Multicenter Phase II Study of Lurbinectedin in BRCA-Mutated and Unselected Metastatic Advanced Breast Cancer and Biomarker Assessment Substudy

Purpose This multicenter phase II trial evaluated lurbinectedin (PM01183), a selective inhibitor of active transcription of protein-coding genes, in patients with metastatic breast cancer. A unicenter translational substudy assessed potential mechanisms of lurbinectedin resistance. Patients and Methods Two arms were evaluated according to germline BRCA1/2 status: BRCA1/2 mutated (arm A; n = 54) and unselected (BRCA1/2 wild-type or unknown status; arm B; n = 35). Lurbinectedin starting dose was a 7-mg flat dose and later, 3.5 mg/m2 in arm A. The primary end point was objective response rate (ORR) per Response Evaluation Criteria in Solid Tumors (RECIST). The translational substudy of resistance mechanisms included exome sequencing (n = 13) and in vivo experiments with patient-derived xenografts (n = 11) from BRCA1/2-mutated tumors. Results ORR was 41% (95% CI, 28% to 55%) in arm A and 9% (95% CI, 2% to 24%) in arm B. In arm A, median progression-free survival was 4.6 months (95% CI, 3.0 to 6.0 months), and median overall survival was 20.0 months (95% CI, 11.8 to 26.6 months). Patients with BRCA2 mutations showed an ORR of 61%, median progression-free survival of 5.9 months, and median overall survival of 26.6 months. The safety profile improved with lurbinectedin dose adjustment to body surface area. The most common nonhematologic adverse events seen at 3.5 mg/m2 were nausea (74%; grade 3, 5%) and fatigue (74%; grade 3, 21%). Neutropenia was the most common severe hematologic adverse event (grade 3, 47%; grade 4, 10%). Exome sequencing showed mutations in genes related to the nucleotide excision repair pathway in four of seven tumors at primary or acquired resistance and in one patient with short-term stable disease. In vivo, sensitivity to cisplatin and lurbinectedin was evidenced in lurbinectedin-resistant (one of two) and cisplatin-resistant (two of three) patient-derived xenografts. Conclusion Lurbinectedin showed noteworthy activity in patients with BRCA1/2 mutations. Response and survival was notable in those with BRCA2 mutations. Additional clinical development in this subset of patients with metastatic breast cancer is warranted.


INTRODUCTION
Metastatic breast cancer (MBC) is a heterogeneous disease. Some new therapeutic approaches offer a tailored therapy on the basis of tumor characteristics. Beyond hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2) status, a better knowledge of the DNA damage response pathway has allowed the identification of actionable targets for MBC.
Approximately 3% to 5% of unselected patients with MBC carry a germline mutation in BRCA1 or BRCA2 genes. These genes encode for two tumor suppressor proteins essential in homologous recombination repair (HRR), a vital DNA repair process that uses the undamaged sister chromatid to carry out high-fidelity repair of DNA double-strand breaks. 1 Trabectedin, an antitumor agent that relies on an efficient nucleotide excision repair (NER) and a deficient HRR pathway, has shown remarkable activity in heavily pretreated patients with MBC with a germline BRCA1/2 mutation. [2][3][4] Lurbinectedin, a trabectedin analog, is a selective inhibitor of the active transcription of protein-coding genes. The mechanism involves the irreversible stalling of elongating RNA polymerase II on the DNA template and its specific degradation by the ubiquitin/ proteasome machinery. Subsequently, recruitment of DNA repair factors, including XPF nuclease, induces the accumulation of double-strand breaks and apoptosis as downstream events. 5 These effects are increased in HRR-deficient cells. Indeed, in BRCA2mutated cells, this could be related to the concurrence of deficient DNA repair and formation of R-loops that occurs during the elongation step of transcription by RNA polymerase II. 6,7 Both antitumor activity observed with lurbinectedin against HRR-deficient cell lines 8,9 and clinical activity observed with trabectedin prompted the conduct of this phase II trial to evaluate the efficacy and safety of lurbinectedin in patients with MBC with deleterious germline BRCA1/2 mutations or unselected disease. Two arms were evaluated according to germline BRCA1/2 status: BRCA1/2 mutated (arm A) and unselected (BRCA1/2 wild-type or unknown status; arm B). In parallel, a correlative translational research study was undertaken to identify predictive biomarkers of response and resistance through exome sequencing of patients' biopsy samples and in vivo efficacy analyses in preclinical models.

PATIENTS AND METHODS
Patients were recruited from 11 investigational sites in the United States and Spain. The study protocol was approved by the independent local ethics committee at each participating center and was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines, and local regulations for clinical trials. Signed informed consent was obtained from all patients before any study-specific procedure.

Eligibility Criteria
Patients eligible for this study were 18 to 75 years old with a histologically proven diagnosis of MBC, no more than three prior chemotherapycontaining regimens in the advanced setting (including at least one prior trastuzumab-containing regimen in those with known HER2-overexpressing tumors), measurable disease per Response Criteria in Solid Tumors (RECIST) version 1.1, 10 Eastern Cooperative Oncology Group performance status # 1, and adequate major organ function. Patients were excluded if they had previously received lurbinectedin, trabectedin, or radiotherapy to . 35% of bone marrow; prior or concurrent other malignant disease unless in complete remission for . 5 years; symptomatic, corticosteroid-requiring, or progressive CNS involvement; exclusively bone-limited disease; concomitant unstable or serious medical condition, or impending need for radiotherapy.

Treatment
All patients were treated with lurbinectedin administered as a 1-hour intravenous (IV) infusion every 3 weeks. The starting dose was a 7.0-mg flat dose (FD) on the basis of results from the first-in-human phase I trial. 11 On the basis of new (unpublished) data from this and other lurbinectedin studies that suggested body surface area (BSA)-related toxicity, the study was amended to introduce BSA-based dosing. After an amendment, patients received lurbinectedin as a 1-hour IV infusion on day 1 every 3 weeks at a starting dose of 3.5 mg/m 2 (capped at BSA 2.0 m 2 ). All patients received antiemetic prophylaxis. Granulocyte colony-stimulating factors were allowed for secondary prevention of neutropenia.

Efficacy Assessment
The primary efficacy end point was objective response rate (ORR) according to RECIST version 1.1. Secondary efficacy end points were duration of response, clinical benefit (ORR or stable disease . 3 months), progression-free survival (PFS), and overall survival (OS). Patients evaluable for efficacy received at least one complete lurbinectedin infusion and had at least one tumor assessment. Radiologic tumor evaluation was performed every 6 weeks (two cycles) until cycle six, and every 9 weeks (three cycles) thereafter. Objective response was confirmed at least 4 weeks later. Patients in arm B who achieved a confirmed response and whose BRCA1/2 status was unknown were tested for BRCA1/2 germline variants.

Safety Assessment
Safety was evaluated in all patients who received at least one lurbinectedin infusion by assessment of adverse events (AEs), clinical laboratory test results, physical examinations, and vital signs. AEs were recorded and coded with the Medical Dictionary for Regulatory Activities version 12.0. AEs and laboratory values were graded according to the National Cancer Institute Common Toxicity Criteria for Adverse Events (version 4.0). All patients were followed until recovery from any lurbinectedin-related AE.

Pharmacogenomic Substudy
Single nucleotide polymorphisms and DNA mutations in a panel of 151 cancer-related genes were centrally analyzed (Appendix, online only).

Fresh Biopsy Cohort, Exome Sequencing, Implantation of Patient-Derived Xenograft Models, and In Vivo Experiments
All patients from one of the investigational sites were offered to participate in a translational substudy (Appendix).

Statistical Methods
A futility analysis on the basis of the primary end point (ORR) was planned after 20 and 30 evaluable patients had been treated in arms A and B, respectively. If fewer than four of the 20 patients in arm A or fewer than three of the 30 patients in arm B achieved an objective confirmed response, recruitment into that arm was halted. Otherwise, recruitment was planned until at least 53 and 64 evaluable patients were included in each arm, respectively (Appendix).
In the clinical part of the study, binomial exact estimator and 95% CI were calculated for ORR. Kaplan-Meier method was used to analyze PFS and OS (compared in subgroups by unstratified log-rank test). Logistic regression was used in the ORR multivariable analysis.
Statistical analysis for in vitro data consisted of t test or one-way analysis of variance (Dunnett's post hoc test). Two-way analysis of variance (Tukey's post hoc test) was used for in vivo patient-derived xenograft (PDX) data. Multiplicity-adjusted P values are reported, and statistical significance was achieved if P , .05.
In arm B, 74% of patients had unknown BRCA1/2 status (26% were BRCA1/2 negative); 49% and 51% had triple-negative and HR+ disease, respectively; and 40% had more than two tumor sites (liver metastasis, 60%). The median number of advanced chemotherapy lines was one (range, one to three), and 6% and 20% of patients received prior platinum-based chemotherapy in the adjuvant/neoadjuvant and advanced setting, respectively (Table 1). No relevant differences between arms were observed (Appendix Table A1, online only).

Dosing
In arm A, 35 patients received lurbinectedin at a starting dose of 7 mg FD (median of six cycles per patient [range, one to 24 cycles]), and 19 patients received 3.5 mg/m 2 (median of nine cycles per patient [range, two to 30 cycles]). In arm B, all 35 patients received lurbinectedin at 7 mg FD (median of three cycles [range, one to 27 cycles]; Fig 1).

Efficacy in Arm B (BRCA1/2 Wild-Type or Unknown Status)
Futility analysis conducted on the first 30 evaluable patients showed two objective responses (below the minimum of three responses required for recruitment extension), and this cohort was closed. The ORR in 34 total evaluable patients was 9% (95% CI, 2% to 24%), including three PRs (Table 2; Appendix Fig A1). Median duration of response was 5 months (95% CI, 2 to 18 months). Disease control rate and clinical benefit rate were 59% and 26%, respectively. Germline BRCA1/2 status of the three patients with PRs was wild type. No other candidate genes for sensitivity were found in the pharmacogenomic substudy of tumor samples (Appendix Table A2, online only). Median PFS was 2.5 months (95% CI, 1.3 to 3.4 months; Fig 3A), and median OS was 12.5 months (95% CI, 6.6 to 17.9 months; Fig 3B).

Safety
All 89 treated patients were evaluable for safety (Table 3). Severe AEs and laboratory abnormalities occurred at lower incidences after dose adjustment according to BSA in arm A. Grade 4 hematologic abnormalities (mainly neutropenia and thrombocytopenia) and febrile neutropenia (from 29% to 5% of patients) were notably reduced. The most common nonhematologic AEs were nausea and fatigue (74%).
Treatment-related discontinuations occurred in three patients treated before the amendment that adjusted dose to BSA (3.5 mg/m 2 ): one in arm A (because of grade 3 dyspnea with multiple pleural metastases at baseline, talc pleurodesis, and pericardial window) and two in arm B (because of grade 3 pneumonitis and grade 3 failure to thrive/vomiting). Furthermore, one patient in arm B who was treated with the 7-mg FD died during cycle 1 as a result of septic shock with unknown relationship, which was concomitant with treatmentrelated grade 4 ALT/AST increase (extensive miliary liver metastases and grade 2 ALT/AST increase were present at baseline; Fig 1).

Exome Sequencing Analysis
A parallel unicenter translational research study included whole-exome sequencing on five paired pre-and postlurbinectedin fresh biopsy samples that met sufficient cellularity (n = 10; Appendix Table A3, online only). No secondary BRCA1/2 mutations were identified at disease progression (PD). Because prior in vitro data reported transcription-coupled NER gene mutations as a mechanism of resistance to trabectedin, 4 this analysis was focused on these and other DNA repair genes. An acquired mutation in the NER gene ERCC4 encodes the nuclease XPF involved in DNA damage accumulation by lurbinectedin. 5 The c.A583T change affects a splicing donor site, which could generate a premature stop codon and protein truncation (XPF p.R195*; Fig 4B). Alternatively, the c.A583T variant could encode a missense amino acid substitution (XPF p.R195W), the functional effect of which was evaluated in vitro. Complementation of XPF-deficient human fibroblasts (GM08437) with wild-type XPF sensitized these cells to lurbinectedin, whereas complementation with XPF p.R195W behaved as the empty vector ( Fig 4C). Similarly, two somatic mutations in NER genes were identified in another patient at PD (patient 4; PFS, 5.9 months): XPA (p.Q216E; VAF, 38%) and GTF2H5 (p.C12R; VAF, 29%; Appendix Table A3). Copy number variation analysis of PDX252 derived from patient 6 at lurbinectedin progression identified a complete loss of NER gene ERCC8/CSA, a previously reported mechanism of trabectedin resistance. 4 In the remaining two patients with paired biopsy samples, exome sequencing did not reveal NER mutations at resistance. Secondary BRCA1/2 mutations and NER-related alterations also were searched in nonpaired fresh tumor samples and in all archival tumor samples available (n = 7 primary; n = 5 metastatic). No secondary BRCA1/2 mutations were found. A heterozygous germline mutation in ERCC4 (XPF p.Q300H; VAF, 46%) was identified in patient 12, who had primary lurbinectedin resistance. Of note, this mutation was enriched to homozygosity in the primary tumor (VAF, 69%) and in the metastatic relapse (VAF, 60%; Appendix Figs A4A and A4B, online only). None of the long-responder patients carried an ERCC4 mutation. A mutation in the XPF scaffold protein SLX4/ FANCP was found in patient 10, who showed stable disease (PFS, 2.7 months; p.A952M; VAF, 91%). Thus, these results suggest that NER-related alterations may arise as a mechanism of resistance to lurbinectedin.

In Vivo Studies With Lurbinectedin and Cisplatin
Prior data suggested that NER alterations could induce resistance to lurbinectedin but increase sensitivity to cisplatin, 4,5 and we confirmed this in vitro (Appendix Fig A3B, online only). Antitumor activity of cisplatin and lurbinectedin was investigated in five coclinical PDX models derived from patients in arm A (Appendix Table A3; Appendix Fig A5A, online only). All PDX models recapitulated the clinical response to lurbinectedin. Three of five models showed opposite responses for lurbinectedin and cisplatin. Of note, one PDX (PDX252) generated at progression to lurbinectedin harbored an ERCC8/CSA loss and responded to cisplatin. Six additional PDX models from a collection of BRCA1/ 2-mutated PDXs 12 were tested, and the spectrum of activity of lurbinectedin and cisplatin only partially overlapped (Appendix Fig A5B,online only).
Supporting evidence for clinical cisplatin sensitivity after progression to lurbinectedin is exemplified in Appendix Fig A6  (online only). The patient who harbored mutations in XPA and GTF2H5 in the liver metastasis at lurbinectedin progression was  The most common nonhematologic (nausea and fatigue) and hematologic (neutropenia) AEs observed in patients with MBC were already reported in a phase I study conducted at 7 mg FD 11 and in a phase II study in patients with platinum-resistant/ refractory advanced ovarian cancer. 13 Nonetheless, the lurbinectedin safety profile was improved after implementation of BSAbased dosing. Severe events occurred at lower incidences, with a notable reduction in the rate of grade 4 hematologic abnormalities and febrile neutropenia. Cumulative toxicity was not observed, and most patients achieved long-time exposure. Therefore, lurbinectedin 3.5 mg/m 2 seems to be a safe and active dose for future trials.
The most remarkable antitumor activity was found in patients with BRCA2 mutation (ORR, 61%; PFS, 5.9 months; OS, 26.6 months) in whom ORR increased to 72% and 71% in those without prior PARPi and platinum therapy, respectively. Trabectedin also had shown higher efficacy in patients with BRCA2 MBC versus BRCA1 (ORR, 33% v 9%; median, PFS 4.7 v 2.5 months). 3 ORR for PARPi therapy in patients with BRCA2 MBC in phase I/II trials was 22% with olaparib, 14 36% with veliparib, 15 and 34% with talazoparib. 16 The reasons for the different efficacy of lurbinectedin in BRCA2 versus BRCA1 disease are under investigation. In addition to its well-known role in HRR, BRCA2 prevents the formation of RNA-DNA hybrids (R-loops) that occurs during the elongation step of transcription by RNA polymerase II. 6,7 One hypothesis to explain the differential activity of trabectedin and lurbinectedin observed in BRCA2compared with BRCA1-mutated MBC is the concurrence of deficient DNA repair and the formation of R-loops. Recognition of HRR deficiency as a biomarker of sensitivity to platinum agents has led to their reconsideration for the treatment of BRCA1/2 mutation-associated tumors. Indeed, the Triple-Negative Breast Cancer Trial showed germline BRCA1/2 tumors to be more sensitive to carboplatin than docetaxel and provided clinical evidence to treat these patients with platinum in the metastatic setting. 17 Nevertheless, previous platinum exposure decreases the benefit to other BRCA1/2-directed therapies, such as PARPi. 16,18 In this regard, in a phase II study with talazoparib, patients who responded to platinum and did not progress until at least 8 weeks after treatment had an ORR of 21% versus 37% among those who were platinum naïve in the metastatic setting. 16 In the phase III OlympiAD 18 trial with olaparib, the ORR (unconfirmed response) decreased from 66% in platinum-naïve patients to 46% in patients with prior platinum therapy and no progression during treatment. Responses to lurbinectedin also decreased in patients previously exposed to platinum, although comparisons with PARPi trials 14,16 cannot be made because of different  jco.org inclusion criteria in the platinum-free interval. Although clinical data show decreased sensitivity to lurbinectedin in platinumpretreated patients, preclinical data in PDXs suggest a partial overlap of efficacy and resistance mechanisms. Additional research is needed to delineate the most appropriate therapeutic sequence to minimize cross-resistances. Secondary mutations in BRCA1 and BRCA2 that re-establish the reading frame may restore HRR proficiency and render cancer cells resistant to agents that target DNA damage. 19 In this study, exome sequencing showed no secondary BRCA1/2 mutations in either primary or acquired resistance. In contrast, alterations in NER-related genes were found. Somatic mutations in ERCC4 could explain primary resistance to lurbinectedin in a nonresponder patient and acquired resistance in another long-responder patient. VAF in two mutated genes involved in active transcription and NER (XPA and GTF2H5) increased at lurbinectedin progression in another patient. Of note, the acquisition of resistance to lurbinectedin did not preclude a subsequent response to platinum-based chemotherapy in this patient, as was also observed in a coclinical acquired-resistance model (PDX252). Limitations of this study were the small size of some of the subsets evaluated and the optional participation for the translational unicentric substudy, which resulted in a limited sampling. Additional studies will have to focus on lurbinectedin activity as a transcription inhibitor in HRR-deficient/NER-active tumors.
In conclusion, lurbinectedin has a unique mechanism of action, with promising efficacy observed in BRCA1/2-mutated MBC. The noteworthy specific activity of lurbinectedin in patients with BRCA2 mutation warrants additional clinical development.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Disclosures provided by the authors are available with this article at jco.org.  To evaluate the sensitivity to the drugs, tumor-bearing mice were equally distributed into treatment groups, with tumors ranging in size from 100 to 300 mm 3 . Lurbinectedin (0.18 mg/kg) and cisplatin (6 mg/kg) were administered intravenously weekly unless relative tumor volume , 0.5 or weight loss . 20%. Tumor growth was measured with caliper biweekly since the first day of treatment, and tumor volume was calculated as V = 4p / 3 / L 3 l 2 , where L is the largest diameter and l is the smallest. The antitumor activity was determined by comparing tumor volume after three to four cycles or at best response (in sensitive patient-derived xenograft [PDX], minimum value of percent tumor volume change sustained for at least 10 days) to its baseline: percent tumor volume change = (V -V initial ) / V initial 3 100. To classify the response of the subcutaneous implants, we modified the RECIST to be based on the percent tumor volume change: complete response, best response , 295%; partial response, best response , 230%; stable disease 230% , best response , 20; and disease progression, best response . 20%. Experiments were conducted using the European Union animal care directive (2010/63/EU) and were approved by the Ethical Committee of Animal Experimentation of the Vall d'Hebron Research Institute.

Statistical Methods: Information From the Phase II Trial
Sample size was calculated using East version 5.4 software (Cytel, Cambridge, MA) for each arm as a single proportion test, including a futility analysis in each one. The next hypotheses were selected as follows: • Arm A (BRCA): At least 53 evaluable patients had to be recruited to test the null hypothesis that objective response rate (ORR) was # 20% (P #.20) versus the alternative hypothesis that $ 40% patients have an objective response (P $.4). With these assumptions, if the number of evaluable patients with an objective response is $ 17, then this would allow the rejection of the null hypothesis. • Arm B (BRCA wild-type or unknown status): At least 64 evaluable patients had to be recruited to test the null hypothesis that ORR is # 10% (P # .10) versus the alternative hypothesis that $ 25% patients have an objective response (P $ .25). With these assumptions, if the number of evaluable patients with an objective response is $ 12, then this would allow the rejection of the null hypothesis.   T TT  TTT  T  T T T T T  T T  TT T T  T      *All human samples were obtained and implanted before lurbinectedin therapy, except PDX252 (at acquired resistance) and PDX221 (from primary tumor). †Gene mutations with a VAF # 5% are excluded by the predefined filtering for variant calling; however, if they are found in the post-treatment sample at VAF . 5%, then they are included and considered subclonal variants in the pretreatment sample. ‡CN variant excluded by the predefined filtering but with a log2 copy ratio of 20.7728, which suggests a subclonal presence that is enriched in the PDX sample. §This sample from patient 12 was obtained from the metastasis at first relapse and received one line of CT before treatment with lurbinectedin. jco.org