Phase I Study of the Antiangiogenic Antibody Bevacizumab and the mTOR/Hypoxia-Inducible Factor Inhibitor Temsirolimus Combined with Liposomal Doxorubicin: Tolerance and Biological Activity

Tumor hypoxia may be a double-edged sword. Lower tissue oxygen concentrations can exert antitumor effects by inhibiting proliferation, limiting metastases, promoting differentiation, and inducing apoptosis and necrosis (1–4). In contrast, some tumor clones, under conditions of hypoxic stress, develop adaptive processes through modification of gene expression that confer an aggressive phenotype, promoting locoregional and distant tumor growth (5–7). Tumor hypoxia can be caused by antiangiogenic therapy (8), which then mediates resistance to antiangiogenesis (9–11). The hypoxia-mediated increase of hypoxia-inducible factors (HIF) is critical to the establishment and progression of many cancers via HIF-dependent activation of genes that allow cancer cells to survive, metastasize, and develop resistance to chemotherapeutic agents (12–17).

In addition to inhibiting the expression of proteins critical to cell cycling, mTOR inactivation suppresses angiogenesis by reducing the expression of HIFs (18). Such dual inhibition would facilitate maintaining the benefits of attenuating angiogenesis while avoiding the negative consequences of increased hypoxia, including an induced aggressive change in tumor biology (19); decreasing HIFs may also resensitize cancer cells to doxorubicin (20, 21). Importantly, mTOR inhibitors interfere with signaling via the PI3K/AKT/mTOR axis, a pathway critical in many types of cancers. Finally, clinical experience indicates that antiangiogenic agents, mTOR inhibitors, and doxorubicin have nonoverlapping side effects, suggesting that they might combine well without producing excessive toxicity.

We hypothesized that simultaneous inhibition of the VEGF and tumor hypoxia–mediated adaptation pathways might provide opportunities to improve outcomes compared with antiangiogenic agents alone, and that these agents, together with doxorubicin, might augment sensitivity to the latter drug. Toward that end, we conducted a clinical study (NCT00761644) that combined bevacizumab and temsirolimus plus liposomal doxorubicin in patients with advanced malignancies.

Patients 12 years of age or older were eligible if they had a histologically confirmed advanced malignancy, with no standard therapy that improved survival for at least 3 months. Children (n = 3) were included as the study regimen might have potential therapeutic value in pediatric cancers. All participants of both genders had measurable or evaluable disease that had progressed before study entry and an Eastern Cooperative Oncology Group (ECOG) performance status of 2 or better (22). Additional eligibility criteria included adequate marrow function (absolute neutrophil count ≥ 1,500/μL and platelet count ≥ 100,000/μL), serum creatinine ≤ 3 times the upper limit of normal, total bilirubin ≤ 2.0 mg/dL, alanine transaminase (ALT) ≤ 5 times the upper limit of normal or 8 times the upper limit of normal if liver metastases present, and cardiac left ventricular ejection fraction ≥ 50%. Patients with the following conditions were excluded: poorly controlled hypertension, defined as systolic blood pressure more than 150 mm Hg and/or diastolic blood pressure more than 100 mm Hg; prior cumulative doxorubicin dose more than 300 mg/m²; clinically significant cardiovascular disease; pregnant or lactating women; and unwillingness or inability to sign an informed consent, as previously described (23).

The objectives of this study were to define the safety and biologic activity of this regimen. This phase I clinical trial was conducted using a modified 3 + 3 design. There were at least 2 patients entered on each cohort for initial assessment of safety. An additional 3 patients were allowed on cohorts as needed for safety assessments and, if benefit was observed in a specific type of cancer, a mini-expansion of up to 14 patients was permitted at the highest dose level considered to be safe at the time of patient entry. Therefore, each dose level might enroll up to 14 × N patients with specific tumor types (N) who displayed antitumor activity defined as below, which ended up with various enrollments into different dose levels as patients were enrolled at the highest dose level deemed safe at the time of enrollment.

Treatment was administered on an outpatient basis with intravenous temsirolimus weekly plus intravenous bevacizumab and liposomal doxorubicin once on day 1 every 21 days, as long as the patient had no evidence of tumor progression or prohibitive toxicity. Each cycle was 21 days. The trial was conducted at The University of Texas MD Anderson Cancer Center (Houston, TX) after approval by the Institutional Review Board (IRB) in accordance with the IRB guidelines. All patients signed an informed consent.

All patients who received one dose of any of the study agents were considered evaluable for safety. The severity of adverse events was graded according to the Common Terminology Criteria for Adverse Events v3.0 (24). All patients underwent close cardiac monitoring: baseline clinical evaluations and then as frequently as indicated, baseline 12-lead electrocardiogram, and then as frequently as indicated, baseline cardiac scan (MUGA: multi-gated acquisition scan) or 2-dimensional echocardiogram, and then once after every 100 mg/m² increment of liposomal doxorubicin whenever cumulative anthracycline and/or liposomal doxorubicin dose was higher than 300 mg/m², with cardiology consultation as indicated. Dose-limiting toxicity (DLT) was defined as (first cycle) any grade 3 or higher study agent–related (possibly, probably, or definitely) toxicity, including fatigue, with the following exceptions: hematologic toxicities that were required to be grade 4 lasting 2 weeks or longer despite supportive care and nausea or vomiting that were required to be grade 4 lasting for more than 5 days despite maximum antiemetic treatment. Also, DLT included symptoms/signs of vascular leak or cytokine release syndrome; or any severe or life-threatening complication or abnormality not defined in the NCI-CTCAE v3.0 that was possibly, probably, or definitely related to the therapy. Correctable electrolyte imbalance and alopecia were not considered DLTs. Maximum tolerated dose (MTD) was defined as the dose level below the dose at which 33% or more of patients experienced drug-related DLT in their first treatment cycle.

All patients who received one dose of any of the study agents were considered evaluable for efficacy. All histologies were centrally reviewed at MD Anderson Cancer Center. Radiographic imaging studies were repeated approximately every 2 cycles (6 weeks) of therapy. Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 were used to characterize tumor responses (25). Patients who were removed from the study before the first scheduled restaging workup because of progression, serious drug-related adverse events, or any other reasons were considered treatment failures and arbitrarily designated as having 21% progression in a waterfall plot showing best tumor responses (Figs. 1 and 2).

Testing for genetic aberrations, such as HER2 amplification, PIK3CA, PTEN, KRAS, BRAF, EGFR, C-MET, and p53 mutations, and protein expression (immunohistochemistry), such as PTEN loss (DAKO antibody), estrogen receptor, progesterone receptor, and HER2, was conducted in a Clinical Laboratory Improvement Amendment (CLIA)-certified molecular diagnostic laboratory at MD Anderson Cancer Center using archival formalin-fixed, paraffin-embedded tissue blocks or material from fine-needle aspiration of tumor tissues, as described previously (23, 26). Mutation testing was conducted by analyzing extracted DNA with a PCR-based DNA sequencing method. For PIK3CA mutations, codons (c)532 to c554 of exon 9 (helical domain) and c1011 to c1062 of exon 20 (kinase domain), which included the mutation hotspot region of the PIK3CA protooncogene, were examined. Sanger sequencing was conducted after amplification of 276– and 198–base pair amplicons, respectively, using primers designed by the MD Anderson Molecular Diagnostic Laboratory.

Expansion was generally added as antitumor activity was seen. For the purpose of mini-expansions, as planned in advance, of up to 14 participants with specific tumor types in which activity was seen, a tumor response signal was defined as one or more of the following: stable disease (SD) for 4 months or more, decrease in measurable tumor by 20% or more, decrease in tumor markers by 25% or more, or a partial response (PR) according to the Choi response criteria (27), that is, decrease in size by ≥ 10%, or a decrease in tumor density, as measured by Hounsfield units (HU), by 15% or more. A sample size of 14 patients will have 82% power to detect effects sizes of 0.85, based on the 2-sided Wilcoxon signed rank test at a significance level of 0.05, where effect size is defined as mean change divided by standard deviation.

Descriptive summary statistics were used to assess demographics, safety, and antitumor activity. Categorical data were summarized using frequency and percentages. Continuous data were summarized by mean, median, range, and coefficient of variation ± standard deviation. Differences in categorical variables were determined by Fisher exact tests. The median duration of responses was estimated using the Kaplan–Meier method. Statistical inferences were based on 2-sided tests at a significance level of P < 0.05. Statistical analyses were carried out using GraphPad Prism 5 software (GraphPad Software, Inc).

Patient characteristics are listed in Table 1. A total of 136 patients (median age, 53 years; range, 12–75 years) were recruited to 8 dose levels during dose escalation (n = 39), as well as during dose expansion (n = 97). Patients were heavily pretreated with a median of 4 prior chemotherapy regimens. The greater enrollment of women was directly related to increased dose expansions in patients with breast cancer and gynecologic malignancies, which was reported previously (23, 26). The planned dose expansions in patients of each tumor type that showed tumor response signals were eventually discontinued because of the ongoing liposomal doxorubicin shortage in the United States (28).

All 136 patients were evaluable for toxicity. There were no treatment-related deaths. One DLT of grade 4 fistula was observed in one patient at dose level 3, and 2 DLTs consisting of one grade 4 thrombocytopenia and one grade 4 neutropenia were observed in 2 of the 3 patients treated at dose level 7, which was deemed beyond the MTD. Six patients showed decreased left ventricular ejection fraction including grade 2 (n = 4) and grade 3 (n = 2) heart failure (Table 2). Two dose levels were expanded as seen in Table 3. Initially, a safety dose expansion with an additional 14 patients at dose level 6 was conducted: no DLTs were observed. As per study design, up to 14 patients with the same tumor type who showed clinical response signals were allowed to be enrolled onto the highest dose level shown to be safe. A total of 57 such patients were enrolled onto dose level 6 without DLTs being observed. However, 20 patients (35%) eventually required a dose reduction, mainly because of grade 2 or higher hand–foot syndrome, mucositis, thrombocytopenia, or fatigue beyond the DLT window of the first cycle. Thus, another dose expansion was conducted at dose level 5 to include patients with diseases in which responses had been seen. A total of 49 such patients were enrolled: no DLTs were observed. At dose level 5, 14 patients (29%) eventually required a dose reduction, mainly for mucositis, which was felt to be due to temsirolimus. Therefore, a new intermediate dose level, designated dose level 5A, was created to administer higher doses of liposomal doxorubicin and lower doses of temsirolimus, as seen in Table 2. Because of the national shortage of liposomal doxorubicin (28), only 3 patients were enrolled: no DLTs and additional toxicities were observed. Therefore, dose level 6 was determined to be the MTD.

Overall, approximately 32% of patients (n = 43) required a dose reduction, generally after the first 2 cycles. The most common reasons for dose reduction were thrombocytopenia (n = 11, 26%), fatigue (n = 8, 19%), mucositis (n = 7, 16%), neutropenia (n = 6, 14%), hand–foot syndrome (n = 5, 12%), infusion reactions (n = 5, 12%), increased creatinine (n = 2, 5%), uncontrolled hypertension (n = 1, 2%), decreased left ventricular ejection fraction (n = 1, 2%), and nausea (n = 1, 2%). Grade 3 or higher treatment-related adverse events (>5%) included anemia, neutropenia, thrombocytopenia, mucositis, hand–foot syndrome, hypertension, and fistula, as shown in Table 2. No Grade 3 or higher adverse events in blood glucose and lipid profiles were seen. The most common grade 2 side effects at dose level 6 (MTD) were anemia (n = 26, 46%), mucositis (n = 10, 18%), hand–foot syndrome (n = 8, 14%), thrombocytopenia (n = 7, 12%), hypercholesterolemia (n = 7, 12%), hypertriglyceridemia (n = 6, 11%), neutropenia (n = 3, 5%), and increased creatinine (n = 3, 5%).

All patients were included for efficacy evaluation. As shown in Fig. 1, of 136 patients treated, 28 patients (21%) attained SD at 6 months or more, 27 patients (20%) a PR, and 2 patients (1%) achieved a CR [total SD ≥ 6 months/PR/CR = 57 (42%); Fig. 1]. The median duration of PR/CR was 9 months (range: 4–36+ months) as determined using the Kaplan–Meier method. It was of great interest to note that only patients with metaplastic breast cancer achieved a CR. One patient received a total of 8 months of study therapy with resolution of her perihepatic implants, right retrocrural node, nodular pleural disease, and right subcarinal node. The other patient who had a PI3K mutation (H1047R) received a total of 12 months of study therapy with biopsy-proven right lower lobe metastasis resolved. For personal reasons, these 2 patients then continued taking an mTOR inhibitor (temsirolimus or everolimus) as a single agent without evidence of tumor progression for 36+ and 18+ months, respectively.

As shown in Table 4, patients without a history of bevacizumab exposure had a significantly greater chance of achieving PR/CR than those with prior bevacizumab exposure (24/90, 26% vs. 5/46, 11%; P = 0.045). If SD at 6 months or more was included, there was no significantly statistical difference between those with and without prior bevacizumab exposure.

Of the 76 patients treated at dose levels 1 to 5A, 33 (43%) achieved SD ≥ 6 months/PR/CR; of the 60 patients treated at dose level 6 or above, 24 (40%) achieved SD ≥ 6 months/PR/CR (P = 0.73); 13 (17%) patients treated at dose levels 1 to 5A achieved PR/CR compared with 16 (27%) at dose level 6 or higher (P = 0.2). These results suggested that there was no statistical difference between the SD ≥ 6 months/PR/CR rate at higher versus lower dose levels. Furthermore, some tumor regression was seen in 4 of 5 patients treated at the lowest dose level, and 3 of 5 patients at that dose level achieved SD ≥ 6 months/PR/CR, in spite of the small doses used (bevacizumab 5 mg/kg IV every 3 weeks, temsirolimus 12.5 mg IV weekly and liposomal doxorubicin 10 mg/m² every 3 weeks). These results indicate that there is no clear relationship between dose and rate of SD ≥ 6 months/PR/CR, and that salutary effects can be achieved even at very low dose levels. On the other hand, there was a numerically higher rate of PR/CR at dose level 6 or more (albeit statistically insignificant) and the CRs only occurred at dose levels 5 and 6.

Table 4 shows PR/CR rates of 67% (5/8), 42% (5/12), and 40% (6/15), in parotid gland adenocarcinoma, metaplastic breast cancer, and endometrial endometrioid carcinoma, respectively (Fig. 2). Among 4 patients with clear cell carcinoma of the ovary, 2 achieved SD ≥ 6 months/PR/CR (50%). In non-metaplastic breast cancer, the rate of SD ≥ 6 months/PR/CR was 30% (5/17); in epithelial ovarian cancer, 43% (10/23); in carcinoma of the cervix, 33% (5/15); in colorectal cancer, 47% (8/17); in high-grade sarcoma, 22% (2/9); and in thymoma, 57% (4/7). Therefore, salutary effects were achieved in a wide variety of tumors.

The majority of patients (n = 120) had at least one baseline molecular biomarker (Table 4). Among these biomarkers, only PIK3CA mutation and/or PTEN loss/mutation was associated with significantly higher PR/CR rates [39%, 11/28 patients with PIK3CA mutation and/or PTEN loss/mutation versus 16%, 12/74 patients without PIK3CA mutation and/or PTEN loss/mutation (P = 0.018)]. However, if SD of 6 or more months was included, there was no difference in rates of SD ≥ 6 months/PR/CR (50%, 14/28 versus 42%, 31/74; P = 0.51). Of 29 patients who achieved PR/CR, 11 (38%) had a PIK3CA mutation and/or PTEN loss/mutation. Of 15 patients with endometrial endometrioid carcinoma, 13 were tested for PIK3CA mutation and/or PTEN loss/mutation and 8 were found to be positive (62%). Five out of 8 patients (63%) with a PIK3CA mutation and/or PTEN loss/mutation achieved SD ≥ 6 months/PR/CR, whereas 1 of 5 patients (20%) without these aberrations achieved SD ≥ 6 months/PR/CR (P = 0.27). Six out of 9 tested patients with metaplastic breast cancer had a PIK3CA mutation or PTEN loss/mutation (67%); 3 of 6 patients with a PIK3CA mutation and/or PTEN loss/mutation achieved SD ≥ 6 months/PR/CR, whereas 2 of 3 patients without a PIK3CA mutation and/or PTEN loss/mutation achieved SD ≥ 6 months/PR/CR. Among 5 patients with parotid cancer tested, none had a PIK3CA mutation and/or PTEN loss/mutation.

Activated redundant pathways in advanced malignancies along with evolving self-induced protective mechanisms in response to single-agent therapy contribute to low clinical response rates. It has become increasingly clear that simultaneous inhibition of multiple intracellular signaling pathways is required to improve clinical efficacy. This can be best accomplished with combination therapy composed of multiple agents. Accordingly, we conducted this phase I study using combined bevacizumab and temsirolimus plus liposomal doxorubicin in patients with advanced malignancies to define the safety profile and identify tumor response signals. Overall, patients tolerated the combined regimen with bevacizumab and temsirolimus at their FDA-approved dosages plus liposomal doxorubicin at 30 mg/m² every 3 weeks. Although the number of patients in each subgroup is small, this regimen led to a high rate of SD ≥ 6 months/PR/CR in several tumor types, including parotid, metaplastic breast, endometrial, ovarian, thymoma, and colorectal cancer (all over 40%) and in cervical, nonmetaplastic breast cancer, and sarcoma (between 22% and 33%; Table 4), despite patients having a median of 4 prior chemotherapy regimens. The overall rate of SD ≥ 6 months/PR/CR was 42% (57/136 participants; Table 4). Responses were durable and the median duration of PR/CR was 6 months (range, 1.5–22 months).

Of interest, patients with a PIK3CA mutation and/or PTEN loss/mutation did well. Among patients with a PIK3CA mutation and/or PTEN loss/mutation, the PR/CR rates were 39% compared with only 16% among those without these aberrations (P = 0.018). Furthermore, of the 29 patients who achieved PR/CR, 11 (38%) had a PIK3CA mutation and/or PTEN loss/mutation. PTEN loss, which often reflects a PTEN mutation, as well as PIK3CA mutations, activates the PI3K/AKT/mTOR axis (29). Because mTOR is downstream of these pathways, the role of the mTOR inhibitor temsirolimus in attaining a response may be important (30, 31). However, in many tumor types, even patients without a PIK3CA mutation and/or PTEN loss/mutation responded, and statistically significant differences between patients with and without molecular aberrations were not apparent. This finding could be explained by one or more of the following: (i) the small number of patients with individual histologies precluded a robust analysis; (ii) the existence of other pathway aberrations that were not tested (e.g., AKT or mTOR mutations); (iii) the presence of PIK3CA mutations in areas of the gene not assayed by our CLIA laboratory (as our assay was limited to exons 9 and 20); (iv) the actions of bevacizumab or liposomal doxorubicin unrelated to the PI3K/AKT/mTOR axis; (v) the synergy among the 3 agents.

Several other observations were noted. First, several agents with different toxicity profiles when used individually could be combined at their U.S. Food and Drug Administration (FDA)-approved dosages with good tolerance. These data support the practical strategy of using these agents in combination to simultaneously target multiple pathways, thus producing improved antitumor activity without excessive toxicity. Accordingly, in this clinical trial, we were able to administer dosages of bevacizumab and temsirolimus at their FDA-approved maximum doses (32, 33) and schedules while liposomal doxorubicin was given at a dose of 30 mg/m² once every 3 weeks (34), equivalent to the dose currently used as a single agent at 40 mg/m² once every 4 weeks (35). Second, there was no statistical relationship between dose level and response. Indeed, 4 of 5 patients at the lowest dose level showed some tumor regression. These findings are reminiscent of those previously reported by our group showing that patients taking lower doses of phase I–targeted agents do not fare worse than individuals taking higher doses (36). A caveat is, however, that the response (PR/CR) rate was numerically higher (albeit not statistically significant) at the higher doses, and that the CRs occurred at dose levels 5 and 6.

When considering the clinical relevance of our findings, several limitations should be borne in mind. First, selection bias based on eligibility criteria may limit the generalizability of our findings, as it does for many clinical trials. However, it should be noted that while these patients mostly had a high performance status and intact organ function, they had also been heavily pretreated. Second, we had a limited sample size available for subgroup analyses, which confounded the ability to validate statistical significance in individual histologies. Third, the recommended phase II dose was difficult to establish, in part because of the nationwide shortage of liposomal doxorubicin. On the other hand, a substantial number of patients were treated at various dose levels, producing significant safety data. Further, the MTD was found to be dose level 6. The MTD was based on first cycle toxicity, as is standard practice. The recommended phase II dose may differ from the MTD because it depends on toxicity that emerges over time. With active regimens such as this one, patients may stay on therapy for months or even years. We found that, with time, 35% (20/57) of patients at dose level 6 eventually required dose reductions, mainly because of grade 2 or higher hand–foot syndrome, mucositis, thrombocytopenia, or fatigue. At dose level 5, 29% (14/49) of patients eventually required a dose reduction, in general for mucositis (probably due to temsirolimus). Further analyses showed that dose reduction was significantly associated with SD ≥ 6 months/PR/CR both at dose level 5 (P = 0.0031) and dose level 6 (P = 0.006) or combined (P < 0.0001). This observation may have emerged because patients who had antitumor activity stayed on drug longer and, therefore, eventually required dose reduction; alternatively it could be that toxicities reflect target impact and hence that toxicity and response are correlated. Regardless, it seems that approximately one-third of patients on either dose level 5 or 6 will, with time, require a dose reduction, but that the side effects that surface are reversible and not life threatening. Hence, it seems reasonable to start patients on dose level 6 (the MTD) and reduce the dose if chronic side effects necessitate such a change. However, it should be noted that liberal criteria used to define DLT within the first cycle (21 days) in heavily pretreated patients with advanced solid tumors might result in establishing a higher MTD.

In conclusion, the study regimen that combines bevacizumab at 15 mg/kg once every 3 weeks and the mTOR inhibitor temsirolimus at 25 mg once every week plus liposomal doxorubicin at 30 mg/m² once every 3 weeks was well tolerated. The overall rate of SD ≥ 6 months/PR/CR was over 50% in heavily pretreated metaplastic breast cancer, parotid gland tumor, thymoma, and endometrial endometrioid cancer. Furthermore patients with ovarian cancer, colorectal cancer, sarcoma, cervical and nonmetaplastic breast cancer all achieved rates of SD ≥ 6 months/PR/CR of 22% to 47%, albeit in small numbers of individuals. Patients with a PIK3CA mutation and/or PTEN loss/mutation had a rate of SD ≥ 6 months/PR/CR of 50% (PR/CR rate of 39%). The median duration of PR/CR for all 29 patients who achieved tumor responses was 9 months (range: 4–36+ months). These observations suggest that further clinical evaluation of this regimen is warranted.

R. Kurzrock: commercial research grants from Genentech/Roche, Wyeth, and Janssen; and honoraria from speakers' bureau from Genentech/Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Moroney, S. Fu, T. Helgason, R. Kurzrock

Development of methodology: J. Moroney, S. Fu, T. Helgason

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Moroney, S. Fu, S.L. Moulder, G.S. Falchook, T. Helgason, C.F. Levenback, D.S. Hong, A. Naing

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Moroney, S. Fu, S.L. Moulder, T. Helgason, C.F. Levenback, D.S. Hong

Writing, review, and/or revision of the manuscript: J. Moroney, S. Fu, S.L. Moulder, G.S. Falchook, T. Helgason, C.F. Levenback, D.S. Hong, A. Naing, J.J. Wheler, R. Kurzrock

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Moroney, S. Fu, T. Helgason

Study supervision: J. Moroney, S. Fu, S.L. Moulder, T. Helgason

The authors thank Joann Aaron in the Department of Investigational Cancer Therapeutics at MD Anderson Cancer Center for editing our manuscript.

This study was supported in part by Grant Number RR024148 from the National Center for Research Resources, a component of the NIH Roadmap for Medical Research to R. Kurzrock.

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.