Abstract
Background Activating mutations in the estrogen receptor 1 (ESR1) gene are recurrent mechanisms of acquired resistance to aromatase inhibitors (AI), and may be the target of other selective estrogen receptor down-regulators. To assess the clinical utility of monitoring ESR1 resistant mutations, a droplet digital PCR (ddPCR)-based assay compatible with body fluids is ideal due to its cost-effectiveness and quick turnaround.
Methods We designed a multiplex ddPCR, which combines a drop-off assay, targeting the clustered hotspot mutations found in exon 8, with another pair of probes interrogating the E380Q mutation in exon 5. We assessed its sensitivity in vitro using synthetic oligonucleotides, harboring E380Q, L536R, Y537C, Y537N, Y537S or D538G mutations. Validation of the assay was performed on plasma samples from a prospective study and compared to next generation sequencing (NGS) data.
Results The multiplex ESR1-ddPCR showed a high sensitivity with a limit of detection ranging from 0.07 to 0.19% in mutant allele frequency depending on the mutation tested. The screening of plasma samples from patients with AI-resistant metastatic breast cancer identified ESR1 mutations in 29% of them with perfect concordance (and higher sensitivity) to NGS data obtained in parallel. Additionally, this test identifies patients harboring polyclonal alterations. Furthermore, the monitoring of ctDNA using this technique during treatment follow-up predicts the radiological response to palbociclib-fulvestrant.
Conclusion The multiplex ESR1-ddPCR detects, in a single reaction, the most frequent ESR1 activating mutations and is compatible with plasma samples. This method is thus suitable for real-time ESR1 mutation monitoring in large cohorts of patients.
Statement of translational relevance Exons 5 and 8 mutations in ESR1 are recurrent mechanisms of resistance to aromatase inhibitors (AI) in estrogen receptor (ER)-positive metastatic breast cancer and may be targeted by selective ER down-regulators. We implemented a novel droplet digital PCR, which allows for the detection of the most frequent ESR1 mutations in circulating cell-free DNA. In prospectively collected plasma samples, ESR1 mutations were found in 29% of AI-resistant patients, with excellent concordance and higher sensitivity to next generation sequencing. Moreover, circulating ESR1 mutations appear to be reliable markers for ctDNA monitoring in order to predict treatment response. Ultimately, the short turnaround time, high sensitivity and limited cost of the ESR1-ddPCR are compatible with repeated samplings to detect the onset of resistance to AI before the radiological progression. This opens a window of opportunity to develop new clinical strategies for breast cancer hormone therapy, as tested in an ongoing phase 3 trial.
- AI
- Aromatase Inhibitor
- cfDNA
- Cell-free DNA
- ctDNA
- Circulating tumor DNA
- ddPCR
- Droplet digital PCR
- ER+ HER2-MBC
- ER+ HER2-negative Metastatic Breast Cancer
- ER
- Estrogen Receptor
- ER+
- Estrogen Receptor positive
- LOB
- Limit of blank
- LOD
- Limit of detection
- MAF
- Mutant Allele Frequency
- PBMC
- Peripheral blood mononuclear cells
- PD
- Progressive disease
- SD
- Standard deviation
- ToP
- Time of progression
- WT
- Wild type
ESR1: Estrogen Receptor 1
HER2: Human Epidermal Growth Factor Receptor 2
EGFR: Epithelial Growth Factor Receptor
KRAS: KRAS proto-oncogene, GTPase
BRAF: B-Raf Proto-Oncogene, Serine/Threonine kinase
INTRODUCTION
In most breast cancer cases, the estrogen receptor (ER) exerts a key survival signaling, and ER-positive (ER+) immunostaining is a well-characterized predictive biomarker of hormone therapy efficiency at both early and advanced stages. In 2013, several studies reported activating point mutations in the estrogen receptor gene, ESR1 (1–4). These oncogenic mutations activate the ER in a ligand-independent manner and were proposed to be a potential mechanism of resistance to hormone depletion, by aromatase inhibitors (AI) for example. In ER+ HER2-negative metastatic breast cancer (ER+ HER2-MBC) patients, ESR1 mutations are rarely detected in primary tumors or at first metastatic relapse (5,6). However, their overall incidence increases alongside the exposure to AI (3–5,7), suggesting that ESR1 mutations are a positively selected resistance mechanism to this treatment. It has been reported that 30 to 40% of ER+ HER2-MBC patients who are resistant to AI carry ESR1 mutations, some even harboring polyclonal ESR1 alterations (8,9). These mutations occur in the ligand-binding domain, typically in two hotspot locations: codon 380 (exon 5) and codons 536, 537 and 538 (exon 8) (10–12). Importantly, other hormone therapy agents, such as selective ER down-regulators (e.g., fulvestrant, elacestrant…), retain a significant activity on mutated ER (Supplementary Table 1) (12). This indicates that mutations in ESR1 could be a predictive biomarker in the management of ER+ MBC (11). Accurate and sensitive detection tools are therefore needed to initiate clinical trials that investigate the clinical utility of ESR1 mutation monitoring. In that context, a liquid biopsy approach based on circulating tumor DNA (ctDNA) detection is a most promising tool to detect the onset of resistance acquisition.
Our group, and others, have shown that drop-off droplet digital PCR (ddPCR) is an efficient tool to screen for clustered mutations that can be covered by a single TaqMan probe, unlike conventional ddPCR, which requires a specific probe for each targeted mutation. The drop-off assay relies on the concept that a single mismatch is sufficient to destabilize the complex ‘probe / target’ sequence. Hence, any mutation in the region covered by the ‘drop-off’ probe leads to a loss of signal when compared to wild-type (WT) samples. All previous studies reported high sensitivity and reproducibility for hotspot mutations in the EGFR, KRAS, and BRAF genes (13–15).
Here, we present the validation of a novel ddPCR assay, which combines a drop-off ddPCR, targeting the clustered hotspot mutations found in ESR1 exon 8, with another pair of probes interrogating specifically the E380Q mutation, located in exon 5. This multiplex test detects, in a single reaction, up to 95% of the described activating ESR1 mutations, based on previous studies (3,4,10,12,16), and is compatible with body fluid samples.
MATERIALS AND METHODS
Droplet digital PCR assay
TaqMan® probes and primer design
For the E380Q mutation in exon 5, the assay used a probe specifically targeting the E380Q mutation (E380Q probe, FAM-labeled) and a reference probe (REFex5 probe, VIC-labeled) annealing to an adjacent invariant region (Figure 1A). For the clustered hotspot mutations in exon 8, a drop-off assay was designed using a probe targeting the WT sequence of codons 536, 537 and 538 where mutations are likely to be found (Hotspot probe, VIC-labeled) and a reference probe (REFex8 probe, FAM-labeled) annealing to an adjacent invariant region (Figure 1B). For compatibility with ctDNA detection, primers were designed to generate amplicons under 150 bp (101 bp and 125 bp for exons 5 and 8, respectively). The following primers were used to amplify (a) exon 5: Fwd primer: 5’-TTGCTTGTTTTCAGGCTTTGTGGA-3’; Rev primer 5’-AGCGCCAGACGAGACCAATCAT-3’; and (b) exon 8: Fwd primer: 5’-ACAGCATGAAGTGCAAGAACGT-3’; Rev primer 5’-TGGCTTTGGTCCGTCTCCTC-3’. The following TaqMan probes (Life Technologies) with a 5’ fluorophore and a 3’ non-fluorescent quencher (NFQ) were designed as follows: REFex5 5’-(VIC)-TGACCCTCCATGATC-(MGB NFQ)-3’; E380Q 5’-(6-FAM)-ACCTTCTACAATGTGCCTG-(MGB NFQ)-3’; REFex8 5’-(6-FAM)-CTAGCCGTGGAGGGGC-(MGB NFQ)-3’; Hotspot 5’-(VIC)-CCTCTATGACCTGCTGC-(MGB NFQ)-3’.
ddPCRmix preparation
ddPCR reactions were performed in 20 μL according to the manufacturer’s protocol. Briefly, 20 μL mastermix solution containing ddPCR™ Supermix for probes without dUTP (Bio-Rad Laboratories) at a final concentration of 1x, relevant primers at 450 nM each and relevant TaqMan® probes (E380Q + REFex5 or Hotspot + REFex8 in simplex conditions or both assays combined in multiplex conditions) at 250 nM each (Applied Biosystems), DNA template (up to 8 μL) and nuclease-free water were loaded into a disposable droplet generator cartridge (Bio-Rad). After adding 70 μL of droplet generation oil (Bio-Rad), the cartridge was loaded into a QX100 Droplet Generator (Bio-Rad). Generated droplets were transferred to a 96-well PCR plate and PCR reactions were run on a C1000 thermal cycler (Bio-Rad) under the following program: 95°C 10 min, 40 cycles of (94°C 30 sec, 58°C 60 sec), 98°C 10 min. For optimization experiments, we used 20 ng (6,060 genome equivalent) of DNA per reaction. For the mutation screening in patient samples, we used 8 μL of cell-free DNA (cfDNA). Negative controls with no DNA and positive controls (E380Q and D538G mutations) with 6,060 WT genome-equivalent from peripheral blood mononucleated cells (PBMC) were included at each run. Reactions were analyzed on the Bio-Rad QX100 droplet reader.
ddPCR data analysis
The concentration of mutant DNA copies was estimated using the dedicated workflow available in the QuantaSoft v1.7.4 software. For the E380Q mutation, the number of mutant copies per droplet is equivalent to the number of VIC+/FAM+ droplets in simplex conditions and to clouds #4 and #6 in multiplex conditions (Figure 1). The mutant allele frequency (MAF) was calculated as follows: (VIC+/FAM+ droplets / (VIC+/FAM+ droplets + VIC+/FAM− droplets)) for simplex conditions and (clouds #4 + #6/(clouds #4 + #6 + #2 + #5)) for multiplex conditions. For mutations in exon 8, the number of mutant copies per droplet is equivalent to the number of VIClow/FAM+ droplets in simplex conditions and to cloud #4 in multiplex conditions. The MAF was calculated as follows: (VIClow/FAM+ droplets / (VIClow/FAM+ droplets + VIC+/FAM+ droplets)) for simplex conditions and (cloud #4 /(clouds #4 + #3 + #5)) for multiplex conditions. Samples were run in triplicates and were considered to be positive if the merged replicates presented a minimum of 3 mutant droplets for E380Q or 8 mutant droplets for exon 8 and if the average MAF was higher than the defined LOD of the multiplex assay.
In vitro performance
The limit of blank (LOB) was determined as previously reported (13,15,17,18). Briefly, we defined the false-positive mean and associated standard deviation (SD) of the E380Q and Drop-off Ex8 assays, in simplex or multiplex conditions by analyzing 48 replicates of WT genomic DNA extracted from PBMC. Then, the calculated 95% confidence interval was used to define the LOB. To assess the limit of detection (LOD) of the E380Q and Drop-off Ex8 assays, we used synthetic oligonucleotides harboring E380Q, or the most frequent mutations in exon 8 (L536R, Y537C, Y537N, Y537S or D538G). Serial dilutions in 10 ng (3,030 genome equivalent) of WT DNA from PBMC, reproducing MAFs from 0.8% to 0.04%, were analyzed in 8 replicates for each mutated oligonucleotide.
Validation in clinical samples
All plasma samples analyzed for ESR1 mutation status were collected from patients with AI-resistant ER+ HER2-MBC. Patients were prospectively enrolled at the Institut Curie (Paris, France) in the ethically-approved ALCINA study (NCT02866149, cohort 6) after signed informed consent and that were included prior to the initiation of a new line of therapy with palbociclib and fulvestrant. Progression-free survival (PFS), defined as the time from inclusion in the study to progression disease (PD) or death from any cause, was collected prospectively. Survival analysis was performed using Kaplan–Meier plots with significance tested using the log-rank test. Plasma was isolated from fresh blood collected in EDTA blood collection tubes (BD Vacutainer®) within 3 hours as previously performed in the laboratory (19,20) and stored at −80°C until needed. For ESR1 screening, 2 mL of plasma were thawed and cfDNA extracted using the QIAamp® Circulating Nucleic Acid Kit (Qiagen) according to the manufacturer’s protocol and stored at −20°C until use. In parallel to ddPCR experiments, targeted-NGS on a panel of 39 cancer-related genes was performed in a blind fashion to allow head-to-head comparison with the ESR1-ddPCR assay. Following library preparation, samples were subjected to ultra-deep sequencing on Illumina HiSeq2500 using a 2 × 100 bp paired-end configuration. Read depth obtained for the two ESR1 hotspots regions was higher than 5,800X. Paired-end read alignments were performed on GRCh37 (hg19) human reference with Bowtie2 (v2.1.0). Once aligned, paired-end reads that map to multiple locations or with poor mapping quality (score <6) were removed. Pileup files were generated using samtools (v.1.1) and variant calling was performed using Varscan2 (v2.3.6). A minimum base quality of 15 was required to count for a read at a position, and only variants supported by a minimum of 5 mutated reads at a position with a minimum read depth of 12 were selected. Additionally, the nucleotide composition for each position of the regions containing the two ESR1 hotspots was extracted from the BAM alignment files using GATK (v.3.5).
RESULTS
Screening of multiple ESR1 hotspot mutations by ddPCR
Based on the evidence that more than 80% of the activating ESR1 mutations found in ER+ HER2-MBC patients resistant to AI alters codon 380 in exon 5 and the codons 536, 537 and 538 in exon 8 (19% and 64% respectively, see Supplementary Table 1), we developed a ddPCR assay targeting these two regions. This assay is composed of only two pairs of TaqMan probes, which can be combined in a single ddPCR reaction. The first pair targets specifically the E380Q mutation in exon 5, using a probe complementary to the E380Q mutant allele (E380Q probe, FAM-labeled) and a reference probe spanning an adjacent invariable region (REFex5 probe, VIC-labeled) (Figure 1A, left panel). In this unconventional E380Q assay, droplets containing WT alleles are VIC-positive only, because the E380Q probe cannot hybridize onto WT sequences, whereas droplets containing E380Q mutant alleles are double positive (VIC+/FAM+) (Figure 1A, right panel). The second pair of probes, constituting the Drop-off Ex8 assay, targets the clustered mutations in exon 8 using a probe complementary to the WT sequence of the altered region (Hotspot probe, VIC-labeled), which detects all the mutations occurring at codons 536, 537 and 538, together with a reference probe spanning an invariable region within the same amplicon (REFex8 probe, FAM-labeled) (Figure 1B, left panel). Therefore, droplets containing WT alleles are double positive (FAM+/VIC+) whereas the mismatch induced by the mutations leads to a VIC signal decrease (FAM+/VIClow), resulting in a shift of the droplet cloud (MUT Ex8) toward a single FAM-positive population (Figure 1B, right panel).
The multiplexed assay, combining primers and probes from the E380Q and the Drop-off Ex8 assays, allows to screen for mutations in exons 5 and 8 in a single reaction. In a test using synthetic E380Q or D538G oligonucleotides, we could distinguish each WT and mutant clouds for exons 5 and 8 in multiplex conditions (Figure 1C). We observed an additional cloud of droplets containing both WT Ex5 and WT Ex8 amplicons (Figure 1C, Cloud #5), which we also find in WT samples (data not shown), as well as a cloud containing E380Q and WT copies when testing for the E380Q mutation (Figure 1C, left panel, Cloud #6), and not observed when testing for the D538G mutation.
Detection of polyclonal alterations
In addition to the D538G mutation, we tested four other synthetic oligonucleotides harboring the Y537C, Y537N, Y537C or L536R mutations found in exon 8. We observed that the position of the MUT Ex8 cloud was dependent on the mutation tested (Figure 2A). This suggests that each nucleotide change does not equally destabilize the complex ‘probe / target sequence’. The largest shifts were observed for the mutations D538G and Y537C. Both clouds of droplets containing these mutant alleles were localized in the same area and were therefore indistinguishable from each other; whereas the clouds for Y537N, Y537S and L536R displayed smaller shifts and were distinct from the D538G/Y537C cloud. We further tested combinations of two mutant oligonucleotides in multiplex conditions confirming that multiple mutations in exon 8 (e.g., D538G + Y537N, D538G + Y537S, D538G + L536R or Y357N + L536R) display a specific pattern (Figure 2B). We clearly detected the combination of one mutation in exon 8 (e.g., D538G) with E380Q (Figure 2B). The multiplex ESR1-ddPCR can thus identify patients carrying polyclonal alterations, as previously reported (8, 9).
In vitro performances
We further estimated the specificity and sensitivity of the ESR1-ddPCR by analyzing 48 replicates of pure WT DNA and serial dilutions of the mutant synthetic oligonucleotides recapitulating MAFs from 0.8% to 0.04% (see Methods section for more details). The ESR1-ddPCR assay showed high specificity with a maximum of one false-positive event observed per reaction (Figure 3A and Supplementary Table 2). The limit of blank (LOB) was estimated at 0.004% for exon 5 mutations and 0.008% for exon 8 mutations. The limit of detection (LOD), defined as the lowest MAF with all replicates presenting values above the LOB, was estimated at 0.19% in mutant allele frequency for E380Q (Figure 3B and Supplementary Table 1) and ranged from 0.07 to 0.13% depending on the exon 8 mutation tested (Figure 3C and Supplementary Table 1). LOB and LOD were also tested in simplex conditions and similar values were observed (LOB were 0.003% for the E380Q assay and 0.01% for the Drop-off Ex8 assay; LOD were 0.09% for mutation E380Q and ranging from 0.06 to 0.1% depending on the exon 8 mutation tested, see Supplementary Figure 1A).
Validation in clinical samples
To validate the performance of the ESR1 multiplex ddPCR assay, we tested a series of 43 plasma samples from a prospective cohort of patients with HR+ HER2-MBC progressing under hormone therapy. We successfully detected ESR1 mutations in 11 out of the 42 (26%) informative patient samples (Table 1). Four cases (P-05, P-17, P-37 and P-43) harbored an E380Q mutation (36% of the mutant cases, Figure 4A) and 8 cases (P-08, P-18, P-20, P-25, P-28, P35, P-39 and P-43) carried at least one mutation in exon 8 (73% of the mutant cases, Figure 4B). Interestingly, based on the shape and the location of the MUT Ex8 clouds, we detected that samples P-18 and P-28 harbored multiple mutations (Figure 4C). We also observed that P-43 harbored ESR1 mutations in both exons 5 and 8 (Figure 4C). In parallel, targeted-NGS, including the ESR1 gene, was performed on 32 plasma samples, for which we had sufficient cfDNA, to allow head-to-head comparison with the ESR1-ddPCR assay. Importantly, NGS analysis was performed in a blinded fashion, with no prior knowledge of ddPCR results. The samples sequenced included 10 mutant and 22 WT cases according to ddPCR analysis. All ESR1 mutations detected by ddPCR were confirmed by NGS except for 2 cases (P-17 and P-20), in which the ESR1 MAF was found to be <1% by ddPCR (Table 1, Supplementary Figure 2A). P-25 was not tested with NGS due to lack of cfDNA (Table 1, Supplementary Figure 2B). We also confirmed the WT status of the 22 remaining plasma samples. In case P-18, NGS results indicated that the mutations were D538G and L536H, a combination we did not test before (Table 1). In case P-28, we observed 3 mutations in exon 8: L536H, Y537N and Y537S. Interestingly, for case P-35, NGS showed that the mutation in exon 8 was Y537N, which did not result from the usual C>T substitution in position 1609, but from a small deletion/insertion at nucleotides 1608-1609 (c.1608_1609delinsTA). Overall, the ESR1 multiplex ddPCR performs better than standard targeted NGS in identifying ESR1 mutation carriers owing to its improved sensitivity.
Monitoring of circulating ESR1 mutant copies to predict response to palbociclib-fulvestrant therapy
We next analyzed the impact of the ESR1 mutant status, detected with the ESR1-ddPCR, in plasma samples collected at baseline and during treatment follow-up. To perform this analysis, we extended the cohort to 60 patients. For each patient, four blood samples were collected: before treatment (D0), after 15 days (D15) and 30 days (D30) of treatment and at the time of progression (ToP). Among the 59 patients screened with contributive results, ESR1 mutations were detected in 17 (28.8%), which is in line with proportions of patients progressing under AI treatment previously reported (8, 9, 11, 24). Out of the 17 patients carrying an ESR1 mutation, 15 had an evaluable PFS (2 patients were withdrawn from the study shortly after the treatment initiation, Figure 5A). First, we observed no impact of the ESR1 mutational status (mutant vs wild-type) detected at baseline on the PFS (Figure S3A). This is in agreement with previous data from the PALOMA-3 trial reporting no improved PFS for patients carrying ESR1 mutations at baseline and treated with fulvestrant (11, 24). After 3 months of treatment, 6 patients presented a disease progression according to RECIST criteria (Figure 5A). We further analyzed the dynamics of ESR1 mutant copy levels in regards to the 3-months disease progression status (PD versus non-PD). At baseline, we observed a median number of ctDNA copies of 54 per ml of plasma (mean = 188.5, range [3 - 1233]) and there was no apparent association with disease progression at 3 months (Figure 5B). After 15 days of treatment, we uncovered a dramatic decrease (median = 0, mean = 14.3, range [0 - 74]) with 58% (7/12) of patients reaching undetectable ESR1 mutant levels (Figure 5B). The evolution of the number of ctDNA copies between D0 and D15, for each patient, demonstrates a decrease for all (Figure 5C, Supplementary Table 3). To a minor extent, a significant decrease was also observed on total cfDNA at D15 (Figure S3B-C), which was previously described and related to the cytostatic growth arrest triggered by palbociclib (24). At D30, the majority of patients with early disease progression had increased ctDNA levels whereas the majority of patients with longer PFS displayed decreased or stable ctDNA levels (Figure 5C). Overall, a lower proportion of patients displayed undetectable ctDNA levels (53% = 8/15, Figure 5B), the majority of which (87.5% = 7/8) had no early disease progression. Patients with early disease progression had significantly more ctDNA detected at D30 than patients with longer PFS (p < 0.0001; Figure 5D). Only one out of 6 patients with early disease progression had no ctDNA detected whereas only 2 out of 9 patients without early disease progression showed detectable ctDNA (Figure 5B). This ctDNA positivity at D30 correlates with PFS (p = 0.0063; Figure 5E-F). Here, we show that the detection of ctDNA after 30 days of palbociclib-fulvestrant is a promising dynamic biomarker associated with PFS.
DISCUSSION
We successfully developed a ddPCR assay detecting the most frequent activating ESR1 mutations at once that is compatible with liquid biopsies. By using an unconventional design, which includes a drop-off assay, we targeted, in a single reaction, the E380Q mutation and all the mutations occurring at codons 536 to 540. The multiplex ESR1-ddPCR covers >80% of the currently described ESR1 mutations and >90% of functionally characterized activating ESR1 mutations. Several teams have previously developed ddPCR assays which target only the most frequent ESR1 mutations found in exon 8: D538G, Y537S, Y537N and Y537C (6,21–23). These assays were designed following the conventional ddPCR method containing specific TaqMan probes complementary to each mutant or WT allele. This implies that each mutation is tested in a separate reaction. Thus, multiplex assays were developed to reduce the number of reactions (9, 11). However, these assays cannot identify more than 4 ESR1 mutations in a single reaction and the mutant samples were usually confirmed by singleplex tests (24). To our knowledge, we are the first to have developed a ddPCR assay which can detect, in a single reaction, at least eight different mutations in ESR1, namely: E380Q, L536H, L536R, Y537C, Y537N (T>A), Y537N (delinsTA), Y537S and D538G. In addition, our system can identify samples harboring multiple ESR1 mutations (e.g., E380Q combined with one or more mutations in exon 8). Polyclonal ESR1 mutations are well-described events (9,10) and the ESR1-ddPCR assay would be useful in monitoring the dynamics of each mutation during treatment follow up as seen for P-43. The multiplex ESR1-ddPCR assay is highly sensitive, detecting all tested mutations at frequencies lower than 0.19%, an improvement as compared with NGS. We also demonstrated that the ESR1-ddPCR is highly specific by cross-validation with NGS experiments. Lupini et al. recently developed an assay based on an ‘enhanced-ice-COLD-PCR followed by NGS’ with a sensitivity reaching 0.01% (25). However, this ddPCR assay targets specifically the Y537S mutation and involves an enrichment step of the mutant copies preceding the ddPCR assay. Yet, in a context of patient monitoring by liquid biopsy, biological samples are of limited quantity and must be tested rapidly at a low cost. The multiplex ESR1-ddPCR can detect most ESR1 mutations in a single reaction faster and at a lower cost than NGS or any other currently available technology.
Interestingly, we observed that exons 5 and 8 mutations can be easily distinguished. Moreover, any nucleotide change covered by the Drop-off Ex8 assay can be identified, as confirmed by the detection of the previously unreported mutation Y537N (delinsTA). We observed that, among exon 8 mutations, the shift in clouds is unique depending on the mutation, indicating if the mutation is more likely to be a D538G or Y537C allele versus mutations in codon 536 or other changes in codon 537. However, the fact that this assay does not recover the exact nucleotide change in exon 8 constitutes its most prominent limitation. From a clinical perspective, all mutations in codons 536-538 have been equally associated with resistance to AI therapy. Nonetheless, preclinical data suggest that the Y537S mutation, which accounts for about 10% of all ESR1 mutations, may be less sensitive to fulvestrant than other mutations (12). If this observation is confirmed to be clinically relevant, the ESR1-ddPCR could be used as a first screening tool, since the shift associated to Y537S is distinguishable from the most frequent mutation: D538G, followed by subsequent sequencing of exon 8, to distinguish Y537S from other 536/537 mutations. A second theoretical limitation is that synonymous mutations may also be detected by this assay. However, ESR1 synonymous mutations, as well as polymorphisms in codons 536-540, are extremely rare. According to the Exome Aggregation Consortium, only two synonymous mutations, L536L and D538D were described at very low frequencies, 6.303e−05 and 8.251e−06, respectively.
The improved analytical sensitivity of the ESR1-ddPCR is particularly useful to monitor ctDNA during treatment follow-up. We demonstrated that ESR1 mutations are good markers for ctDNA dynamics exploration and prediction of treatment response. Indeed, we observed that detection of ctDNA after 30 days of palbociclib-fulvestrant, using the ESR1-ddPCR, correlates with the treatment response and has an impact on PFS.
In conclusion, this method presents the advantage to screen for at least 80% of the ESR1 mutations in a single reaction, as required by large screening studies involving plasma samples. This assay is currently implemented in the randomized phase 3 PADA-1 trial (NCT03079011) in which it allows for real-time monitoring of ESR1 mutation rising in 1000 ER+ HER2-MBC patients treated with palbociclib and AI.
ACKNOWLEDGEMENTS
We would like to thank members of the Circulating Tumor Biomarkers laboratory for helpful discussions and especially Amanda Silveira for critical reading of the manuscript. High-throughput sequencing has been performed by the ICGex NGS platform of the Institut Curie supported by the grants ANR-10-EQPX-03 (Equipex) and ANR-10-INBS-09-08 (France Génomique Consortium) from the Agence Nationale de la Recherche (“Investissements d’Avenir” program), by the Canceropole Ile-de-France and by the SiRIC-Curie program - SiRIC Grant « INCa-DGOS-Inserm_12554 ».
Footnotes
↵* These authors share senior co-authorship
Financial support: SIRIC 2 Curie (grant INCa-DGOS-Inserm_12554) and the Innovative Medicines Initiative Joint Undertaking under grant agreement no. 115749 (project Cancer ID). The funding sources of the study had no role in the design of the study, collection, analysis or interpretation of the data or in the writing of this report.
Disclosure: C.P., E.J., M.H.S., and F.C.B. have ongoing patent applications relating to circulating tumor DNA analysis.