A phase II study of talazoparib monotherapy in patients with wild-type BRCA1 and BRCA2 with a mutation in other homologous recombination genes

Abstract

Talazoparib, a PARP inhibitor, is active in germline BRCA1 and BRCA2 (gBRCA1/2)-mutant advanced breast cancer, but its activity beyond gBRCA1/2 is poorly understood. We conducted Talazoparib Beyond BRCA ( NCT02401347 ), an open-label phase II trial, to evaluate talazoparib in patients with pretreated advanced HER2-negative breast cancer (n = 13) or other solid tumors (n = 7) with mutations in homologous recombination (HR) pathway genes other than BRCA1 and BRCA2. In patients with breast cancer, four patients had a Response Evaluation Criteria in Solid Tumors (RECIST) partial response (overall response rate, 31%), and three additional patients had stable disease of ≥6 months (clinical benefit rate, 54%). All patients with germline mutations in PALB2 (gPALB2; encoding partner and localizer of BRCA2) had treatment-associated tumor regression. Tumor or plasma circulating tumor DNA (ctDNA) HR deficiency (HRD) scores were correlated with treatment outcomes and were increased in all gPALB2 tumors. In addition, a gPALB2-associated mutational signature was associated with tumor response. Thus, talazoparib has been demonstrated to have efficacy in patients with advanced breast cancer who have gPALB2 mutations, showing activity in the context of HR pathway gene mutations beyond gBRCA1/2.

Conflict of interest statement

J.J.G. had contracted research with Curis, consults for Sharma Therapeutics, LLC and Guidepoint and reports research support (to his institution) from Hummingbird Biosciences. J.M.F. has received research support (to his institution) from AstraZeneca, Genentech, Incyte, Merus, Pfizer and PUMA. M.L.T. reports research support (to her institution) from AbbVie, Arvinas, Bayer, Biothera, Calithera Biosciences, EMD Serono, Genentech, GlaxoSmithKline, Hummingbird Biosciences, Medivation, Merck, Novartis, OncoSec, Pfizer, PharmaMar, Tesaro and Vertex and consulting/advisory fees from AbbVie, Aduro Biotech, AstraZeneca, Blueprint Medicines, Celgene, Daiichi Sankyo, Genentech/Roche, Gilead, GlaxoSmithKline, G1 Therapeutics, Guardant, Immunomedics, Lilly, Merck, Natera, Novartis, OncoSec, Pfizer, RefleXion and Sanofi. Funding support was provided by Pfizer in the form of an unrestricted research grant for drug and clinical trial budget. V.N.A., H.W., M.B. and H.S. are full-time employees of Natera with stock or options to own stock in the company. K.T. is an employee and shareholder of Myriad Genetics. The remaining authors declare no competing interests.

© 2022. The Author(s).

Figures

Fig. 1. Treatment response and duration for therapy for all patients.

a, Waterfall plot of best change in the sum of longest diameters (SLD) of target lesions by RECIST v.1.1 for all patients (n = 20) treated in cohort B, colored by tumor type. Germline (g) or somatic (s) mutations in genes used for enrollment are indicated. The dashed line represents a 30% decrease in tumor size. b, Duration of therapy by swimmers plot indicating duration of talazoparib therapy. Tumor responses assessments are noted. n = 20 patients. Source data

Fig. 2. Tumor HRD scores and correlation with treatment response.

HRD score was obtained by NGS assay (Myriad myChoice) from tumor biopsies (n = 15 patients; 2 patients excluded for insufficient sample and 2 patients excluded for assay failure). a, HRD score is plotted in primary and metastatic samples for all evaluable patients with lines connecting paired samples. Horizontal dashed lines indicate an HRD score threshold of ≥33 (captures 99% of known BRCA1- and/or BRCA2-deficient ovarian cancers) or ≥42 (captures 95% of known BRCA1- and/or BRCA2-deficient ovarian cancers). Significance testing by paired t-test for primary versus metastatic HRD scores. b, Correlation of HRD score with talazoparib treatment response. HRD score is plotted against best change in SLD by RECIST. Pearson’s correlation is indicated by the solid line, with associated r and two-sided P value denoted. Vertical dashed lines indicate an HRD threshold of ≥33 or ≥42. In cases where patients had more than one HRD score (for example, due to assay of primary and metastatic tumors), the higher score was used. Source data

Fig. 3. Somatic mutations and LOH identified by NGS.

Heatmap of next-generation panel sequencing (108 genes interrogated) for HRD-associated and cancer-associated mutations in tumor specimens from patients treated in cohort B (n = 18 patients, 2 excluded due to insufficient sample). Germline (g) and somatic (s) mutations used for study enrollment are indicated. Hierarchical clustering is by Euclidean distance. Bar graphs depict numbers of mutations of each type detected across the cohort. Tumor type is indicated by color. See Supplementary Table 3 for a list of genes assayed. Source data

Fig. 4. ctDNA variants and ctHRD score correlating with treatment outcomes.

a, Variants detected by Signatera assay of baseline and progression ctDNA samples plotted by VAF. Dashed lines represent fivefold difference cutoffs, n = 14 patients. b, pWES-identified indels (n = 18 patients) at baseline and progression. c, SNVs identified by pWES at baseline and progression (n = 18 patients). d, Deleterious SNVs enriched fivefold in progression samples (n = 6 patients) from responders (decrease of >20% in SLD) compared to baseline. Variants identified in tumor WES are shown (an anterisk indicates insufficient sample for tumor WES). e, ctDNA SNV counts are significantly increased in progression samples compared to control samples (n = 18 patients), whereas there is no significant difference in indel counts (two-way ANOVA with Tukey’s correction, *P = 0.033; median indicated by horizontal line). NS, not significant. f, scarHRD was used to calculate ctHRD score from baseline pWES and correlated with time to progression in days. The correlation coefficient (r) and two-sided P value were calculated with Pearson’s method and are depicted as a dashed line. n = 12 patients. Source data

Fig. 5. Mutational signature analysis and correlation with treatment outcome.

a, De novo mutational signatures were computed using all available mutation data including tumor exome and plasma ctDNA exome sequencing. The predominant signature detected in the gPALB2 samples is shown. n = 19 patients. b, Mutational signatures were binned by whether they included the gPALB2-associated signature (HRD) or other signatures and plotted as a percentage of total SNVs. n = 19 patients. c, Treatment response to talazoparib (best change in SLD by RECIST) is plotted according to percentage contribution of HRD signature derived from tumor exome and plasma ctDNA sequencing. Correlation by Pearson’s method is depicted by the dashed line, with the two-sided P value. n = 19 patients. Source data

Extended Data Fig. 1. Talazoparib Beyond BRCA study schema for cohort B.

Cohort B enrolled patients with advanced HER2-negative breast cancer or other solid tumors lacking a germline BRCA1 or BRCA2 mutation with either a germline or somatic mutation in a gene associated with HR-deficiency. Patients were treated with single-agent talazoparib in an initial cohort of 10 patients. On the basis of observed responses in the first 10 patients, an additional 10 patients were enrolled and treated.

Extended Data Fig. 2. Overlap of variant detection by tumor WES and plasma WES.

Number of detected variants by tumor WES (red), plasma WES at baseline (green) and plasma WES at progression (blue) depicted as proportional venn diagrams. Patient study number and qualifying genetic mutation are above and below, respectively. Source data

Extended Data Fig. 3. Deleterious ctDNA SNVs enriched at progression in non-responders.

Deleterious ctDNA SNVs enriched 5-fold in progression samples from non-responders (change of SLD > -20%) compared to baseline. Deleterious variants identified in tumor WES are also shown. N = 9 patients. Source data