Chromosome 12p Amplification in Triple-Negative/BRCA1-Mutated Breast Cancer Associates with Emergence of Docetaxel Resistance and Carboplatin Sensitivity

Triple-negative breast cancer (TNBC) is the most aggressive subtype of breast cancer and most tumors display basal-like features (1). As no targetable therapies are available, chemotherapy, mostly taxanes, either alone or in combination with other drugs, is the most common treatment for patients with TNBC (2). Tumors can accumulate mutations during cancer progression and treatment, leading to acquired resistance. In addition, de novo or intrinsic chemoresistance to taxanes consistently occurs in TNBCs, leading to resistant relapse or distant metastasis and mortality (2). A form of de novo resistance occurs when a preexisting resistant subpopulation already present in the tumor is selected as a consequence of treatment (3, 4). In such cases, resistance can be thought of as “emergent” in otherwise responsive tumors, leading to partial response. Some mechanisms of resistance to taxanes described as potentially targetable (5, 6) have shown little relevance in the clinic. Thus, a better understanding of these mechanisms is essential to improve patient outcome.

Many mutations and rearrangements involving different chromosomal locations have been identified in breast cancer (7, 8). Tumors from the basal-like and HER2⁺ subtype display more chromosomal aberrations and mutational rates per megabase than do luminal subtypes (7) and mutated cancer driver genes are relatively subtype-specific (7). Germline mutations in BRCA1 and BRCA2 genes have been associated with up to 15% of TNBC cases, and TNBC accounts for 70% of breast tumors arising in BRCA1 mutation carriers (9, 10).

Some genomic alterations identified in TNBC tumors are associated with various responses to chemotherapy and targeted therapies (11, 12). However, no mutations or chromosomal rearrangements in TNBC tumors have been found that confer resistance to specific chemotherapeutic agents.

Gene expression signatures extracted from sensitive and resistant breast cancer cell lines may be useful for elucidating mechanisms of resistance (13), but most of the cell line–derived signatures are unable to predict response in the clinical setting (14). Patient-derived xenografts (PDX) represent a useful model for the study of tumor progression, clonal evolution, and for assessing novel therapies, because they maintain the main features of originating patient's tumors: transcriptomic, genomic, and histologic patterns, and drug responses and provide unlimited tumor samples pre-/posttreatment that are difficult to obtain in the clinical setting (15, 16).

We have recently shown that PDX models are suitable tools for the study of chemoresistance. We found that unlike luminal tumors, basal-like PDX models were initially sensitive to docetaxel (IDB-01S and IDB-02S) but they progressively develop resistance after continuous in vivo docetaxel treatment (IDB-01R and IDB-02R), mimicking the clinical scenario (17). A gene expression signature predicting residual disease after anthracycline/taxane-based therapy in patients with basal-like disease was identified using our paired docetaxel sensitive and resistant basal-like PDX models, highlighting the clinical relevance of these models (17).

In this study, we hypothesized that acquisition, or selection of genomic changes could be associated with docetaxel resistance emergence in patients with TNBC. Using matched docetaxel-sensitive and docetaxel-resistant TNBC PDX models we investigated the association of point mutations, small insertions/deletions (INDEL), and copy number alterations (CNA) with resistance. We found that a subpopulation of tumor cells harboring amplification in chr12p is preferentially found in BRCA1-mutated tumors and became enriched upon docetaxel treatment and the emergence of chemoresistance. These results indicate that despite initial responses, the presence of chr12p amplification may indicate the rapid emergence of resistance to taxanes in TNBC/BRCA1-mutated disease. Importantly, detection of chr12p amplification in TNBC is associated with better responses to carboplatin, suggesting that sequential treatment with docetaxel and carboplatin would improve survival of these patients.

We assessed mutations and CNAs in paired docetaxel-sensitive and -resistant TNBC PDX models using exome-genome sequencing (WES). Once chr12p amplification was elucidated, additional TNBC PDX models, breast cancer cell lines, breast cancer patients who progressed to docetaxel and public databases with clinical breast cancer data were screened. All human samples were obtained following institutional guidelines. Written informed consent for PDX generation and WES was obtained from all subjects and the study received approval from the corresponding institutional Ethics Committee in accordance with the Declaration of Helsinki. All research involving animals was performed in compliance with protocols approved by the Institutional Committees on Animal Care and adhering to European Union and international regulations. PDX models were generated as described (17). Findings from the IDB models were evaluated in an interim analysis of 2 ongoing PDX-based preclinical chemotherapy trials being conducted at Baylor College of Medicine to be described in full elsewhere upon completion. For in vivo experiments, 3 to 12 mice randomized by tumor size were used per treatment group. Mice that died before the end of the experiment for reasons unrelated to treatment were excluded. WES, expression arrays, and chemotherapy treatment in the PDX models are described in Supplementary Data. All the TNBC PDX models and breast cancer cell lines used in this study are shown in Supplementary Table S1.1.

All cell lines were purchased from the ATCC, except for UACC3199, which was obtained from the Arizona Cancer Center (Tucson, AZ). ATCC provides molecular authentication in support of their collection through their genomics, immunology, and proteomic cores, as described by using DNA barcoding and species identification, quantitative gene expression, and transcriptomic analyses. UACC3199 was authenticated by its ability to reexpress BRCA1 after DNA demethylation treatment with 5-aza-2′-deoxycytidine. All lines were expanded and frozen within 2 weeks of purchase and used for a maximum of 4 months after resuscitation of frozen aliquots. All cell lines were routinely tested for mycoplasma (Biotools; B & M Labs; #4542) every month and before each experiment and were shown to be free of contamination.

The METABRIC dataset (n = 2,509; refs. 8, 18) was downloaded from cBioPortal (http://www.cbioportal.org/) and The Cancer Genome Atlas (TCGA) data (7) were downloaded from the Genomics Data Commons portal (https://portal.gdc.cancer.gov/).

The association between the copy number of chromosome 12p and the drug response was estimated by the Spearman rank correlation test. All data are expressed as mean ± SEM. Group differences were compared using Student unpaired samples t test and GraphPad Prism version 5.04. Values of P ≤ 0.05 were considered statistically significant. Levels of significance are expressed as: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.001 < P < 0.0001; ****, P < 0.0001.

WES was performed using DNA from matched sensitive (untreated) and resistant IDB-01 and IDB-02 basal-like PDX models (17) at different passages and “branches” (Supplementary Fig. S1A), the metastatic sample of origin in both models, peritumoral tissue and normal lymphocytic DNA from the patient from whom the IDB-02 model was derived. At the time, the metastatic samples were collected (pleural effusion enriched in tumor cells), both patients had been heavily treated with chemotherapy including taxanes (Supplementary Fig. S1B; ref. 17). Mean coverage for all samples was 95× and more than 84% of target bases had 15× coverage in all samples.

The metastatic sample of origin of the IDB-01 model had a mutation in the most significantly mutated gene in basal-like breast cancer (7), TP53 (p.N239Tfs), and a specific mutation in PIK3CA (p.E545K; Supplementary Table S1.2). These are recurrent mutations present in the COSMIC database, suggesting that they are bona fide somatic mutations.

The comparison of point mutations between the metastatic sample of origin of IDB-02 and normal lymphocytic DNA revealed that 91 somatic point mutations accumulated during tumor progression in the exome (Supplementary Table S1.3). Analysis of germline variants revealed the presence of a heterozygous frameshift mutation in BRCA1 (p.Q1756Pfs) in both normal lymphocytic DNA and in peritumoral tissue from the patient (Supplementary Table S1.3), which was validated by Sanger sequencing (Fig. 1A). The BRCA1 mutation was found to be homozygous in the human metastasis and in the IDB-02S PDX model (Fig. 1A). The mutation burden of the basal-ike metastatic sample from which IDB-02 was derived (1.82 muts/Mb) was similar to the mutational rate described for primary basal-like tumors in the TCGA data (1.68 muts/Mb; Supplementary Table S1.3; ref. 7).

A close similarity of point mutations and INDELs was observed between the metastatic human sample of origin (M) and the IDB-01 PDX model during tumor engraftment (96%–98%); whereas in the case of the IDB-02 PDX model, a first clonal selection step took place during xenografting (86% of similarity; Supplementary Fig. S1C). In both PDX models, the genome was stable during serial transplantation in mice (96%–99%; Supplementary Fig. S1C).

To identify point mutations or small INDELs that might act as drivers of docetaxel chemoresistance, WES data of paired sensitive and resistant samples from each TNBC PDX model were compared. More than 95% of the genotype was shared by sensitive and resistant tumors in both models (Supplementary Fig. S1C). Very few point mutations or small INDELs were shared by all the resistant tumors analyzed in each model while being absent in the corresponding sensitive tumors and even fewer potentially affecting protein function (nonsynonymous changes, frameshifts, or insertion/deletion of amino acids; Supplementary Fig. S1D and S1E). Only 1 heterozygous point mutation in KLHL42 was differentially detected between resistant and sensitive tumors but this was not confirmed by Sanger sequencing in additional tumor samples (Supplementary Fig. S1F).

These results reveal that, despite the high mutational load (and the presence of BRCA1 germline mutations), TNBC PDX models do not accumulate de novo mutations or small INDELs during serial passages in mice or drug exposure, and no recurrent mutations were identified that could account for the chemoresistance. Consistent with previous observations (19–21), both basal-like PDX models had a stable genome during xenotransplantation in mice and chemotherapy exposure, although some clonal selection was observed during engraftment.

Given previous evidences of the involvement of CNAs in the acquisition of drug resistance in breast cancer (6, 11), CNAs were inferred from WES data for IDB-01 and IDB-02 (22, 23). In the case of the IDB-01 model, for which patient-matched normal DNA was not available, comparison of the human metastatic sample with a normal human DNA reference sample showed an aberrant chromosomal landscape (Supplementary Fig. S2A; Supplementary Table S1.4). Analysis of CNAs in the BRCA1-mutated human metastasis revealed even more genomic alterations compared with those in the normal lymphocytic DNA, a known genotype in BRCA1-mutated tumors (Fig. 1B; Supplementary Table S1.5; refs. 24–26).

Most chromosomal alterations detected in the human metastatic samples of origin were maintained in the corresponding TNBC PDX models, consistent with previous data (Fig. 1B; Supplementary Fig. S2A and S2B; Supplementary Tables S1.4 and S1.5; ref. 19); the metastatic patient's samples contained more diploid stromal cells, but stroma is lost in the PDX. Thus, the CNAs also remains largely stable during PDX generation and passage in mice. All tumors from the IDB-01 PDX model showed a 35 Mb amplification in chromosome 5p15.2-p12 compared with the metastatic sample of origin and duplication in ERBB2, suggesting that these changes are selected during xenograft expansion (Supplementary Table S1.4). The IDB-02 PDX model showed, among other CNAs in common with the metastatic human sample of origin, amplifications in 1q, 8q, and 10p and losses of 1p, 4p, 5q, 10q, 15q, and Xp (Fig. 1B; Supplementary Table S1.5), a profile associated with TNBC/basal-like tumors (7, 27).

Comparison of IDB-01 sensitive and resistant tumors revealed very few CNA changes in IDB-01R tumors and none was shared between the 3 IDB-01R and absent in IDB-01S tumors (Supplementary Fig. S2A; Supplementary Table S1.4). However, in IDB-02, 2 CNAs were consistently shared between IDB-02R tumors but undetectable in IDB-02S tumors. These were amplifications of a region of chromosome 3p, including around 50 exons from 3 genes, and a region of chromosome 12p (chr12p), including 23 Mb and around 225 genes, from 12p13.31 to 12p11.21. Amplification of the same chr12p region was also detected in the human metastatic sample of origin (Fig. 1C; Supplementary Table S1.5), which may be explained by the fact that the patient progressed on docetaxel treatment (Supplementary Fig. S1B). Chr12p amplification includes 2 small regions that were initially amplified in IDB-02S tumors with 3 copies, but they were overamplified in IDB-02R tumors and in the human metastasis, including a central region and 2 telomeric and centromeric flanking regions that were amplified only in the IDB-02R tumors and the human metastasis (Fig. 1C and D; Supplementary Table S1.5).

DNA amplification was directly validated in additional IDB-02S and IDB-02R tumors using exonic TaqMan probes for genes located at different positions in the chr12p-amplified region: GABARAPL1, ETV6, and KLHL42 (Fig. 1D). ETV6 and KLHL42 were already amplified in IDB-02S tumors but one additional copy was found in the IDB-02R variants, whereas GABARAPL1 was amplified with an extra copy in IDB-02R tumors but not in IDB-02S (Fig. 1E). Thus, there was a significant increase in copy number of all 3 genes detected in IDB-02R compared with those in IDB-02S tumors, thereby validating the amplification of the chr12p region.

Chr12p amplification was indirectly validated using gene expression microarray data obtained from 3 IDB-02S and 3 IDB-02R independent tumors (Supplementary Tables S2.1 and S2.2). Of the 25 genes located on chromosome 12 and overexpressed in IDB-02R compared with IDB-02S tumors (FC > 1.5 and adjusted P-value < 0.05), 16 were located in the chr12p-amplified region (Supplementary Fig. S2C; Supplementary Table S2.1 and S2.2), indicating a strong correlation between amplification and overexpression (P < 0.0001). Gene expression analyses by qRT-PCR of 9 genes located in the chr12p amplified region confirmed higher mRNA expression levels in IDB-02R tumors compared with those of IDB-02S, which were statistically significant for GABARAPL1, ETV6, and MGST1 (Fig. 1F; Supplementary Fig. S2D). Given that amplification of chr12p in IDB-02R tumors implies a gain of only one copy, small changes in gene expression are expected. Overall, these results demonstrate that amplification of most of the chr12p arm occurs after emergence of resistance to docetaxel in the TNBC/BRCA1-mutated IDB-02 PDX.

The presence of the amplification in the docetaxel-treated human metastasis from which the IDB-02S model was derived prompted us to hypothesize that the population harboring chr12p amplification is diluted during engraftment and serial passages in the absence of docetaxel, so that it is undetectable in IDB-02S tumors but enriched after continuous exposure to docetaxel treatment. To test this hypothesis, DNA copy number and mRNA expression levels of GABARAPL1, ETV6, and KLHL42 were analyzed, in IDB-02S tumors after 4 doses of docetaxel treatment, when the tumors were shrinking (residual disease; IDB-02 RD; Fig. 2A). An increase in copy number of the 3 surrogate genes, together with increased gene expression of some chr12p-located genes was observed in IDB-02S RD (Fig. 2B; Supplementary Fig. S3A), which suggests that the population harboring chr12p amplification is initially present in IDB-02S tumors and selected after docetaxel treatment.

To establish whether the enrichment of chr12p after docetaxel treatment is an isolated event or could be a general effect in TNBC/BRCA1-mutated disease, we used 6 additional TNBC PDX models, 3 of them BRCA1 mutated, all sensitive to docetaxel (see Supplementary Table S1.1 for all models used; ref. 17). No changes in chr12p copy number were found in residual disease after short-term docetaxel treatment in the non-BRCA1–mutated PDX models: VHIO-98, HCI-001, IDB-01, and VHIO-270 (Supplementary Fig. S3B and S3C). A preexisting amplification of chr12p was detected in the BRCA1-deficient VHIO-127S, as well as in BCM-4664. Significant copy number gains for the surrogate genes were observed in residual disease after docetaxel treatment in these 2 models, as well as in the BRCA1-deficient BCM-9161 and AB559 (Fig. 2C–F). At the time that VHIO-127S (Supplementary Fig. S3D) and BCM-4664 PDX were established, the corresponding patients had progressed on docetaxel treatment. We also generated resistant variants of the initially sensitive VHIO-98 and VHIO-127 as described previously (17), in addition to the IDB-01R (Fig. 2C; Supplementary Fig. S3B). The BRCA1-mutated model VHIO-127 was sensitive to docetaxel but rapidly develop resistance (Fig. 2C). VHIO-98 showed a more heterogeneous response but most tumors developed resistance by passage 4 (Supplementary Fig. S3B). Again, no changes in chr12p copy number were observed in the docetaxel resistant, non-BRCA1–mutated models: VHIO-98R and IDB-01R (Supplementary Fig. S3C). However, in the BRCA1-mutated VHIO-127R additional copy gains were found (Fig. 2D), recapitulating the results obtained in the BRCA1-mutant IDB-02R tumors. These result indicate that the population bearing chr12p amplification is present in the initially sensitive tumors and is enriched after short-term treatment with docetaxel and during the emergence of resistance.

CNA of chr12p was also analyzed in 6 independent TNBC cell lines, including 2 BRCA1-mutant cell lines (MDA-MB-436 and HCC1937) and 1 with BRCA1 promoter hypermethylation (UACC3199), and 3 non-BRCA1–mutated lines (MDA-MB-231, MDA-MB-468, and HCC1143; Supplementary Table S1.1). MDA-MB-436 showed amplification mainly of ETV6, but also gain of GABARAPL1, which is indicative of chr12p amplification, whereas most other TNBC cell lines showed partial deletion of these genes (Supplementary Fig. S3E). CNAs in cells surviving docetaxel treatments (72 hours) revealed that the population with the GABARAPL1/ETV6 amplification was only enriched in the BRCA1-mutated MDA-MB-436 TNBC cell line (Fig. 2G; Supplementary Fig. S3F), mimicking the results obtained in the BRCA1-mutated PDX models. Finally, we analyzed chr12p in clinical samples from a TNBC 1731C>T BRCA1 carrier, collected before and after docetaxel treatment. After diagnosis of TNBC, the patient received 6 courses of adjuvant docetaxel, and metastatic disease was confirmed 5 years later. Exome sequencing was performed in blood, and in primary tumor at diagnosis, metastatic relapse, and metastatic progression. Chr12p gain was detected in the relapse after docetaxel treatment (seg. p mean = 0.2, Fig. 2H and I) and a greater gain was found at metastatic progression (seg. p mean = 0.45; Fig. 2H and I). Together, these results demonstrate that chr12p amplification is frequently found in docetaxel sensitive TNBC/BRCA1-mutated patients, in which a subpopulation of tumor cells harboring chr12p amplification is enriched after docetaxel treatment and the emergence of docetaxel resistance.

To explore the potential relevance of chr12p amplification in clinical disease, we queried the cBioPortal (28, 29) about alterations in genes located at different regions of chr12p. Amplification of GABARAPL1, ETV6, and KLHL42 occurred in a high proportion of testicular germ cell, ovarian and lung tumors, and in 4% of breast tumors, with relative consistency between breast cancer datasets (Supplementary Fig. S4A).

Analyses of breast cancer samples from the METABRIC cohort (8, 18) revealed that the amplification and gain of GABARAPL1, ETV6, and KLHL42 tends to co-occur in these patients (Supplementary Fig. S4B; Supplementary Table S2.3), together with most genes located in chr12p (Supplementary Table S2.4). Analyses of these breast cancer samples provided no evidence of amplification of a minimal region within chr12p, suggesting that the entire chr12p tends to be coamplified (Fig. 3A and B). To investigate whether chr12p amplification was enriched in any molecular breast cancer subtype, the PAM50+claudin-low subtype and 3-gene classifier subtype were used in METABRIC patients with breast cancer. Chr12p amplification was strongly associated with basal-like tumors (OR = 9.072; P < 0.0001); similarly, amplification of GABARAPL1, ETV6, and KLHL42 was associated to ER⁻/HER2⁻ tumors (OR = 3.968; P < 0.0001; Supplementary Fig. S4C). Conversely, a statistically significant negative correlation of chr12p amplification with luminal A tumors (OR = 0.319; P < 0.0001) and amplification of the surrogate genes with ER⁺/HER2⁻ low proliferative tumors (OR = 0.207; P < 0.0001) was shown (Fig. 3A; Supplementary Fig. S4C). Amplification and gain of these genes were significantly correlated with the increased expression of the corresponding genes (P < 0.0001; Supplementary Fig. S4D) and cooccurred with mutations associated with basal-like breast cancer, including those of BRCA1 and TP53 (Supplementary Table S2.5). Comparison of transcriptomic profiles and associated pathways in TNBC tumors with or without chr12p amplification revealed several pathways like those enriched in IDB-02R tumors (e.g., regulation of mitosis, replication, and cell cycle), demonstrating the reliability of our resistant model to address the relevance of chr12p amplification in the clinical setting (Supplementary Fig. S4E; Supplementary Tables S2.6 and S2.7).

Analysis of COSMIC-defined mutational signatures in the TCGA dataset revealed that chr12p amplification is significantly enriched in tumors displaying a genomic mutational signature characteristic of cases with BRCA1/2 mutations and/or defects in homologous recombination (signature S3, OR = 3.73; P < 0.0001; Fig. 3C; refs. 30–32). Overall survival was predicted to be lower in breast cancer patients with an amplification in chr12p (P = 0.001; Fig. 3D), and most importantly, survival analyses restricted to the subset of S3 breast tumors had a significantly poorer outcome in BRCA1-like tumors with chr12p amplification (P = 0.02; Fig. 3E). These results imply that in breast cancer, chr12p amplification is associated with basal-like/TNBC tumors and BRCAness mutational signatures, and poor survival even in aggressive BRCA1-like phenotypes.

Selection of the most effective chemotherapeutic agents is often based on empirical arbitrary decisions. Thus, we next aimed to investigate the relevance of chr12p in TNBC disease and primary response to the standard of care, docetaxel but also carboplatin, a platinum agent with promising responses in a subset of TNBC tumors (33). The copy number of chr12p and docetaxel and carboplatin responses (according to modified RECIST 1.1 criteria; ref. 34) were determined in an independent cohort of 24 TNBC PDX tumors (including ten BRCA1 mutant) from Baylor College of Medicine, many of which have been reported previously (Supplementary Table S1.1; ref. 21). Analysis (inferred from whole-exome sequencing data) revealed that 14 of the 24 PDX models (58%), including 7 of 10 with BRCA1 mutations, showed chr12p amplification mimicking associations found in clinical samples (Supplementary Table S1.1; Fig. 4A). CNA analysis using TaqMan probes of the 3 surrogate genes GABARAPL1, ETV6, and KLHL42 confirmed amplification in 7 of the 24 PDX models tested, 4 of them BRCA1 mutated (Supplementary Fig. S5A).

Most tumors with chr12p amplification/gain showed a partial response to docetaxel, but no significant association with primary response to docetaxel was found (Fig. 4A and B). In contrast, chr12p amplification/gain was associated with a better response to carboplatin, as demonstrated by the negative correlation between copy number and carboplatin response (Fig. 4A and B). Unlike the enrichment observed after docetaxel treatment, no chr12p copy number changes were observed after short-term treatment with carboplatin, either in IDB-02S or in MDA-MB-436 cells (Supplementary Fig. S5B–S5E), suggesting that the population harboring chr12p amplification is not resistant to carboplatin. These results suggest that IDB-02R, despite displaying resistance to docetaxel will remain sensitive to carboplatin. To test this hypothesis, IDB-02S and IDB-02R tumors were treated with different doses of carboplatin. Strikingly, IDB-02R tumors were more sensitive to carboplatin than were IDB-02S tumors. IDB-02S tumors were resistant to weekly treatments with 10 mg/kg of carboplatin and a response was only observed after 5 doses of 50 mg/kg of carboplatin. In contrast, IDB-02R tumors showed partial response to 10 mg/kg of carboplatin and full regression was observed after 5 doses of 50 mg/kg (Fig. 4C).

Finally, aiming to directly address the causality of genes located in chr12p amplification in drug responses, docetaxel sensitivity was evaluated in ETV6 and GABARAPL1 overexpressing cells (3 different TNBC breast cancer cell lines where the amplification was not detected; Supplementary Fig. S6A–S6C). In addition, sensitivity to docetaxel and carboplatin was evaluated in MDA-MB-436 cells with ETV6 and GABARAPL1 overexpression and knockdown, which harbor chr12p amplification (Supplementary Fig. S6D–S6G). No significant differences in the response to docetaxel or carboplatin were found between overexpressing/knockdown variants and corresponding controls in any of the TNBC cell lines tested (Supplementary Fig. S6C, S6E, and S6G), suggesting that emergence of docetaxel resistance or sensitivity to carboplatin is not driven by increased expression of these 2 genes.

Together, these results confirmed the presence of chr12p amplification in TNBC, mainly but not exclusively in BRCA1 mutated tumors. The chr12p amplification, despite being selected upon docetaxel treatment and emergence of resistance, cannot predict primary response to docetaxel; in contrast, chr12p amplification is significantly associated with a better response to carboplatin, suggesting that sequential treatment with docetaxel and carboplatin may improve survival in TNBC/BRCA1-mutated tumors.

The genomic landscape of breast cancer has been extensively studied in recent years, taking advantage of next-generation sequencing technologies and platforms (7, 35). Despite the multiple genomic alterations present in TNBC (7), only a few clinical studies have associated them with chemoresistance (6, 11, 36). In attempting to overcome these limitations, this study investigates the contribution of genomics as a mechanism for developing resistance to docetaxel in patients with TNBC using matched sensitive and resistant TNBC PDX models.

Breast cancer PDXs maintain genomic features present in the initial breast tumors from patients, with different percentages of subclonal intratumor selection during xenografting and serial engraftment in mice (19, 20). Our results confirm that, despite the high rate of mutations and chromosomal aberrations that characterize TNBC, most of these changes are maintained during serial transplantation in mice and even upon heavy exposure and emergence of resistance to docetaxel, although some clonal dynamics are observed during engraftment and drug treatments. Our PDX models were established from metastatic pleural effusions, suggesting that the first step in selecting an aggressive tumor population had occurred during the metastatic process in the patient and a second, minor round of selection occurred during xenografting in mice.

We found that chr12p amplification is associated to TNBC/basal-like tumors, the mutational signature S3 (BRCAness) and TP53 and BRCA1 mutations, among other chromosomal aberrations. Analyses of large clinical breast cancer datasets did not reveal a minimal amplified region, and the whole chr12p tends to be coamplified. Gene expression pathways enriched in IDB-02R model converge with those TNBCs from METABRIC featuring chr12p amplification in pathways involved in mitosis, cell cycle, and cell division. These are key biological processes in tumorigenesis and resistance acquisition (37), all of which are mediated by microtubules, kinesins, and cell division cycle-associated proteins (38, 39), among others.

Amplification of chr12p is a common chromosomal alteration detected in various types of cancer: testicular germ cell tumors (40), ovarian cancer (7), and some breast tumors (27). Basal-like tumors with amplification of chr12p show other chromosomal alterations, such as chr5q deletion (27), which is also observed in the IDB-02 PDX model. Moreover, we found that chr12p amplification is associated with poor survival in breast cancer and even within the subgroup of tumors with the S3 mutational signature (BRCAness).

Intratumoral genetic heterogeneity in breast cancer is an established phenomenon and breast tumor subclones with different genomic profiles respond differently to treatment (6, 19). The question remains as to whether resistant mutations appear de novo in the presence of treatment or if they arise from preexisting mutations present in subclonal populations in the initially sensitive tumor. The observation that chr12p amplification is present in the human metastasis of origin, diluted in sensitive IDB-02S tumors, and selected after docetaxel treatment suggests that the amplification identifies a subclonal population with chemoresistant properties that is overcome by the nonamplified chr12p population during long-term drug interruption. Gains in chr12p observed after short-term docetaxel treatment in residual disease of several docetaxel-sensitive TNBC PDX models and cell lines provide further evidence that enrichment of a preexisting population takes place. This result is consistent with recent reports of the emergence of preexisting populations in metastasis, relapses, or after chemotherapy treatment (6, 11, 36) and is an additional evidence of the contribution of the “drug holidays effect” that we recently described in TNBC PDX docetaxel-resistant tumors during the absence of selective drug pressure (17).

A major challenge for oncologists is to select the most effective chemotherapeutic agents and schedule for individual patients. Single agents or combinations are given to patients following variable schedules and mostly in an arbitrary empirical manner (41). The administration of ineffective chemotherapeutic agents increases side effects and decreases the quality of life of patients. Moreover, suboptimal therapies or dosage may result in the early emergence of resistance or even in the selection of a more aggressive and metastatic population of cells. Thus, we urgently need to identify biomarkers of response to current chemotherapies. Analyses of chr12p copy number and primary response to single-agent docetaxel or carboplatin in a collection of 24 TNBC PDX models revealed that the presence of chr12p amplification alone cannot predict primary response to docetaxel, but it is associated with high sensitivity to carboplatin. Unlike after docetaxel treatment, no enrichment is observed in the chr12p-amplified population after short-term treatment with carboplatin, indicating that it remains sensitive to this drug. The gain of additional copies of chr12p is indicative of greater sensitivity to carboplatin, which may provide a window of opportunity for the treatment of patients with TNBC who no longer respond to taxanes.

Patients with breast and ovarian cancer with BRCA1 mutations are more sensitive to platinum-based agents, such as carboplatin, than other patients (42, 43), due to the deficiency in the homologous repair machinery (44). BRCA1-mutated breast and ovarian tumors have been associated with an unstable chromosomal phenotype (45), which, in turn, has been associated in vitro and in vivo with taxane resistance (46, 47). An intratumor population present in BRCA1-mutated or BRCAness breast tumors harboring chr12p amplification seems to be enriched in mitotic pathways and thus involved in resistance to taxanes, by bypassing its cytotoxic effect. However, as a gain in vulnerability, this selected population is more sensitive to carboplatin. Our study offers an explanation of the association between BRCA1-mutated tumors with the aforementioned unstable chromosomal phenotype and taxane resistance/carboplatin sensitivity.

In a retrospective study patients with TNBC who received carboplatin with anthracycline/taxane-based neoadjuvant therapy have significantly higher rate of complete regression compared with those who received anthracycline/taxane alone (48), which substantiate the efficacy of docetaxel and carboplatin combinations in the clinical setting (49, 50).

Studying the CNAs of chr12p, the response rate and progression-free survival in patients with TNBC treated with single-agent docetaxel or carboplatin, such as those of the TNT trial (42), and sequential treatment schedules with docetaxel and carboplatin in the clinics will be important for confirming the clinical relevance of our results.

Our results using TNBC and BRCA1-mutated cell lines, PDX models, and clinical samples provide compelling evidence that a preexisting tumor subpopulation harboring chromosome 12p amplification is enriched after short-course docetaxel treatment and the emergence of resistance. Importantly, enrichment in chr12p amplification is indicative of carboplatin sensitivity, suggesting that sequential treatments first with docetaxel and then with carboplatin could improve survival in TNBC and BRCA1-mutated patients.

L.E. Dobrolecki is a project manager at StemMed. A. Prat is a consultant/advisory board member of Nanostring Technologies, Roche, Pfizer, Novartis, MSD, Amgen, and Oncolytics Biotech. C. Caldas reports receiving commercial research grants from AstraZeneca, Servier, Genentech, and Roche. J. Arribas is a consultant/advisory board member of Menarini Biotech. J. Balmaña is a consultant/advisory board member of AZ, Bristol Myers, and Pfizer. M.T. Lewis is a limited partner at StemMed Ltd. and StemMed Holdings, reports receiving other commercial research support from Ventana Roche, has ownership interest (including stock, patents, etc.) in StemMed Ltd. and StemMed Holdings, and is a consultant/advisory board member of Tvardi Therapeutics. X.S. Puente has ownership interest (including stock, patents, etc.) in DREAMgenics. E. González-Suárez reports of receiving other commercial research support from Amgen Inc. (denosumab) and is a consultant/advisory board member at Amgen Inc. G.M. Wulf has a patent Application 14/348810, Compositions and Methods for the Treatment of proliferative diseases pending, and a patent US 20090258352 A1, Pin1 as a marker for abnormal cell growth licensed to Cell Signaling Technology; R&D Systems.

Conception and design: M.T. Lewis, X.S. Puente, E. González-Suárez

Development of methodology: J. Gómez-Miragaya, A. Collado-Sole, E.M. Trinidad, V. Serra, E. González-Suárez

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Gómez-Miragaya, M. Palafox, L.E. Dobrolecki, G.M. Wulf, A. Collado-Sole, E.M. Trinidad, A. Prat, A. Bruna, C. Caldas, J. Arribas, M.T. Soler-Monso, A. Petit, J. Balmaña, C. Cruz, M.T. Lewis, E. González-Suárez

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Gómez-Miragaya, A. Díaz-Navarro, R. Tonda, S. Beltran, L. Palomero, C. Huang, S. Vasaikar, B. Zhang, A. Collado-Sole, L. Paré, A. Prat, C. Caldas, C. Cruz, M.A. Pujana, M.T. Lewis, X.S. Puente, E. González-Suárez

Writing, review, and/or revision of the manuscript: J. Gómez-Miragaya, R. Tonda, L.E. Dobrolecki, C. Huang, S. Vasaikar, B. Zhang, P. Muñoz, A. Prat, C. Caldas, J. Balmaña, M.A. Pujana, M.T. Lewis, X.S. Puente, E. González-Suárez

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Gómez-Miragaya, C. Huang, E. González-Suárez

Study supervision: B. Zhang, M.T. Lewis, X.S. Puente, E. González-Suárez

We acknowledge Conxi Lázaro, Ander Urruticoecha, Idoia Morilla, Adela Fernandez (Catalan Institute of Oncology), Josefina Climent (University Hospital of Bellvitge) for advice, and Alana Welm and Y. DeRose (Huntsman Institute) for sharing PDX models (Huntsman Institute). Cristina Saura (VHIO) and Isabel T. Rubio (Vall d'Hebron Hospital), and Albert Gris-Oliver (VHIO) for providing study materials. The IDIBELL animal facility for experimental support. The authors also acknowledge the joint participation by Diana Henry Helis Medical Research Foundation through its direct engagement in the continuous active conduct of medical research in conjunction with Baylor College of Medicine and its “Blood-borne BioMarkers for Detection of Breast Cancer” Program.

This work was supported by grants to E. González-Suárez by the Spanish Ministerio de Ciencia, Innovación y Universidades, which is part of Agencia Estatal de Investigación (AEI), through the projects (SAF2008-01975, SAF2011-22893, SAF2014-55997-R, SAF2017-86117-R), and ISCIII PIE13/00022, co-funded by European Regional Development Fund. ERDF, a way to build Europe, by a Career Catalyst Grant from the Susan Komen Foundation CCR13262449. The laboratory of E. Gonzalez-Suarez is funded by an ERC Consolidator grant LS4-682935. J. Gómez-Miragaya was recipient of a predoctoral grant from the MICINN. A. Collado-Sole is recipient of a grant from the FI programme of the Secretariat for Universities and Research of the Department of Business and Knowledge of the Government of Catalonia. We thank CERCA Programme/Generalitat de Catalunya for institutional support. With the support of the European Social Fund (ESF) “ESF, Investing in your future.” The PDX from VHIO were supported by a “GHD-pink” research support via the FERO Foundation to V. Serra, Breast Cancer Research Foundation (BCRF-17-008) and Instituto de Salud Carlos III (PI16/00253 and CB16/12/00449) to J. Arribas. V. Serra is recipient of ISCIII grant CP14/0028, C. Cruz is recipient of an AECC grant (AIOC15152806CRUZ), M.A. Pujana of PI15/00854 and X.S. Puente of Spanish Ministry of Economy and Competitivity MINECO (SAF2017-87811-R), J. Balmaña of FIS PI12-02606, and A. Prat of PI13/01718 from the Instituto de Salud Carlos III. C. Caldas and A. Bruna are supported by Cancer Research UK. M.T. Lewis and L. Dobrolecki are supported in part by the Breast Cancer Research Foundation, the Susan G. Komen Foundation, The V Foundation, NIH/NCI grant U54CA224076, BCM Breast Cancer SPORE P50 CA186784, BCM Cancer Center grant P30 CA125123, and a generous gift from the Korell family. B. Zhang, C. Huang, and S. Vasaikar are supported by Cancer Prevention & Research Institutes of Texas (CPRIT RR160027) and the McNair Medical Institute Scholar. G.M. Wulf reports grants from a Stand Up To Cancer-American Association for Cancer Research Dream Team Translational Cancer Research Grant (Grant Number SU2C-AACR-DT0209). Stand Up To Cancer is a division of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the scientific partner of SU2C. He also reports grants from Mary Kay Ash Foundation, grants from Ovarian Cancer Research Foundation, grants from Breast Cancer Alliance, grants from Breast Cancer Research Foundation, grants from NIH RO1 1R01CA226776-01, and grants from Merck&Co. during the conduct of the study.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.