Decreased HER2 expression in endometrial cancer following anti-HER2 therapy

Abstract

Trastuzumab has demonstrated clinical efficacy in the treatment of HER2-positive serous endometrial cancer (EC), which led to its incorporation into standard-of-care management of this aggressive disease. Acquired resistance remains an important challenge, however, and its underlying mechanisms in EC are unknown. To define the molecular changes that occur in response to anti-HER2 therapy in EC, targeted next-generation sequencing, HER2 immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) were performed on pre- and post-treatment tumour samples from 14 patients with EC treated with trastuzumab or trastuzumab emtansine. Recurrent tumours after anti-HER2 therapy acquired additional genetic alterations compared to matched pre-treatment ECs and frequently showed decreased HER2 protein expression by IHC (7/14, 50%). Complete/near-complete absence of HER2 protein expression (score 0/1+) observed post-treatment (4/14, 29%) was associated with retained HER2 gene amplification (n=3) or copy number neutral status (n=1). Whole-exome sequencing performed on primary and recurrent tumours from the latter case, which exhibited genetic heterogeneity of HER2 amplification in the primary tumour, revealed selection of an early HER2-non-amplified clone following therapy. Our findings demonstrate that loss of target expression, by selection of HER2 non-amplified clones or more commonly, by downregulation of expression, may constitute a mechanism of resistance to anti-HER2 therapy in HER2-positive EC.

Introduction

Trastuzumab, a monoclonal antibody targeting HER2, has revolutionized the treatment of patients with HER2-positive malignancies, in particular breast and gastroesophageal cancers. In high-grade endometrial cancer (EC), of which serous carcinoma is the archetype, HER2 amplification or overexpression has been reported in 10–40% of cases across various studies [1–3]. A randomized phase II trial has demonstrated improved survival outcomes in advanced stage and recurrent HER2-overexpressing serous ECs treated with trastuzumab combined with carboplatin/paclitaxel, compared to chemotherapy alone [4]. This has led to incorporation of trastuzumab into clinical treatment guidelines for HER2-positive serous EC and development of newer anti-HER2 agents, including trastuzumab emtansine (T-DM1), for this disease.

Whilst HER2 is undeniably an effective drug target, most patients eventually develop resistance to targeted therapy. Various resistance mechanisms have been proposed based on pre-clinical studies, with some validated in clinical samples of breast or gastroesophageal cancer [5,6]. These include changes at the receptor level, including altered level of expression or acquisition of HER2 mutations, or involving downstream signalling, leading to re-activation of PI3K and MAPK signalling, or crosstalk with other signalling pathways [5,6]. Limited experience with anti-HER2 therapies in high-grade ECs, given their relative rarity and only recent incorporation of this therapeutic option into clinical practice, has, until now, precluded investigation into the molecular changes associated with HER2 blockade in this tumour type. In the present study, we performed molecular analyses of paired pre- and post-treatment tumour samples from high-grade EC patients treated with trastuzumab or T-DM1 to elucidate putative mechanisms of acquired resistance.

Materials and methods

This study was approved by the institutional review board of Memorial Sloan Kettering Cancer Center.

HER2 immunohistochemistry and fluorescence in situ hybridization

Following institutional review board approval, 14 patients with high-grade EC were identified with archival formalin-fixed paraffin-embedded (FFPE) tissue specimens obtained pre-treatment and post-treatment with trastuzumab or T-DM1 (Table 1). HER2 immunohistochemical (IHC) analyses were performed (4B5, pre-diluted; Ventana, Tucson, AZ, USA)[1,3]. HER2 IHC was scored according to recently proposed criteria for EC, endorsed by the College of American Pathologists (CAP)[7]. For a more continuous measure, the percentage of positive tumour cells and staining intensity were also used to calculate a HER2 H-score: (0 x percentage of cells with absent membranous staining) + (1 x percentage of “1+” cells) + (2 x percentage of “2+” cells) + (3 x percentage of “3+” cells).

Table 1.

Clinicopathologic features of high-grade endometrial cancers treated with anti-HER2 therapy.

Case	Histological type	Treatment setting	Anti-HER2 therapy	Best radiological response	Pre-treatment tumour			Post-treatment recurrent tumour
					Specimen	HER2 IHC score	FISH (HER2/CEP17)	Specimen	HER2 IHC score	FISH (HER2/CEP17)
EC01	Serous	Recurrent	T-DM1	SD	Primary tumour (res) Lung recurrence (bx)	3+ 3+	3.9 N/A	Thoracentesis (bx)	1+	3.3
EC02	HGEC	Recurrent	T-DM1	SD	Pelvic mass (bx)	3+	5.5	Vagina (bx)	3+	6.4
EC03	HGEC	Recurrent	Trastuzumab	PD	Primary tumour (res) Adrenal recurrence (bx)	2+ 3+	1.7 1.8	Pleura (bx)	2+	2.7
EC04	Serous	Recurrent	Trastuzumab	CR	Primary tumour (res)	2+	3.0	Vagina (bx)	2+	3.1
EC05	HGEC	Recurrent	T-DM1	PR	Primary tumour (res)	3+	N/A	Liver (bx)	3+	N/A
EC06	Serous	Recurrent	Trastuzumab, T-DM1, other	PD	Primary tumour (res)	3+	7.8	1. Vagina (bx) 2. Bladder (bx)	3+ 0	11.5 1.5
EC07	Serous	Adjuvant	Trastuzumab	N/A	Primary tumour (res)	3+	5.3	Vagina (bx)	3+	4.1
EC08	Serous	Recurrent	Trastuzumab	PR	Primary tumour (res)	3+	1.8	Abdominal wall (bx)	3+	1.8
EC09	Serous	Adjuvant	Trastuzumab	N/A	Primary tumour (res)	3+	7.3	Lymph node (bx)	3+	7.0
EC10	Serous	Recurrent	Trastuzumab	PR	Primary tumour (bx)	3+	5.1	Ileum (res)	3+	7.0
EC11	HGEC	Recurrent	Trastuzumab	SD	Abdominal wall (bx)	3+	3.6	Omentum (res)	3+	3.7
EC12	Serous	Adjuvant	Trastuzumab	N/A	Primary tumour (res)	3+	6.8	Peritoneum (bx)	3+	8.4
EC13	Serous	Recurrent	Trastuzumab	PR	Primary tumour (res)	3+	3.1	Pleura (bx)	0	3.0
EC14	Serous	Recurrent	Trastuzumab	SD	Primary tumour (res) Lung recurrence (bx)	3+ 2+	3.4 2.7	Lung (bx)	0	3.0

HGEC – high-grade endometrial carcinoma with ambiguous features; CR – complete response; PR – partial response; SD – stable disease; PD – progression of disease; IHC – immunohistochemistry; FISH – fluorescence in situ hybridization; res – resection; bx – biopsy; N/A – not applicable or not available.

HER2 fluorescence in situ hybridization (FISH) was performed on available archival tissue using the HER2 IQFISH pharmDx assay (Dako, Carpinteria, CA, USA). HER2 (red) and chromosome enumeration probe 17 (CEP17; green) signals were enumerated in at least 20 tumour cell nuclei. HER2 amplification by FISH was defined as HER2/CEP17 ratio of ≥ 2.0.

Tumour DNA sequencing and analysis

Tumour samples were microdissected under a stereomicroscope to ensure tumour purity >80%. DNA extracted from tumour and matched normal tissue was subjected to targeted panel sequencing using the MSK-IMPACT assay (n=10) or whole-exome sequencing (n=1). Sequencing data were analyzed as described previously [8]. In one case, HER2-positive and HER2-negative areas by IHC were separately microdissected and subjected to whole-exome sequencing [9]. To estimate the clonal architecture and composition of different lesions and construction of phylogenetic tree based on mutations, mutant allelic fractions from all somatic mutations were adjusted for tumour cell content, ploidy, local copy number, and sequencing errors using PyClone (https://github.com/Roth-Lab/pyclone ). For the construction of phylogenetic trees based on copy number alterations (CNAs), major and minor copy numbers computed by FACETS, were modelled using transducer-based pairwise comparison functions using MEDICC, assuming a diploid state with no CNAs to root the phylogenies [8].

Gene expression analysis

Histological review was performed to ensure tumour purity >80% and comparable between matched pre- and post-treatment samples, and percentage area occupied by stromal immune infiltrates was estimated. RNA extracted from FFPE tissue sections was subjected to RT-qPCR and NanoString nCounter assays. Levels of HER2 mRNA expression relative to ACTB (β-actin) control was assessed by RT-qPCR reactions in triplicate, using TaqMan Gene Expression Assay probes (ERBB2: Hs01001580_m1, ACTB: Hs01060665_g1; ThermoFisher, Waltham, MA, USA). The NanoString nCounter PanCancer Progression Panel was employed, which assesses 770 genes related to cancer progression and metastasis (NanoString Technologies Inc, Seattle, WA, USA). Data analysis was performed using the NanoString nSolver 4.0 software and differential gene expression analysis was performed on raw counts using DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html).

Statistical analysis

Comparisons of matched pre- and post-treatment samples (fraction of genome altered, HER2 IHC H-scores, stromal immune infiltrates) were performed by paired t-tests, two-sided.

Results and Discussion

Genetic profiling of ECs before and after anti-HER2 therapy

We sought to investigate whether genetic alterations would be acquired following anti-HER2 therapy (trastuzumab or T-DM1) through somatic genetic profiling of matched tumour samples obtained pre-treatment, and post-treatment at progression/recurrence (n=11, Figure 1A). Sequencing analysis revealed that all truncal/clonal somatic mutations affecting cancer-related genes, including TP53, PIK3CA, FBXW7, PPP2R1A, RB1, ERBB2, and NF1, in pre-treatment tumours were retained in matched post-treatment tumours (Figure 1B). In EC08, the post-treatment sample acquired a BRAF p.Q609K variant of unknown significance. EC04 harboured two PIK3CA hotspot mutations (p.H1047R and p.E545K). While the p.H1047R mutation was clonal in both samples, a subclonal-to-clonal shift in the cancer cell fraction for the p.E545K mutation was observed from pre- to post-treatment. Acquired mutations in MAPK pathway components and PIK3CA mutations have been previously implicated in trastuzumab resistance in breast cancer [10,11].

Figure 1. Molecular profiling of matched endometrial cancer samples prior to anti-HER2 therapy and post-treatment at recurrence.

(A) Schematic illustrating temporal relationships between pre-treatment and post-treatment tissue samples collected and courses of anti-HER2 therapy, annotated according to the legend. IHC – HER2 immunohistochemistry; FISH – HER2 fluorescence in situ hybridization; NGS – targeted panel next generation sequencing. (B) Pathogenic somatic genetic alterations affecting cancer-related genes in matched pre- and post-treatment tumour samples. Somatic mutations (top), cancer cell fractions (CCFs) of somatic mutations identified (middle) and gene copy number alterations (bottom), color-coded according to the legend. For EC04, two separate hotspot mutations in each of TP53 and PIK3CA are represented by half of the corresponding boxes on the CCF plot. For EC06, multiple pre- and post-treatment samples were sequenced (see Figure 3 for details).

Additional CNAs were observed in post- compared to pre-treatment tumours, including PIK3CA (n=1) and CCNE1 amplification (n=4). Chromosomal instability was consistently increased post-treatment (median fraction of genome altered of 40% versus 60%, for pre- versus post-treatment tumours, respectively, p=0.0004, supplementary material, Figure S1A), consistent with previous work showing tumour progression to be associated with a generalized increase in CNAs in EC [12]. Thus, while CCNE1 amplification has been associated with trastuzumab resistance in breast cancer [13], further work is necessary to determine whether anti-HER2 therapy directly selects for CCNE1 amplification in EC. Overall, while targeted genetic sequencing did not definitively identify a predominant driver of therapeutic resistance, the results are consistent with prior studies comparing the genetic alterations between primary ECs and their metastases, which demonstrated shared truncal driver mutations across all lesions [14,15].

Decreased HER2 expression following anti-HER2 therapy

Previous studies, most notably in breast cancer, reported loss of HER2 amplification and/or expression following trastuzumab-based therapy [16,17]. In our EC cohort, IHC revealed decreased HER2 protein expression (i.e. decreased IHC score) in post-treatment relative to pre-treatment ECs from 7/14 (50%) patients (Figure 2A, Table 1). Of these, complete or near-complete absence of HER2 protein expression was observed in 4 cases (29%; Figure 2B,C). In the remaining 3 cases, pre-treatment tumours exhibited 3+ HER2 expression, whilst matched post-treatment samples were scored as 2+, due to enrichment of tumour cells with weaker HER2 expression. Quantitation of HER2 IHC using the H-score method also revealed a generalized decrease in HER2 protein expression post-treatment, across the cohort (median pre-treatment versus post-treatment: 245 versus 188, p=0.005; supplementary material, Figure S1B).

Figure 2. Decreased HER2 expression in endometrial carcinomas following anti-HER2 therapy.

(A) Proportions of tumour cells at varying levels of HER2 expression, estimated from immunohistochemical (IHC) stained sections, and overall HER2 IHC scores for matched pre- and post-anti-HER2-treatment endometrial cancer tissue samples. (B) Photomicrographs of HER2 IHC performed on pre- and post-treatment tumours, and corresponding copy number plots, from EC01. Despite the post-treatment sample retaining a high level of HER2 amplification, there is loss of HER2 protein expression following T-DM1 therapy. (C) Photomicrographs of HER2 IHC performed on pre- and post-treatment tumours from EC14. Corresponding HER2 fluorescence in situ hybridization on the post-treatment sample shows retained HER2 amplification. Scale bar, 10 μm. (D) HER2 mRNA expression levels, quantified by RT-qPCR, in evaluable pre-treatment and post-anti-HER2-treatment endometrial cancer samples.

HER2 amplification status, by FISH, was concordant between matched samples, in all cases except EC06. HER2 (ERBB2) mRNA expression, assessed using RT-qPCR, was evaluable in 5 cases with RNA of sufficient quantity and quality extracted from matched samples. Post-treatment samples from EC01 and EC06, with HER2 IHC scores of 1+ and 0, respectively, showed the lowest levels of HER2 mRNA expression (Figure 2D).

Reduction or loss of HER2 following anti-HER2 therapy has been well described in breast cancer patients and associated with inferior prognosis [16,17]. Several non-genetic mechanisms have been implicated, including receptor internalization induced by drugs targeting the extracellular domain [18], and transcriptional downregulation mediated by epithelial-mesenchymal transition or activation of STAT1 signalling due to increased IFN-gamma production by immune cells [19]. While it is likely that complete loss of HER2 expression would render tumour cells resistant to anti-HER2 therapy, it is unclear whether smaller decreases (e.g., from 3+ to 2+) is clinically significant. Halle et al have reported that 62% of primary endometrial cancers with high HER2 expression (3+) are associated with metastases with lower HER2 protein expression (0 to 2+) [20]; therefore we cannot definitively conclude that decreased HER2 expression was the direct result of anti-HER2 therapy. However, therapy-induced transcript downregulation is still the most parsimonious explanation for those cases with discordant IHC/FISH results in post-treatment samples (0 or 1+ IHC, but with HER2 amplification, as observed in EC01, EC13 and EC14). In the series by Halle et al, all ECs with HER2 amplification were IHC 3+ [20]. Whether chemotherapy alone can induce decreased HER2 expression remains to be determined.

Transcriptomic analyses reveal upregulation of an immune/inflammatory response following trastuzumab therapy

To further characterize the transcriptomic changes following anti-HER2 therapy, 4 matched pre- and post-treatment samples with sufficient RNA quality, similar tumour purity between matched samples (EC04, EC09, EC11 and EC06 - primary versus first recurrence only) and uniform treatment (trastuzumab), were subjected to gene expression profiling using a NanoString assay targeting genes involved in cancer progression. This exploratory analysis showed that post-trastuzumab-treated tumours displayed increased expression of immune/inflammatory genes compared to the matched pre-treatment samples, with IL6, CXCL8, CXCL10, and PRF1 being the most highly upregulated genes (supplementary material, Figure S2). Morphological assessment did not reveal appreciable differences in densities of stromal immune infiltrates in pre-treatment versus post-treatment samples (29% versus 33%, p=0.58). It is unclear whether the transcript changes observed are due to therapy or a generalized response associated with disease progression.

Selection of HER2-negative tumour clones following multiple courses of anti-HER2 therapy

For one case, EC06, multi-regional sampling of the primary tumour and biopsies of recurrences at separate timepoints were subjected to whole-exome sequencing (WES) to characterize the evolutionary progression of disease. The primary resection specimen showed a stage IIIC (pT2N1) endometrial serous carcinoma with heterogeneous HER2 amplification and expression (Figure 3A). A CT scan at 2 months post-surgery revealed disease progression in the lung and abdomen. The patient received trastuzumab with paclitaxel, which was discontinued after 4 cycles due to progressive disease. Recurrent high-grade carcinoma, detected on vaginal biopsy at 12 months after initial diagnosis, was homogeneously HER2-amplified with overexpression. Multiple recurrences were observed over the next 3 years. Subsequent treatment regimens included T-DM1 and an investigational bi-specific monoclonal antibody against HER2, on clinical trial. An enlarging pelvic mass encasing the bladder which was recalcitrant to treatment was then biopsied. This tumour was negative for HER2 amplification/expression. The patient died 4 months later.

Figure 3. Clonal decomposition and phylogenetic analysis of multiple primary and recurrent samples before and after anti-HER2 therapy from Case EC06.

(A) HER2 immunohistochemistry and fluorescence in situ hybridization on the pre-treatment primary endometrial tumour (T1, T2) and pelvic lymph node metastasis (LN) from the resection specimen, and post-treatment vaginal recurrence (Post1) and subsequent bladder recurrence (Post2). Scale bar, 10 μm. (B) Clonal frequency heatmap of mutations in the different samples of EC06, grouped by clonal/subclonal clusters. Cancer cell fractions (CCFs) of identified mutations are shown, colour coded according to the legend. Clusters are colour-coded below the heatmap, with the number of somatic mutations per cluster indicated. (C) Phylogenetic tree based on mutations; the numbers of somatic mutations resulting in the divergence of a clone/subclone from its ancestor are shown. (D) Phylogenetic tree based on copy number alterations; the numbers of copy number alterations resulting in the divergence of a clone/subclone from its ancestor are shown. WGD, whole-genome duplication.

WES was performed on HER2-non-amplified (T1) and HER2-amplified (T2) regions of the primary EC, pelvic lymph node metastasis from the original resection specimen (LN) with HER2 copy number gain, the recurrent HER2-amplified vaginal tumour following trastuzumab treatment (Post1), and the subsequent HER2-non-amplified bladder recurrence after additional courses of investigational anti-HER2 therapies (Post2; Figure 3A). Clonal decomposition and phylogenetic analyses based on mutations revealed all tumour samples to be clonally related, sharing truncal mutations, including TP53 (C135Y) and RB1 (E97*) somatic mutations (Figure 3B,C and supplementary material, Figure S3). Notably, the HER2-non-amplified bladder recurrence (Post2) stemmed from an ancestral clone with early clonal divergence, indicating that the disseminated tumour cells that gave rise to Post2 were present even at the time of initial diagnosis. Independent phylogenetic reconstruction based on CNAs demonstrated similar relationships and showed whole-genome duplication events accompanying tumour progression (Figure 3D and supplementary material, Figure S4).

Previous genomic analyses of matched primary ECs and matched metastases generally supported monophyletic evolution, with all metastases originating from a single subclone within the primary tumour [14,15]. In contrast, EC06 exhibited polyphyly, as HER2-amplified and non-amplified metastases were derived from distinct subclones within the primary tumour. Our data suggest HER2 genetic heterogeneity within the primary tumour may give rise to inter-tumour heterogeneity across metastatic lesions, and strategies to treat HER2-negative metastases are an unmet clinical need.

The findings of the present study are limited by the small sample size and heterogeneity with respect to treatment setting, regimen, and duration. Nevertheless, we show that loss of HER2 expression is a common occurrence in ECs treated with trastuzumab/T-DM1, which occurs via transcript downregulation or selection of HER2-non-amplified tumour cells. Due to lack of a chemotherapy-only group, we cannot definitively conclude that these changes are specific to anti-HER2 therapy, though loss of target expression after anti-HER2 therapy has been observed in other tumour types and linked to trastuzumab resistance [5,6]. Emerging HER2 antibody drug conjugates with bystander effects, including trastuzumab deruxtecan, may therefore represent a promising strategy to target HER2-negative tumour subpopulations in HER2-expressing ECs.

Supplementary Material

Supinfo2

Table S1. Raw counts from the NanoString nCounter PanCancer Progression Panel

Supinfo1

Figure S1. Chromosomal instability is increased, while HER2 expression is decreased, following anti-HER2 therapy

Figure S2. Differentially expressed genes in paired pre- and post-trastuzumab treated endometrial cancer samples

Figure S3. Somatic mutations and cancer cell fractions of mutations identified in primary and recurrent samples of case E06 subjected to whole-exome sequencing

Figure S4. Copy number profiles for all specimens from case EC06.

Data availability statement

The gene expression dataset is provided in the supplementary material. Table S1.