Germline Mutation Status, Pathological Complete Response, and Disease-Free Survival in Triple-Negative Breast Cancer: Secondary Analysis of the GeparSixto Randomized Clinical Trial

Key Points

Question  Does BRCA1 and BRCA2 germline mutation status predict therapy response in patients with triple-negative breast cancer enrolled in the GeparSixto trial?

Findings  In this secondary analysis of a randomized clinical trial of 291 patients with triple-negative breast cancer, patients with BRCA1 and BRCA2 mutations showed superior response rates, without additive effects observed for carboplatin. Patients without BRCA1 and BRCA2 germline mutations benefited from the addition of carboplatin to a regimen of paclitaxel, low-dose doxorubicin, and bevacizumab.

Meaning  A less-intense treatment regimen might be considered for BRCA1 and BRCA2 mutation carriers, but further prospective studies are needed to identify the optimal regimen.

Importance  The GeparSixto trial provided evidence that the addition of neoadjuvant carboplatin to a regimen consisting of anthracycline, taxane, and bevacizumab increases pathological complete response (pCR) rates in patients with triple-negative breast cancer (TNBC). Whether BRCA1 and BRCA2 germline mutation status affects treatment outcome remains elusive.

Objective  To determine whether BRCA1 and BRCA2 germline mutation status affects therapy response in patients with TNBC.

Design, Setting, and Participants  This secondary analysis of a randomized clinical trial used archived DNA samples and cancer family history of 315 patients with TNBC enrolled between August 1, 2011, and December 31, 2012, in the GeparSixto trial. In all, 291 participants (92.4%) were included in this multicenter prospective investigation. DNA samples were analyzed for germline mutations in BRCA1, BRCA2, and 16 other cancer predisposition genes. The pCR rates between the carboplatin and noncarboplatin arms were compared. Genetic analyses were performed at the Center for Familial Breast and Ovarian Cancer in Cologne, Germany; data analysis, November 1 through December 31, 2015.

Main Outcomes and Measures  Proportion of patients who achieved pCR and disease-free survival after neoadjuvant treatment according to BRCA1 and BRCA2 germline mutation status. For pCR rates, the ypT0/is ypN0 definition was used as a primary end point.

Results  Of the 291 patients with TNBC, all were women; the mean (SD) age was 48 (11) years. The pCR rate in the carboplatin group was 56.8% (83 of 146) and 41.4% (60 of 145) in the noncarboplatin group (odds ratio [OR], 1.87; 95% CI, 1.17-2.97; P = .009). Pathogenic BRCA1 and BRCA2 germline mutations were present in 50 of the 291 patients (17.2%). In the noncarboplatin arm, the pCR rate was 66.7% (16 of 24) for patients with BRCA1 and BRCA2 mutations and 36.4% (44 of 121) for patients without (OR, 3.50; 95% CI, 1.39-8.84; P = .008). The high pCR rate observed in BRCA1 and BRCA2 mutation carriers (16 of 24 [66.7%]) was not increased further by adding carboplatin (17 of 26 [65.4%]). In contrast, carboplatin increased response rates in patients without BRCA1 and BRCA2 mutations: 66 of the 120 patients (55%) without BRCA1 and BRCA2 mutations achieved pCR in the carboplatin arm vs 44 of the 121 patients (36.4%) in the noncarboplatin arm (OR, 2.14; 95% CI, 1.28-3.58; P = .004). Patients without pathogenic BRCA1 and BRCA2 alterations showed elevated disease-free survival rates when carboplatin was added (without carboplatin, 73.5%; 95% CI, 64.1%-80.8% vs with carboplatin, 85.3%; 95% CI, 77.0%-90.8%; hazard ratio, 0.53; 95% CI, 0.29-0.96; P = .04).

Conclusions and Relevance  Under the nonstandard GeparSixto polychemotherapy regimen, patients without BRCA1 and BRCA2 germline mutations benefited from the addition of carboplatin and those with BRCA1 and BRCA2 mutations showed superior response rates without additive effects observed for carboplatin.

Trial Registration  clinicaltrials.gov Identifier: NCT01426880

The GeparSixto randomized clinical trial¹ assessed the efficacy of adding neoadjuvant carboplatin to a regimen of paclitaxel, non–pegylated doxorubicin hydrochloride, and targeted therapy for triple-negative breast cancer (TNBC) and ERBB2/HER2 (OMIM 164870)–positive breast cancer. Targeted therapy included lapatinib and trastuzumab for ERBB2/HER2-positive breast cancer and bevacizumab for TNBC. Of the patients with TNBC, 90 of 158 (57%) achieved a pathological complete response (pCR) with carboplatin therapy compared with 67 of 157 patients (42.7%) without carboplatin therapy (P = .015; ypT0/is ypN0 definition).¹ Of the patients with ERBB2/HER2-positive tumors, 72 of 137 (52.6%) achieved a pCR with carboplatin compared with 67 of 136 patients (49.3%) without carboplatin (P = .58; ypT0/is ypN0 definition).¹ Thus, the addition of neoadjuvant carboplatin to the anthracycline and taxane–containing regimen substantially increased pCR rates in patients with TNBC but not in patients with ERBB2/HER2-positive breast cancer. The GeparSixto trial used a nonstandard neoadjuvant chemotherapy regimen that included low-dose doxorubicin and no cyclophosphamide. In the Cancer and Leukemia Group B (CALGB 40603 Alliance) trial, standard neoadjuvant chemotherapy (paclitaxel, dose-dense doxorubicin, and cyclophosphamide) of patients with TNBC revealed elevated pCR rates when carboplatin was added,² but an event-free survival benefit was not observed.³

The triple-negative tumor phenotype accounts for up to 17% of all breast cancers⁴ and appears to be associated with a hereditary disease cause. Approximately 70% of breast cancers arising in BRCA1 (OMIM 113705) mutation carriers and up to 23% of breast cancers in BRCA2 (OMIM 600185) carriers are triple negative.⁵ In line with these findings, mutational screening of TNBC cases for deleterious germline mutations in BRCA1 and BRCA2 revealed comparatively high mutation frequencies. While germline BRCA1 and BRCA2 mutations were found in 5.3% of all breast cancers according to The Cancer Genome Atlas,⁶ a recent study showed that 11.2% of unselected TNBC cases had deleterious mutations in the BRCA1 (8.5%) and BRCA2 (2.7%) genes. Mutations in additional 15 non–BRCA1 and BRCA2 cancer predisposition genes were detected in 3.7% of the patients.⁷

BRCA1 and BRCA2 are critical genes in the homologous recombination repair of double-stranded DNA breaks. Many of the other genes involved in homologous recombination repair are now recognized to also contribute to hereditary breast cancer risk and/or ovarian cancer risk, including ATM, BRIP1, CHEK2, NBN, PALB2, RAD51C, and RAD51D; limited evidence is available for BARD1, FANCM, MRE11A, and RAD50.^8-12 Among these genes, only BRCA1, BRCA2, and PALB2 so far have been associated with the TNBC tumor phenotype.⁸ Heterozygous germline inactivation of homologous recombination genes may be accompanied by a somatic inactivation of the second allele by mutation, loss of heterozygosity, or promoter methylation and result in a homologous recombination deficiency and limited DNA repair capacities of the tumor cells.¹³ This functional role in DNA repair could be exploited in the treatment of homologous recombination–deficient cancers by targeting the tumors with drugs that create DNA damage that is highly reliant on these genes for repair.¹⁴ There is increasing evidence that breast and ovarian cancers arising in BRCA1 and BRCA2 germline mutation carriers are associated with a better response to DNA-damaging treatment regimens.^9,15-19 These data prompted us to conduct this prospective-retrospective secondary analysis of the germline mutation status using archived DNA samples and cancer family history of patients with TNBC enrolled in the GeparSixto trial.

The GeparSixto trial cohort, randomization process, clinical procedures, and statistical analyses were described in the initial trial publication.¹ Of the 315 patients with TNBC enrolled between August 1, 2011, and December 31, 2012, in the GeparSixto study, 24 (7.6%) were excluded from this secondary analysis because of unavailable or insufficient amounts of DNA samples (Figure 1). Genomic DNA samples isolated from venous blood samples were derived from the other 291 patients with TNBC (92.4%) and were successfully analyzed for germline mutations. The treatment regimen for these 291 patients is shown in Figure 1. Data on cancer family history for all 291 patients were available and were considered positive when the inclusion criteria of the German Consortium for Hereditary Breast and Ovarian Cancer for genetic germline testing were fulfilled (eTable 1 in the Supplement). Data analysis took place from November 1, 2015, to December 31, 2015. Ethical approval for this secondary analysis was granted by the ethics committee of the University of Cologne, and written informed consent was obtained from all patients.

Genetic analyses were performed at the Center for Familial Breast and Ovarian Cancer in Cologne, Germany, the coordinating entity of the German Consortium for Hereditary Breast and Ovarian Cancer involved in diagnostic BRCA1 and BRCA2 germline testing since 1996.²⁰ The diagnostic pipeline is certified by the European Molecular Genetics Quality Network. Genomic DNA samples were isolated from venous blood samples using standard methods. All samples (n = 291) were screened for gross genomic aberrations in the BRCA1 and BRCA2 genes by multiplex ligation-dependent probe amplification (MLPA) using probe mixes (SALSA MLPA probe mixes P002 [BRCA1] and P045 [BRCA2]; MRC-Holland) according to the manufacturer’s protocol. Data were analyzed using the Coffalyzer.Net software, version 140429.1057 (MRC-Holland). All BRCA1 and BRCA2 deletions or duplications were verified using probe mixes (SALSA MLPA probe mixes P087 [BRCA1] and P077 [BRCA2]; MRC-Holland). In parallel, all samples were screened for the predominant pathogenic BRCA1 and BRCA2 mutations identified within the framework of the German Consortium for Hereditary Breast and Ovarian Cancer. For these analyses, a customized single-nucleotide polymorphism (SNP) genotyping assay (SNP Type Assay; Fluidigm) covering 90 distinct BRCA1 and BRCA2 alterations was established (eTable 2 in the Supplement). Specific target amplification was performed according to the assay manufacturer’s potocol using 25 ng of genomic DNA. Samples and SNP type assay mixes were loaded on nanofluid chips (96.96 Dynamic Arrays; Fluidigm), run on a thermocycler (FC1-Cycler; Fluidigm), and analyzed using a fluorecence imager (EP1-System; Fluidigm). For variant calling, the SNP Genotyping Analysis Software, version 3.1.3 (Fluidigm) was used. All mutations identified by this approach were verified by Sanger sequencing.

All samples that tested negative for pathogenic BRCA1 and BRCA2 mutations by MLPA/SNP type assay (n = 250) were subsequently analyzed by next-generation sequencing covering the entire coding regions and exon-flanking sequences (±25 nucleotides) of BRCA1, BRCA2, and 16 non–BRCA1 and BRCA2 cancer predisposition genes (ATM [OMIM 607585], BARD1 [OMIM 601593], BRIP1 [OMIM 605882], CDH1 [OMIM 192090], CHEK2 [OMIM 604373], FANCM [OMIM 609644], MRE11A [OMIM 600814], NBN [OMIM 602667], PALB2 [OMIM 610355], PTEN [OMIM 601728], RAD50 [OMIM 604040], RAD51C [OMIM 602774], RAD51D [OMIM 602954], STK11 [OMIM 602216], TP53 [OMIM 191170], and XRCC2 [OMIM 600375]).⁸ For next-generation sequencing, a customer-tailored gene panel protocol optimized for 200 ng of genomic DNA was used (SureSelect^XT Target Enrichment for Illumina Paired-End Multiplexed Sequencing; Agilent Technologies). Sequencing was performed using a sequencing platform (HiSeq 2000; Illumina). Bioinformatic analyses were carried out using the VARBANK, version 2.10 pipeline of the Cologne Center for Genomics. A detailed description of the variant calling is given in eTable 3 in the Supplement. Variant classification was performed in accordance with the regulations of the international ENIGMA consortium (https://enigmaconsortium.org).

The primary outcome of this biomarker study was the proportion of patients who achieved a pCR and disease-free survival (DFS) after neoadjuvant treatment according to BRCA1 and BRCA2 germline mutation status and family history (eTable 1 in the Supplement). Regarding the pCR rates, the ypT0/is ypN0 definition was used as a primary end point and the more stringent ypT0 ypN0 definition as a secondary end point.²¹ Disease-free survival was defined according to the description by Hudis and colleagues²² as time in months from randomization until any invasive locoregional (ipsilateral breast, local/regional lymph nodes) recurrence of disease, any invasive contralateral breast cancer, any distant recurrence of disease, any secondary malignant neoplasm, or death from any cause, whichever occurs first. Disease progression under therapy was not considered as an event for DFS. Patients without an event (236 of 291 [81%]) were censored at the date of their last contact with the GeparSixto study.

The Pearson χ² test was used to compare pCR rates between groups. Univariate logistic regressions were performed to estimate odds ratios (ORs) and 95% CIs. Multivariate logistic regressions adjusting for baseline variables (age, tumor stage, nodal status, grading, Ki67 staining level, and BRCA risk according to family history) were performed, including interaction between mutation status and carboplatin treatment. The Kaplan-Meier product-limit method was used to estimate DFS. A Cox proportional hazards model was used to estimate hazard ratios (HRs) and 95% CIs, with a 2-sided Wald P value.

Detailed data on cancer family history and blood-derived DNA samples were available from 291 of 315 patients (92.4%) with TNBC enrolled in the GeparSixto trial. Of the 291 patients with TNBC, 100% were women, with a mean (SD) age of 48 (11) years. Compatible with the data on the initially published entire cohort (n = 315), the pCR rate (ypT0/is ypN0 definition) in the carboplatin group was 56.8% (83 of 146 patients) and was 41.4% (60 of 145) in the noncarboplatin group (OR, 1.87; 95% CI, 1.17-2.97; P = .009; Table 1 and Table 2), with differences reaching levels of significance in the multivariable analyses (OR, 2.08; 95% CI, 1.19-3.63; P = .01; eTable 4 in the Supplement).

We screened for deleterious germline mutations in BRCA1, BRCA2, and 16 non–BRCA1 and BRCA2 cancer predisposition genes. Besides homologous recombination and other DNA repair genes, cancer predisposition genes not belonging to the DNA repair machinery (CDH1, PTEN, STK11, and TP53), all of which were associated with rare cancer predisposition syndromes, were included.⁸ Pathogenic BRCA1 mutations were present in 43 of 291 cases (14.8%), BRCA2 mutations were found in 7 of 291 cases (2.4%), and another 10 cases (3.4%) carried deleterious non–BRCA1 and BRCA2 gene mutations (BARD1 [n = 1], FANCM [n = 2], MRE11A [n = 1], NBN [n = 2], RAD50 [n = 2], RAD51C [n = 1], and XRCC2 [n = 1]). Overall, 50 of 291 patients (17.2%) with TNBC carried germline BRCA1 and BRCA2 alterations, and 10 of 291 patients (3.4%) carried deleterious alterations in other putative cancer predisposition genes (eTable 3 in the Supplement). Because of the small number of patients with deleterious mutations in non–BRCA1 and BRCA2 genes (n = 10) and the unclear association of these genes with the TNBC tumor phenotype, we refrained from calculating pCR rates for this small and heterogeneous subgroup and considered these patients as BRCA1 and BRCA2–negative.

For 23 of 50 patients (46%) with pathogenic BRCA1 and BRCA2 mutations, a disease onset before age 40 years was observed, compared with only 42 of 241 patients (17.4%) without deleterious BRCA1 and BRCA2 alterations (eTable 5 in the Supplement). An association with a positive BRCA1 and BRCA2 mutation status was also observed for family history: 31 of 50 patients (62%) carrying pathogenic BRCA1 and BRCA2 mutations reported a positive family history, compared with 79 of 241 patients (32.8%) without (eTable 5 in the Supplement). No significant associations were observed between BRCA1 and BRCA2 mutation status and tumor stage at baseline, nodal status at baseline, grading, or Ki67 staining level (eTable 5 in the Supplement). In the noncarboplatin arm, 16 of 24 patients (66.7%) with BRCA1 and BRCA2 mutation showed a pCR, compared with only 44 of 121 patients (36.4%) without BRCA1 and BRCA2 mutations (OR, 3.50; 95% CI, 1.39-8.84; P = .008; Table 1). Adding carboplatin did not enhance overall pCR rates in the subgroup of germline BRCA1 and BRCA2 mutation carriers: 17 of 26 patients (65.4%) carrying the BRCA1 and BRCA2 mutations achieved a pCR with adjuvant carboplatin therapy compared with 16 of 24 patients (66.7%) without carboplatin therapy (Table 1). In an interaction test, the interaction between the mutation and carboplatin revealed nonsignificant results (OR, 0.68; 95% CI, 0.17-2.68; P = .58; eTable 4 in the Supplement). The overall increased pCR rate with carboplatin therapy appeared to be driven by elevated response rates in patients with TNBC not carrying germline BRCA1 and BRCA2 mutations: patients with TNBC without pathogenic BRCA1 and BRCA2 alterations showed a 36.4% response rate (44 of 121 patients), which increased to 55% (66 of 120) when carboplatin was added to the regimen (OR, 2.14; 95% CI, 1.28-3.58; P = .004; Tables 1 and 2).

Patients with TNBC without a family history of cancer (181 of the 291 patients), a subgroup not enriched for BRCA1 and BRCA2 mutation carriers (19 of 181 patients [10.5%]; eTable 5 in the Supplement), showed significantly elevated pCR rates in the carboplatin arm, which increased from 37% (34 of 92 patients) without carboplatin to 53.9% (48 of 89) with carboplatin (OR, 2.0; 95% CI, 1.10-3.62; P = .02; eTable 6 in the Supplement). Thus, the absence of a positive cancer family history is associated with carboplatin response in this trial. In all, 110 of 291 patients (37.8%) with TNBC reported a positive cancer family history (eTable 6 in the Supplement). Patients with a positive cancer family history, a subgroup enriched for BRCA1 and BRCA2 germline mutation carriers (31 of 110 [28.2%]; eTable 5 in the Supplement), showed comparatively high response rates in the noncarboplatin arm (26 of 53 [49.1%]), which moderately increased to 61.4% (35 of 57) when carboplatin was added (OR, 1.65; 95% CI, 0.77-3.53; P = .19; eTable 6 in the Supplement). Of note, 9 of 10 patients carrying non–BRCA1 and BRCA2 germline mutations did not show a cancer family history, and 4 of 10 patients achieved a pCR (eTable 7 in the Supplement).

With a median follow-up of 35 months, superior DFS rates were observed in the carboplatin group vs the noncarboplatin group (HR, 0.55; 95% CI, 0.32-0.95; P = .03; Figure 2A). These data were compatible with the early DFS rates described for the entire TNBC cohort (n = 315).²³ Patients with TNBC without pathogenic BRCA1 and BRCA2 alterations showed elevated DFS rates when carboplatin was added to the treatment regimen: without carboplatin, 73.5%; 95% CI, 64.1%-80.8%; with carboplatin, 85.3%; 95% CI 77.0%-90.8% (HR, 0.53; 95% CI, 0.29-0.96; P = .04; Figure 2B). Regardless of the treatment regimen, the DFS rate was generally high in BRCA1 and BRCA2 mutation carriers, with differences separated by study arm that did not reach levels of significance: without carboplatin, 82.5%; 95% CI, 59.6%-93.1%; with carboplatin, 86.3%; 95% CI, 63.1%-95.4% (Figure 2B). We observed a significant correlation of pCR rates with DFS rates (log-rank P < .001) irrespective of the BRCA1 and BRCA2 mutation status (eFigure in the Supplement).

In the noncarboplatin arm, this investigation suggests superior pCR rates in patients with germline BRCA1 and BRCA2 mutations compared with patients without BRCA1 and BRCA2 mutations, with differences translating into clinical benefit when considering the respective DFS rates. This finding may be the result of better treatment response of BRCA1 and BRCA2 mutation carriers to either of the chemotherapeutic agents used in the noncarboplatin arm. The primary mechanism of action of doxorubicin is thought to be via DNA intercalation and stabilization of the topoisomerase IIa/DNA complex, ultimately promoting the formation of single-stranded and double-stranded DNA breaks.^24,25 Thus, it appears plausible that BRCA1 and BRCA2 mutation carriers achieve higher response rates under therapy with antracyclines because of limited DNA repair capacities of the tumors. On the basis of initial data presented by the randomized Triple Negative Trial,²⁶BRCA1 and BRCA2 mutation status is unlikely to be correlated with therapy response to docetaxel. The Triple Negative Trial included patients with metastatic disease, and pretreatment may have changed their responsiveness to chemotherapeutic agents. However, the trial did not suggest that the differences observed between BRCA1 and BRCA2 mutation carriers and noncarriers in the noncarboplatin arm were driven by treatment with taxane. Bevacizumab, a vascular endothelial growth factor inhibitor, has been shown to elevate pCR rates in patients with TNBC (OR, 1.36; 95% CI, 1.11-1.66).²⁷ Vascular endothelial growth factor expression in tumors of BRCA1 and BRCA2 mutation carriers is demonstrably higher than in sporadic tumors.²⁸ Concordantly, an investigation of the GeparQuinto trial recently revealed higher pCR rates in response to bevacizumab therapy for BRCA1 and BRCA2 mutation carriers than for noncarriers. However, a superior outcome of survival probability could not be demonstrated.²⁹

Because the TNBC tumor phenotype is closely associated with hereditary breast cancer,⁵ the use of platinum agents has received a new impetus. The cytotoxic actions of platinum drugs are mediated by covalent binding of platinum to DNA, interfering with DNA replication and transcription and ultimately inducing cell death.³⁰ It seems likely that partially processed cross-links cause replication fork stalling when encountered by the DNA replication machinery during S phase, which may degenerate into double-stranded DNA breaks.³¹ Thus, tumor cells with limited DNA repair capacities are hypersensitive against platinum, as demonstrated in preclinical studies.³² In a recent neoadjuvant trial, platinum-based chemotherapy was shown to be highly effective in BRCA1 germline mutation carriers: a total of 50 of 82 patients (61%) with TNBC experienced a pCR following cisplatin single-agent therapy.¹⁵ The high sensitivity of BRCA1 and BRCA2 mutation carriers to platin-based chemotherapy is in line with data in the metastatic or recurrent locally advanced setting in which carboplatin monotherapy revealed significantly higher response rates in BRCA1 and BRCA2 mutation carriers than in patients without BRCA1 and BRCA2 mutations.²⁶ In summary, carboplatin therapy has been proven effective in BRCA1 and BRCA2 mutation carriers.

The GeparSixto treatment regimen for TNBC cases included 2 DNA-damaging compounds, doxorubicin and carboplatin, both of which challenge the DNA repair machinery. This finding could explain why, in germline mutation carriers, the addition of carboplatin to the treatment regimen does not further increase pCR or DFS rates above those observed for the combination of paclitaxel, doxorubicin, and bevacizumab. Given that the more intense regimen significantly increases hematological and nonhematological adverse effects in the GeparSixto trial,¹ our findings may have implications for personalized therapy regimens that consider BRCA1 and BRCA2 germline mutation status. Because of the lack of additive effects observed in this study and the finding that elevated pCR rates translate into a clinical benefit,²¹ a less intense therapy regimen might be considered for BRCA1 and BRCA2 germline mutation carriers.

This study has several limitations. First, the GeparSixto trial was not powered for long-term end points such as DFS and overall survival. Second, there was a small number of patients carrying BRCA1 and BRCA2 (n = 50), especially non–BRCA1 and BRCA2 gene mutations (n = 10). Thus, additional trials are necessary to assess the clinical benefit of a combination use of DNA-damaging compounds, especially in BRCA1 and BRCA2 mutation carriers.

In contrast to the GeparSixto trial, the CALGB 40603 trial revealed that the addition of carboplatin to a standard neoadjuvant chemotherapy (paclitaxel, dose-dense doxorubicin, and cyclophosphamide) did not result in an event-free survival benefit.^2,3 The GeparSixto trial used a nonstandard neoadjuvant chemotherapy regimen, including low-dose doxorubicin and no cyclophosphamide. Thus, the differences in response rates between the GeparSixto and CALGB 40603 trials might be caused by the different doxorubicin (low dose vs dose dense) and/or cyclophosphamide exposures. It would be interesting to stratify the CALGB 40603 response rates by BRCA1 and BRCA2 mutation status. It appears likely that BRCA1 and BRCA2 mutation carriers show superior response rates following standard neoadjuvant chemotherapy, while patients without BRCA1 and BRCA2 mutations may benefit from the addition of carboplatin. Byrski and colleagues^15,33 suggested that a chemotherapy regimen with doxorubicin and cyclophosphamide or platinum may result in the highest benefit for BRCA1 germline mutation carriers. This suggestion is based on a limited number of patients, but future trials may evaluate this hypothesis. Under the nonstandard GeparSixto polychemotherapy regimen, however, patients without BRCA1 and BRCA2 germline mutations benefit from the addition of carboplatin while those with BRCA1 and BRCA2 mutations show superior response rates without additive effects observed for carboplatin. Additional prospective studies stratified by BRCA1 and BRCA2 mutation status are needed to elucidate the effect of carboplatin in polychemotherapy regimens.

Corresponding Author: Eric Hahnen, PhD, Center for Hereditary Breast and Ovarian Cancer, University Hospital Cologne, Kerpener Strasse 34, 50931 Cologne, Germany (eric.hahnen@uk-koeln.de).

Accepted for Publication: February 24, 2017.

Published Online: July 13, 2017. doi:10.1001/jamaoncol.2017.1007

Author Contributions: Drs Nekljudova and Hahnen had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Hahnen, Loibl, Kröber, Schneeweiss, Blohmer, Jackisch, Gerber, Kümmel, Huober, Costa, Hanusch, Untch, von Minckwitz, Schmutzler.

Acquisition, analysis, or interpretation of data: Hahnen, Lederer, Hauke, Loibl, Kröber, Schneeweiss, Denkert, Fasching, Blohmer, Jackisch, Paepke, Gerber, Kümmel, Schem, Neidhardt, Rhiem, Costa, Altmüller, Hanusch, Thiele, Müller, Nürnberg, Karn, Nekljudova, Untch, von Minckwitz, Schmutzler.

Drafting of the manuscript: Hahnen, Lederer, Loibl, Schneeweiss, Jackisch, Gerber, Schem, Hanusch, Müller, Untch.

Critical revision of the manuscript for important intellectual content: Hahnen, Hauke, Loibl, Kröber, Schneeweiss, Denkert, Fasching, Blohmer, Jackisch, Paepke, Gerber, Kümmel, Schem, Neidhardt, Huober, Rhiem, Costa, Altmüller, Hanusch, Thiele, Müller, Nürnberg, Karn, Nekljudova, Untch, von Minckwitz, Schmutzler.

Statistical analysis: Lederer, Loibl, Denkert, Jackisch, Thiele, Nekljudova.

Obtained funding: Gerber, Schmutzler.

Administrative, technical, or material support: Hahnen, Lederer, Kröber, Schneeweiss, Denkert, Fasching, Blohmer, Jackisch, Paepke, Gerber, Huober, Rhiem, Costa, Altmüller, Nürnberg, Untch, von Minckwitz, Schmutzler.

Study supervision: Hahnen, Schneeweiss, Denkert, Jackisch, Paepke, Gerber, Kümmel, Nürnberg, Untch, von Minckwitz, Schmutzler.

Conflict of Interest Disclosures: Drs Loibl and von Minckwitz report that their institution has received research funding from Amgen, AstraZeneca, Celgene, Hexal, Novartis, Pfizer, Roche, and Teva. No other disclosures were reported.

Funding/Support: The molecular genetic analyses were supported by grant No. 110837 from the German Cancer Aid.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We are thankful to all patients who participated in this study.