A Phase I Study of Veliparib in Combination with Metronomic Cyclophosphamide in Adults with Refractory Solid Tumors and Lymphomas

PARP enzymes are characterized by the ability to poly(ADP-ribosyl)ate protein substrates (1, 2). PARP-1 and PARP-2 help maintain genomic stability by regulating the repair of DNA damage; PARP inhibition potentiates the DNA-damaging effects of alkylating agents, including cyclophosphamide, by interfering with the repair of DNA damage (3, 4). PARP inhibition in homozygous BRCA-deficient cells produces synthetic lethality, making PARP inhibitors attractive therapeutic agents for patients with BRCA mutations (5–7). Veliparib (ABT-888), a potent, oral small molecule, inhibits PARP activity significantly in tumors at clinically achievable concentrations (4, 8). We hypothesized that coadministration of a PARP inhibitor with metronomic cyclophosphamide would be well tolerated and might enhance the therapeutic index because of the known modest toxicity of cyclophosphamide administered on a daily, oral schedule.

Chronic administration of the alkylating agent cyclophosphamide at low doses, known as metronomic dosing, is effective in lymphoma and multiple tumor types, including ovarian, prostate, and breast cancer (9–12). In addition to inducing DNA damage, metronomic cyclophosphamide may target tumor angiogenesis by reducing tumor endothelial cell proliferation through activation of thrombospondin 1 (13, 14). It has also been reported to reduce the population of CD4⁺CD25⁺ regulatory T cells that suppress antitumor immunity (15).

We conducted a phase I trial of the combination of veliparib with metronomic oral cyclophosphamide in patients with refractory solid tumors and lymphoid malignancies. The objectives were to establish the safety, tolerability, and maximum tolerated dose (MTD) of the combination; to determine the PK of veliparib; and to examine the effects on PARP activity and phosphorylated histone H2AX (γH2AX) levels, a marker of DNA double-strand breaks, in peripheral blood mononuclear cells (PBMC), tumor biopsies, and circulating tumor cells (CTC).

Patients (age ≥ 18 years) were eligible if they had pathologically confirmed metastatic solid tumor or low-grade lymphoma for which there was no acceptable standard therapy; a Karnofsky performance status of 60% or more; and adequate liver, kidney, and marrow function defined as absolute neutrophil count of 1,500/μL or more, platelets of 100,000/μL or more, total bilirubin of 1.5 × the upper limit of normal (ULN), aspartate aminotransferase and/or alanine aminotransferase less than 2.5 × ULN, creatinine less than 1.5 × ULN. Prior exposure to PARP inhibitors or cyclophosphamide was allowed.

Previous anticancer therapy or surgery must have been completed at least 4 weeks before enrollment; patients were required to have evidence of disease progression on previous therapy by staging scans. Patients unable to swallow pills or those with uncontrolled intercurrent illness, brain metastases within the past 3 months, history of seizures (high-dose veliparib caused seizures in a preclinical model), or gastrointestinal conditions that might predispose to drug intolerability or poor drug absorption were excluded. Documentation of BRCA mutation status was not required.

This trial was conducted under a National Cancer Institute (NCI)-sponsored Investigational New Drug Application with Institutional Review Board approval at each participating site. Protocol design and conduct followed all applicable regulations, guidances, and local policies. [ClinicalTrials.gov Identifier: NCT00810966].

This was an open-label, multicenter, single-arm phase I combination study of veliparib and metronomic oral cyclophosphamide in patients with advanced malignancies. The Division of Cancer Treatment and Diagnosis, NCI, supplied veliparib under a Collaborative Research and Development Agreement with Abbott Laboratories. Cyclophosphamide was obtained through commercial sources.

Cyclophosphamide was administered orally once daily throughout a 21-day cycle. Veliparib was administered orally once daily, with cyclophosphamide, for the first 7, 14, or 21 days of the cycle, depending on DL. Starting doses were veliparib 20 mg daily and cyclophosphamide 50 mg daily (Table 1). Dose escalation did not follow the Fibonacci schema due to low anticipated toxicity and desire to provide uninterrupted PARP inhibition coverage to the majority of patients given the potential for clinical benefit.

A standard phase I dose escalation design (3 + 3) was employed. Higher DLs were not opened until the last patient in the previous cohort had completed 1 cycle. Intrapatient dose escalation was allowed once 3 new patients completed that DL without grade 2 or higher toxicity and the given patient was tolerating therapy well. Adverse events were graded according to NCI Common Toxicity Criteria (version 3.0). Dose-limiting toxicity (DLT) was defined as an adverse event that occurred in the first cycle, was felt to be related to the study drugs, and fulfilled one of the following criteria: grade 3 or greater nonhematologic toxicity (except grade 3 nausea/vomiting and diarrhea without maximal symptomatic treatment, grade 3 creatinine and electrolyte abnormalities that corrected to grade 1 or baseline within 24 hours); grade 4 neutropenia; grade 4 thrombocytopenia; or a 2-week or more delay in starting the next cycle due to toxicity. Any degree of anemia, leucopenia in the absence of neutropenia, or lymphopenia was not considered dose limiting. Patients were considered evaluable for cohort dose escalation decisions if they either experienced DLT or completed 1 full cycle without DLT and received at least 80% of the planned dose.

Toxicities had to resolve to grade 1 or less for nonhematologic toxicities (except electrolyte abnormalities, which had to resolve to grade 2 or less) and grade 2 or less for hematologic toxicities, before starting the next cycle. Treatment could be delayed for a maximum of 2 weeks to allow resolution of toxicities. If toxicities did not resolve as defined above, patients were taken off study treatment. DLT resulted in a reduction in DL. The MTD was defined as the DL at which none or 1 of 6 patients experienced DLT, and the DL below 1 in which 2 or more patients experienced DLT. Six additional patients were accrued at the MTD to further evaluate PK and pharmacodynamics (PD).

History and physical examination, including performance status and vital signs, were carried out at baseline and repeated at the start of every cycle. Complete blood counts with differential and serum chemistries were conducted at baseline, weekly during cycle 1, and the start of every cycle from cycle 2 onward. In the studies conducted with veliparib, there has been no indication of cardiac toxicity related to study drug, therefore, electrocardiogram was done at baseline only and repeated as clinically indicated. Radiographic evaluation was carried out at baseline and every 2 cycles to assess tumor response based on the Response Evaluation Criteria in Solid Tumors (RECIST; version 1.0; ref. 16).

Veliparib PK analysis was conducted over a 24-hour period (samples collected predose and at 0.5, 1, 1.5, 2, 3, 4, 6, 8, and 24 hours postdose) on day 1 of cycle 1 and on the last day veliparib was administered in cycle 1 (day 7, 14, or 21, depending on DL). Blood samples were centrifuged at 1,000 × g and plasma stored at −70°C until analysis. Plasma veliparib concentrations were quantified with a liquid chromatography–mass spectrometry assay validated to U.S. Food and Drug Administration guidelines (17). See Supplementary Data for additional details.

PBMC samples for PD analysis were obtained in 8 mL Cell Prep Tubes (Becton Dickinson) on days 1 and 21 of the first cycle over a 24-hour period (predose and 2, 4, 6, and 24 hours postdose). Optional tumor biopsies were collected predose and 2 to 6 hours (or 24 hours for the MTD expansion cohort) after the last dose of veliparib in cycle 1. Poly(ADP-ribose) (PAR), a product of PARP, was measured in PBMCs and tumor biopsies using a validated immunoassay as previously described (8, 18). Levels of γH2AX were measured in PBMCs and CTCs with standard operating procedures previously described for the validated immunofluorescence assays (19–21).

Statistical analyses for PK parameters and concentration values were conducted with SPSS 17.0 for Windows (SPSS Inc.). Values were compared by a 2-tailed, paired exact Wilcoxon signed-rank test; P < 0.05 was considered significant.

Blyth-Still-Casella 95% CIs were calculated for the true probability across patients for change in PAR levels using StatXact 4.0 (Cytel, Inc.; refs. 22, 23). These CIs were for descriptive purposes only, as PAR threshold values were influenced by the observations.

Thirty-five patients with advanced malignancies were enrolled (Table 2). Two patients on DL 7 did not complete a full cycle of therapy and were not included in the MTD determination. One patient developed a small bowel obstruction shortly after enrollment that was considered secondary to disease progression and adhesions from prior surgeries, and was taken off study. One patient required intervention for pericardial involvement by tumor, necessitating discontinuation of study therapy. Patients were heavily pretreated; 16 had received prior intravenous cyclophosphamide and 3 received prior oral cyclophosphamide, in combination with other therapeutic agents. Three patients had prior single-agent PARP inhibitor therapy (1 patient received veliparib and 2 received olaparib).

The study regimen was well tolerated. Grade 2 myelosuppression (with some higher grade lymphopenia) was the most common toxicity across DLs (Table 1). At DL 8, 2 patients developed DLTs. One patient with advanced breast cancer and baseline dyspnea on exertion developed worsening dyspnea, hypoxia, and eventually acute respiratory failure during the first week of treatment. No clear evidence of cardiac dysfunction or disease progression in the lungs was documented. This patient died a few days later despite best supportive care. The treating physician felt the cause of death was likely disease progression, but the possibility that the study drug combination may have contributed to the respiratory failure could not be excluded. A second patient with metastatic breast cancer and extensive liver involvement presented with worsening abdominal pain and distension in the middle of cycle 1; radiologic assessment was consistent with mild bowel loop dilatation without evidence of obstruction. The patient did not have a prior history of ileus or small bowel obstruction, and there was no known intestinal or peritoneal involvement by disease. Study drugs were held, and the patient was managed conservatively, with resolution of symptoms. Study therapy was reinitiated at the next lower DL with no recurrence of symptoms. Due to the 2 DLTs at DL 8, additional patients entered on DL 7 (veliparib 60 mg daily with cyclophosphamide 50 mg daily), which was established as the MTD.

Seven patients experienced a partial response, and an additional 6 patients had prolonged stable disease (6 or more cycles), including 1 patient (patient 1) with low-grade lymphoma who received a total of 42 cycles of study treatment with resolution of B symptoms (Table 3, Fig. 1). He was enrolled on DL 1 and eventually escalated to DL 7 following discussions with the study sponsor, lack of significant toxicity, and establishment of the safety of DL 7, to assess whether chronic PARP inhibition would provide additional clinical benefit. One patient with BRCA2-positive ovarian cancer (patient 23; DL 7), who had received multiple lines of prior chemotherapy (including cisplatin, paclitaxel, liposomal doxorubicin, and irinotecan), had disappearance of target lesions on follow-up scans, and received a total of 17 cycles with eventual disease progression. However, CA 125 levels, although improved, remained above normal; thus, this patient was not considered a complete response by RECIST 1.0. One of the patients with urothelial malignancy and Muir-Torre syndrome had prolonged disease stabilization for 17 cycles.

Thirteen patients had known BRCA mutations; of these, 6 had partial responses and 3 had prolonged stable disease. The BRCA status of the remaining patients was unknown; therefore, no comparisons to wild-type BRCA could be made.

Veliparib PK data were available for 35 patients, and data were available for 30 patients on both day 1 and the last day of veliparib dosing (day 7, 14, or 21; Supplementary Table S1). Dose linearity assessment for maximum plasma concentration (C_max) resulted in a coefficient of 1.098 (95% CI: 0.784–1.41; 90% CI: 0.720–1.48). Dose linearity assessment for area under the curve (AUC) resulted in a coefficient of 1.15 (95% CI: 0.882–1.42; 90% CI: 0.828–1.47).

No statistically significant changes in PK parameters between baseline and day 7, 14, or 21 were observed. The accumulation index was 1.04 (SD ± 0.23) for AUC_0–24 and 1.12 (SD ± 0.40) for C_max; these were not statistically significant. In the 3 patients on DL 5 (100 mg cyclophosphamide), PK behavior of veliparib was not different from other DLs.

Mean PAR levels in patient PBMCs over the first 24 hours of cycle 1 are shown in Fig. 2A. Of the 33 evaluable patients with PBMCs available at 4 hours postdose, 21 (64%) had at least a 50% decrease in PAR; 6 of these 21 patients had PAR levels below the level quantifiable by the assay. The Blyth-Still-Casella 95% CI, indicating the true probability of such a decrease across patients, without accounting for DL (a limitation imposed by the small sample numbers), ranged from 0.45 to 0.79. PBMC PAR levels also decreased by at least 50% from baseline on day 21 of cycle 1 in all 7 patients for whom samples were available; these patients were on DL 6 to 8 (Supplementary Fig. S1). Five of the 7 patients were known to have a BRCA mutation, but no conclusions can be made about BRCA status due to the small sample size. Due to the small number of patients treated per DL, statistical comparisons of the degree of inhibition at day 21 compared with baseline could not be carried out.

All 5 patients with pre- and postdose (after 7 or 21 days of treatment) paired tumor biopsies had more than 80% decreases in PAR levels (Fig. 2B). The Blyth-Still-Casella 95% CI, indicating the true fraction of patients who had more than an 80% decrease in tumor biopsy PAR levels on either day 7 or 21, ranged from 0.50 to 1.0.

We measured γH2AX levels, expressed as percentage nuclear area positive (%NAP), in PBMCs collected over 24 hours from 17 patients on day 1. Fifteen of 17 patients had posttreatment γH2AX levels at or below baseline (Supplementary Fig. S2). Two patients had a slight increase in γH2AX %NAP at the 4-hour time point that returned to near baseline by 6 hours. We also measured γH2AX in CTCs. CTCs were detectable in 9 patients; most samples had fewer than 10 detectable CTCs, a number that remained relatively constant over the first 5 days of cycle 1 (Fig. 2C). Seven of 9 patients had an increased percentage of γH2AX-positive CTCs on day 2 (Fig. 2D), but these data are difficult to interpret in regard to response given the small numbers of CTCs. For patient 8, the number of CTCs increased from 0 at baseline to 33 at day 2; 64% of these cells were γH2AX positive. Patient 17, a 56-year-old man with colonic adenocarcinoma, had considerably more CTCs than other patients, and received 2 cycles of therapy before disease progression. The percentage of γH2AX-positive CTCs for this patient increased slightly from 3.9% at baseline to 6.2% on day 2.

In 1 patient, a 57-year-old man with small lymphocytic lymphoma/chronic lymphoid leukemia who had prolonged stable disease (patient 1), PAR levels in PBMCs and tumor, and γH2AX in PBMCs were measured (Fig. 3). This patient was on DL 1 for cycle 1.

This is the first clinical trial to evaluate once-daily dosing of veliparib in combination with a chemotherapeutic agent. Given its short half-life (5.0 ± 1.5 hours), veliparib has been administered twice daily in all other combination trials. We evaluated once-daily dosing based on PD data from our phase 0 trial, which showed more than 48% inhibition of PARP activity in tumor biopsies 24 hours after a single 50 mg dose (8), and because of better patient compliance with once-daily dosing (24, 25). Enhanced myelotoxicity has been observed with PARP inhibitor combination regimens with cytotoxic chemotherapy, requiring dose reduction of the chemotherapy (26–28). This has raised questions about the relative contribution of the addition of PARP inhibitors to combination regimens and whether administration of full doses of chemotherapy alone would provide similar benefit. Therefore, we designed a regimen that would be well tolerated and have low levels of toxicity and good patient compliance, a potentially better therapeutic ratio, and we were able to safely escalate the PARP inhibitor dose with metronomic cyclophosphamide dosing. The regimen was well tolerated, and the MTD was established at 60 mg veliparib with 50 mg cyclophosphamide administered daily in 21-day cycles.

Six of 7 partial responses were observed in patients with known BRCA mutations, and an additional 3 patients with BRCA mutations had prolonged disease stabilization. Evidence of clinical benefit and PARP inhibition was observed across DLs, suggesting that even at lower doses, veliparib produced sufficient inhibition of PARP activity to provide benefit in BRCA-positive patients receiving DNA-damaging chemotherapy. Two of the partial responses, and 2 of the prolonged stable diseases, occurred in patients who had received prior cyclophosphamide. One patient with metastatic ovarian cancer and a known BRCA mutation had complete disappearance of radiologic evidence of disease, even though tumor markers remained elevated. To our knowledge, this is the first report of the disappearance of radiologic evidence of disease in a patient treated with veliparib. Due to the unknown BRCA status of the remaining patients, no comparative analysis with wild-type BRCA patients can be conducted. However, these preliminary data support the hypothesis that tumors with DNA repair defects may be sensitive to PARP inhibitors. Interestingly, a patient with Muir-Torre syndrome, an autosomal dominant genetic disorder likely caused by microsatellite instability (29), and urothelial malignancy had prolonged disease stabilization for 17 cycles, further supporting the potential therapeutic role of PARP inhibitors in the treatment of tumors with DNA repair defects.

We observed a statistically significant inhibition of PARP activity in PBMCs and tumor biopsy samples across DLs. The degree and duration of PARP inhibition in tumor required for clinical benefit has not been established. A comparative study of the effects of treatment with other PARP inhibitors, MK-4827, olaparib, and PF-01367338, on PBMCs has recently been reported, and decreases in PAR levels ranged from 65% to 92% (30). The PAR levels at baseline and the degree of inhibition in PBMCs is variable as evidenced by our observations in this study. In the small subset of patients who underwent tumor biopsies and PBMC sampling on the same day (patients 20, 22, and 34), there seemed to be concordance in the inhibition of PAR in both sample sets. Although there is inherent variability in the baseline levels of PAR in PBMCs and the degree of PARP inhibition for a given DL as shown in Fig. 2A, overall PAR levels were significantly decreased in PBMCs across all DLs.

We did not observe consistent increases in γH2AX, a sensitive marker of DNA damage (20, 31), in PBMCs. As previously reported in our phase 0 study of veliparib (8), PARP activity is more easily inhibited in tumor cells than PBMCs; therefore, we evaluated the number of CTCs and the presence of γH2AX as a marker of drug effect in tumor. Though we observed increases in the fraction of CTCs positive for γH2AX after treatment, definitive conclusions cannot be made due to the few patients who underwent CTC sampling (as this was added later to the study) and low number of CTCs recovered per sample. We did not observe increases in γH2AX in PBMCs in the majority of patients evaluated, but did observe increases in γH2AX in CTCs, which is consistent with the minimal myelosuppression and promising antitumor activity observed with this regimen.

We were concerned about the potential for coadministration of cyclophosphamide to increase PARP inhibitor metabolism, because cyclophosphamide can induce CYP3A4 expression in human hepatocytes and liver slices (32). However, no effect of metronomic cyclophosphamide on veliparib PK was observed between the first and last days of treatment (days 1 and 7, 14, or 21). These results are consistent with our previous report in which 31% to 115% of the veliparib dose was recovered in urine as unchanged parent drug (8).

Because of the encouraging activity and tolerability of this combination in patients with DNA repair deficiencies, the activity of metronomic cyclophosphamide alone and in combination with veliparib is being compared in a multicenter, randomized phase II study in patients with advanced ovarian cancer and BRCA mutations, high-grade serous ovarian cancers, triple-negative breast cancers, and low-grade lymphomas (ClinicalTrials.gov identifier: NCT01306032). The trial includes a detailed genetic analysis of underlying DNA repair defects for the ovarian cancer cohort. Veliparib as a single agent is being evaluated in a separate phase I clinical trial in patients with cancer carrying the BRCA mutation (ClinicalTrials.gov identifier: NCT00892736). Data from the single-agent trial were not available at the time of designing the phase II trial of the combination to add single-agent veliparib as a comparator arm.

The phase II trial, as currently designed, should help define the contribution of PARP inhibition to the activity of this combination and will begin to establish whether a true increase in the therapeutic ratio of a cytotoxic is possible with the use of PARP inhibitors in the clinic.

R. J. Kinders serves as a consultant for Trevigen and owns Abbott Laboratories common stock (less than $10,000). The other authors disclosed no potential conflicts of interest. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

The authors thank Gina Uhlenbrauck, Yvonne Evrard, and Heather Gorby, SAIC-Frederick, Inc., for editorial assistance in the preparation of this article. The authors also thank Michael Minchik, Edward Grant, Ravithat Putvatana, Donna Ketchum, Lan Tran, and Will Yutzy, SAIC-Frederick, Inc., for their work on PBMC samples for PD analysis.

This manuscript is dedicated to the memory of Dr. Merrill J. Egorin, University of Pittsburgh School of Medicine.

This work has been supported in whole or in part with federal funds from the NCI, NIH, under contract no. HHSN261200800001E and in part by the Intramural Research Program of the NIH, NCI, Center for Cancer Research. Additional support was provided by NCI, Cancer Therapy Evaluation Program cooperative agreements U01-CA062505 and U01-CA099168 and NCI grant P30-CA47904.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.