PIK3CA copy-number gain and inhibitors of the PI3K/AKT/mTOR pathway in triple-negative breast cancer

Abstract

As wider insights are gained on the molecular landscape of triple-negative breast cancer (TNBC), novel targeted therapeutic strategies might become an option in this setting as well. Activating mutations of PIK3CA represent the second most common alteration in TNBC after the TP53 mutation, with a prevalence of ∼10%–15%. Considering the well-established predictive role of PIK3CA mutations for response to agents targeting the PI3K/AKT/mTOR pathway, several clinical trials are currently evaluating these drugs in patients with advanced TNBC. However, much less is known regarding the actionability of PIK3CA copy-number gains, which represent a thoroughly common molecular alteration in TNBC, with a prevalence estimated at 6%–20%, and are listed as “likely gain-of-function” alterations in the OncoKB database. In the present paper, we describe two clinical cases in which patients harboring PIK3CA-amplified TNBC received a targeted treatment with the mTOR-inhibitor everolimus and the PI3K-inhibitor alpelisib, respectively, with evidence of disease response on 18F-FDG positron-emission tomography (PET) imaging. Hence, we discuss the evidence presently available regarding a possible predictive value of PIK3CA amplification for response to targeted treatment strategies, suggesting that this molecular alteration might represent an intriguing biomarker in this sense. Considering that few of the currently active clinical trials assessing agents targeting the PI3K/AKT/mTOR pathway in TNBC select patients based on tumor molecular characterization, and none of these based on PIK3CA copy-number status, we urge for the introduction of PIK3CA amplification as a criterion for patient selection in future clinical trials in this setting.

INTRODUCTION

Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) mutations occur in ∼40% of all breast cancers (BCs) and are known to predict the response to phosphatidylinositol 3-kinase (PI3K)-targeted agents in the hormone receptor-positive (HR+)/HER2-negative subset (Lefebvre et al. 2016; André et al. 2019; Martínez-Sáez et al. 2020). Although gene mutations are an established biomarker for treatment efficacy, the actionability of PIK3CA copy-number alterations still needs to be demonstrated, and no compounds are currently approved for patients with PIK3CA-amplified BC (Chakravarty et al. 2017; Condorelli et al. 2019). Recent insights into the molecular landscape of triple-negative breast cancer (TNBC) have demonstrated that oncogenic activation of the PI3K/AKT/mTOR pathway is enriched in this subtype as well, resulting from multiple molecular mechanisms, though without any impact on clinical practice, at present (Fuso et al. 2022). Here we describe two clinical cases in which TNBC harboring PIK3CA copy-number gains responded to PI3K/AKT/mTOR-targeted treatments.

RESULTS

Case Presentation 1

A lady aged 31 underwent sequential anthracycline-taxane-based neoadjuvant chemotherapy (NACT), left mastectomy with axillary lymph node dissection, and postoperative regional radiation therapy for a 63-mm multinodular TNBC, showing poor response to NACT, with ypT2N2MX pathological staging. Testing for germline BRCA1/2 mutations was negative. No additional genes were tested for germline mutations. Nine months later, as four liver metastases appeared, she started the first of several lines of systemic treatment, as diagrammed in Figure 1A. A liver biopsy confirmed the TNBC phenotype. Nearly 20 mo after the distant relapse, we observed a complete metabolic response of hepatic and nodal lesions following six cycles of fourth-line cisplatin + gemcitabine. After 3 mo, the patient underwent surgical resection of the residual hepatic lesions. One month later, pathologic right axillary and retroperitoneal lymph nodes appeared on a positron emission tomography (PET)/computed tomography (CT) scan. Following a multidisciplinary discussion, DNA extracted from the resected liver metastases was tested using the OncoDEEP clinical cancer panel (Ion Torrent with AmpliSeq library kit 2.0 for 50 genes, immunohistochemistry and fluorescent in situ hybridization (FISH) on formalin-fixed paraffin-embedded (FFPE) tissue; cfr Methods), which revealed a copy-number amplification of PIK3CA and a high expression of phosphorylated AKT and mTOR proteins—full results are listed in Table 1. Given the unavailability neither of clinical trials to enroll the patient into, nor of access to PI3K inhibitors, the patient was a candidate for off-label treatment with the mTOR-inhibitor everolimus (Afinitor), 10 mg daily, which was complicated by stomatitis (grade 1) and severe bilateral dependent lower limb edema. After 1 mo, a new PET/CT scan showed a complete metabolic response of retroperitoneal lymphadenopathies and stable metabolic disease for the right axillary lymph nodes (Fig. 2). After 5 mo on everolimus, the disease progressed at the right axillary level only, with complete response maintained for hepatic and retroperitoneal nodal lesions. The patient received epirubicin + cyclophosphamide, but developed peritoneal carcinomatosis and biliary tract compression, leading to her demise.

Figure 1.

Timeline of treatments and patient's response. (A) Timeline of treatment and patient response relating to case 1, in which PIK3CA-amplified TNBC was treated with everolimus. (B) Timeline of treatments and patient response relating to case 2, in which metaplastic PIK3CA-mutated and -amplified TNBC was treated with alpelisib. (TNBC) Triple negative breast cancer, (Adj RT) adjuvant radiotherapy, (5-FU) 5-fluorouracil, (BR) best response, (PR) partial response, (PD) progressive disease, (CR) complete response, (ctDNA) circulating tumoral DNA, (M) metastases, (CMF-VP) cyclophosphamide + methotrexate + 5-fluorouracile + vincristine + prednisone, (RT) radiotherapy, (LFTs) liver function tests.

Table 1.

Results on liver metastatic tissue of the OncoDEEP panel performed in case 1, comprehensive of targeted NGS, in situ hybridization, and immunohistochemistry analyses, commercially available from OncoDNA

Genomic analyses 1:
Gene alterations in liver metastatic tissue detected by the OncoDEEP targeted NGS panel (50 genes), liver metastasis FFPE histology slides, neoplastic cellularity = 50%
Single-nucleotide variants analysis (NGS)
Gene	Chr	Exon	HGVS DNA reference	HGVS protein reference	Variant type	VAF	Clinical classificationa	Clinical database information
TP53	17p13.1	4	c.398T > A	p.Met133Lys	SNV, missense	51%	Tier II	Likely oncogenicb
Copy-Number Variations Analysis (ISH)
PIK3CA amplification	Analysis on 50 nuclei: 225 PIK3CA copies, 90 centromeres, 4.5 PIK3CA copies per nucleus, 1.8 centromeres per nucleus, 2.5 PIK3CA copies per centromere; 3–6 copies in 49 nuclei + 9 copies in 1 nucleus
Immunohistochemistry for activated effector proteins
Phospho-AKT	3+, high expression
Phospho-mTOR	3+, high expression

See Methods for further details.

(NGS) Next-generation sequencing, (FFPE) formalin-fixed paraffin-embedded, (SNV) single-nucleotide variation, (VAF) variant allele frequency, (ISH) in situ hybridization.

^aClinical classification as per AMP criteria (Li et al. 2017).

^bAs per OncoKB (Chakravarty et al. 2017).

Figure 2.

Baseline and follow-up 18F-FDG positron emission tomography (PET)/computed tomography (CT) scans after the start of everolimus treatment in case 1, showing partial response to therapy. (A) Maximum intensity projection (MIP) PET image and two axial fused PET/CT slices of the abdomen just before starting everolimus, demonstrating 18F-FDG uptake in right axillary (see MIP) and retroperitoneal lymph nodes (see slices). (B) MIP PET image and two axial fused PET/CT slices of the abdomen after 4 wk on everolimus, showing complete metabolic response of the retroperitoneal lymph nodes (see slices), whereas the right axillary nodes were considered metabolically stable (see MIP).

Case Presentation 2

After discovering a 3-cm lump in her breast, a 30-yr-old woman was diagnosed with metaplastic TNBC, with Ki-67 90%, PDL1+ by SP142, androgen-receptor expression not determined. She tested negative for germline mutations across 25 genes associated with inherited cancer syndromes, including BRCA1/2, PALB2, TP53, and CHEK2. Whole-body imaging demonstrated axillary nodal and liver metastases, so she received systemic carboplatin, paclitaxel and atezolizumab followed by maintenance atezolizumab + bevacizumab, with partial response for all disease sites. In the meantime, DNA extracted from FFPE tissue from primary tumor and liver metastases, as well as peripheral blood (for germline alterations) and plasma, were tested using the OncoDEEP clinical cancer panel (Ion Torrent with AmpliSeq library kit 2.0 for 411 genes for FFPE tissue and for 40 genes for circulating tumor DNA [ctDNA]; cfr Methods). A mutation of the PIK3CA kinase domain (H1047R) was identified in all samples, and a PIK3CA copy-number gain in both FFPE samples (Table 2). Copy-number variations (CNVs) were not tested in ctDNA. After 2 mo on maintenance treatment, the disease progressed at the nodal, hepatic, bone, and pulmonary level, so the patient received two further chemotherapy lines (Fig. 1B), with relentless liver progression. Ten months after the initial diagnosis, based on molecular tests results, she started off-label the α-specific PI3K inhibitor alpelisib (Piqray) at 300 mg daily, reduced to 200 mg after one month for recurrent skin rashes (grade 3) despite antihistaminic treatment, with good tolerance thereafter. After 5 wk on alpelisib, a PET/CT scan demonstrated excellent partial metabolic and volumetric response of the hepatic lesions, but bone disease oligoprogression (Fig. 3). The patient received palliative radiation therapy on the spine, continuing her systemic treatment for 4 mo altogether. Eventually, further disease progression and clinical deterioration ensued, and she received no more active treatments before passing away, 1 mo later.

Table 2.

Results of the OncoDEEP targeted NGS panel on primary tumor tissue from core needle biopsy, liver biopsy tissue and ctDNA performed in case 2, commercially available from OncoDNA

Genomic analyses 2:
Gene alterations detected by the OncoDEEP targeted NGS panel (411 genes) on tissue from the primary breast tumor, from liver metastasis and on ctDNA (40 genes)
Single-nucleotide variants analysis
Gene	Chr	Exon	HGVS DNA reference	HGVS protein reference	Variant type	VAF	Clinical classificationa	Clinical database information
Primary breast tumor biopsy neoplastic cellularity = 20%
ADAMTS20	12q12	38	c.5574G > T	p.Met1858Ile	SNV, missense	16.3%	Tier III	NAb
TP53	17p13.1	6	c.578A > T	p.His193Leu	SNV, missense	18.3%	Tier III	Unknown oncogenic potential/effect on protein functionc
PIK3CA	3q26.32	21	c.3140A > G	p.His1047Arg	SNV, missense	61.8%	Tier I	Oncogenicc
Liver metastasis biopsy neoplastic cellularity = 30%
ADAMTS20	12q12	38	c.5574G > T	p.Met1858Ile	SNV, missense	71.4%	Tier III	NAb
TP53	17p13.1	6	c.578A > T	p.His193Leu	SNV, missense	66.5%	Tier III	Unknown oncogenic potential/effect on protein functionc
PIK3CA	3q26.32	21	c.3140A > G	p.His1047Arg	SNV, missense	88.1%	Tier I	Oncogenicc
ctDNA
TP53	17p13.1	6	c.578A > T	p.His193Leu	SNV, missense	10.3%	Tier III	Unknown oncogenic potential/effect on protein functionc
PIK3CA	3q26.32	21	c.3140A > G	p.His1047Arg	SNV, missense	56.8%	Tier I	Oncogenicc
Copy-number gains analysis (copy number ≥ 4)
Primary breast tumor biopsy	PIK3CA	4.0
	PDE4DIP	3.0
	SOX2	4.0
	BCL6	4.0
	LPP	4.0
	TNK2	4.0
Liver metastasis biopsy	PIK3CA	7.0
	PDE4DIP	4.0
	SOX2	7.0
	BCL6	7.0
	LPP	7.0
	TNK2	5.5

See Methods for further details.

(NGS) Next-generation sequencing, (ctDNA) circulating tumoral DNA, (Chr) chromosome, (VAF) variant allele frequency.

^aClinical classification as per AMP criteria (Li et al. 2017).

^bNot listed in any of OncoKB, ClinVar, or JAX-CKB public databases (Chakravarty et al. 2017; Landrum et al. 2018; Patterson et al. 2019).

^cAs per OncoKB (Chakravarty et al. 2017).

Figure 3.

Baseline and follow-up 18F-FDG positron emission tomography (PET)/computed tomography (CT) scans after the start of alpelisib treatment in case 2, showing transient partial response to therapy. (A) Maximum intensity projection (MIP) PET image just before starting the treatment with alpelisib, demonstrating 18F-FDG uptake in extensive bilobar hepatic metastases. (B) MIP PET image after 5 wk on alpelisib, showing excellent partial metabolic and volumetric response of hepatic metastases but bone oligoprogression. (C–E) Axial PET, CT, and fused slices showing the new lytic bone metastasis in the right iliac bone from the same PET/CT scan after 5 wk on alpelisib. (F) MIP PET image after 16 wk on alpelisib, demonstrating widespread liver and bone disease progression.

DISCUSSION

In the last decade, vast insights have been achieved regarding the molecular characterization of TNBC, demonstrating its wide molecular heterogeneity (Metzger-Filho et al. 2012; Bianchini et al. 2016). In 2011, Lehmann and colleagues suggested a molecular classification of TNBC into six entities—namely two basal-like-related subgroups, basal-like 1 (BL1) and 2 (BL2), two mesenchymal-related subgroups, mesenchymal (M) and mesenchymal stem–like (MSL), and the immune-modulatory subgroup (IM) and luminal androgen receptor group (LAR)—characterized by different gene-expression profiles and correlating with clinical outcomes (Lehmann et al. 2011; Masuda et al. 2013). Activating mutations of PIK3CA represent the second most common alteration in TNBC after the TP53 mutation, with a prevalence of ∼10%–15%, and they are especially found in the LAR and mesenchymal-related subtypes, the latter including metaplastic cancers as the one presented in case 2, characterized by aggressive behavior and chemo-refractoriness (Lehmann et al. 2011; Bianchini et al. 2016; Basho et al. 2017; Piscuoglio et al. 2017; Zhang et al. 2017; Bareche et al. 2018; Pascual and Turner 2019; Martínez-Sáez et al. 2020; Fuso et al. 2022). PIK3CA alterations are mainly represented by activating mutations affecting “hotspot” regions encoding the PI3K catalytic subunit α, but several other mechanisms of oncogenic activation of the PI3K/AKT/mTOR pathway have been described in TNBC, such as amplification of PIK3CA or other upstream regulators, loss of function of the pathway down-regulators phosphatase and tensin homolog (PTEN) or inositol polyphosphate-4-phosphatase (INPP4B), activating mutations of AKT1 or MTOR, or overexpression or phosphorylation of the mTOR protein, globally occurring in 25%–70% of TNBCs (López-Knowles et al. 2010; Koboldt et al. 2012; Costa et al. 2018; Goncalves et al. 2018; Fuso et al. 2022; Gupta et al. 2022; Zagami and Carey 2022). There are scant reports regarding the prevalence of PIK3CA copy-number alterations (CNAs), mainly referring to all BC subtypes as a whole and suggesting a prevalence of 6%–20%, whereas a search on the GENIE database (GENIE Cohort v11.0) yielded a prevalence of 14% across all BC subtypes (López-Knowles et al. 2010; Koboldt et al. 2012; Firoozinia et al. 2014; Thorpe et al. 2014; AACR Project GENIE Consortium 2017; Gerratana et al. 2022; Migliaccio et al. 2022). A study on ctDNA in TNBC reported PIK3CA amplification being the second most common CNA in this subset of patients, after MYC amplification (Davis et al. 2020).

The antitumor effect of inhibiting PI3K in TNBC has been demonstrated in both in vitro and in patient-derived xenograft models, and several clinical trials are currently ongoing in this setting (Table 3); nevertheless, none of these selects patients according to PIK3CA copy-number status, but rather based on PIK3CA single-nucleotide variations (SNVs) or loss of PTEN expression (Lehmann et al. 2011; Kim et al. 2017; Khan et al. 2019; Coussy et al. 2020; Schmid et al. 2020; Fuso et al. 2022). PIK3CA mutations are now listed as Tier I genomic alterations according to the ESMO Scale for Clinical Actionability of molecular Targets (ESCAT) in HR+/HER2− BC following the SOLAR-1 study, which led to the approval of alpelisib in that setting when a PIK3CA mutation is detected on therascreen, the FDA-approved companion diagnostic test (André et al. 2019; Condorelli et al. 2019). Conversely, PIK3CA CNAs are not considered targetable alterations even in luminal BC, and no compounds are specifically approved for PIK3CA-amplified BC at present (Chakravarty et al. 2017; Condorelli et al. 2019). Nevertheless, two pan-cancer genomic studies have suggested that a copy-number gain in PIK3CA impacts on PI3K protein expression and on phosphor-AKT levels, whereas the OncoKB database lists the amplification as likely gain-of-function, providing rationale for a predictive value of CNAs for response to PI3K inhibitors (Chakravarty et al. 2017; Zhang et al. 2017; Smith and Sheltzer 2018).

Table 3.

List of the main clinical trials currently active involving drugs targeting the PI3K/AKT/mTOR pathway in advanced triple-negative breast cancer

Trial identifier	Phase	Patient population	Biomarker for PI3K/AKT/mTOR pathway activation	Drug	Molecular target/Drug class
NCT04251533 (EPIK-B3)	Phase 3	Advanced TNBC	PIK3CA mutation or PTEN loss	Alpelisib + Nab-paclitaxel vs Placebo + Nab-paclitaxel	PI3K-α inhibitor + chemotherapy
NCT03961698 (MARIO-3)	Phase 2	Advanced TNBC	None	Eganelisib (IPI-549) + Atezolizumab + Nab-paclitaxel	PI3K-γ inhibitor + immune checkpoint inhibitor + chemotherapy
NCT03911973	Phase 2	Advanced TNBC or BRCA1/2-mutated HER2-negative BC	None	Gedatolisib + Talazoparib	PI3K and mTOR inhibitor + PARP inhibitor
NCT02531932	Phase 2	Advanced TNBC	None	Carboplatin + Everolimus vs Carboplatin + Placebo	mTOR inhibitor + chemotherapy
NCT03193853	Phase 2	Advanced TNBC	None	Serabelisib (TAK-117) + Sapanisertib (TAK-228) + Cisplatin + Nab-paclitaxel	PI3K-α inhibitor + mTORC1/2 inhibitor + chemotherapy
NCT03801369	Phase 2	Advanced TNBC	None	Olaparib + Capivasertib (AZD5363) or Olaparib + Durvalumab or Olaparib + Selumetinib (AZD6244) or Ceralasertib (AZD6738)	PARP inhibitor + AKT inhibitor or PARP inhibitor + immune checkpoint inhibitor or MEK inhibitor + PARP inhibitor or ATR inhibitor
NCT02208375	Phase 1b/2	Advanced TNBC and gynecological malignancies	None	Olaparib + Vistusertib (AZD2014) or Olaparib + Capivasertib (AZD5363)	mTORC1/2 inhibitor + PARP inhibitor or AKT inhibitor + PARP inhibitor
NCT02457910	Phase 1b/2	Advanced TNBC androgen receptor positive	None	Taselisib + Enzalutamide	PI3K inhibitor + Androgen receptor inhibitor
NCT03805399 (FUTURE)	Phase 1b/2	Advanced TNBC further classified in four subtypes (LAR, IM, BLIS, MES)	LAR subtype without ERBB2 mutation and with PIK3CA mutation, MES subtype with PIK3CA mutation	Everolimus + SHR3680 (Arm B1) or Everolimus + Nab-paclitaxel (Arm G) or other arms	mTOR inhibitor + Androgen receptor inhibitor or mTOR inhibitor + chemotherapy
NCT04395989 (FUTURE SUPER)	Phase 2, Umbrella trial	Advanced TNBC further classified in four subtypes (LAR, IM, BLIS, MES)	LAR subtype without ERBB2 mutation and with PI3K/AKT/mTOR pathway mutation, MES subtype with PI3K/AKT/mTOR pathway mutation	Everolimus + Nab-paclitaxel (Arms B1 and E1) or other arms	mTOR inhibitor + chemotherapy
NCT03742102 (BEGONIA)	Phase 1b/2	Advanced TNBC	None	Capivasertib (AZD5363) + Durvalumab + Paclitaxel or other arms	AKT inhibitor + immune checkpoint inhibitor + chemotherapy
NCT02583542	Phase 1b/2a	TNBC and lung cancer	None	AZD2014 + Selumetinib (AZD6244)	mTORC1/2 inhibitor + MEK inhibitor
NCT03207529	Phase 1b	Advanced HR + BC or TNBC androgen receptor–positive and PTEN-positive (on IHC)	PTEN nuclear staining >0% in IHC	Alpelisib + Enzalutamide	PI3K-α inhibitor + Androgen receptor inhibitor
NCT02637531	Phase 1/1b	Advanced solid tumors, among which TNBC	None	Eganelisib (IPI-549) + Nivolumab	PI3K-γ inhibitor + immune checkpoint inhibitor
NCT03218826	Phase 1	Advanced solid tumors, among which HER2-negative BC and TNBC	PTEN loss of function mutation or PIK3CB gain of function mutation	AZD8186 + Docetaxel	PI3K-β inhibitor + chemotherapy

(TNBC) Triple-negative breast cancer, (BC) breast cancer, (LAR) luminal androgen receptor, (IM) immunomodulatory, (BLIS) basal-like immune suppressed, (MES) mesenchymal, (HR) hormone receptor, (IHC) immunohistochemistry.

Migliaccio and colleagues recently published an analysis of publicly available data sets focusing on PIK3CA CNAs in HR+/HER2− BC, demonstrating that PIK3CA copy-number gains induce a significant increase in PIK3CA mRNA expression, and that cancer lines harboring both PIK3CA CNAs and SNVs have more aggressive clinicopathological characteristics and bring a shorter disease-free survival (Migliaccio et al. 2022). However, they could not demonstrate any difference in the response to alpelisib between clones with both alterations and those PIK3CA-mutated but not amplified (Migliaccio et al. 2022). Other studies suggest that cells harboring a double PIK3CA mutation may be more sensitive to specific inhibitors, even without any data on the contribution of gene amplification to drug sensitivity (Vasan et al. 2019; Saito et al. 2020).

The two cases we present suggest that PIK3CA CNAs may represent a biomarker predictive for response to agents targeting the PI3K/AKT/mTOR pathway in TNBC. These compounds provided a valuable objective response in both clinical cases, and lead to a progression-free survival of 5 and 4 mo, respectively, in patients with heavily pretreated and highly aggressive disease. In case 1, a high expression of phosphorylated AKT and mTOR was observed in immunohistochemistry, which is consistent with an increased phosphorylation activity by PI3K, supporting the hypothesis that PIK3CA amplifications indeed have an impact on activation of the PI3K/AKT/mTOR pathway. In this case, we treated the patient with everolimus, whose clinical efficacy is not influenced by PIK3CA mutations nor by PI3K pathway hyperactivation in HR+ BC, according to a post hoc analysis of the BOLERO-2 study (Hortobagyi et al. 2016). However, it should be noted that the authors did not include PIK3CA CNAs among the criteria used to define PI3K pathway hyperactivation, and they suggest that other genomic alterations could indeed impact everolimus activity (Hortobagyi et al. 2016). In case 2, we observed the coexistence of the H1047R mutation of PIK3CA, at very high VAF on all the evaluated samples, with a PIK3CA amplification (seven gene copies in liver metastatic tissue). The evidence available from literature is insufficient to drive conclusions on the relative contribution of the two molecular alterations to generate alpelisib sensitivity in the tumor. Anyway, in their study focusing on HR+ BC, Migliaccio et al. observed a greater benefit of alpelisib in pan-cancer xenograft models harboring both a PIK3CA mutation and copy-number gain, suggesting that efforts to further elucidate the role of the double hit in TNBC subtypes might be warranted, considering its biological difference from HR+ BC (Migliaccio et al. 2022). In both clinical cases the targeted treatment produced an uneven response among different disease sites. We hypothesize that this could reflect both a polyclonal metastatic seeding and a remodeling of clonal architecture induced by the treatment itself (Hu et al. 2020). Given the strong presence of diverse PI3K/AKT/mTOR alterations in TNBC, especially in the metaplastic subset, and taken into account the paucity of targeted treatments in this setting, PIK3CA CNAs appear as an intriguing therapeutic target, needing further validation in ad hoc clinical trials.

METHODS

Case 1

At the commercial lab (OncoDNA) samples underwent pathology review, reporting a moderate lymphocytic inflammatory infiltration and 10% necrosis, and standard IHC for estrogen receptor (ER), progesterone receptor (PR), and HER2. Targeted gene sequencing (TGS), CNV screening + ISH confirmation, and additional immunohistochemical (IHC) analyses were performed on FFPE histology slides of metastatic tumor tissue with a tumor cellularity of 50%. The test is certified by ISO15189 (Medical laboratories-Requirements for quality and competence), ISO17025 (Testing and calibration laboratories), CE-IVD (In vitro diagnostic devices complied to be sold in Europe), ISO 27001 (Information security management), and ISO 13485:2016 (Quality Management System).

For next-generation sequencing (NGS) testing, genomic DNA was extracted from the FFPE samples using the QIAamp DNA FFPE tumor tissue kit, according to the manufacturer's instructions (QIAGEN). The DNA quantity was measured using the Qubit 2.0 Fluorometer (Life Technologies). Targeted sequencing libraries were generated using the Ion AmpliSeq Library kit 2.0 according to the manufacturer's instructions (Life Technologies). The starting material consisted of 10 ng DNA. The primers used for amplification were partially digested by Pfu enzyme. The product of digestion was then ligated with corresponding barcoded adapters and purified using Ampure Beads. The product of purification has been amplified for five more cycles and purified using Ampure Beads. The quality of the libraries was assessed using the 2100 Bioanalyzer instrument (Agilent Technologies) and a high sensitivity chip. 10 pm of each library were put on the OneTouch 2 system for the emulsion polymerase chain reaction (PCR). The samples were sequenced via the PGM instrument.

Data were analyzed using the Variant Caller 4.0 software, using the somatic high stringency parameters and the targeted and hotspot pipelines. The bed files used were associated with the panels provided by Life Technologies. All the variants identified were confirmed by visualizing the data through IGV 2.3 (Broad Institute). CNV analyses were performed using the OncoCNV software (Boeva et al. 2014). Libraries were sequenced at a median coverage of 1000× to be able to detect variants at 5% frequency.

The NGS test targeted the 50 genes listed in Table 4, and was validated to detect mutations with a sensitivity of 5%. A variant was accepted when meeting the following criteria: coverage > 100, VAF > 5%, sequenced read in both senses at a minimum ratio of 25%/75%. Table 5 reports sequencing details.

Table 4.

List of the genes targeted by the OncoDEEP targeted next-generation sequencing panel utilized for genomic analyses in Case 1

ABL1	GNAQ	PTPN11	ERBB2	MET
ALK	HNF1A	RET	EZH2	MPL
ATM	IDH1	SMARCB1	FGFR1	NPM1
CDH1	JAK2	SRC	FGFR3	PDGFRA
CSF1R	KDR	VHL	GNA11	PTEN
EGFR	KRAS	AKT1	GNAS	RB1
ERBB4	MLH1	APC	HRAS	SMAD4
FBXW7	NOTCH1	BRAF	IDH2	SMO
FGFR2	NRAS	CDKN2A	JAK3	STK11
FLT3	PIK3CA	CTNNB1	KIT	TP53

Table 5.

Sequencing details for targeted genome sequencing tests performed in Case 1

Sequencing details for Case 1
TP53	Tumor protein p53
Genomic Primary	Chr 17:7578532 T/A VAF 51%	Read depth >100

(VAF) Variant allele frequency.

A screening for CNV was performed on deep sequencing data by means of the OncoCNV software, evaluating gene amplification and loss for 50 genes. A copy-number gain was identified for PIK3CA only, and was hence confirmed by means of in situ hybridization.

Additional IHC analyses were performed to evaluate phospho-AKT levels, by means of the p-AKT Rabbit monoclonal anti-phospho-Akt473 (Ser473) (clone D9E) mAb CST#4060, and to evaluate phospho-mTOR levels, by means of the Rabbit monoclonal anti-phospho-mTOR (Ser2448) (clone 49F9) mAb CST#5536.

Case 2

At the commercial lab (OncoDNA, Gosselies, Belgium) FFPE samples from primary tumor and metastases underwent pathology review and standard IHC for estrogen receptor (ER), progesterone receptor (PR), HER2, and Ki67, using the ER/PR pharmDx kit (Dako), the HercepTest kit (Dako), and the clone MIB-1 (Dako), respectively. ER/PR staining was interpreted according to ASCO/CAP 2010 guidelines, whereas HER2 staining was interpreted according to ASCO/CAP 2013 guidelines.

Regarding liquid biopsy, three samples of 9 mL blood in EDTA were withdrawn, one for germline DNA sequencing and two for plasma cell-free DNA (cfDNA) sequencing. EDTA tubes were centrifuged at 820g for 10 min at 4°C or within 30 min of blood withdrawal at room temperature, in order to separate plasma.

DNA was extracted from FFPE samples, with a tumor cellularity of 20% for primary tumor and 30% for metastatic tissue, using the QIAamp DNA FFPE tissue kit (QIAGEN). For blood samples, DNA was extracted by means of the QIAamp DNeasy Blood and Tissue kit (QIAGEN), following the manufacturer's instructions. DNA concentrations were measured using the Qubit fluorometer (Life Technologies). The cutoff values for tumor content and DNA quantity were 10% and 400 ng, respectively. cfDNA was extracted from plasma using the QIAsymphony DSP Circulating DNA Kit (QIAGEN) and quantified using the Thermo Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), with readings done on a Berthold TriStar fluorometer.

Somatic mutations were assessed using the OncoDEEP clinical cancer panel (OncoDNA), a validated AmpliSeq design panel targeting the exonic regions of 409 cancer related genes, to which probes for the BRCA1 and BRCA2 genes were added. The test is certified by ISO15189 (Medical laboratories—Requirements for quality and competence), CE-IVD (in vitro diagnostic devices complied to be sold in Europe), ISO 27001 (Information security management), and ISO 13485:2016 (Quality Management System). Targeted sequencing libraries were generated using the Ion AmpliSeq library kit 2.0 according to the manufacturer's instructions (Life Technologies) using 80 ng of genomic DNA. The primers used for amplification were partially digested by Pfu restriction enzyme and the digestion three products were ligated to barcoded adaptors and purified using Ampure Beads. The purified products were amplified for five more cycles and purified again using Ampure Beads. The quality of the libraries was assessed using a quantitative PCR (qPCR), following which 10 pM of each library underwent emulsion PCR using an IonChef system. The chips were loaded on an Ion PGM and were sequenced at a target coverage of 500×. The NGS panel used for tissue samples targeted the 411 genes reported in Table 6.

Table 6.

List of the genes targeted by the next-generation sequencing panel utilized for genomic analyses on tissue samples in Case 2

ABL1 ABL2 ACVR2A ADAMTS20 AFF1 AFF3 AKAP9 AKT1 AKT2 AKT3 ALK APC AR ARID1A ARID2 ARNT ASXL1 ATF1 ATM ATR ATRX AURKA AURKB AURKC AXL BAI3 BAP1 BCL10 BCL11A BCL11B BCL2 BCL2L1 BCL2L2 BCL3 BCL6 BCL9 BCR BIRC2 BIRC3 BIRC5 BLM BLNK BMPR1A BRAF BRCA1 BRCA2 BRD3 BRIP1 BTK BUB1B CARD11 CASC5

CBL CCND1 CCND2 CCNE1 CD79A CD79B CDC73 CDH1 CDH11 CDH2 CDH20 CDH5 CDK12 CDK4 CDK6 CDK8 CDKN2A CDKN2B CDKN2C CEBPA CHEK1 CHEK2 CIC CKS1B CMPK1 COL1A1 CRBN CREB1 CREBBP CRKL CRTC1 CSF1R CSMD3 CTNNA1 CTNNB1 CYLD CYP2C19 CYP2D6 DAXX DCC DDB2 DDIT3 DDR2 DEK DICER1 DNMT3A DPYD DST EGFR EML4 EP300 EP400

EPHA3 EPHA7 EPHB1 EPHB4 EPHB6 ERBB2v ERBB3 ERBB4 ERCC1 ERCC2 ERCC3 ERCC4 ERCC5 ERG ESR1 ETS1 ETV1 ETV4 EXT1 EXT2 EZH2 FAM123B FANCA FANCC FANCD2 FANCF FANCG FAS FBXW7 FGFR1 FGFR2 FGFR3 FGFR4 FH FLCN FLI1 FLT1 FLT3 FLT4 FN1 FOXL2 FOXO1 FOXO3 FOXP1 FOXP4 FZR1 G6PD GATA1 GATA2 GATA3 GDNF GNA11

GNAQ GNAS GPR124 GRM8 GUCY1A2 HCAR1 HIF1A HLF HNF1A HOOK3v HRAS HSP90AA1 HSP90AB1 ICK IDH1 IDH2 IGF1R IGF2 IGF2R IKBKB 19 IKBKE IKZF1 IL2 IL21R IL6ST IL7R ING4 IRF4 IRS2 ITGA10 ITGA9 ITGB2 ITGB3 JAK1 JAK2 JAK3 JUN KAT6A KAT6B KDM5C KDM6A KDR KEAP1 KIT KLF6 KRAS LAMP1 LCK LIFR LPHN3 LPP

LRP1B LTF LTK MAF MAFB MAGEA1 MAGI1 MALT1 MAML2 MAP2K1 MAP2K2 MAP2K4 MAP3K7 MAPK1 MAPK8 MARK1 MARK4 MDB1 MCL1 MDM2 MDM4 MEN1 MET MITF MLH1 MLL MLL2 MLL3 MLLT10 MMP2 MN1 MPL MRE11A MSH2 MSH6 MTOR MTR MTRR MUC1 MUTYH MYB MYC MYCL1 MYCN MYD88 MYH11 MYH9 NBN NCOA1 NCOA2 NCOA4 NF1

NF2 NFE2L2 NFKB1 NFKB2 NIN NKX2 NLRP1 NOTCH1 NOTCH2 NOTCH4 NPM1 NRAS NSD1 NTRK1 NTRK3 NUMA1 NUP214 NUP98 PAK3 PALB2 PARP1 PAX3 PAX5 PAX7 PAX8 PBRM1 PBX1 PDE4DIP PDGFB PDGFRA PDGFRB PER1 PGAP3 PHOX2B PIK3C2B PIK3CA PIK3CB PIK3CD PIK3CG PIK3R1 PIK3R2 PIM1 PKHD1 PLAG1 PLCG1 PLEKHG5 PML PMS1 PMS2 POT1 POU5F1 PPARG

PPP2R1A PRDM1 PRKAR1A PRKDC PSIP1 PTCH1 PTEN PTGS2 PTPN11 PTPRD PTPRT RAD50 RAF1 RALGDS RARA RB1 RECQL4 REL RET RHOH RNASEL RNF2 RNF213 ROS1 RPS6KA2 RRM1 RUNX1 RUNX1T1 SAMD9 SBDS SDHA SDHB SDHC SDHD SEPT9 SETD2 SF3B1 SGK1 SH2D1A SMAD2 SMAD4 SMARCA4 SMARCB1 SMO SMUG1 SOCS1 SOX11 SOX2 SRC SSX1 STK11 STK36

SUFU SYK SYNE1 TAF1 TAF1L TAL1 TBX22 TCF12 TCF3 TCF7L1 TCF7L2 TCL1A TET1 TET2 TFE3 TGFBR2 TGM7 THBS1 TIMP3 TLR4 TLX1 TNFAIP3 TNFRSF14 TNK2 TOP1 TP53 TPR TRIM24 TRIM33 TRIP11 TRRAP TSC1 TSC2 TSHR UBR5 UGT1A1 USP9X VHL WAS WHSC1 WRN WT1 XPA XPC XPO1 XRCC2 ZNF384 ZNF521

For cfDNA sequencing, libraries were prepared using the AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific) with a combination of an AmpliSeq custom panel and the OncoTrace core panel (OncoDNA). Libraries were quantified using the Thermo Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), with readings done on a Berthold TriStar flurorometer. Sequencing was done using the Ion 540 Kit—Chef in combination with the Ion 540 Chip Kit. The NGS panel used for liquid biopsies targeted the 40 genes reported in Table 7.

Table 7.

List of the genes targeted by the next-generation sequencing panel utilized for genomic analyses on cell-free DNA in Case 2

AKT1	ERBB2	GNA11	KIT	NRAS
ALK	ESR1	GNAQ	KRAS	PDGFRA
AR	EZH2	GNAS	MAP2K1	PIK3CA
BRAF	FBXW7	HRAS	MAP2K2	PTEN
BTK	FGFR1	IDH1	MET	RAF1
CTNNB1	FGFR2	IDH2	MPL	RET
DDR2	FGFR3	JAK2	MTOR	ROS1
EGFR	FOXL2	JAK3	NPM1	TP53

Sequence reads from the tumor and matched normal samples were aligned against the human genome reference version hg19/GRCh37, using the Ion Torrent TMAP aligner with default parameter settings. Mutations were called from the resulting BAM files using the Torrent Suite variant caller (Life Technologies) with the default settings of the “Somatic High Stringency” pipeline and cross-checked using the NextGENe software (Softgenetics) using the “Ion Torrent” predefined pipeline. Germline mutations were filtered by subtracting variants found in the matched blood sample (with variant allele fractions—VAF > 5%) from those called in the corresponding tumor sample. The resulting somatic mutation calls were further cleaned by excluding those that were not sequenced in both sense with a minimum ratio of 10/90%, those with a depth inferior to 100 reads depth, those with VAF < 10% in the tumor sample or present at 1% or more in ExAC. Several categories of variants were manually evaluated in IGV to rule out artefacts, including all BRCA1, BRCA2, and PALB2 variants, and all indels. Table 8 reports sequencing details.

Table 8.

Sequencing details for targeted genome sequencing tests performed in Case 2

Sequencing details for Case 2
ADAMTS20	ADAM metallopeptidase with thrombospondin type 1 motif 20
Genomic  Primary  Metastasis  ctDNA	Chr 12:43750356 C/A VAF 16.3% VAF 71.4% Not tested	Read depth 2195 Read depth 490
TP53	Tumor protein p53
Genomic  Primary  Metastasis  ctDNA	Chr 17:7578271 T/A VAF 18.3% VAF 66.5% VAF 10.3%	Read depth 1161 Read depth 602 Read depth 49146
PIK3CA	Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit α
Genomic  Primary  Metastasis  ctDNA	Chr 3:178952085 A/G VAF 61.8% VAF 88.1% VAF 56.8%	Read depth 16144 Read depth 19988 Read depth 41593

(ctDNA) Circulating tumor DNA, (VAF) variant allele frequency.

Copy numbers were extracted from TGS data of tissue samples using FACETS (Shen and Seshan 2016). Pileups were calculated at every single-nucleotide polymorphism (SNP) and every 50 bp, with no maximum depth, and minimum depth of 20 bp for the normal. For positions that were not SNPs, a rolling window smoothing of width 3 was used twice. FACETS preprocessing was done with a maximum depth of 105 and cval of 10, the processing with cval of 50 and min.nhet of 3. Fifty such fits were obtained with FACETS, and their qualities were recorded. The best fit among the primary and the metastatic sample was kept, and for the other sample the fit that matched best was kept. From the FACET fits, copy numbers were derived for each gene of the targeted panel.

To compare CNV between samples, we scaled those to get pseudo-diploid samples, hence we divided by half of their ploidy as estimated from the median CN across the genome. Three copy-number aberration categories were considered: deletions (CN < 1.5), gain (CN > 2.5), and amplifications (CN > 4).

ADDITIONAL INFORMATION

Data Deposition and Access

The data sets that support the findings of this report are not publicly available in order to protect patient privacy. The consent documentation signed by the patient does not expressly allow submission of full sequencing data (FASTQ, BAM/BAI, VCF) to external data repositories. The variants were submitted to ClinVar (https://www.ncbi.nlm.nih./gov/clinvar/) and can be found under accession numbers SCV003803731 and SCV003803732.

Ethics Statement

Written consent for the release of health information was obtained from the patients referenced in these reports. Written consent for the release of health information was obtained from the patients referenced in these reports. Patient 1 provided written consent for samples storage in the Institut Bordet Tumour Bank, whose activity is subject to strict legislation. Patient 2 provided written consent for participation in the AURORA study (NCT02102165), under the approval of Institut Bordet's IRB.

Acknowledgments

The authors acknowledge the American Association for Cancer Research and its financial and material support in the development of the AACR Project GENIE registry, as well as members of the consortium for their commitment to data sharing. Interpretations are the responsibility of study authors.

Authors Contributions

All authors made substantial contributions to this paper: (1) substantial contributions to the conception or design of the work or the acquisition, analysis, or interpretation of the data; (2) drafting the work or revising it critically for important intellectual content; (3) final approval of the completed version; and (4) accountability for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Each author approved the submitted version and agreed to be personally accountable for the author's own contributions.

Funding

No funding was received for the present work.

Competing Interest Statement

The authors have declared no competing interest.