Highlights:

NRG1 fusions result in ErbB-mediated pathway activation and present a rational therapeutic target

ErbB-targeted treatments, such as afatinib, can be effective therapeutic options in tumors harboring NRG1 fusions

The incidence of NRG1 fusion-driven tumors is low, occurring in <1% of cancers, but is highest (≈10‒30%) in IMA