Abstract
About 50% of breast cancers are defined as HER2-low and may benefit from HER2-directed antibody–drug conjugates. While tissue sequencing has evaluated potential differences in genomic profiles for patients with HER2-low breast cancer, genetic alterations in circulating tumor DNA (ctDNA) have not been well described.
We retrospectively analyzed 749 patients with metastatic breast cancer (MBC) and ctDNA evaluation by Guardant360 from three academic medical centers. Tumors were classified as HER2-low, HER2-0 (IHC 0) or HER2-positive. Single-nucleotide variants, copy-number variants, and oncogenic pathways were compared across the spectrum of HER2 expression. Overall survival (OS) was evaluated by HER2 status and according to oncogenic pathways.
Patients with HER2-low had higher rates of PIK3CA mutations [relative risk ratio (RRR), 1.57; P = 0.024] compared with HER2-0 MBC. There were no differences in ERBB2 alterations or oncogenic pathways between HER2-low and HER2-0 MBC. Patients with HER2-positive MBC had more ERBB2 alterations (RRR, 12.43; P = 0.002 for amplification; RRR, 3.22; P = 0.047 for mutations, in the hormone receptor–positive cohort), fewer ERS1 mutations (RRR, 0.458; P = 0.029), and fewer ER pathway alterations (RRR, 0.321; P < 0.001). There was no difference in OS for HER2-low and HER2-0 MBC [HR, 1.01; 95% confidence interval (CI), 0.79–1.29], while OS was improved in HER2-positive MBC (HR, 0.32; 95% CI, 0.21–0.49; P < 0.001).
We observed a higher rate of PIK3CA mutations, but no significant difference in ERBB2 alterations, oncogenic pathways, or prognosis, between patients with HER2-low and HER2-0 MBC. If validated, our findings support the conclusion that HER2-low MBC does not represent a unique biological subtype.
HER2 low is an area of growing interest given groundbreaking results of DESTINY-Breast04, and subsequent approval of trastuzumab deruxtecan, a novel HER2-directed antibody–drug conjugate, for patients with HER2-low metastatic breast cancer (MBC). Prior studies in tissue have aimed to determine the prognostic significance of HER2 low and to evaluate whether HER2-low represents a unique biologic subtype. We present the first study to evaluate molecular alterations by circulating tumor DNA (ctDNA) across the spectrum of HER2 expression to evaluate HER2-low MBC. Our analysis using a large clinically annotated multi-institutional dataset indicates that HER2-low breast cancers do not have a unique subtype based on ctDNA. We only identified a higher rate of PIK3CA mutations among hormone receptor–positive HER2-low MBC and, if validated in other datasets, PIK3CA could be an important target for combination treatment trials in HER2-low MBC. HER2-low appears to be a heterogenous group with outcomes driven by hormone-receptor expression.
Introduction
In breast cancer, 10% to 20% of tumors are characterized by overexpression or amplification of HER2 and considered HER2-positive by current ASCO/CAP guidelines (1). HER2 overexpression/amplification is assessed using IHC and/or ISH-based techniques (2, 3). Within HER2-negative disease, substantial heterogeneity exists regarding the expression of HER2. Approximately 50% of all breast cancers will have low HER2 expression, defined as an IHC score of 1+ or 2+ without gene amplification by ISH (HER2-low; refs. 1, 3, 4). Traditionally, patients with HER2-low tumors did not benefit from HER2-targeted therapies, such as trastuzumab (5). However, novel HER2-directed antibody–drug conjugates (ADC) with bystander effect have shown promising therapeutic activity (6, 7). The pivotal phase III trial of trastuzumab deruxtecan demonstrated significant improvement in overall survival (OS) for patients with pretreated HER2-low metastatic breast cancer (MBC), leading to FDA approval in August 2022 (8). These findings have led to consideration of HER2-low MBC as a new subtype and led to significant interest in understanding the biology of these tumors.
HER2 overexpression in breast cancer, defined as IHC score 3+ or IHC 2+ with gene amplification using ISH-based techniques, is associated with more aggressive tumor biology and, prior to the advent of HER2-targeted therapies, decreased patient survival (9). Heterogenous HER2 expression is associated with resistance to HER2-targeted therapies, shorter time to progression, and lower OS (10, 11). The clinical significance of HER2-low breast cancer is currently under investigation. Several retrospective studies have described the prognostic significance of HER2-low breast cancer, with conflicting results. While some studies identified differences in clinical outcomes between HER2-low and HER2-0 breast cancers, including decreased pCR rates with neoadjuvant chemotherapy (12–14), other studies indicate no prognostic significance (15–19). Overall, no solid evidence supports HER2-low status as an independent prognostic factor. Differences in outcomes may be driven by the presence or absence of hormone-receptor expression (15).
Recent retrospective cohort studies have aimed to evaluate the molecular landscape of HER2-low breast cancer using gene-expression profiling from tumor tissue (4, 14, 16, 20). By gene expression profiling, HER2-low breast cancers encompass heterogenous tumor types representing all four intrinsic subtypes, and the proportion of luminal and basal-like tumors is primarily driven by hormone receptor status (20, 21). Agostinetto and colleagues reported a higher proportion of HER2-enriched tumors (20), and Schettini and colleagues reported higher expression of ERBB2 in HER2-low breast cancers compared with HER2-0 (16). These findings could indicate that HER2-low breast cancers have some HER2-driven cancer biology. However, other studies have failed to confirm such associations (4, 14). It remains unclear whether HER2-low represents a unique biologic subtype.
Circulating tumor DNA (ctDNA) has emerged as a noninvasive technique to detect cancer-specific gene aberrations and select patients for targeted therapies (22–25). Prior studies have demonstrated a relatively high concordance between blood and tissue next-generation sequencing (NGS), with potential differences in the detection of specific alterations between blood and tissue limited by spatial and temporal heterogeneity, differences in sequencing technologies, and sensitivity for detecting low-frequency variants (26–28). Prior work from our group characterized genetic alterations detected by ctDNA across the different clinical subtypes and histologies of breast cancer (24, 29). The ctDNA profile of HER2-low breast cancers has not been previously described.
In this study, we describe the landscape of genetic alterations detected by ctDNA among patients with HER2-low MBC using a large, multi-institutional cohort of patients. The primary objective of this work was to define differences in mutation frequency, copy-number variations (CNV) and oncogenic pathways across the spectrum of HER2 expression.
Materials and Methods
Study population
This retrospective cohort included 749 patients with MBC who received treatment at three academic centers (Washington University in St. Louis, MO, Northwestern University in Chicago, IL, and Massachusetts General Hospital in Boston, MA) and underwent ctDNA analysis prior to initiating a new treatment between 2015 and 2020. Samples were collected as standard-of-care, and ctDNA analysis was performed using the commercially available Guardant360 assay (Guardant Health, Redwood City, CA). No selection was made on the basis of prior therapies. Baseline imaging was performed before treatment start and ctDNA sampling. Type of imaging was selected according to investigator's choice. Sites of metastasis were categorized on the basis of the presence of specific organ involvement (e.g., whether or not there was liver involvement) independently from other sites of metastasis.
Data collection and sharing
The study was approved by the Institutional Review Boards (IRB) of the three sites: Washington University School of Medicine (St. Louis, MO; IRB No. 202101147), Northwestern University (Chicago, IL; IRB No. STU00214133) and Massachusetts General Hospital (Boston, MA; IRB No. 2013P000848). The requirement for informed consent was waived by the IRB for this deidentified analysis. Data were shared using a data use agreement that was signed by each participating site. The study was performed in concordance with the Health Insurance Portability and Accountability Act and the Declaration of Helsinki.
Dates (e.g., date of birth, date of progression) were removed from each database prior to the merger and only years (e.g., age) and number of days (e.g., for OS) were shared.
Determining HER2 status
The patients had tissue biopsies collected from either the metastatic site (majority of cases) or primary breast cancer, which were evaluated for HER2 expression using current pathologic assessment methods. HER2-positive was defined as IHC 3+ or IHC 2+/ISH positive. HER2-negative cases were further categorized as HER2-low, defined as IHC 1+ or IHC 2+/ISH negative or HER2-0, defined as IHC 0. For patients with multiple tissue biopsies, the tissue biopsy that was collected in closest proximity to the date of blood sampling for ctDNA analysis was used to determine the HER2 status. Pathologic determination was determined at each site per standard guidelines.
ctDNA sample collection and analysis
Two 10-mL samples of whole blood were collected for each patient using standard stabilizing tubes (Streck, NE) and were analyzed using the commercial Guardant360 assay. The NGS testing was performed as part of standard clinical care. We restricted our analysis to include only ctDNA sampled at the time of metastatic diagnosis or at the time of disease progression. Plasma samples were evaluated for somatic single-nucleotide variants (SNV), insertions/deletions (indels), gene fusions/rearrangements (fus), and CNV. The allele fraction reporting threshold was 0.04% for SNVs, 0.02% for indels, and 0.04% for fusions. The threshold for reporting copy-number alterations was 2.12 copies for all genes examined in the assay (30–32). Mutations were annotated through the OncoKB database according to their effect (loss of function, gain of function) and pathogenicity (33). Oncogenic pathways (RTK, RAS, RAF, MEK, NRF2, ER, WNT, MYC, P53, cell cycle, notch, PI3K) were defined on the basis of prior work generated using The Cancer Genome Atlas (34). Our analysis was modified to include only those genes assessed from the Guardant360 assay (Supplementary Table S1).
Statistical analysis
Clinical and pathologic variables were reported using descriptive analyses. Categorical variables were reported as frequency distributions, whereas continuous variables were described through median and interquartile ranges. Mutational profiles were compared using χ2 and Fisher exact test, where appropriate, to assess differences in alteration frequency across the spectrum of HER2 expression and by hormone receptor status. Univariable and multivariable multinomial logistic regression was performed to determine relative risk ratios (RRR) and 95% confidence intervals (CI) taking HER2-0 as the reference category.
OS was defined from the time of first ctDNA collection to death from any cause. Patients who were still living at the date of last follow-up were censored. Hormone receptor status, lines of therapy and sites of disease were included in the multivariable model. Differences in OS were assessed using the log-rank test and Cox regression models and displayed using Kaplan–Meier plots. Interaction with HER2 status and oncogenic pathways was investigated through subgroup analysis.
Statistical analysis was performed using STATA [StataCorp. (2019) Stata Statistical Software: Release 16.1. College Station, TX: StataCorp LP], JMP [SAS Institute Inc. (2019), version 16. Cary, NC], and R [R Core Team (2019), version 4.1.0. R Foundation for Statistical Computing, Vienna, Austria].
Data availability
The data generated in this study are available upon reasonable request from the corresponding author and with a data-sharing agreement from all participating sites. Supplementary data for this article are also available online.
Results
Cohort characteristics
The combined cohort consisted of 749 patients with MBC who received treatment at Washington University in St. Louis, Northwestern University, or Massachusetts General Hospital and underwent ctDNA testing using the commercially available Guardant360 assay. Testing was performed at the time of diagnosis for patients with de novo metastatic disease or in close proximity to clinical or radiographic disease progression. All patients had tissue biopsies performed for HER2 expression analysis including 49 samples (7%) from the primary tumor and 700 samples (93%) from a metastatic site.
The cohort consisted of 224 patients with HER2-0 MBC (29.9%), 353 patients with HER2-low MBC (47.1%), and 172 patients with HER2-positive MBC (23.0%). Of the HER2-low MBC, 55% were IHC 1+, while 45% were IHC 2+/ISH negative. Most HER2-low MBC were hormone receptor–positive (85.3%). Triple-negative breast cancer (TNBC) occurred in 14.7% of HER2-low and 18.8% of HER2-0 MBC (P = 0.203). The most common histologic subtype in our cohort was invasive ductal carcinoma (IDC; 80.7%). Invasive lobular carcinoma (ILC) occurred in 13.1% of HER2-low and 11.4% of HER2-0 MBC (P = 0.625; Table 1). There was no significant difference in clinical or pathologic characteristics between HER2-low and HER2-0 MBC (Table 1).
. | HER2-0 N 224 (%) . | HER2-low N 353 (%) . | HER2-positive N 172 (%) . | P value (Overall) . | P value (HER2-0 vs. low) . |
---|---|---|---|---|---|
HER2 status | P < 0.001 | ||||
0 | 224 (100%) | 0 | 0 | ||
1+ | 0 | 194 (54.96%) | 0 | ||
2+ | 0 | 159 (45.04%) | 15 (8.72%) | ||
3+ | 0 | 0 | 46 (26.74%) | ||
Unknown | 0 | 0 | 111 (64.53%) | ||
Hormone receptor status | P < 0.0001 | P = 0.203 | |||
Negative | 42 (18.75%) | 52 (14.73%) | 68 (39.53%) | ||
Positive | 182 (81.25%) | 301 (85.27%) | 104 (60.47%) | ||
Histology | P = 0.001 | P = 0.625 | |||
IDC | 155 (77.11%) | 255 (77.74%) | 120 (93.75%) | ||
ILC | 23 (11.44%) | 43 (13.11%) | 4 (3.12%) | ||
Mixed histology | 23 (11.44%) | 30 (9.15%) | 4 (3.12%) | ||
Bone metastasis | P < 0.0001 | P = 0.319 | |||
Yes | 163 (73.09%) | 271 (76.77%) | 79 (45.95%) | ||
No | 60 (26.91%) | 82 (23.23%) | |||
Liver metastasis | P = 0.032 | P = 0.157 | |||
Yes | 84 (37.67%) | 154 (43.63%) | 55 (31.98%) | ||
No | 139 (62.33%) | 199 (56.37%) | 117 (68.02%) | ||
Nodal metastasis | P = 0.801 | P = 0.586 | |||
Yes | 84 (37.67%) | 141 (39.94%) | 70 (40.70%) | ||
No | 139 (62.33%) | 212 (60.06%) | 102 (59.30%) | ||
Lung metastasis | P = 0.749 | P = 0.563 | |||
Yes | 70 (31.39%) | 119 (33.71%) | 53 (30.81%) | ||
No | 153 (68.61%) | 234 (66.29%) | 119 (69.19%) | ||
Soft tissue metastasis | P = 0.154 | P = 0.554 | |||
Yes | 41 (18.39%) | 72 (20.40%) | 45 (26.16%) | ||
No | 182 (81.61%) | 281 (79.60%) | 127 (73.84%) | ||
CNS metastasis | P = 0.001 | P = 0.466 | |||
Yes | 16 (7.17%) | 20 (5.67%) | 26 (15.12%) | ||
No | 207 (92.83%) | 333 (94.33%) | 146 (84.88%) | ||
Line of treatment | P = 0.05 | ||||
1 | 73 (33.80%) | 104 (29.97%) | 30 (25.64%) | ||
2 | 41 (18.98%) | 88 (25.36%) | 16 (13.68%) | ||
≥3 | 102 (45.53%) | 155 (43.91%) | 71 (60.68%) | ||
Prior ET | P = 0.047 | P = 0.307 | |||
Yes | 132 (58.93%) | 223 (63.17%) | 43 (48.86%) | ||
No | 92 (41.07%) | 130 (36.83%) | 45 (51.14) | ||
Prior CT | P < 0.0001 | P = 0.390 | |||
Yes | 95 (41.07%) | 137 (38.81%) | 60 (68.18%) | ||
No | 129 (58.93%) | 216 (61.19%) | 28 (31.82%) | ||
Prior IO | P = 0.857 | P = 0.645 | |||
Yes | 8 (3.57%) | 10 (2.83%) | 2 (2.27%) | ||
No | 216 (96.43%) | 342 (96.88%) | 86 (97.73%) | ||
Prior CDK4/6i | P < 0.0001 | P = 0.864 | |||
Yes | 98 (43.75%) | 157 (44.48%) | 11 (12.5%) | ||
No | 126 (56.25%) | 196 (55.52%) | 77 (87.5%) | ||
Prior PIK3i | P = 0.621 | P = 0.707 | |||
Yes | 10 (4.46%) | 21 (5.95%) | 2 (2.27%) | ||
No | 213 (95.09%) | 331 (93.77%) | 86 (97.73%) | ||
Prior mTORi | P = 0.108 | P = 0.379 | |||
Yes | 35 (15.62%) | 49 (13.88%) | 5 (5.68%) | ||
No | 188 (83.93%) | 304 (86.12%) | 83 (94.32%) |
. | HER2-0 N 224 (%) . | HER2-low N 353 (%) . | HER2-positive N 172 (%) . | P value (Overall) . | P value (HER2-0 vs. low) . |
---|---|---|---|---|---|
HER2 status | P < 0.001 | ||||
0 | 224 (100%) | 0 | 0 | ||
1+ | 0 | 194 (54.96%) | 0 | ||
2+ | 0 | 159 (45.04%) | 15 (8.72%) | ||
3+ | 0 | 0 | 46 (26.74%) | ||
Unknown | 0 | 0 | 111 (64.53%) | ||
Hormone receptor status | P < 0.0001 | P = 0.203 | |||
Negative | 42 (18.75%) | 52 (14.73%) | 68 (39.53%) | ||
Positive | 182 (81.25%) | 301 (85.27%) | 104 (60.47%) | ||
Histology | P = 0.001 | P = 0.625 | |||
IDC | 155 (77.11%) | 255 (77.74%) | 120 (93.75%) | ||
ILC | 23 (11.44%) | 43 (13.11%) | 4 (3.12%) | ||
Mixed histology | 23 (11.44%) | 30 (9.15%) | 4 (3.12%) | ||
Bone metastasis | P < 0.0001 | P = 0.319 | |||
Yes | 163 (73.09%) | 271 (76.77%) | 79 (45.95%) | ||
No | 60 (26.91%) | 82 (23.23%) | |||
Liver metastasis | P = 0.032 | P = 0.157 | |||
Yes | 84 (37.67%) | 154 (43.63%) | 55 (31.98%) | ||
No | 139 (62.33%) | 199 (56.37%) | 117 (68.02%) | ||
Nodal metastasis | P = 0.801 | P = 0.586 | |||
Yes | 84 (37.67%) | 141 (39.94%) | 70 (40.70%) | ||
No | 139 (62.33%) | 212 (60.06%) | 102 (59.30%) | ||
Lung metastasis | P = 0.749 | P = 0.563 | |||
Yes | 70 (31.39%) | 119 (33.71%) | 53 (30.81%) | ||
No | 153 (68.61%) | 234 (66.29%) | 119 (69.19%) | ||
Soft tissue metastasis | P = 0.154 | P = 0.554 | |||
Yes | 41 (18.39%) | 72 (20.40%) | 45 (26.16%) | ||
No | 182 (81.61%) | 281 (79.60%) | 127 (73.84%) | ||
CNS metastasis | P = 0.001 | P = 0.466 | |||
Yes | 16 (7.17%) | 20 (5.67%) | 26 (15.12%) | ||
No | 207 (92.83%) | 333 (94.33%) | 146 (84.88%) | ||
Line of treatment | P = 0.05 | ||||
1 | 73 (33.80%) | 104 (29.97%) | 30 (25.64%) | ||
2 | 41 (18.98%) | 88 (25.36%) | 16 (13.68%) | ||
≥3 | 102 (45.53%) | 155 (43.91%) | 71 (60.68%) | ||
Prior ET | P = 0.047 | P = 0.307 | |||
Yes | 132 (58.93%) | 223 (63.17%) | 43 (48.86%) | ||
No | 92 (41.07%) | 130 (36.83%) | 45 (51.14) | ||
Prior CT | P < 0.0001 | P = 0.390 | |||
Yes | 95 (41.07%) | 137 (38.81%) | 60 (68.18%) | ||
No | 129 (58.93%) | 216 (61.19%) | 28 (31.82%) | ||
Prior IO | P = 0.857 | P = 0.645 | |||
Yes | 8 (3.57%) | 10 (2.83%) | 2 (2.27%) | ||
No | 216 (96.43%) | 342 (96.88%) | 86 (97.73%) | ||
Prior CDK4/6i | P < 0.0001 | P = 0.864 | |||
Yes | 98 (43.75%) | 157 (44.48%) | 11 (12.5%) | ||
No | 126 (56.25%) | 196 (55.52%) | 77 (87.5%) | ||
Prior PIK3i | P = 0.621 | P = 0.707 | |||
Yes | 10 (4.46%) | 21 (5.95%) | 2 (2.27%) | ||
No | 213 (95.09%) | 331 (93.77%) | 86 (97.73%) | ||
Prior mTORi | P = 0.108 | P = 0.379 | |||
Yes | 35 (15.62%) | 49 (13.88%) | 5 (5.68%) | ||
No | 188 (83.93%) | 304 (86.12%) | 83 (94.32%) |
Abbreviations: Endocrine therapy (ET), chemotherapy (CT), immunotherapy (IO), CDK4/6 inhibitors (CDK4/6i), invasive ductal carcinoma (IDC), invasive lobular carcinoma (ILC), PIK3 inhibitors (PIK3i), mTOR inhibitors (mTORi). All variables do not add up to total N due to missing data.
Genetic alterations detected by ctDNA
Alterations in ctDNA were evaluated in patients with HER2-0, HER2-low, and HER2-positive breast cancer. The landscape of alterations across the spectrum of HER2 expression is shown in Fig. 1. When combining pathogenic SNV and CNV based on OncoKB, the most common alterations among HER2-0 and HER2-low tumors were PIK3CA, TP53, and ESR1 (Fig. 1A and B). The most common alterations among HER2-positive tumors were TP53, ERBB2, and PIK3CA (Fig. 1C). EGFR, FGFR1, NF1, and MYC alterations were also common amongst HER2-0 and HER2-low tumors with varied frequency (Fig. 1A and B).
The total number of ctDNA alterations was compared in patients with HER2-0, HER2-low, and HER2-positive MBC. No differences were observed in terms of MAF of the dominant clone or when combining pathogenic SNV and CNV (Fig. 2).
Genetic alterations were compared by HER2 status for all patients and among the cohort of patients with hormone receptor–positive MBC. In the overall population, PIK3CA SNVs were detected in 52 (23%), 115 (33%), and 43 (25%) patients in HER2-0, HER2-low, and HER2-positive subgroups, respectively; ERBB2 CNV were observed in three (1%), five (1%), and 46 (27%); ESR1 SNVs were detected in 39 (17%), 67 (19%), and 14 (8%); MYC CNV were detected in 16 (7%), 39 (11%) and 24 (14%); and PDGFRA SNVs were observed in one (0.5%), 12 (3%) and six (4%). In multivariable analysis, PIK3CA mutations were significantly more common for HER2-low compared with HER2-0 in both the overall population (RRR, 1.57; P = 0.024; Table 2) and in the cohort of patients with hormone receptor–positive MBC (RRR, 1.58; P = 0.03; Supplementary Table S2). The distribution of alterations in PIK3CA by HER2 status is shown in Supplementary Fig. S1 and Supplementary Table S3. PIK3CA mutations also occur in hormone receptor–negative MBC, and we detected 33 PIK3CA alterations in this cohort. There was no statistically significant difference based on HER2 subgroup (P = 0.178; Supplementary Tables S4 and S5). There was no difference in the frequency of ERBB2 alterations, including amplifications or mutations, between HER2-low and HER2-0 MBC. As expected, patients with HER2-positive MBC had higher rates of ERBB2 amplification, compared with HER2-0 MBC, in both the overall population (RRR, 12.43; P = 0.002; Table 2) and in patients with hormone receptor–positive MBC (RRR, 9.97; P = 0.014; Supplementary Table S2). ERBB2 mutations were also more common in HER2-positive, compared with HER2-0, in the overall population (RRR, 2.54; P = 0.098), although not statistically significant, and in the cohort of patients with hormone receptor–positive MBC (RRR, 3.22; P = 0.047; Supplementary Table S2). Patients with HER2-positive MBC were significantly less likely to have mutations in ESR1 (RRR, 0.458; P = 0.029), compared with HER2-0, in the overall population (Table 2). However, the frequency of ESR1 mutations did not differ per HER2 status in the cohort of patients with hormone receptor–positive MBC (Supplementary Table S2).
Total cohort (N = 749) . | ||||
---|---|---|---|---|
. | RRR . | 95% CI . | P value . | |
Gene alterations | ||||
HER2-0 | ||||
reference | 1.00 | |||
HER2-low | ||||
PIK3CA mutationsa | 1.571 | 1.062 | 2.325 | 0.024 |
ERBB2 amplification | 0.971 | 0.156 | 6.026 | 0.975 |
ERBB2 mutations | 0.698 | 0.233 | 2.090 | 0.520 |
ESR1 mutations | 1.012 | 0.647 | 1.582 | 0.958 |
MYC amplification | 1.287 | 0.677 | 2.448 | 0.442 |
PDGFRA mutations | 6.680 | 0.835 | 53.460 | 0.074 |
HER2-positive | ||||
PIK3CA mutations | 0.713 | 0.411 | 1.237 | 0.229 |
ERBB2 amplificationa | 12.429 | 2.453 | 62.963 | 0.002 |
ERBB2 mutations | 2.539 | 0.842 | 7.650 | 0.098 |
ESR1 mutationsa | 0.458 | 0.226 | 0.924 | 0.029 |
MYC amplification | 0.813 | 0.334 | 1.977 | 0.648 |
PDGFRA mutations | 2.709 | 0.252 | 29.165 | 0.411 |
Oncogenic pathways | ||||
HER2-0 | ||||
Reference | 1.00 | |||
HER2-low | ||||
PIK3CA SNV | 1.498 | 1.029 | 2.182 | 0.035 |
ER SNV | 0.885 | 0.588 | 1.333 | 0.560 |
MYC CNV | 1.496 | 0.768 | 2.914 | 0.237 |
RTK CNV | 0.705 | 0.245 | 2.030 | 0.517 |
RTK SNV | 1.506 | 0.568 | 3.995 | 0.411 |
HER2-positive | ||||
PIK3CA SNV | 0.725 | 0.448 | 1.174 | 0.191 |
ER SNVa | 0.321 | 0.177 | 0.582 | <0.001 |
MYC CNV | 1.596 | 0.745 | 3.419 | 0.229 |
RTK CNV | 1.231 | 0.379 | 4.004 | 0.729 |
RTK SNV | 2.338 | 0.777 | 7.036 | 0.131 |
Total cohort (N = 749) . | ||||
---|---|---|---|---|
. | RRR . | 95% CI . | P value . | |
Gene alterations | ||||
HER2-0 | ||||
reference | 1.00 | |||
HER2-low | ||||
PIK3CA mutationsa | 1.571 | 1.062 | 2.325 | 0.024 |
ERBB2 amplification | 0.971 | 0.156 | 6.026 | 0.975 |
ERBB2 mutations | 0.698 | 0.233 | 2.090 | 0.520 |
ESR1 mutations | 1.012 | 0.647 | 1.582 | 0.958 |
MYC amplification | 1.287 | 0.677 | 2.448 | 0.442 |
PDGFRA mutations | 6.680 | 0.835 | 53.460 | 0.074 |
HER2-positive | ||||
PIK3CA mutations | 0.713 | 0.411 | 1.237 | 0.229 |
ERBB2 amplificationa | 12.429 | 2.453 | 62.963 | 0.002 |
ERBB2 mutations | 2.539 | 0.842 | 7.650 | 0.098 |
ESR1 mutationsa | 0.458 | 0.226 | 0.924 | 0.029 |
MYC amplification | 0.813 | 0.334 | 1.977 | 0.648 |
PDGFRA mutations | 2.709 | 0.252 | 29.165 | 0.411 |
Oncogenic pathways | ||||
HER2-0 | ||||
Reference | 1.00 | |||
HER2-low | ||||
PIK3CA SNV | 1.498 | 1.029 | 2.182 | 0.035 |
ER SNV | 0.885 | 0.588 | 1.333 | 0.560 |
MYC CNV | 1.496 | 0.768 | 2.914 | 0.237 |
RTK CNV | 0.705 | 0.245 | 2.030 | 0.517 |
RTK SNV | 1.506 | 0.568 | 3.995 | 0.411 |
HER2-positive | ||||
PIK3CA SNV | 0.725 | 0.448 | 1.174 | 0.191 |
ER SNVa | 0.321 | 0.177 | 0.582 | <0.001 |
MYC CNV | 1.596 | 0.745 | 3.419 | 0.229 |
RTK CNV | 1.231 | 0.379 | 4.004 | 0.729 |
RTK SNV | 2.338 | 0.777 | 7.036 | 0.131 |
Multivariable analysis of genetic alterations and oncogenic pathways using HER2-0 as reference.
aStatistically significant (P < 0.05).
Oncogenic pathways
On the basis of prior work defining canonical oncogenic pathways, the following pathways were compared across the spectrum of HER2 expression: RTK, RAS, RAF, MEK, NRF2, ER, WNT, MYC, P53, cell cycle, notch, and PI3K (ref. 34; Supplementary Table S1). Patients with HER2-positive MBC were significantly less likely to have oncogenic alterations for SNV in the estrogen receptor (ER) pathway in both the overall population (RRR, 0.321; P < 0.001) and the hormone receptor–positive MBC cohort (RRR, 0.447; P = 0.013; Table 2; Supplementary Table S2). There were no significant differences in oncogenic pathway alterations between HER2-0 and HER2-low MBC (Table 2; Supplementary Table S2).
Survival analysis
OS was evaluated across the spectrum for HER2 expression for 716 patients (95.6% of the total cohort) with outcome data available (210 HER2-0, 340 HER2-low, 166 HER2-positive). In multivariate analysis, including hormone receptor status, sites of metastasis and prior lines of therapy, no significant difference in survival was observed for HER2-low MBC, compared with HER2-0 MBC (HR, 1.01; 95% CI, 0.79–1.29; Fig. 3; Supplementary Table S6). Further, we found no significant interactions between oncogenic pathways and survival for patients with HER2-low compared with HER2-0 MBC (Supplementary Fig. S2).
Patients with HER2-positive MBC had significantly improved survival compared with HER2-0 MBC (HR, 0.32; 95% CI, 0.21–0.49; P < 0.001; Fig. 3; Supplementary Table S6). When analyzing interaction between oncogenic pathways and HER2 status, we found significant interaction between PIK3CA and cell cycle pathway alterations for HER2-positive MBC compared with HER2-0 MBC (Supplementary Fig. S3).
Discussion
In this study, we evaluated genetic alterations and associated clinical data from a large multi-institutional cohort of patients with MBC who underwent ctDNA testing. To our knowledge, this is the first study to describe genetic alterations detected by ctDNA across the spectrum of HER2 expression in breast cancer, including HER2-low MBC. The goals of this analysis were to identify the landscape of genetic alterations in HER2-low breast cancers in plasma and to evaluate any significant differences between HER2-low and HER2-0 tumors.
The landscape of genetic alterations was similar for HER2-low and HER2-0 MBC in the overall population. The most common alterations in our study were TP53 (35%), PIK3CA (28%), and ESR1 (16%). These are the most common mutations seen in prior ctDNA studies (22, 24).
As expected, the majority of HER2-low MBC were also hormone receptor–positive. There was an increased incidence of PIK3CA mutations among hormone receptor–positive HER2-low MBC compared with hormone receptor–positive HER2-0 MBC. These mutations have been implicated as potential mechanisms of resistance to hormonal and HER2-targeted therapies. Prior studies have identified a higher incidence of PIK3CA mutations detected by tissue-based NGS in HER2-low breast cancers compared with HER2-0 (13, 14). PIK3CA mutations are potentially clinically actionable and were detected at a frequency of 35% among hormone receptor–positive HER2-low tumors, compared with 26% in hormone receptor–positive HER2-0. At present, there is an FDA-approved targeted agent for PIK3CA (alpelisib) and many ongoing clinical trials of PIK3CA inhibitors. Our findings could have important clinical implications when considering mechanisms of resistance or combinations with novel anti-HER2 therapies in hormone receptor–positive HER2-low MBC. Further work is needed to confirm this finding from our retrospective analysis.
ERBB2 alterations (mutation or amplification) were not common among patients with HER2-low MBC (4%), which could suggest that the HER2 pathway is not a primary driver of tumor growth for these cancers. However, there are limitations with respect to copy-number changes, such as ERBB2, in ctDNA compared with tissue NGS. In an analysis by Schettini and colleagues using gene expression profiling of tumor tissue, the researchers identified higher ERBB2 gene expression with HER2-low compared with HER2-0 breast cancers (16). The role of the HER2 pathway as a driver of tumorigenesis in HER2-low breast cancers is still under investigation.
HER2-low breast cancers represent a heterogeneous group of tumors, including both hormone receptor–positive and TNBC. Some retrospective analyses have indicated differences in clinical outcomes between HER2-low and HER2-0 MBC (12–14). However, it is possible that these differences are not due to unique biology but rather a differential distribution of hormone receptor–positive and TNBC between HER2-low and HER2-0 breast cancers in these analyses. In a recent study by Tarantino and colleagues, ERBB2-low (HER2-low) was positively associated with ER expression, and ER-low tumors were enriched among ERBB2–0 (HER2-0), suggesting that the worse prognosis of ER-low tumors was confounding the prognostic significance of ERBB2-low (HER2-low; ref. 15). In our retrospective cohort of patients with MBC, we found that low HER2 expression did not affect prognosis. Further, when we performed an analysis for interaction between oncogenic pathways and HER2 status on OS, we found no significant interactions between the HER2-0 and HER2-low groups. HER2-positive breast cancers had improved survival compared with HER2-0 and HER2-low in our cohort, and there were several oncogenic pathways that helped to explain the difference in survival between HER2-0 and HER2-positive groups. Therefore, our findings support the idea that, while HER2-positive breast cancers have unique cancer biology, HER2-0 and HER2-low breast cancers do not. HER2-0 and HER2-low breast cancers have similar genomic profiles in ctDNA and survival outcomes driven primarily by the level of hormone receptor expression.
There were several limitations to our study. First, this was a retrospective analysis and therefore prone to unknown bias. Second, tumor tissue for HER2 expression was obtained per standard of care. The majority of tissue biopsies were from a metastatic site and correlated closely with the time of blood sampling for ctDNA analysis. However, a small proportion of biopsies (7%) were from primary breast tissue. Third, there may have been variability across institutions and pathologists regarding the assessment of HER2 expression, and no central assessment of HER2 IHC was performed. While there is excellent agreement across pathologists for HER2 IHC 3+ versus HER2 IHC 0, there is limited concordance for HER2 IHC 1+ and HER2 IHC 0. In a recent series reported by Fernandez and colleagues, concordance between pathologists for HER2 IHC 1+ versus HER2 IHC 0 was only 26% (35). It is uncertain how central testing may or may not have overcome this. Further, institutions included in this cohort study are all large academic centers with dedicated breast pathologists, which increases the generalizability of our findings.
Our study evaluating molecular alterations from ctDNA support the idea that HER2-low tumors are not reliant on HER2 pathway signaling and, instead, response to novel HER2-directed ADC is likely related to delivery of the cytotoxic payload. Following FDA approval of trastuzumab deruxtecan for HER2-low MBC, there is a need to better define HER2 status. In the phase II DAISY trial, trastuzumab deruxtecan was associated with a 30% response rate among heavily pretreated patients with HER2-0 MBC (36), and in the randomized phase III trial, DESTINY-Breast06, trastuzumab deruxtecan will be investigated in a cohort of patients with HER2 ultra-low breast cancer (defined as IHC >0 but <1+). Novel quantitative methods to determine HER2 expression, such as RT-PCR and mass spectrometry, are in development, and these methods will potentially help to expand the population of patients eligible for HER2-directed ADC (37, 38). More work is needed to understand the mechanism of action for novel HER2-directed ADC in tumors with low or ultra-low HER2 expression, and most importantly, the mechanisms of drug resistance.
Conclusion
Overall, our results suggest HER2-low tumors based on assessment of ctDNA are more likely to have PIK3CA mutations compared with HER2-0 MBC. However, we found no difference in ERBB2 alterations, oncogenic pathways, or prognosis between tumors without HER2 expression (HER2 0) and those with low HER2 expression. Our findings support the idea that HER2-low does not represent a unique biologic subtype but rather a clinical subtype defined by drug delivery and response to HER2-directed ADC.
Authors' Disclosures
L. Gerratana reports personal fees from Eli Lilly, Novartis, AstraZeneca, GSK, and Incyte outside the submitted work. K. Clifton reports grant funding from Cancer and Aging Research Group and consulting fees from Pfizer, Biotheranostics, and Guidepoint Consulting. A.J. Medford reports personal fees from Illumina and Natera outside the submitted work. A.N. Shah reports personal fees from AstraZeneca, Gilead, and Pfizer outside the submitted work. P. D’Amico reports other support from the American Italian Cancer Foundation during the conduct of the study, grants from Roche outside the submitted work, and reports employment at Merck & Co. C. Reduzzi reports grants from Menarini Silicon Biosystems outside the submitted work. N.A. Bagegni reports grants and other support from Sermonix Pharmaceuticals, AstraZeneca, Xcovery Holdings LLC, Novartis, Seattle Genetics, Daiichi Sankyo, Ambrx, Sarah Cannon Research Institute, and Biovica outside the submitted work. M. Opyrchal reports grants from Eli Lilly, grants and personal fees from Pfizer, personal fees from Novartis, AZ, and grants from Alphageneron outside the submitted work. F.O. Ademuyiwa reports grants and personal fees from Pfizer, personal fees from Teladoc Health, AstraZeneca, Cardinal Health, Biotheranostics, and Gilead, and grants from Astellas and RNA Diagnostics outside the submitted work. R. Bose reports personal fees from Genentech and grants from Puma Biotechnology, Inc. outside the submitted work. A. Behdad reports personal fees from Lilly, Leica, Caris, and Foundation Medicine China outside the submitted work. C.X. Ma reports personal fees from Agendia, AstraZeneca, Athenex, Bayer Healthcare, Eli-Lilly & Co., Guardant Health, Novartis Pharma AG, Olaris Inc., Pfizer Inc., PlusOne Health GmbH, Sanofi-Genzyme, Seattle Genetics Inc, Tempus, UpToDate/Wolters Kluwer, Biovica Inc., Eisai, Gilead Sciences, Inivata, Jacobio, Natera, Inc., Philips Electronics North America, The Lancet Oncology, and SAGA Diagnostics AB outside the submitted work. A. Bardia reports grants and personal fees from Genentech, Novartis, Pfizer, Merck, Sanofi, Radius Health/Menarini, Immunomedics/Gilead, Daiichi Pharma/AstraZeneca, Eli Lilly, Mersana, Guardant, and Natera during the conduct of the study. M. Cristofanilli reports personal fees from Pfizer, Menarini, Sermonix, Datar Genomics, and Tempus, nonfinancial support from Syantra, and personal fees from AZ outside the submitted work. A.A. Davis reports personal fees from Pfizer and Biotheranostics outside the submitted work. No disclosures were reported by the other authors.
Authors' Contributions
W.L. Hensing: Conceptualization, data curation, investigation, writing–original draft. L. Gerratana: Data curation, formal analysis, investigation, writing–review and editing. K. Clifton: Data curation, writing–review and editing. A.J. Medford: Data curation, writing–review and editing. M. Velimirovic: Data curation, writing–review and editing. A.N. Shah: Writing–review and editing. P. D'Amico: Writing–review and editing. C. Reduzzi: Writing–review and editing. Q. Zhang: Writing–review and editing. C.S. Dai: Writing–review and editing. E.N. Denault: Writing–review and editing. N.A. Bagegni: Investigation, writing–review and editing. M. Opyrchal: Writing–review and editing. F.O. Ademuyiwa: Writing–review and editing. R. Bose: Writing–review and editing. A. Behdad: Writing–review and editing. C.X. Ma: Conceptualization, resources, writing–review and editing. A. Bardia: Conceptualization, resources, writing–review and editing. M. Cristofanilli: Conceptualization, resources, writing–review and editing. A.A. Davis: Conceptualization, resources, data curation, investigation, writing–review and editing.
The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).