Palbociclib in combination with endocrine therapy versus capecitabine in hormonal receptor-positive, human epidermal growth factor 2-negative, aromatase inhibitor-resistant metastatic breast cancer: a phase III randomised controlled trial—PEARL

ORIGINAL ARTICLE Palbociclib in combination with endocrine therapy versus capecitabine in hormonal receptor-positive, human epidermal growth factor 2-negative, aromatase inhibitor-resistant metastatic breast cancer: a phase III randomised controlled trialdPEARL 5 M. Martin 1,2,3*y , C. Zielinski 4,5y , M. Ruiz-Borrego 3,6 , E. Carrasco 3 , N. Turner 7 , E. M. Ciruelos 3,8,9,10 , M. Muñoz 3,11,12 , B. Bermejo 2,3,13,14 , M. Margeli 3,15 , A. Anton 2,3,16 , Z. Kahan 17 , T. Csöszi 18 , M. I. Casas 3 , L. Murillo 3,19 , S. Morales 3,20 , E. Alba 2,3,21 , E. Gal-Yam 22 , A. Guerrero-Zotano 3,23 , L. Calvo 3,24 , J. de la Haba-Rodriguez 2,3,25 , M. Ramos 3,26 , I. Alvarez 3,27 , A. Garcia-Palomo 3,28 , C. Huang Bartlett 29 , M. Koehler 29,30 , R. Caballero 3 , M. Corsaro 31 , X. Huang 29 , J. A. Garcia-Sáenz 3,32 , J. I. Chacón 3,33 , C. Swift 34 , C. Thallinger 5,35 & M. Gil-Gil 3,36 1 Medical Oncology, Instituto de Investigación Sanitaria Gregorio Marañón, Medicine Department, Universidad Complutense, Madrid; 2 Oncology Biomedical Research National Network (CIBERONC-ISCIII), Madrid; 3 GEICAM Spanish Breast Cancer Group, Madrid, Spain; 4 Medical Oncology, Central European Cancer Center, Wiener Privatklinik Hospital, Vienna; 5 CECOG Central European Cooperative Oncology Group, Vienna, Austria; 6 Medical Oncology, Hospital Universitario Virgen del Rocio, Sevilla, Spain; 7 Institute of Cancer Research and Royal Marsden, London, UK; 8 Medical Oncology, Hospital Universitario 12 de Octubre, Madrid; 9 Medical Oncology, HM Hospitales Madrid, Madrid; 10 SOLTI Group on Breast Cancer Research, Barcelona; 11 Medical Oncology, Hospital Clinic de Barcelona, Barcelona; 12 Translational Genomics and Targeted Therapeutics in Solid Tumors (IDIBAPS), Barcelona; 13 Medical Oncology, Hospital Clínico Universitario de Valencia, Valencia; 14 Biomedical Research Institute INCLIVA, Valencia; 15 B-ARGO Group, Catalan Institute of Oncology, Hospital Universitari Germans Trias i Pujol, Badalona; 16 Medical Oncology, Hospital Universitario Miguel Servet, Zaragoza, Spain; 17 Department of Oncotherapy, University of Szeged, Szeged; 18 Department of Oncology, Jasz-Nagykun-Szolnok Megyei Hetenyi Geza Korhaz-Rendel} ointezet, Szolnok, Hungary; 19 Medical Oncology, Hospital Clínico de Zaragoza Lozano Blesa, Zaragoza; 20 Medical Oncology, Hospital Universitario Arnau de Vilanova, Lleida; 21 UGCI Medical Oncology, Hospitales Regional y Virgen de la Victoria, IBIMA, Málaga, Spain; 22 Department of Oncology, Institute of Oncology, Sheba Medical Center, Tel-Hashomer, Israel; 23 Medical Oncology, Instituto Valenciano de Oncología, Valencia; 24 Medical Oncology, Complejo Hospitalario A Coruña, Coruña; 25 Medical Oncology, Hospital Universitario Reina Sofia, Córdoba; Instituto Maimonides de Investigación Biomédica (IMIBIC); Universidad de Córdoba, Córdoba; 26 Centro Oncológico de Galicia, A Coruña, Coruña; 27 Medical Oncology, Hospital Universitario Donostia-Biodonostia, San Sebastián; 28 Medical Oncology, Hospital de León, León, Spain; 29 Pfizer, New York, USA; 30 Repare Therapeutics, Cambridge, USA; 31 Pfizer, Milano, Italy; 32 Medical Oncology, Hospital Clínico Universitario San Carlos, Madrid; 33 Medical Oncology, Hospital Virgen de la Salud, Toledo, Spain; 34 Ralph Lauren Centre for Breast Cancer Research, Royal Marsden, London, UK; 35 Department of Oncology, Medical University of Vienna, Department of Oncology, Vienna, Austria; 36 Institut Català d’Oncologia (ICO) & IDIBELL, L’Hospitalet, Barcelona, Spain Available online XXX Background: Palbociclib plus endocrine therapy (ET) is the standard treatment of hormone receptor-positive and human epidermal growth factor receptor 2-negative, metastatic breast cancer (MBC). However, its efficacy has not been compared with that of chemotherapy in a phase III trial. Patients and methods: PEARL is a multicentre, phase III randomised study in which patients with aromatase inhibitor (AI)-resistant MBC were included in two consecutive cohorts. In cohort 1, patients were randomised 1 : 1 to palbociclib plus exemestane or capecitabine. On discovering new evidence about estrogen receptor-1 (ESR1) mutations inducing resistance to AIs, the trial was amended to include cohort 2, in which patients were randomised 1 : 1 between palbociclib plus fulvestrant and capecitabine. The stratification criteria were disease site, prior sensitivity to ET, prior chemotherapy for MBC, and country of origin. Co-primary endpoints were progression-free survival (PFS) in cohort 2 and in wild-type ESR1 patients (cohort 1 þcohort 2). ESR1 hotspot mutations were analysed in baseline circulating tumour DNA. Results: From March 2014 to July 2018, 296 and 305 patients were included in cohort 1 and cohort 2, respectively. Palbociclib plus ET was not superior to capecitabine in both cohort 2 [median PFS: 7.5 versus 10.0 months; adjusted *Correspondence to: Professor Miguel Martín, Medical Oncology, Instituto de Investigación Sanitaria Gregorio Marañón, CIBERONC-ISCIII GEICAM Spanish Breast Cancer Group, Doctor Esquerdo, 46, Madrid, Spain, 28007. Tel: þ34-91659-28-70 E-mail: [email protected] (M. Martin). y These authors contributed equally to this study. 5 This study has been previously presented at San Antonio Breast Cancer Symposium; 10-14 December 2019; San Antonio, TX. Philadelphia (PA): AACR; Published at Cancer Res 2020;80(4 Suppl): Abstract number GS2-07. Cancer Res February 14 2020;80 (4 Supplement) GS2-07-GS2-07; https://doi.org/10.1158/1538-7445. SABCS19-GS2-07 Published February 2020. Final PFS results were presented as a poster discussion at the American Society of Clinical Oncology (ASCO) virtual meeting and the quality-of-life data has been presented as a poster at the virtual European Society of Medical Oncology (ESMO) Breast Cancer Meeting. 0923-7534/© 2020 The Author(s). Published by Elsevier Ltd on behalf of European Society for Medical Oncology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Volume xxx -Issue xxx -2021 https://doi.org/10.1016/j.annonc.2020.12.013 1 hazard ratio (aHR): 1.13; 95% confidence interval (CI): 0.85-1.50] and wild-type ESR1 patients (median PFS: 8.0 versus 10.6 months; aHR: 1.11; 95% CI: 0.87-1.41). The most frequent grade 3-4 toxicities with palbociclib plus exemestane, palbociclib plus fulvestrant and capecitabine, respectively, were neutropenia (57.4%, 55.7% and 5.5%), hand/foot syndrome (0%, 0% and 23.5%), and diarrhoea (1.3%, 1.3% and 7.6%). Palbociclib plus ET offered better quality of life (aHR for time to deterioration of global health status: 0.67; 95% CI: 0.53-0.85). Conclusions: There was no statistical superiority of palbociclib plus ET over capecitabine with respect to PFS in MBC patients resistant to AIs. Palbociclib plus ET showed a better safety profile and improved quality of life. Key words: palbociclib, capecitabine, metastatic breast cancer, hormone receptor-positive, HER2-negative, endocrine therapy INTRODUCTION Until recently, single-agent endocrine therapy (ET) was the recommended choice of treatment of most women with hormone receptor-positive and human epidermal growth factor receptor 2 (HER2)-negative metastatic breast cancer (MBC). Unfortunately, not all patients respond to ET due to primary or acquired resistance. In the past decade, new targeted therapies, mainly cyclin-dependent kinase 4/6 (CDK4/6) inhibitors, in combination with ET have significantly improved progression-free survival (PFS) 1-7 and overall survival (OS) 8-10 compared with ET alone in patients with treatment-naive or pretreated MBC. The PALOMA-3 trial 4 showed that palbociclib plus fulvestrant significantly improved PFS as opposed to fulvestrant plus placebo [hazard ratio (HR): 0.46; P<0.0001] in patients who experienced cancer relapse or progression during or within 12 months of completing adjuvant ET or while they were on ET or within 1 month of prior ET for MBC. Consequently, palbociclib plus fulvestrant was approved by the Food and Drug Administration and the European Medicines Agency for these patients. That trial showed that adding palbociclib to fulvestrant significantly delayed disease progression compared with fulvestrant alone in patients resistant to aromatase inhibitors (AIs). However, we still considered it necessary to analyse the efficacy differences between palbociclib plus ET and other current standards of care in MBC patients resistant to AIs, such as chemotherapy. In 2014, the GEICAM Spanish Breast Cancer Group started the PEARL trial in collaboration with the Central European Cooperative Oncology Group (CECOG). This trial compared palbociclib plus ET with capecitabine in a population of postmenopausal patients very similar to those in the PALOMA-3 trial. We selected capecitabine as the chemotherapy agent as it is considered to be one of the most active drugs available for MBC, with median PFS ranging from 2.8 to 5.9 months (which was even higher in patients with hormone receptor-positive disease) and OS times of 9.3-18.1 months in previously treated MBC patients. 11-14 We combined palbociclib with exemestane in the initial study design; however, after the emerging evidence that patients pretreated with AIs may develop ESR1 mutations that generate resistance to AIs, we introduced a second cohort in which palbociclib was combined with fulvestrant. 15,16 METHODS Study design The PEARL trial (clinical trial registration number: ClinTrials. gov reference NCT02028507), a multicentre, international, open-label, controlled, randomised phase III study with two successive cohorts of similar characteristics, was carried out in four countries (37 sites): Spain (GEICAM), Austria, Hungary, and Israel (CECOG). Cohort 1 patients were randomised 1 : 1 to receive palbociclib (125 mg/day for 3 weeks followed by 1 week off) plus exemestane (25 mg/ day) or capecitabine [according to the approved label: 2500 mg/m 2 /day (2000 mg/m 2 /day in patients aged >70 years old) for 2 weeks followed by 1 week off]. The study hypothesis endorsed the superiority of palbociclib plus exemestane over capecitabine (expected PFS, HR: 0.686, with a 5% significance level). In December 2015, new data suggested that exemestane in patients who have progressed on AIs could be a suboptimal option because ESR1 mutations may confer AI therapy resistance in patients previously exposed to AIs (with a frequency of mutations of 29%-37%). 15-17 One of the studies suggested that fulvestrant may be effective in patients with ESR1 mutationpositive tumours. 16 In May 2016, a protocol amendment with a modification of trial design and objectives was approved before any efficacy data were available. Therefore, a subsequent cohort 2 was introduced, in which patients were randomised 1 : 1 to receive palbociclib (same schedule as cohort 1) plus fulvestrant (500 mg intramuscular injection on days 1 and 15 of cycle 1; then on day 1 of subsequent 28-day cycles) or capecitabine (same schedule as cohort 1). At that time, 296 patients were already recruited in cohort 1 (from an initial planned sample of 348 patients). The new study hypotheses endorsed the superiority of palbociclib plus fulvestrant over capecitabine and palbociclib plus ET over capecitabine in patients with wild-type ESR1 (expected PFS, HR: 0.667, with a 5% significance level) (Supplementary Material S1, available at https://doi. org/10.1016/j.annonc.2020.12.013). Randomisation was carried out centrally at the GEICAM headquarters. In both the cohorts, stratification criteria were disease site (visceral/non-visceral), sensitivity to prior ET [relapse after 24 months of adjuvant ET or response (complete or partial) or stabilisation after 24 weeks of the most recent ET in the context of advanced disease (yes/ Annals of Oncology M. Martin et al. 2https://doi.org/10.1016/j.annonc.2020.12.013 Volume xxx -Issue xxx -2021 no)], prior chemotherapy for MBC (yes/no), and country of origin. The treatment continued until either objective disease progression, according to the RECIST v1.1, 18 symptomatic deterioration, unacceptable toxicity, death, or withdrawal of consent, whichever occurred first. As-perprotocol dose reductions of palbociclib and capecitabine were allowed in case of toxicity. Upon completion of the study treatment, patients were monitored for survival every 6 months. Research protocol was approved by every site’s institutional review board and every country’s regulatory agency. All the patients signed written informed consents. Safety and efficacy data were continuously evaluated by an independent data monitoring committee. The data were analysed by a statistician employed by GEICAM. Patients Postmenopausal women with hormone receptor-positive and HER2-negative AI-resistant MBC (defined as recurrence: while on or within 12 months after the end of adjuvant treatment or progression; while on or within 1 month after the end of treatment of advanced disease) were included. Patients had to have measurable disease assessable by computed tomography (CT)/magnetic imaging resonance (MRI) according to RECIST v1.1 or at least one lytic or mixed bone lesion. One chemotherapy line for MBC was permitted. Additional inclusion criteria included Eastern Cooperative Oncology Group performance status (ECOG) of 0 or 1, life expectancy of 12 weeks or more, and adequate organ function. Patients who received prior treatment with CDK4/6, mammalian target of rapamycin (mTOR) or phosphoinositide 3-kinase (PI3K) inhibitors, capecitabine, or patients with visceral crisis were excluded. Patients were required to have a corrected QT interval (QTc) <480 ms and no family or personal history of long or short QT syndrome, Brugada syndrome, Torsade de Pointes, or known history of QTc prolongation. Trial assessments Baseline disease assessments (carried out within 4 weeks before randomisation), required a CT or MRI scan of the chest, abdomen, and pelvis. Assessments were carried out every 8 weeks for 120 weeks and then every 12 weeks until documented progressive disease, initiation of a new anticancer therapy, or patient dropout. Patients who discontinued study treatment for reasons other than progressive disease had tumour assessments every 12 weeks. Haematology and biochemistry tests were carried out before each cycle; haematology testing was additionally carried out on day 14 of cycles 1 and 2 in the palbociclib arms. Adverse events (AEs) were assessed and graded at each cycle according to National Cancer Institute common terminology criteria for adverse events (NCI-CTCAE) version 4.0. Patients completed the European Organization for Research and Treatment of Cancer core quality-of-life (EORTC QLQ-C30; v3.0), 19 BC-specific (EORTC QLQ-BR23; v1.0), 20 and the EuroQoL Health Utilities Index EQ-5D-3L 21 questionnaires at baseline, at every two cycles for the first seven cycles, then at every three cycles until the end of treatment, and once again at the visit after treatment. Patients were required to have a mandatory plasma sample drawn for exploratory biomarker analyses in circulating tumour DNA (ctDNA) obtained before treatment onset. With the protocol amendment to include cohort 2, the ESR1 mutational status assessment was a predefined analysis required to evaluate the primary objective of the study. The results were blinded to patients and investigators (Supplementary Material S2, available at https://doi.org/ 10.1016/j.annonc.2020.12.013). In addition, formalin-fixed paraffin-embedded tumour samples were collected before study entry to genetically identify intrinsic BC subtypes (Luminal A and B, HER2enriched, basal-like, and normal-like) using the HTG EdgeSeq Oncology Biomarker Panel (Supplementary Material S3, available at https://doi.org/10.1016/j.annonc.2020.12.013). Objectives and endpoints The initial primary objective was to compare PFS with palbociclib plus exemestane and that with capecitabine treatment. After the protocol amendment to include cohort 2, the two new co-primary objectives were to compare PFS of patients treated with (i) palbociclib plus fulvestrant versus capecitabine regardless of ESR1 mutational status and (ii) palbociclib plus ET (exemestane or fulvestrant) versus capecitabine in patients with wild-type ESR1 in ctDNA at study entry. PFS was defined as the time from randomisation to the first documentation of progressive disease based on investigators’assessments according to RECIST v1.1 or to death from any cause. Secondary objectives included, among others, PFS with palbociclib plus ET versus capecitabine regardless of ESR1 mutational status, objective response rate (ORR), clinical benefit rate (CBR) (defined as ORR plus stable disease rate of at least 24 weeks duration), response duration (RD), OS, safety, and patient-reported outcomes (PROs). Concerning PROs, we reported the time to deterioration for the global health status from the EORTC QLQ-C30, defined as the time from randomisation to first detection of a deterioration event (marked with a decrease of 10 points from the baseline). Additionally, we explored the independent prognostic and predictive value of intrinsic subtypes. Statistical analysis A total of 193 PFS events were required in cohort 2 to have 80% power to detect a difference between capecitabine (estimated median PFS of 6 months) and palbociclib plus fulvestrant (median PFS of 9 months 4 ), for an HR of 0.667, with a 5% significance level. The target sample size was 300 patients. To detect the same difference between M. Martin et al. Annals of Oncology Volume xxx -Issue xxx -2021 https://doi.org/10.1016/j.annonc.2020.12.013 3 capecitabine and palbociclib plus ET in patients with wildtype ESR1 and assuming an 80% ctDNA collection/detection rate and 30% of the patients with ESR1 mutations, the required sample size was also 300 patients. The study was designed to have two interim analyses and a final analysis. The final PFS analysis was planned when 193 events in cohort 2 were observed. A modification of Hochberg’s method 22 was used for two primary treatment comparisons to provide the control of experiment-wise type 1 error rate at a two-sided 5% significance level. The KaplaneMeier method was used to estimate the median PFS; 95% confidence intervals (CIs) were provided for estimates of interest. The Cox proportional-hazards model was used to calculate the unadjusted and adjusted HR (aHR) (by stratification factors and number of involved sites) and 95% CI. Efficacy analyses were based on two populations: all randomised patients [intention-to-treat (ITT) population] and all randomised patients with wildtype ESR1 in ctDNA at study entry (wild-type ESR1 population). Safety analysis was carried out on all patients who received one or more dose of study therapy. PROs analysis was carried out on patients with baseline and one or more quality of life (QoL) questionnaires completed. Time to deterioration was analysed using Cox regression models. RESULTS Patients and treatment A total of 601 patients were included in this study from March 2014 to July 2018. Cohort 1 included 296 patients (153 on palbociclib plus exemestane and 143 on capecitabine) and cohort 2 included 305 patients (149 on palbociclib plus fulvestrant and 156 on capecitabine). Efficacy analyses included all patients, but safety analyses excluded 13 patients (10 on capecitabine and 3 on palbociclib plus ET) never receiving study treatment. ESR1 mutations were assessed in 557 patients (92.7%), 91% of who were from the capecitabine arms and 94% were from the palbociclib plus ET arms; 164 of them (29%) had ESR1 mutations (Figure 1). All the baseline demographics and disease characteristics were balanced between the arms across both the cohorts, except for the number of involved sites (greater in the capecitabine arm in cohort 2) (Table 1). Cohort 1 N = 296 Randomised N = 601 Cohort 2 N = 305 Palbociclib + exemestane N = 153 Capecitabine N = 143 Palbociclib + fulvestrant N = 149 Capecitabine N = 156 Screening failure N = 92 No treatment N = 3 No treatment N = 6 No treatment N = 4 Registered N = 693 Mutant N = 48 (32.9%) WT N = 98 (67.1%) Mutant N = 38 (27.1%) WT N = 102 (72.8%) Mutant N = 37 (29.4%) WT N = 89 (70.6%) Mutant N = 41 (28.3%) WT N = 104 (71.7%) Safety N = 150 (98.0%) Safety N = 137 (95.8%) ESR1a N = 145 (94.8%) ESR1b N = 126 (88.1%) Safety N = 149 (100%) ESR1c N = 140 (94%) Safety N = 152 (97.4%) ESR1d N = 146 (93.6%) Figure 1. Consort diagram. ESR1, estrogen receptor 1; WT, wild-type ESR1. a No treatment n¼2 and sample not available n¼6. b No treatment n¼6 and sample not available n¼11. c Sample not available n¼9. d No treatment n¼3 and sample not available n¼7. Annals of Oncology M. Martin et al. 4https://doi.org/10.1016/j.annonc.2020.12.013 Volume xxx -Issue xxx -2021 Table 1. Patients’baseline characteristics (intention-to-treat population) Variables Cohort 1 Cohort 2 Palbociclib plus exemestane Capecitabine Palbociclib plus fulvestrant Capecitabine n¼153 n¼143 Pvalue n¼149 n¼156 Pvalue Demographics and disease characteristics Median age, years (range) 60 (31-89) 60 (38-87) 0.5574 62 (38-86) 60 (33-85) 0.2618 ECOG performance status c ,n(%) 0 85 (55.6) 84 (58.7) 0.5800 90 (60.4) 93 (59.6) 0.8884 1 68 (44.4) 59 (41.3) 59 (39.6) 63 (40.4) Visceral disease, n(%) Yes 103 (67.3) 94 (65.7) 0.8379 97 (65.1) 102 (65.4) 0.9585 No 50 (32.7) 48 (33.6) 52 (34.9) 54 (34.6) Most frequent disease sites, n(%) Bone 107 (69.9) 101 (70.6) 0.8960 97 (65.1) 114 (73.1) 0.1316 Liver 67 (43.8) 61 (42.7) 0.8441 60 (40.3) 68 (43.6) 0.5569 Breast/skin/subcutaneous/lymph node 62 (40.5) 72 (50.3) 0.0896 63 (42.3) 84 (53.8) 0.0433 a,b Lung 44 (28.8) 38 (26.6) 0.6747 40 (26.8) 44 (28.2) 0.7905 Pleura 19 (12.4) 20 (14.0) 0.6903 12 (8.1) 23 (14.7) 0.0669 Number of involved sites, n(%) 1 47 (30.7) 32 (22.4) 0.2196 56 (37.6) 35 (22.4) 0.0147 a 2 62 (40.5) 59 (41.3) 48 (32.2) 60 (38.5) 3 44 (28.8) 51 (35.7) 45 (30.2) 61 (39.1) Tumour characteristics Hormone receptor status, n(%) d ERþPRþ114 (74.5) 103 (72.0) 114 (76.5) 118 (75.6) ERþPRe36 (23.5) 38 (26.6) 33 (22.2) 33 (21.2) ERePRþor ERþPR not available e 2 (1.3) 2 (1.4) 2 (1.3) 5 (3.2) ESR1 mutational status, n(%) Wild-type 104 (68.0) 89 (62.2) 0.8434 102 (68.5) 98 (62.8) 0.2905 Mutant 41 (26.8) 37 (25.9) 38 (25.5) 48 (30.8) Not available 8 (5.2) 17 (11.9) 9 (6.0) 10 (6.4) Sensitivity to prior endocrine therapy, n(%) f Yes 107 (69.9) 104 (72.7) 0.5956 119 (79.9) 122 (78.2) 0.7218 No 46 (30.1) 39 (27.3) 30 (20.1) 34 (21.8) Genomic subtype, n(%) g n[117 n[107 n[112 n[119 Luminal A 61 (52.1) 61 (57.0) 58 (51.8) 52 (43.7) Luminal B 49 (41.9) 42 (39.3) 43 (38.4) 58 (48.7) HER2-enriched 5 (4.3) 4 (3.7) 11 (9.8) 9 (7.6) Basal-like 2 (1.7) 0 0 0 Prior therapy Number of prior lines of endocrine therapy for MBC, n(%) No prior endocrine therapy for MBC 30 (19.6) 31 (21.7) 38 (25.5) 44 (28.2) 1 82 (53.6) 70 (49.0) 85 (57.0) 90 (57.7) 2 35 (22.9) 34 (23.8) 12 (8.1) 9 (5.8) 3 3 (2.0) 4 (2.8) 1 (0.7) 1 (0.6) Maintenance after chemotherapy 3 (2.0) 4 (2.8) 12 (8.1) 12 (7.7) Prior endocrine therapy for MBC, n(%) Aromatase inhibitor 106 (69.3) 105 (73.4) 0.4308 111 (74.5) 109 (69.9) 0.3679 Fulvestrant 44 (28.8) 35 (24.5) 0.4052 0 1 (0.6) 0.3276 Other selective estrogen receptor degrader 0 0 - 2 (1.3) 1 (0.6) 0.5350 Tamoxifen 16 (10.5) 17 (11.9) 0.6960 12 (8.1) 16 (10.3) 0.5054 Luteinising hormone-releasing hormone analogues 10 (6.5) 11 (7.7) 0.6986 8 (5.4) 14 (9.0) 0.2238 Prior chemotherapy for MBC, n(%) Yes 48 (31.4) 41 (28.7) 0.6125 41 (27.5) 41 (26.3) 0.8079 No 105 (68.6) 102 (71.3) 108 (72.5) 115 (73.7) Line at study entry, h n(%) 1st line 27 (17.6) 31 (21.7) 0.5498 38 (25.5) 43 (27.6) 0.8477 2nd line 61 (39.9) 50 (35.0) 76 (51.0) 79 (50.6) 3rd line 62 (40.5) 62 (43.3) 35 (23.5) 34 (21.8) Continued M. Martin et al. Annals of Oncology Volume xxx -Issue xxx -2021 https://doi.org/10.1016/j.annonc.2020.12.013 5 At the cut-off date for the primary analysis (14 January 2019), 80 patients were still on the study treatment: 10 (6.7%) were on palbociclib plus exemestane, 37 (24.8%) on palbociclib plus fulvestrant, and 33 (11%) on capecitabine. The median relative dose-intensity in cohort 1 was 82.6% for capecitabine, 100% for exemestane, and 95.2% for palbociclib, and that in cohort 2 was 79.5% for capecitabine, 100% for fulvestrant, and 92.9% for palbociclib. The median time on study therapy in cohort 1 was higher for capecitabine, 7.9 months (range: 0.2-50.5), than for palbociclib plus exemestane, 6.3 months (range: 0.5-52.3). However, in cohort 2 the median time on study therapy was 6.3 months for capecitabine (range: 0.2-26.4) and 7.8 months for palbociclib plus fulvestrant (range: 0.8-31.1). The main reason for permanent discontinuation of the treatment was disease progression. In both cohorts, the proportion of patients who discontinued due to progressive disease was smaller in the capecitabine arm (65.7% in cohort 1, 58.6% in cohort 2) than in the palbociclib plus exemestane (81.3%) and palbociclib plus fulvestrant arms (68.5% in cohort 2) (Supplementary Table S1, available at https://doi.org/10. 1016/j.annonc.2020.12.013). Efficacy The median follow-ups of cohort 2 and the wild-type ESR1 population were 13.5 months (range: 0.0-30.7) and 18.9 months (range: 0.0-56.3), respectively. The median PFS in cohort 2 was 7.5 months (95% CI: 5.7-10.9) in the palbociclib plus fulvestrant arm and 10.0 months (95% CI: 6.312.9) in the capecitabine arm (aHR: 1.13; 95% CI: 0.85-1.50; P¼0.398). The median PFS in the wild-type ESR1 population was 8.0 months (95% CI: 6.5-10.9) in the palbociclib plus ET arm and 10.6 months (95% CI: 7.4-13.0) in the capecitabine arm (aHR: 1.11; 95% CI: 0.87-1.41; P¼0.404) (Figure 2). PFS subgroup analyses by stratification factors and other baseline characteristics in cohort 2 and in the wild-type ESR1 population (Figure 3), as well as in the overall population regardless of ESR1 mutational status (Supplementary Figure S1, available at https://doi.org/10. 1016/j.annonc.2020.12.013) confirmed the non-superiority of palbociclib plus ET over capecitabine. Regarding the study’s secondary endpoints of efficacy, the median PFS in all patients from cohort 1 and cohort 2 was 7.4 months (95% CI: 5.9-9.3) in the palbociclib plus ET arm and 9.4 months (95% CI: 7.5-11.3) in the capecitabine arm (aHR: 1.11; 95% CI: 0.92-1.34; P¼0.380) (Supplementary Figure S2, available at https://doi.org/10. 1016/j.annonc.2020.12.013). The aHR for PFS in the mutant ESR1 population was 1.12 (95% CI: 0.78-1.60; P¼ 0.540) as shown in Supplementary Figure S3, available at https://doi.org/10.1016/j.annonc.2020.12.013. The ORR in cohort 2 was 26.7% for palbociclib plus fulvestrant versus 33.3% for capecitabine. In patients with ESR1 wild-type, ORR was 27.8% for palbociclib plus ET versus 36.9% for capecitabine. The CBR was very similar between the arms in cohort 2 and the patients with ESR1 wild-type. The median RD in cohort 2 was 9.4 months in the palbociclib plus fulvestrant arm and 12.9 months in the capecitabine arm (HR: 0.69; 95% CI: 0.33-1.46; P¼0.335). Finally, the median RD in the wild-type ESR1 population was 9.7 months in the palbociclib plus ET arm and 11.2 months in the capecitabine arm (HR: 0.75; 95% CI: 0.44-1.25; P¼0.269) (Supplementary Table S2, available at https://doi.org/10. 1016/j.annonc.2020.12.013). PROs The completion rate of the questionnaires was similar across the arms, surpassing 82% until cycle 13. The median time to deterioration in global health status was 8.6 months in patients treated with palbociclib plus ET versus 6.2 months in those treated with capecitabine (aHR: 0.67, 95% CI: 0.53-0.85; P¼0.001) (Figure 4). Table 1. Continued Variables Cohort 1 Cohort 2 Palbociclib plus exemestane Capecitabine Palbociclib plus fulvestrant Capecitabine n¼153 n¼143 Pvalue n¼149 n¼156 Pvalue Status at initial diagnosis, n(%) M0 127 (83.0) 109 (76.2) 0.1469 115 (77.2) 120 (76.0) 0.9573 M1 (de novo MBC) 26 (17.0) 34 (23.8) 34 (22.8) 36 (23.1) Pvalue statistically significant. ER, estrogen receptor; ESR1, estrogen receptor 1; ET, endocrine therapy; HER2, human epidermal growth factor receptor 2; Lum, luminal; MBC, metastatic breast cancer; PR, progesterone receptor. a No significant differences in patients’baseline characteristics were identified between treatment groups, except for the number of involved sites and breast/skin/subcutaneous/ lymph node disease site in cohort 2. b Hazard ratios were not adjusted by ‘breast/skin/subcutaneous/lymph node’as disease site, because this item is included within stratification factor (visceral versus non-visceral). c Eastern Cooperative Oncology Group (ECOG) performance status scores range from 0 to 5, with 0 indicating no symptoms and higher scores indicating greater disability. d Based on local laboratory determination, positive defined as 1% positive cells by immunohistochemistry for ER and/or PR. Hormone receptor status was evaluated on primary tumours in 62.1% of patients and on metastatic lesions in 37.9% of them. e One patient treated with exemestane þpalbociclib was triple-negative (protocol deviation). f Sensitivity to prior endocrine therapy was defined as relapse after 24 months of adjuvant ET or response (complete or partial) or stabilisation after 24 weeks of the most recent ET in the context of advanced disease. g By HTG EdgeSeq Oncology Biomarker Panel. h Line at study entry means the treatment line received in the study, considering all prior lines of therapy, either chemotherapy and/or endocrine therapy. Annals of Oncology M. Martin et al. 6https://doi.org/10.1016/j.annonc.2020.12.013 Volume xxx -Issue xxx -2021 Safety Safety information is shown in Table 2 and Supplementary Table S3, available at https://doi.org/10.1016/j.annonc. 2020.12.013. The most frequent grade 3-4 toxicities in the palbociclib plus exemestane, palbociclib plus fulvestrant, and capecitabine arms, were neutropenia [(57.4%, 55.7%, 5.5%, respectively) with febrile neutropenia (1.3%, 0.7%, 1.4%, respectively)], hand/foot syndrome (0%, 0%, 23.5%, respectively), diarrhoea (1.3%, 1.3%, 7.6%, respectively), fatigue (1.3%, 0.7%, 5.5%, respectively), and anaemia (0.7%, 2.0%, 3.5%, respectively). The incidence of nonhaematologic toxicity grade 3 was higher for patients on capecitabine (38.8%) than for those on palbociclib plus exemestane (6.7%) or palbociclib plus fulvestrant (6.0%). Notably, grade 1-2 alopecia was reported in 11.0% of the 0 6 12 18 24 30 Time (months) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Progression-free survival probability 156 70 39 12 3 0 149 84 35 18 5 1 Capecitabine Palbo + fulve ACohort 2 No. of patients No. of events (%) No. of censored (%) PFS median no. of months (95% CI) Palbociclib + fulvestrant 149 108 (72.5) 41 (27.5) Capecitabine 156 94 (60.3) 62 (39.7) No. at risk Palbociclib + fulvestrant Capecitabine 0 6 12 18 24 30 36 42 48 Time (months) 187 101 58 27 16 12 6 3 1 206 116 67 41 19 12 7 6 5 Capecitabine Palbo + ET B No. at risk Capecitabine Palbociclib + ET Wild-type ESR1 (Cohort 1 + Cohort 2) Adjusted hazard ratio (95% CI): 1.13 (0.85-1.50), P = 0.398 7.5 (5.7-10.9) 10.0 (6.3-12.9) No. of patients No. of events (%) No. of censored (%) PFS median no. of months (95% CI) Palbociclib + ET 206 161 (78.2) 45 (21.8) Capecitabine 187 126 (67.4) 61 (32.6) 8.0 (6.5-10.9) 10.6 (7.4-13.0) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Progression-free survival probability Adjusted hazard ratio (95% CI): 1.11 (0.87-1.41), P = 0.404 Figure 2. Progression-free survival. KaplaneMeier curves for PFS were represented for (A) patients in cohort 2: palbociclib plus fulvestrant versus capecitabine and (B) patients with wild-type ESR1 from cohort 1 þcohort 2: palbociclib plus endocrine therapy versus capecitabine. Hazard ratios were adjusted by disease site, prior sensitivity to endocrine therapy, prior chemotherapy for metastatic breast cancer, and the number of involved sites. CI, confidence interval; ESR1, estrogen receptor 1; ET, endocrine therapy; fulve, fulvestrant; No, number; Palbo, palbociclib; PFS, progression-free survival. M. Martin et al. Annals of Oncology Volume xxx -Issue xxx -2021 https://doi.org/10.1016/j.annonc.2020.12.013 7 patients on palbociclib plus ET as opposed to 3.8% of the patients on capecitabine. Serious AEs related to the study treatment were reported by 10.4% of the patients on capecitabine, 4.0% of the patients on palbociclib plus exemestane, and 3.4% of the patients on palbociclib plus fulvestrant. A total of 46 patients on capecitabine (15.9%) dropped out due to AEs compared with 9 patients (6%) on palbociclib plus exemestane and 10 patients (6.7%) on palbociclib plus fulvestrant. Of the 17 deaths observed during the study treatment, 11 were due to progressive disease; two serious AEs Subgroup 0.996 0.165 0.634 0.335 0.516 0.678 0.302 0.927 0.260 0.514 0.878 0.078 0.423 0.652 0.191 0.425 0.25 0.5 1 1.5 2 3 1.00 (0.75-1.34) 1.33 (0.89-1.99) 1.07 (0.81-1.42) 1.22 (0.81-1.85) 1.16 (0.74-1.83) 1.06 (0.81-1.39) 1.16 (0.87-1.54) 0.98 (0.65-1.48) 1.30 (0.82-2.07) 1.10 (0.83-1.45) 1.02 (0.79-1.33) 1.63 (0.95-2.80) 1.21 (0.76-1.95) 0.92 (0.65-1.31) 1.34 (0.86-2.08) 1.10 (0.87-1.39) (70.4) (63.4) (63.6) (76.4) (71.7) (66.0) (67.2) (67.7) (64.4) (68.8) (68.5) (64.9) (63.2) (74.3) (64.8) (67.4) 81/115 45/71 84/132 42/55 33/46 93/141 84/125 42/62 29/45 97/141 102/149 24/37 36/57 55/74 35/54 126/187 (78.8) (76.8) (75.0) (86.2) (77.2) (78.5) (85.4) (65.8) (73.0) (81.1) (75.7) (89.2) (75.5) (80.2) (78.3) (78.2) 108/137 53/69 111/148 50/58 44/57 117/149 111/130 50/76 54/74 107/132 128/169 33/37 37/49 77/96 47/60 161/206 Visceral Non-visceral Prior sensitivity to ET: yes Prior sensitivity to ET: no Prior CT for MBC: yes Prior CT for MBC: no Age <65 years Age ≥65 years One site of disease Multiple sites of disease Measurable lesions Non-measurable lesions Treatment line: 1st Treatment line: 2nd Treatment line: ≥3rd ALL Palbociclib + ET better Capecitabine better Palbociclib + ET Capecitabine Palbociclib + fulvestrant Cohort 2 (n = 305) A Wild-Type ESR1 (Cohort 1 + cohort 2) (n = 393) B Capecitabine Hazard ratio (95% CI)Events/N (%)Events/N (%) Hazard ratio (95% CI)Events/N (%)Events/N (%) Subgroup 0.811 0.497 0.974 0.057 0.784 0.419 0.566 0.814 0.155 0.786 0.630 0.558 0.503 0.515 0.786 0.880 0.734 0.599 0.25 0.5 11.5 2 3 1.04 (0.75-1.45) 1.19 (0.72-1.98) 0.99 (0.73-1.36) 1.79 (0.98-3.28) 0.93 (0.55-1.56) 1.15 (0.82-1.59) 1.10 (0.79-1.55) 1.06 (0.65-1.72) 1.54 (0.85-2.81) 1.05 (0.76-1.44) 1.08 (0.79-1.47) 1.20 (0.65-2.24) 0.83 (0.47-1.45) 1.14 (0.77-1.67) 1.09 (0.60-1.96) 1.03 (0.73-1.44) 1.10 (0.64-1.87) 1.08 (0.82-1.42) (48.1) (59.0) (64.7) (68.3) (57.4) (61.2) (58.5) (42.9) (65.3) (61.9) (53.3) (59.5) (64.1) (59.4) (61.2) (62.5) (60.3) (66.7)68/102 26/54 72/122 22/34 28/41 66/115 63/103 31/53 15/35 79/121 78/126 16/30 25/42 50/78 19/32 60/98 30/48 94/156 (69.2) (70.6) (80.0) (75.6) (71.3) (78.5) (62.5) (67.9) (75.3) (69.8) (81.8) (68.4) (71.1) (80.0) (74.5) (65.8) (72.5) (74.2) 72/97 36/52 84/119 24/30 31/41 77/108 73/93 35/56 38/56 70/93 81/116 27/33 26/38 54/76 28/35 76/102 25/38 108/149 Visceral Non-visceral Prior sensitivity to ET: yes Prior sensitivity to ET: no Prior CT for MBC: yes Prior CT for MBC: no Age <65 years Age ≥65 years One site of disease Multiple sites of disease Measurable lesions Non-measurable lesions Treatment line: 1st Treatment line: 2nd Treatment line: ≥3rd Wild-type ESR1 Mutant ESR1 ALL Palbociclib + fulvestrant better Capecitabine better P valuea P valueb Figure 3. Forest plot of progression-free survival hazard ratios by subgroups. Subgroups for progression-free survival and their respective hazard ratios were represented for (A) patients in cohort 2: palbociclib plus fulvestrant versus capecitabine and (B) patients with wild-type ESR1 from cohort 1 þcohort 2: palbociclib plus endocrine therapy versus capecitabine. Pvalues from ManneWhitney test (continuous variables) and chi-square test (categorical variables). CI, confidence interval; CT, chemotherapy; ESR1, estrogen receptor 1; ET, endocrine therapy; MBC, metastatic breast cancer; ALL, all patients of Cohort 2 (Figure A) all patients Wild-Type ESR1 (Cohort 1 + cohort 2) (Figure B). a Unadjusted Cox Pvalue comparing palbociclib plus fulvestrant versus capecitabine in each subgroup. b Unadjusted Cox Pvalue comparing palbociclib plus endocrine therapy versus capecitabine in each subgroup. Annals of Oncology M. Martin et al. 8https://doi.org/10.1016/j.annonc.2020.12.013 Volume xxx -Issue xxx -2021 occurred while patients were on palbociclib plus ET (pneumonitis and sepsis), and four occurred while the patients were on capecitabine (diarrhoea, general health status worsening, colitis, and sudden death). Diarrhoea, general health status worsening, and colitis were considered toxic deaths according to the investigators’ assessments. Exploratory objectives Prognostic/predictive value of intrinsic BC subtypes. Subtypes were obtained for 455 patients (94.4% of the 482 patients assessed) with metastatic (30%) or primary tumour tissue (70%) available (Table 1); 75.7% of cohort 2 and 79.6% of the wild-type ESR1 patient population. Most patients (93.2%) had luminal tumours. Cohort 2 patients with luminal tumours showed a median PFS of 7.7 and 10 months with palbociclib plus fulvestrant and capecitabine, respectively (HR: 1.07; 95% CI: 0.77-1.49; P¼0.681). Patients with non-luminal tumours (n¼20) had a median PFS of 3.3 and 13.7 months with palbociclib plus fulvestrant and capecitabine, respectively (HR: 5.87; 95% CI: 1.60-21.55; P¼0.008). Patients with wild-type ESR1 luminal tumours presented a median PFS of 9.3 and 11.0 months with palbociclib plus fulvestrant and capecitabine, respectively (HR: 1.01; 95% CI: 0.77-1.33; P¼0.930). Patients with nonluminal tumours (n¼25) on palbociclib plus ET and capecitabine had a median PFS of 2.3 and 13.7 months, respectively (HR: 7.36; 95% CI: 2.05-26.37; P¼0.002) (Supplementary Figure S4, available at https://doi.org/10. 1016/j.annonc.2020.12.013). DISCUSSION The PEARL trial did not provide evidence of PFS superiority of palbociclib plus fulvestrant or of palbociclib plus ET in patients without ESR1 mutations over capecitabine in AIresistant MBC patients. However, it is worth noting that compared with capecitabine, palbociclib plus ET was associated with a significant delay in QoL deterioration, less treatment discontinuations due to AEs, and a lower proportion of patients with related serious AEs. The initial study design of the PEARL trial was modified after some compelling evidence that ESR1 mutations (present in up to 37% of patients pretreated with AIs) could produce resistance to additional AI therapy, but not to fulvestrant. 15-17 Since in the initial design the endocrine arm was exemestane plus palbociclib, we added a second cohort of patients in which the endocrine arm was fulvestrant plus palbociclib, to avoid the potential negative influence of ESR1 mutations in patients treated with AIs. In fact, we identified 29% of ESR1 mutations in the patients included in this trial. Of note, this modification was made before any results were available. The combination of palbociclib plus fulvestrant has been approved by several regulatory agencies for the treatment Adjusted hazard ratio (95% CI): 0.67 (0.53-0.85), P = 0.001 No. of patients No. of events (%) No. of censored (%) TTD median no. of months (95% CI) Palbociclib + ET 279 131 (47.0) 140 (53.0) Capecitabine 273 153 (56.0) 120 (44.0) 8.6 (6.4-11.3) 6.2 (4.2-9.6) 0 6 12 18 24 30 Time (months) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 273 95 37 18 8 6 279 108 46 27 11 7 Capecitabine Palbo + ET Capecitabine Palbociclib + ET No. at Risk Event-free probability Figure 4. Time to deterioration based on EORTC core quality-of-life-C30_global health status. The figure shows the median time to deterioration for the global health status from the European Organization for Research and Treatment of Cancer core quality-oflife-C30 (EORTC QLQ-C30). The adjusted hazard ratio was obtained using a stratified Cox proportional hazard model with treatment arm, the stratification factors (visceral, sensitivity to prior ET, prior CT for MBC), and number of involved sites as covariates. CI, confidence interval; CT, chemotherapy; EORTC, European Organization for Research and Treatment of Cancer; ET, endocrine therapy; MBC, metastatic breast cancer; No., number; Palbo, palbociclib; TTD, time to deterioration. M. Martin et al. Annals of Oncology Volume xxx -Issue xxx -2021 https://doi.org/10.1016/j.annonc.2020.12.013 9