Purpose: NGR-hTNF exploits the tumor-homing peptide asparagine-glycine-arginine (NGR) for selectively targeting TNF-α to an aminopeptidase N overexpressed on cancer endothelial cells. Preclinical synergism with cisplatin was displayed even at low doses. This study primarily aimed to explore the safety of low-dose NGR-hTNF combined with cisplatin in resistant/refractory malignancies. Secondary aims included pharmacokinetics (PKs), pharmacodynamics, and activity.
Experimental Design: NGR-hTNF was escalated using a doubling-dose scheme (0.2–0.4–0.8–1.6 μg/m2) in combination with fixed-dose of cisplatin (80 mg/m2), both given intravenously once every three weeks. PKs and circulating TNF-receptors (sTNF-Rs) were assessed over the first three cycles.
Results: Globally, 22 patients (12 pretreated with platinum) received a range of one to ten cycles. Consistently with the low-dose range tested, maximum-tolerated dose was not reached. No dose-limiting toxicities (DLTs) were observed at 0.2 (n = 4) and 0.4 μg/m2 (n = 3). One DLT (grade 3 infusion-related reaction) was observed at 0.8 μg/m2. This dose cohort was expanded to six patients without further DLTs. No DLTs were noted also at 1.6 μg/m2 (n = 3). NGR-hTNF exposure increased dose-proportionally without apparent PK interactions with cisplatin. No shedding of sTNF-Rs was detected up to 0.8 μg/m2. At the dose level of 0.8 μg/m2, expanded to 12 patients for activity assessment, a platinum-pretreated lung cancer patient achieved a partial response lasting more than six months and five patients maintained stable disease for a median time of 5.9 months.
Conclusions: The combination of NGR-hTNF 0.8 μg/m2 with cisplatin 80 mg/m2 showed favorable toxicity profile and promising antitumor activity. Clin Cancer Res; 17(7); 1964–72. ©2011 AACR.
NGR-hTNF consists of human TNF-α fused with the tumor-homing peptide asparagine-glycine-arginine (NGR), which is able to selectively bind to an aminopeptidase N isoform highly expressed on tumor blood vessels. Preclinically, NGR-TNF was found to be strongly more active than untargeted TNF-α. This phase I study was prompted by preclinical models revealing that even low doses of NGR-TNF were able to induce synergistic effects when combined with cisplatin, by damaging the tumor vasculature and thereby increasing the selective intratumoral uptake of the cytotoxic agent. This study confirmed that NGR-hTNF can be safely combined with standard dose of cisplatin in heavily pretreated patients, with evidence of durable disease control. Pharmacodynamic relationships between plasma concentrations of NGR-hTNF and shedding kinetics of circulating TNF-α receptors were also detected, showing a correlation with the duration of therapy. Additional studies of NGR-hTNF plus cisplatin are in progress.
TNF-α is a powerful antitumor cytokine originally identified for its ability to induce in experimental cancers massive hemorrhagic necrosis, which is mainly mediated by apoptosis of tumor-associated endothelial cells (1). The apoptotic pathway is triggered by TNF-α binding with the TNF-receptor 1 (sTNF-R1), whereas receptor TNF-R2 lacks a death domain. Both receptors are also shed, thus competing with cell-surface receptors for free ligand (2).
To fully harness these anticancer effects, increasing doses of TNF-α were progressively tested in humans by using a variety of dosing schedules (3). Disappointingly, the initial dose-finding trials of TNF-α administered systemically were associated to severe toxicities, the maximum tolerated dose (MTD) being significantly lower than the estimated effective dose (3–8). Later on, through isolated limb or hepatic perfusion it was possible to deliver loco-regionally doses of TNF-α 10-fold higher than MTD. Interestingly, when this cytokine was combined with chemotherapy, high response rates were reported in several tumors, with an acceptable toxicity profile (9–12).
Aiming to increase the TNF-α therapeutic index, a ligand-directed vascular-targeting approach was explored. For this purpose, NGR-hTNF was prepared by fusing the N-terminus of recombinant human TNF-α with the C-terminus of the peptide asparagine-glycine-arginine (NGR), which is a selective ligand of an aminopeptidase N (APN, CD13) isoform overexpressed by angiogenic tumor vessels (13–16). The functional role of APN in promoting angiogenesis was recently confirmed in a CD13-null murine model showing that although APN activity is not essential for embryonic and fetal development, including de novo blood vessel formation (i.e., vasculogenesis) and normal adult function, it is crucial for the pathologic development of newly formed blood vessels from preexisting blood vessels (i.e., angiogenesis; ref. 17).
Preclinically, NGR-TNF was found to be at least 10-fold more active than untargeted TNF-α, and more interestingly, showed a biphasic (U-shaped) dose–response curve with significant anticancer activity observed even at very low doses (100 pg/mouse; refs. 13, 18) equivalent in humans to 0.2 μg/m2, the starting dose of the phase I.
In addition to inducing apoptosis of cancer endothelial cells, TNF-α is also able to rapidly decrease the interstitial fluid pressure, a process pivotal to facilitating the tumor-selective uptake of chemotherapeutic agents, by altering the integrity of endothelial cell barrier. Consistently with a selective vascular-targeting activity, even low-dose NGR-TNF displayed synergistic effects when combined with cisplatin in murine tumor models (19). Maximal synergism was sequence- and time-dependent being observed with NGR-hTNF given 2 hours before chemotherapy dosing, whereas reduced effects were noted with longer or shorter intervals (19).
Moreover, CD13-overexpressing tumors, exhibiting reduced sensitivity for cisplatin in vivo, did not show decreased sensitivity in vitro (20). These findings confirmed that other indirect features, such as impaired drug delivery to the tumor rather than direct CD13-induced resistance to cisplatin, were involved.
In early-stage clinical development, a phase I study testing a wide-ranging dose interval (from 0.2 to 60 μg/m2) established the MTD of NGR-hTNF at 45 μg/m2 when given once every 3 weeks, with dose-limiting toxicities (DLTs) being characterized by grade 3 dyspnoea and acute infusion reaction (21). Shedding of soluble TNF-α receptors was not detected at low doses (≤0.8 μg/m2), while thereafter increased proportionally with dose. No objective responses were observed, but of 6 patients with stable disease who received 6 or more cycles, 5 were treated with low doses (21). An additional trial aiming to further explore the low-dose range (from 0.2 to 1.6 μg/m2) selected 0.8 μg/m2 as the optimal biological dose based on dynamic imaging changes, soluble TNF-Rs kinetics, safety, and preliminary activity (22). To overcome the counterregulatory mechanism of the shedding of soluble TNF–α-receptors noted at higher doses and given the apparent similar preclinical and clinical activity between low and high doses, only the low-dose range previously tested in the single-agent trial (ref. 22; 0.2–0.4–0.8–1.6 μg/m2) was used in this phase I study. The study primarily assessed the safety of NGR-hTNF in combination with a fixed dose of cisplatin in patients with refractory solid tumors. Secondary study aims were pharmacokinetics (PKs), pharmacodynamics (PDs), and preliminary antitumor activity.
Materials and Methods
Patients 18 years of age or older with advanced solid tumors not amenable to any clinical improvement by current standard therapies and suitable for cisplatin were enrolled. Additional eligibility requirements included: ECOG performance status of 0 to 1; absolute neutrophil count >1.5 × 109/L; platelet count >100 × 109/L; total bilirubin <1.5 × upper limit of normal (ULN); aspartate and alanine aminotransferase <2.5 × ULN in absence of liver metastasis or <5 × ULN in presence of liver metastasis; serum creatinine <1.5 × ULN. Patients with significant cardiac, infectious, or peripheral vascular diseases were excluded, and patients completing systemic therapy within 4 weeks or having surgery or receiving corticosteroids within 2 weeks before treatment start. Patients with brain metastases were cautiously excluded because at the time of study design the risk of tumor-associated hemorrhage using an antivascular agent was unknown (23). The study protocol was approved by the Ethical Committees of the respective participating Institutions. All patients signed a written informed consent.
Study Design and Treatment Plan
This was a 2-center, phase I, dose-escalation study with a minimum of 3 patients who were administered each of 4 low-dose levels (DLs) of NGR-hTNF escalated with a doubling dose-scheme (0.2–0.4–0.8–1.6 μg/m2). NGR-hTNF (supplied by MolMed) was given intravenously as a 1-hour infusion followed after approximately 1-hour interval by a fixed dose of cisplatin (80 mg/m2), delivered as a 1.5-hour intravenous infusion. Both drugs were given once every 3 weeks, to maintain sequence and timing of administration in combination derived from preclinical models (19).
In case of continued response or stable disease, cisplatin was administered for up to 6 cycles, whereas NGR-hTNF was continued until progression, unacceptable toxicity, patient refusal, or physician decision occurred. According to institutional guidelines, adequate hydration to keep urine output ≥100 mL/h and standard antiemetic protocol for highly emetogenic chemotherapy was used for cisplatin infusion. 5HT3-antagonist and aprepitant (125 mg per os) were given after NGR-hTNF, while dexamethasone (16 mg intravenously) was given after cisplatin to avoid any interaction with TNF-α pathway activities. For retreatment on next cycle, all treatment-related toxicities should be recovered to grade 1 or less (grade 2 or less in case of neurotoxicity). If a patient was unable to meet retreatment criteria, both drugs were delayed for 1 week for up to 3 weeks. No NGR-hTNF dose reduction was allowed. In presence of chills, prophylaxis with acetaminophen/paracetamol (1,000 mg per os) was recommended for next cycles, while it was allowed before first cycle at Investigator's discretion.
Adverse events (AEs) were graded according to the Common Terminology Criteria for Adverse Events, version 3.0. DLT applicable to the study was defined as any grade 3 to 4 toxicity clearly related to NGR-hTNF. Exceptions were nausea, vomiting, chills, and fever, which could be rapidly controlled with appropriate treatments. The well-known toxicity profile of cisplatin had to be taken into account in the decision process. Three patients were enrolled to each cohort. If no DLT was observed during the first cycle, an additional 3 patients were entered at the next higher DL with dose escalation continuing until DLT was observed. If 1 of 3 patients experienced a DLT, an additional 3 patients were entered at that DL. If 2 of 3 to 6 patients experienced a DLT at a given DL, the next lower DL would be the MTD of combination, and an additional 6 patients were to be enrolled into a dose expansion at the MTD. Because MTD was not reached in this study, after the completion of the highest DL planned (1.6 μg/m2), these additional 6 patients were enrolled at the previously defined optimal biologic dose of 0.8 μg/m2.
Pharmacokinetic and Pharmacodynamic Analysis
Intensive PK blood sampling was performed on day 1 of the first 3 treatment cycles with samples drawn at baseline (before NGR-hTNF dosing) and on-treatment at 30, 60, 90 (just before cisplatin administration), 135, 179 (just before the end of cisplatin infusion), 210, 240, 300, and 360 minutes. NGR-hTNF and soluble TNF-Rs (sTNF-R1 and sTNF-R2) levels were computed by using an enzyme-linked immunosorbent assay. For NGR-hTNF and cisplatin, maximum plasma concentration (Cmax), area under the plasma concentration–time curve up to the last detectable concentration (AUC0-t last), and apparent terminal elimination half-life (t1/2) were estimated from plasma concentration–time data using standard noncompartmental methods in WinNonlin (Version 4.1). The clearance (CL) was calculated as dose/AUC and the volume of distribution (Vss) as CL*(MRT – T/2), where MRT is the mean residence time and T is the duration of infusion.
The PD variables determined for soluble TNF-R1 and TNF-R2 were Emax (maximum plasma concentration), AUC (area under the plasma concentration–time curve), and tmax (time of maximum concentration). The levels of sTNF-Rs were either baseline-normalized by subtracting the time-zero value to all other time-points values or dose-normalized by dividing the estimated concentrations for DLs.
Descriptive statistics were provided using medians (with ranges) and means (with standard deviations) for continuous variables and proportions for categorical variables. A Kruskal–Wallis analysis of variance by ranks was used to assess dose proportionality and compare differences in baseline-normalized sTNF-Rs as a function of dose. The degree of association between continuous variables was quantified by Spearman rank correlation coefficient. Tumor evaluation was done every other cycle according to RECIST criteria. Kaplan–Meier estimates were computed for progression-free survival (PFS).
From July 2007 to April 2008, 22 patients (14 men and 8 women) with a median age of 60 years (range, 47–75 years) and a PS of 0 (55%) or 1 (45%) were enrolled. All patients were heavily pretreated: all had earlier received systemic therapy with a median number of 3 treatment lines (range, 1–6). Twelve patients (55%) were previously treated with a platinum-containing regimen (range, 1–3).
A total of 77 cycles of NGR-hTNF (range, 1–10 cycles) and 57 of cisplatin (range, 1–6 cycles) were delivered. In all the 22 enrolled patients, treatment discontinuation was a result of either symptomatic deterioration (n = 5) or radiologically documented progressive disease (n = 17).
As expected testing the low-dose range of NGR-hTNF, MTD was not reached. No DLTs were observed at doses of 0.2 μg/m2 (n = 4) and 0.4 μg/m2 (n = 3). At 0.8 μg/m2, a 65-year-old mesothelioma patient, previously treated with 3 regimens, had a transient acute infusion reaction consisting of grade 3 dyspnoea with hypoxia, considered as a DLT. Two weeks after full resolution of these events and completion of infusion, the same patient experienced grade 4 dyspnoea and respiratory failure. A computed tomography scan revealed pleural progressive disease. Even if this event was not deemed surely dose-related, the corresponding cohort was expanded to 6 patients for full safety check, without any further DLT observed. Consequently, the NGR-hTNF dose was escalated to 1.6 μg/m2 (n = 3), with no DLT detected also in this cohort. At this DL of 1.6 μg/m2, an additional patient had serious respiratory AEs after 4 cycles. This patient, a 67-year-old patient with a metastatic pulmonary carcinoid tumor, developed grade 4 dyspnoea, hypoxia, nonneutropenic pneumonia, atelectasis, and grade 3 hypertension. This patient showed also concurrent radiological tumor progression on the fourth cycle.
Most commonly reported toxicities, over all cycles and regardless of drug relationship, were nausea, pain, increased serum creatinine, and chills (Table 1). Off all recorded AEs (n = 248), the majority (79%) were mild-to-moderate in intensity, with grade 3 and 4 being 17% and 4%, respectively. The combination was well tolerated without apparent difference in either frequency or intensity of AEs by DL. No grade 4 hematologic AEs occurred, while 3 patients had grade 4 nonhematologic AEs, which were however considered unrelated to both study drugs.
Regarding cisplatin-related toxicity, a 25% dose reduction was required in 2 patients (9%), for grade 3 atrial flutter and tinnitus, respectively, and a 1 to 3-week dose delay in 7 cycles (13%). Cisplatin was discontinued early for toxicity in 3 platinum-pretreated patients (after 1, 4, and 5 cycles) because of persistent grade 2-increased serum creatinine and in 2 platinum-naive patients (after 3 and 5 cycles) due to recurrent grade 2 tinnitus and neutropenia, respectively. These 5 patients received further courses of single-agent NGR-hTNF.
The observed toxicities seem to be within the expected range of AEs for single-agent cisplatin (24,25). Overall, renal impairment was mild-to-moderate in 9 patients and severe in 1 case, while grade 1 to 2 and 3 hematologic toxicities were registered in 9 and 2 patients, respectively.
Only 25 (6%) out of all AEs were related to NGR-hTNF and the most frequent was chills, experienced by 9 patients (41%) over 18 cycles (23%). These events were infusion-time related, short-lived, and mild-to-moderate in severity. No grade 4 AEs related to NGR-hTNF were reported.
Seventeen patients had the PK studies completed. NGR-hTNF and cisplatin mean PK parameters based on first-cycle data are summarized in Table 2. The concentration–time profiles and the individual Cmax levels of NGR-hTNF over the first 3 treatment cycles are plotted in Figure 1A and B, respectively. Both Cmax and AUC of NGR-hTNF increased proportionally with dose (P = 0.005 and 0.02, respectively). Conversely, time-to-maximum concentration (range, 0.8–1.0 hour), half-life (range, 2.6–6.0 hours), and systemic clearance (range, 151–274 mL/min/m2) were independent of dose. Across the planned dose cohorts, the volume of distribution (range, 30.8–50.4 L/m2) showed values slightly greater than total plasma volume, thus suggesting moderate binding or distribution of NGR-hTNF. No evidence of accumulation was noted after multiple doses of NGR-hTNF, as indicated by the mean concentrations at day 21 compared to those at day 1. Despite of large interpatient variability, PKs were in reasonable agreement with those found in studies testing low-dose NGR-hTNF given as single agent (22).
Even if slightly increased values of cisplatin Cmax were detectable on cycle 3 versus cycle 1 and 2 (Fig. 1C), mean cisplatin PKs were comparable among the patient cohorts, regardless of the NGR-hTNF dose (Fig. 1D). Comparing the concentration-time curves of cisplatin from either human plasma matrix or ultrafiltrate matrix, a 50% difference in Cmax was noted, accounting for a 50% protein-binding of this agent. Overall, there were no apparent PK interactions between NGR-hTNF and cisplatin.
The plasma levels of soluble TNFα receptors were monitored over the first 3 cycles after NGR-hTNF dosing in 17 patients. The concentration–time curves of sTNF-R1 (Fig. 2A) and sTNF-R2 (Fig. 2B) remained constant over time at 0.2, 0.4, and 0.8 μg/m2, indicating no induction of circulating TNF-α receptors by NGR-hTNF. Conversely, the concentrations of sTNF-R1 and sTNF-R2 at 1.6 μg/m2 were significantly higher than at lower doses (P = 0.001 and 0.0001, respectively). Similarly, the median time-to-peak of both sTNF-R1 and sTNF-R2 at 1.6 μg/m2 (1.6 and 1.5 hours, respectively) resulted faster than at 0.2 (3.8 and 4.3 hours), 0.4 (3.5 and 6.0 hours), and 0.8 μg/m2 (4.5 and 3.0 hours).
After the first cycle, plasma NGR-hTNF exposure significantly correlated with both the AUC and the Emax of sTNF-R2 (r = 0.62, P = 0.008 and r = 0.61, P = 0.009, respectively), whereas no significant relationships were detectable for sTNF-R1 (r = 0.29, P = 0.26 and r = 0.22, P = 0.39, respectively).
In addition, a significant inverse correlation was observed between the levels of sTNF-Rs after the first cycle and the duration of NGR-hTNF therapy as shown in Figure 2E and F. Both the Emax of sTNF-R1 and sTNF-R2 (r = –0.53, P = 0.03 and r = –0.56, P = 0.02, respectively) and the AUC of sTNF-R1 and sTNF-R2 (r = –0.53, P = 0.03 and r = –0.65, P =.005, respectively) inversely correlated with the number of NGR-hTNF cycles administered.
Globally, 17 patients reached their first tumor evaluation and were assessable for response, whereas 5 patients come off study early because of symptomatic deterioration (Table 3). Considering that 0.8 μg/m2 was previously selected in a single-agent trial (22) as optimal biological dose, the corresponding cohort in this study was expanded by an additional 6 patients to a total of 12 patients for antitumor assessment. At this DL, a platinum-pretreated non–small cell lung cancer (NSCLC) patient had a partial response lasting 6.6 months. In addition, 5 patients had stable disease for a median time of 5.9 months. Responding and stable disease patients (n = 8) received a median of 5 cycles of combination therapy (ranges, 4–10 cycles of NGR-hTNF and 1–6 cycles of cisplatin). The 6-month PFS rates for all patients (n = 22), for platinum-pretreated patients (n = 12), and for patients enrolled at 0.8 μg/m2 (n = 12), were 24%, 38%, and 40%, respectively.
This study has demonstrated that NGR-hTNF might be added to full dosing of cisplatin with a well-tolerated and nonoverlapping toxicity profile. Most common toxicities were those expected from each agent independently, consisting of nausea, elevated creatinine, and vomiting for cisplatin and chills for NGR-hTNF. Chills were mild-to-moderate in severity, easily manageable, and unrelated to dose. Notably, only 6% of all registered AEs were attributable to NGR-hTNF. The cardiovascular toxicities previously described for vascular-targeting agents (including acute coronary and thrombophlebitic syndromes; clinically relevant alterations in blood pressure, heart rate and ventricular conduction; ref. 24) did not appear in this study, likely because of targeted delivery of low doses.
The large interpatient variability observed in drug exposure, probably due to low number of patients enrolled in each cohort, did not allow defining an optimal dose based on PK parameters. However, mean PK variables of cisplatin did not change across DLs and over treatment, thus suggesting that NGR-hTNF did not affect cisplatin disposition. This is consistent with the lack of worsening of the cisplatin-related toxicity observed in clinical assessment.
Potential limitations of this study were that the full dose range up to MTD was not explored and the changes in tumor vascularity were not assessed by dynamic imaging. As expected by using low-dose NGR-hTNF, this trial did not formally define an MTD. Indeed, combining biological with chemotherapeutic agents is especially challenging with regard to dose selection. The traditional paradigm of chemotherapy combination, based on careful drug-dose escalation until MTD and avoidance of overlapping toxicity, might be unsuitable for biological agents characterized by toxicity profiles, which are mostly different and lower than those of cytotoxic agents. Moreover, given the cytostatic nature and the lack of dose–effect relationship for most targeted agents, increasing drug dose to MTD may be unnecessary for drug effect (26).
Accordingly, this study was prompted by preclinical models revealing that even low doses of NGR-TNF induced synergistic effects when combined with various chemotherapeutic agents, including cisplatin, likely by damaging the tumor vasculature and thereby maximizing tumor-selective uptake of cytotoxic agents (18,19). Notably, the observed synergy was shown with all chemotherapeutic drugs tested and resulted strongly dependent on the timing of dosing schedule. Indeed, NGR-TNF did not significantly increase the in vitro cytotoxicity of chemotherapy, implying that the synergism detected in vivo might rely on NGR-TNF effects on host cells rather than on tumor cells (19). Consistently, maximal synergism was displayed with a 2-hour interval between NGR-TNF and subsequent chemotherapy delivery, thus suggesting a necessary time for the drug to affect tumor vessels (19).
Although not directly addressed in this study, the antivascular effects of NGR-hTNF were confirmed in earlier clinical trials (21,22), wherein significant reductions of tumor blood flow and perfusion were detected by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The decreased tumor vascularity, as measured by DCE-MRI, might be supportive of the synergistic effects between NGR-hTNF and cisplatin. Otherwise, the synergism could be merely ascribed to targeting 2 different cell populations (27).
Notwithstanding the fact that the selection of an optimal biological dose is highly challenging, it is worthy to note that 0.8 μg/m2 was selected in a previous single-agent trial based on both more pronounced antivascular effects and no shedding of soluble TNF-Rs noted at this DL (28). The circulating receptors might compete with the cell-surface receptors for free TNF-α, thus blocking its bioavailability and activity, with the amount and speed of this shedding being linearly correlated with serum TNF-α level (29). To retest the drug effect on soluble TNF kinetics also in a combination trial, the entire low-dose range (0.2–0.4–0.8–1.6 μg/m2) already explored in the single-agent experience (22) was used in this study. Consistently, also in this study the plasma levels of soluble receptors did not increase up to 0.8 μg/m2 and, of note, the level after the first cycle inversely correlated with NGR-hTNF treatment duration. This interesting finding (i.e., the lower the ratio of soluble receptors to drug dose, the longer the potential clinical benefit) deserves further investigations in larger patient population.
Finally, even though response assessment was not a primary study endpoint, the disease control achieved in patients heavily pretreated with chemotherapy, 55% of which with a platinum compound, seems to be interesting also taking into account the minimal and nonoverlapping toxicity of NGR-hTNF with cisplatin. Specifically, in the cohort of 12 patients enrolled at 0.8 μg/m2 and pretreated with a median of 4 regimens, 6 patients maintained a disease control for a median time of 6 months.
In conclusion, this phase I study showed that the combination of low-dose NGR-hTNF and standard-dose cisplatin can be safely administered to heavily pretreated patients. The observed tolerability profile and preliminary clinical activity warrant further phase II clinical exploration of NGR-hTNF 0.8 μg/m2 and cisplatin 80 mg/m2 in platinum-sensitive solid tumors. A randomized phase II trial (www.clinicaltrials.gov/ct2/show/NCT00994097) testing standard platinum-based chemotherapy regimens with or without NGR-hTNF is currently open to accrual in NSCLC patients.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
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.
- Received May 21, 2010.
- Revision received July 20, 2010.
- Accepted August 5, 2010.
- ©2011 American Association for Cancer Research.