An open-label, pilot study of veliparib and lapatinib in patients with metastatic, triple-negative breast cancer

We conducted a single institution, open-label pilot trial in which eligible patients received veliparib and lapatinib for the treatment of metastatic TNBC. The primary objective was patient safety and tolerability. This study was conducted at The University of Alabama at Birmingham (Birmingham, AL) and was approved by the UAB Institutional Review Board. Written informed consent was obtained from patients prior to any study procedure. The study was registered on ClinicalTrials.​gov as NCT02158507.

Patients with biopsy-proven, metastatic or locally advanced, unresectable TNBC (ER negative, PR negative, HER-2- Neu negative by IHC or FISH) breast cancer (BC) were eligible for participation. Additional eligibility requirements included age ≥ 19 years, ECOG PS of 0–2, adequate organ and marrow function as defined by the study protocol, a left ventricular ejection fraction ≥ 50%, and measurable disease per RECIST criteria v1.1. Negative serum or urine beta-HCG pregnancy test was necessary for patients of child-bearing ages, in addition to an agreement to use an effective means of contraception in this population. Patients were required to have had prior anthracycline and taxane use in the neoadjuvant, adjuvant, or metastatic setting. Patients with more than two prior regimens for metastatic disease were ineligible. Although not mandatory for study entry, a biopsy was requested in patients with reasonably accessible lesions.

All study participants received genetic counseling and had germline testing if indicated as part of the standard of care. No pathogenic germline alterations were found in DNA repair genes. Two participants did not meet guidelines for genetic testing and therefore had low likelihood of harboring germline alterations. Furthermore, this study enrolled patients from 2014 to 2017, a time in which NGS testing was not standard of care at our institution. Therefore, tumor mutation status was not available at the time of enrollment.

Patients were excluded if they had a history of hypersensitivity to veliparib, lapatinib, or drugs of similar chemical composition. Concomitant investigational or commercial agents or therapies administered with the intent to treat the patient’s malignancy were not allowed, with the exception of bisphosphonates and RANKL inhibitors for bone metastases. Patients with germline BRCA1 or BRCA2 mutations, prolonged QTc interval > 470 ms, and untreated or uncontrolled brain metastases were excluded. In the presence of brain metastases, patients must have had at least one site of measurable disease outside of the central nervous system. Those with prior invasive malignant disease within 5 years (with the exception of in situ skin or cervical cancers), history of HIV, or hepatitis B were excluded.

Veliparib was provided by AbbVie and lapatinib was supplied by Novartis Pharmaceuticals Corporation.

Within 4 weeks prior to initiation of therapy, patients underwent baseline clinical evaluation, including medical history, physical exam, performance status, laboratory evaluation, radiographic tumor measurement, EKG, and echocardiogram. Patients with an accessible metastatic site underwent a biopsy.

Twenty patients with evaluable disease were initially enrolled and treated with lapatinib 1250 mg a day by mouth continuously for 28 days, starting at cycle 1 day 1, in combination with veliparib 200 mg every 12 h by mouth for 28 consecutive days, starting on cycle 1 day 2. A cycle of therapy was defined as 28 days. Treatment was administered on an outpatient basis (Fig. 1). Treatment continued in patients with complete response (CR), partial response (PR), or stable disease (SD) until progression of the disease or unacceptable toxicity. SD was defined per protocol as no evidence of progressive disease for > 2 cycles.

Patients were evaluated for response every 2 courses (every 8 weeks) using CT scans according to the RECIST criteria version 1.1. An EKG was obtained on day 1 of each cycle and an echocardiogram was obtained every 3 months. Patients received physical and laboratory evaluation on day 1 of each cycle. AEs were recorded at every visit and graded according to the National Cancer Institute Common Terminology Criteria of Adverse Events (NCI-CTCAE) version 4.03.

A continuous safety assessment was utilized to determine the probability of stopping treatment early. There was a high probability of stopping early if the toxicity rate was unacceptably high (i.e., ≥ 60%), with a low probability of stopping early if the toxicity rate was acceptable, i.e., ≤ 30% by Pocock-type boundary [18]. The starting dose of lapatinib was 1250 mg by mouth daily and veliparib 200 mg by mouth twice daily with plans to de-escalate if toxicities were noted. Toxicities were assessed during day 1 of each cycle of therapy according to the NCI Common Toxicity Criteria, version 4.03. Adverse events were assessed among all enrolled patients for the purpose of dose de-escalation decisions.

Plasma samples were collected for the bioanalysis of lapatinib to investigate potential interactions between lapatinib and veliparib. The dosing schema (lapatinib on day 1, veliparib on day 2) was designed to understand the PK interactions between the drugs as well as to induce a double-stranded break (DSB) repair defect by administering lapatinib first. PK profiles were evaluated once the steady state of both drugs was reached (day 28).

Blood samples (3 mL, collected in lithium heparin tubes) were taken at predose (0) and then 2, 4, 6, 8, and 24 h after dosing for lapatinib. A validated liquid chromatographic mass spectrometric assay was used to measure lapatinib and veliparib in plasma. Standard PK variables were calculated by a non-compartmental method including maximal plasma or serum concentration (C_max), area under the curve to the last collection point (AUC_last), area under the curve for dose interval (AUC_0–t), area under the curve extrapolated to infinity (AUC_inf), time of maximal concentration (T_max), elimination rate constant (k_el), terminal half-life (t_1/2), total clearance (CL), and a volume of distribution (V_ss), using the Phoenix WinNonlin 8.0 (Certara, Princeton, NJ) software. The linearity of PK parameters was explored. Descriptive statistics of the non-compartmental parameters were calculated for cycle 1 and cycle 2.

Tumor core biopsies were snap-frozen in the procedure room prior to treatment and placed in TRIzol reagent immediately before thawing. Total RNA was isolated using the PureLink RNA Mini kit with TRIzol protocol (Invitrogen/Thermo Fisher, Waltham, MA) and concentrated as needed using the Zymo RNA Clean and Concentrator kit (Irvine, CA). The Breast IO360 panel from NanoString Technologies (Seattle, WA) was run using 100 ng of RNA and counted at the maximum FOV setting, as suggested by the manufacturer. This panel of 776 genes across 23 key breast cancer pathways and processes provides a unique 360-degree view of gene expression for the breast tumor microenvironment and immune response. Standards supplied with the kit were also run in order for further analysis to be performed by NanoString Technologies. This analysis included PAM50 breast cancer subtyping, Tumor Inflammation Signatures (TIS) scores, tumor mutation susceptibility (BRCAness, Homologous Recombination Repair Status, HRD), rate of recurrence (ROR), and other signatures (i.e., Claudin-low) predicting tumor sensitivity to immune attack.

The primary endpoint of the trial was the safety of the combination of lapatinib and veliparib. Objective response rate (ORR) and progression-free survival (PFS) were secondary endpoints. A total of 20 patients were planned. The sample size was not determined by the statistical power but by the feasibility and relative precision of estimate. Assuming the acceptable toxicity rate with combination therapy was no more than 30%, a two-sided 95% exact confidence intervals (Clopper–Pearson intervals) of toxicity would be estimated from 11.8 to 54.3% with standards error of the estimation less than or equal to 10%. A continuous assessment of toxicity and stopping rules were implemented using the method of Ivanova et al. [18].

The primary analysis of safety was assessed in all patients who received at least one dose of lapatinib and/or veliparib. Descriptive statistics were presented for both quantitative safety data and response rate. During the study, adverse event and serious adverse event (AE/SAE) reporting was summarized by relationship to study drug and intensity. The number and proportion of AE, SAE, and grade 3 or 4 lab toxicities were reported by the body system. Mean, median, and range were used to describe continuous demographic variables and EGFR expression. Frequency and proportion were used to describe categorical variables. At the completion of the study, the secondary endpoint of ORR rate was estimated along with two-sided 95% confidential intervals with the exact method of Clopper–Pearson intervals.

From August 2014 to September 2017, twenty patients were enrolled; seventeen were evaluable for response by RECIST; and 3 patients experienced clinical progression of disease during cycle 1. Evaluable patients completed at least two cycles of therapy. Patient demographics and disease characteristics are shown in Table 1. Of note, 43% of evaluable patients were African–American. Approximately 46% of the study population had prior systemic therapy for metastatic disease. The median number of prior therapies for advanced breast cancer was 1 (range 0–2). Patients with more than 2 prior lines of chemotherapy in the metastatic setting were excluded.

In this phase 1 study, the combination of lapatinib with veliparib did not result in unexpected toxicities. The MTD of the combination was defined at lapatinib 1250 mg QD (days 1–28) with veliparib 200 mg PO twice daily. No dose level reductions were required. Toxicities related to study drugs were limited to grade 1 or 2. Only 2 patients experienced grade 3 toxicities of headache and abdominal pain, respectively, which were attributed to progression of the disease and not due to study drugs. Most common adverse events in decreasing order of frequency included fatigue (6.4%), diarrhea (5.1%), constipation (5.5%), insomnia (4.5%), vomiting (2.9%), anemia (2.6%), headache (2.6%), dizziness (2.6%), dyspnea (2.3%), and rash (2.3%) (Table 2). All treatment-related adverse events were manageable and did not result in any dose reductions, delays, or discontinuations.

Seventeen patients underwent lapatinib pharmacokinetic analysis. One patient missed a PK evaluation during cycle 2. The maximum tolerated dose of lapatinib was 1250 mg by mouth daily when in combination with veliparib. Lapatinib pharmacokinetic parameters are described in Supplementary Table S1 in Additional file 1. Lapatinib PK parameters demonstrated moderate interpatient variability during cycle 1. At steady state, lapatinib plasma concentrations were consistent with historical data published on lapatinib monotherapy (Supplementary Figure S1 in Additional file 1) [19, 20]. Co-administration of veliparib did not influence lapatinib PK parameters.

Of the 20 patients enrolled in the study, 17 were evaluated for clinical response. The summary of objective response by RECIST criteria v 1.1 is depicted in Fig. 2. Four patients achieved a partial response (PR), 2 patients had stable disease (SD), and 11 patients had progressive disease (PD). Patient 012 had a prolonged PR of over 10 months. Patient 009 achieved a favorable reduction in disease burden after 2 cycles and was rendered resectable by the combination treatment. The patient discontinued the study to have a left radical mastectomy with reconstruction followed by radiation (Fig. 3). At the time of this manuscript, the patient remains disease free for more than 5 years as of her last clinical visit of July 2020.

Of the 17 patients evaluated for clinical response, tumor biopsies were performed for 11 patients prior to study treatment. The sites from which the biopsies were obtained are listed in Supplementary Table S2. For the remaining patients, the procedural biopsies were determined unsafe due to tumor location. We performed Nanostring analysis on the available samples using the Breast360 panel, which contains 48 published signatures of breast cancer tumor biology, including PAM50, Tumor Inflammation Signature (TIS), and risk of recurrence (ROR)/genomic risk scores. Direct comparison of the 2 PR samples (patients 002, 009) with 9 PD samples (patients 004, 005, 007, 008, 010, 013, 014, 022, 023) showed a significant increase in the expression of genes involved in antigen presentation (HLA-DOB (5.0-fold), HLA-DPA1 (4.6-fold), HLA-DRA (3.7-fold), HLA-DMB (3.5-fold)), immune infiltration (CCL5 (5.5-fold), GZMA (4.8-fold), CD8A (3.3-fold), PDCD1ILG2 (3.3-fold), STAT1 (3.0-fold), LAG3 (2.9-fold)), and cytokine and chemokine signaling (CCR5 (3.6-fold), NOD2 (2.9-fold), CCL8 (2.8-fold), CCL4 (2.7-fold), CCL2 (2.7-fold)) (Fig. 4). Additionally, the TIS trended higher in the responders. The TIS scores were 7.6 and 8.8 for the PR samples and 5.2, 6.1, 7.3, 6.0, 6.1, 7.7, 6.0, 6.3, and 5.9 for the PD samples (Supplementary Figure S2 in Additional file 1). These data suggest the potential of pre-therapy immune status to correlate with response.

To verify the advanced analysis performed via the Nanostring platform, we performed analysis using the DEGs generated from Nanostring and Fisher’s exact test for top potential pathways involved by these DEGs between responders vs. non-responders. Similar to the Nanostring advanced analysis, the top canonical pathways differentially regulated as listed in Supplementary Table 3 (Additional file 1) are involved in immune regulation. Specifically, these include Th1, Th2, T cell exhaustion, neuroinflammation, and MSP-RON signaling in macrophages pathways (Supplementary Table 3 in Additional file 1). For detailed results of the full analysis, refer to Additional files 2 and 3. However, these are preliminary and hypothesis generating, due to the limitations of small sample size and limited number of DEGs.

Interestingly, while all 11 samples showed a “BRCAness” phenotype, the PR samples showed a significant reduction in the DNA damage repair genes BRCA1, CRYAB, and CKB genes (4.8-, 4.5-, and 4.4-fold, respectively), as well as a significant change in genes related to other pathways implicated in DNA damage response (Table 3). Specifically, CALML5, MMP7, COLL11A1, RASGRF1, PTCH1, COL27A1, FGFR2, and NOTCH1 all showed decreased expression of at least 2-fold in PR samples, while PLA2G3, LFNG, PIK3R3, SOCS1, GJB2, STAT1, SOCS2, and HGF all showed at least a 2-fold increase in gene expression. A reduction in genes associated with triple-negative biology across the board was also observed (GABRP (19.9-fold), PROM1 (6.9-fold), HBB (5.7-fold), PDE9A (3.9-fold), BCL11A (3.6-fold), data not shown). Indeed, when PAM50 criteria were used, all the samples were classified as basal-like subtype except for one of the PR samples (002) which was HER2 enriched, and one of the PD samples (patient 008) which was luminal A. All of the breast cancer 360 biological signatures are depicted in Supplementary Figure S2 in Additional file 1 except the risk of recurrence (ROR) scores which were as follows: PR samples 82 (patient 002), 72 (patient 009); PD samples 66 (patient 004), 56 (patient 005), 56 (patient 007), 53 (patient 008), 55 (patient 010), 74 (patient 013), 69 (patient 014), 58 (patient 022), and 74 (patient 023).