SU5416 Plus Interferon α in Advanced Renal Cell Carcinoma: A Phase II California Cancer Consortium Study with Biological and Imaging Correlates of Angiogenesis Inhibition

RCC² is diagnosed in ∼30,000 Americans annually, resulting in 11,600 deaths. Many patients present with advanced or unresectable disease and up to 30% of patients treated by nephrectomy for localized disease will relapse (1). The 5-year survival rate for metastatic RCC is estimated to be 0–10% (2). Hormonal, chemotherapeutic, and radiation therapy approaches have failed to significantly improve outcomes, particularly for patients with metastatic disease. Modulation of host immune mechanisms to regulate tumor growth has been extensively evaluated (1). Among the most widely used immunotherapy agents are the IFNs and interleukins, administered at variable doses either alone or in combination. However, clinical benefit has generally been modest. In a randomized clinical trial, tumor response rates were 6.5, 7.5, and 18.6% for patients receiving IL-2, IFN-α-2a, and IL-2 plus IFN-α-2a, respectively, with no survival benefit and severe toxicity associated with the combination. EFS rates at one year for this study were 15, 12, and 20%, respectively (3). The most encouraging results have been with high-dose IL-2, yielding overall response rates of 15–20%, with ∼5% durable CRs in excess of 5 years (4). However, high-dose IL-2 is associated with considerable toxicity and is difficult to deliver in the community setting. New agents with unique mechanisms of action are thus needed.

There is increasing evidence that tumor-associated angiogenesis and neovascularization are critical events in tumor growth, invasion, and metastases (5). Mediators of angiogenesis are therefore potential targets for antitumor therapy. Among these mediators is VEGF, a well-characterized angiogenic cytokine that plays a central role in normal glomerulogenesis and nephrogenesis (6). VEGF is expressed in up to 70% of RCC (70%) and is detectable in patient serum (7, 8, 9). In addition, VEGF mRNA expression correlates significantly with vascularization in RCC (7).

SU5416 is a small organic molecule that interferes with endothelial cell proliferation and neovascularization by inhibiting VEGF-mediated signaling through Flk-1, a transmembrane tyrosine kinase. SU5416 is believed to interfere with the catalytic activity of Flk-1 and, as a putative adenine mimic, binds to the ATP binding pocket of the kinase domain. SU5416 blocks the signaling cascade initiated by binding of VEGF to the extracellular domain of Flk-1, which then results in endothelial cell proliferation and development of new blood vessels (10). The activity and specificity of SU5416 on Flk-1 signaling have been demonstrated in vitro through experiments examining its effects on ligand-stimulated cellular events, which include receptor autophosphorylation and endothelial cell proliferation. In vivo studies have demonstrated the ability of SU5416 to inhibit s.c. growth of a number of tumor types as well inhibiting regrowth of tumor after shrinkage with a traditional cytotoxic agent (11). A Phase I trial of SU5416 demonstrated that it was fairly well tolerated and clinically feasible (12). The maximum tolerated dose was determined to be 145 mg/m² i.v. twice weekly.

IFN-α has been shown to have antiviral, immunomodulatory, and antiproliferative effects. Furthermore, it has also been shown to possess antiangiogenic properties (13). Chronic daily administration of low-dose IFN-α induced the regression of hemangiomas in infants (14), as well as in AIDS-associated Kaposi’s sarcoma (15). It has been reported that in a mouse bladder cancer xenograft model, daily low-dose IFN-α therapy produced the most significant inhibition of tumor growth, tumor vascularization, and down-regulation of basic fibroblast growth factor and matrix metalloproease-9 mRNA and protein expression (16). These investigators observed that maximal angiogenesis inhibition with IFN-α occurred at a daily dose of 5,000–10,000 IU and that daily administration of higher IFN doses did not yield any significant antiangiogenic effects. The recommended (extrapolated) human dose for angiogenesis inhibition appeared to be 1 million units of IFN-α administered s.c. twice daily.³

We therefore conducted a Phase II trial of SU5416 plus low-dose IFN-α 2B in metastatic RCC to explore its tolerability and efficacy. We hypothesized that this combination would prolong the 1-year EFS in this patient cohort to >20%. Secondary end points included overall survival and for patients with measurable disease, response rate. We also performed biological and functional imaging correlative studies, including metabolic and perfusion imaging studies with PET scans, serial plasma levels of VEGF and PAI-1 (an emerging marker for tumor hypoxia), and evaluation of procoagulant and anticoagulant protein levels, as well as platelet activation.

Eligibility requirements included histological proof of RCC of the clear cell type. Patients were required to be at least 18 years of age, to have a Zubrod performance status of 0–2, a life expectancy of ≥12 weeks at the time of study entry, and measurable or evaluable disease. All patients were also required to have a pretreatment WBC of ≥3500/μl, an absolute granulocyte count ≥ 1500, and a pretreatment platelet count of greater than or equal to institutional lower limit of normal obtained within 3 weeks of registration. A pretreatment calculated creatinine clearance of ≥60 ml/min and/or a serum creatinine of ≤1.5 mg/dl was required. Patients must also have had a serum bilirubin ≤ 1.5 mg/dl and aspartate aminotransferase ≤ 2× institutional upper limits of normal. Patients who had a nephrectomy must have had documented progression of disease before study entry.

Patients with known brain metastases, with uncompensated coronary artery disease on electrocardiogram or physical examination, or with a history of myocardial infarction or severe/unstable angina within the past 6 months, or have diabetes mellitus with severe peripheral vascular disease, or have deep venous or arterial thrombosis (including pulmonary embolism) within 3 months of registration were deemed ineligible. In addition, patients who had documented hypersensitivity to any of the SU5416 excipients (Cremophor, ethanol, and polyethylene glycol), and those who had undergone any major surgical procedure within 3 weeks of registration were excluded from protocol entry. No prior malignancy was allowed, except for adequately treated basal cell (or squamous cell) skin cancer, cervical carcinoma in situ, or any other cancer for which the patient had been disease free for 5 years.

All patients were informed of the investigational nature of the study and were required to provide written informed consent in accordance with institutional and federal guidelines.

Because inhibition of angiogenesis may have severe and potentially lethal effects on an unborn fetus as well as in newborn infants, pregnant and breastfeeding women were ineligible. For similar reasons, patients were required to follow contraception practices during the conduct of the trial.

Patients were not permitted to have any prior biological therapy, except for IL-2, the last dose of which had to be delivered at least 8 weeks before study entry with full recovery from all toxicities. No more than one prior chemotherapy regimen was allowed before entry onto this protocol. Prior chemotherapy had to be completed at least 4 weeks before study entry, and patients must have recovered from all toxicities.

Treatment consisted of SU5416 administered i.v. at a dose of 145 mg/m² twice weekly, at a rate of 200 ml/h, at least 3 days apart. The initial infusion of SU5416 was administered at 100 ml/h for the first 15 min to establish tolerance and then increased to the maximal rate of 200 mlc/h. IFN-α 2B was administered by self-injection s.c. at a dose of 1 million units twice daily, 12 h apart. Treatment cycle length was 6 weeks. Initially, there were no planned treatment breaks, but because of grade III fatigue in 2 of the first 3 patients treated, the protocol was amended as follows: the first two cycles of treatment (12 weeks) were given without interruption followed by a one-week break. Subsequently, a one-week break followed every 6-week treatment cycle. Dosage delays and reductions for SU5416 and/or IFN were required for treatment-related toxicities (see below).

Premedications to prevent hypersensitivity reactions to Cremophor (the SU5416 vehicle) included 25–50 mg of diphenhydramine i.v. or p.o. 1 h before SU5416 injection [or an equivalent dose of an alternate oral H1 blocker such as loratadine (10 mg once daily) or fexofenadine (60 mg twice daily)], 20 mg of famotidine i.v. or p.o. 1 h before SU5416 injection (or its equivalent H2-blocker), and dexamethasone 8 mg of p.o. twice daily × 3 doses (later amended to 2 mg p.o. twice daily × 3 doses) beginning 12 h before SU5416 injection. In addition, low-dose dexamethasone (0.5–1.0 mg p.o.) could be administered on the day after SU5416 infusions, if clinically indicated, to ameliorate symptoms of steroid withdrawal. Patients with central venous catheters received warfarin (1 mg p.o./day) as a prophylaxis for line-related thrombosis.

All patients were observed 2–3 h after the first, second, and third infusions of SU5416. Subsequent SU5416 administrations required vital sign monitoring for at least 30 min after the infusion. As dexamethasone was required before SU5416 infusion, ACTH stimulation tests were performed pretherapy and every 6 weeks to monitor for adrenal insufficiency. A post-ACTH cortisol level of <17 μg/dl was defined as biochemical evidence of adrenal insufficiency.

PAI-1 is involved in extracellular matrix degradation and was recently identified as a marker of tumor hypoxia (17). In this trial, serial VEGF and PAI-1 levels were drawn before the initiation of therapy and at the end of each cycle. Standardized ELISAs were performed using commercially available kits following the manufacturer’s instructions: PAI-1 Imulyse kit (Biopool International, Ventura, CA) and human VEGF (BioSource International, Inc., Camarillo, CA). Eighty-two specimens from 25 patients were investigated in quadruplicate for VEGF and 76 of these were additionally analyzed for PAI-1 in duplicate, for a total of 480 determinations.

Metabolic and perfusion imaging studies with PET scans using the FDG-glucose and O-15 H20 isotopes, respectively, were planned. These studies were performed at the Northern California PET Imaging Center in Sacramento, California. The patients were positioned in the scanner using anatomical markers and centered over areas defined by abnormalities on prior CT scans. i.v. injection of 50 mCi of O-15 was immediately followed by a 10-frame dynamic acquisition over 8 min. The patients were then injected with 15 mCi of FDG, and whole body imaging from the base of the skull through the upper thighs was started 45 min after injection. The total imaging time was 40 min. Patients were asked to fast for a minimum of 4 h before arrival. The PET scanning studies were limited to patients enrolling at the University of California at Davis because of financial and geographic access issues.

Because of prior reports of thrombotic complications in patients receiving SU5416 (18), levels of several procoagulant and anticoagulant proteins, as well as platelet activation, were assessed in participating patients at two time points: before and within 4 h of the initial SU5416 infusion. Levels of procoagulants factor VIII, IX, and fibrinogen, vWF activity, and of the anticoagulant proteins protein C, S, and antithrombin were all measured using standard functional assays. Specifically, factor VIII, IX, fibrinogen, vWF activity, protein C (Dade Behring, Deerfield, IL), and functional protein S (Stago International, Parsippany, NJ) were determined by clotting assays using the BCS coagulation analyzer (Dade Behring). Antithrombin was assessed with a chromogenic thrombin inhibition assay (Dade Behring) using the BCS. Platelet activation was measured using fluorescent-labeled monoclonal antibodies to activation-dependent platelet antigens and flow cytometry using methods described in our prior studies (19, 20).

Treatment cycles were repeated every 6 weeks for the combination. Patients were to receive two cycles of SU5416 and IFN-α before the initial response and toxicity evaluation. Response was assessed using standard RECIST criteria. Patients with a partial response or with SD were continued on treatment until a CR was obtained, PD was documented, or unacceptable toxicity occurred. Disease assessments were performed after every two cycles (or every 12 weeks). In patients achieving a CR, therapy could be continued at the discretion of the investigator until PD was documented or if unacceptable toxicity occurred. Because of the presumed cytostatic effects on antiangiogenic therapy and the possibility of delayed responses, patients with PD by standard criteria who were reasonably tolerating therapy were allowed to remain on the protocol until any measurable lesion increased to >100% of baseline (unidimensional) measurement. However, the formal definition for progression per the RECIST criteria was used for results reporting. Toxicity was evaluated using the National Cancer Institute Common Toxicity Criteria version 2.0.

The primary study end point is EFS at 1 year. An event was defined as disease progression or death from any cause. A single-stage Phase II design without a planned interim analysis was used because the required follow-up time is expected to be large compared with the expected open accrual period. We assumed that a 1-year EFS of ≥20% [the best EFS in the Negrier trial (3)] would warrant further study, whereas a 1-year EFS < 5% would not warrant any further study of this regimen. This study was therefore designed to distinguish a 5% EFS rate from a 20% rate. Accrual of 27 patients evaluable for 1-year EFS will permit 80% power for a one-sided test at the 0.05 level of significance, based on a binomial distribution. An additional 3–4 patients would achieve a false negative rate of at least 10%. Response rates were calculated as the percentage of evaluable patients whose best response is a CR or PR, and exact 95% confidence intervals were calculated for this estimate. Time to treatment failure, duration of response, and survival were estimated using the product-limit method of Kaplan and Meier. Accrual began in September 2000 and ended in April 2002 when the thirtieth patient was enrolled.

A total of 30 patients with metastatic RCC of the clear cell type were enrolled onto this study. Patient characteristics are summarized in Table 1. Twenty-five men and 5 women participated, with a median age of 59 years (range, 37–79 years). The majority of patients had Zubrod performance status 0 or 1 (90%). Most patients were treatment naïve: only 7 patients had prior immunotherapy with IL-2 or an IL-2 derivative. Six patients had prior radiotherapy to known disease sites. Twenty patients had a prior nephrectomy. At the time of this analysis, all patients were assessable for toxicity and survival, including the primary end point of 1-year EFS. Only 23 were assessable for response because 7 nonprogressing patients failed to complete the initial two cycles of therapy required for response assessment.

There were no complete or partial responders. Fifteen of the 23 assessable patients had SD at the 12-week assessment. One of these patients had a minor response, characterized by a decrease in the size and number of pulmonary nodules, occurring ∼6 months into protocol therapy. Four patients with SD, including the minor responder, received protocol therapy for >7 months (7.8, 8.4, 9.2, and 9.6 months). None of these patients developed PD by RECIST criteria during protocol therapy. The median number of treatment cycles completed was 2 (range, 1–7).

Kaplan-Meier curves for overall survival and EFS are summarized in Fig. 1. Median survival time was 10 months, with an estimated 1-year survival rate of 38% (95% confidence interval, 23–62). The 1-year EFS was estimated at 6% (95% confidence interval, 1–35), with a median EFS of 5 months.

Treatment toxicities are summarized in Tables 2 and 3. Table 2 describes toxicities of grade 3 or worse that are possibly, probably, or definitely attributed to protocol therapy. The most common toxicities were fatigue/asthenia, lymphopenia, anemia, hyperglycemia (steroid-induced), neutropenia, headache, nausea, and infection. There were 3 patients who died while on protocol treatment: 1 because of treatment-related hepatorenal failure with sepsis, another because of a perforated bowel with sepsis, and another because of a nontreatment-related infection. A minority of patients had grade 3 or worse thromboembolism (pulmonary embolism or myocardial infarction), liver dysfunction, pancreatitis, dyspnea, hypoxia, pneumonitis, anorexia, and dehydration.

Twenty-eight patients underwent serial ACTH stimulation tests. A post-ACTH cortisol level of <17 μg/dl was prospectively identified as biochemical evidence of adrenal insufficiency. Nine patients (32%) met this criterion, 8 of whom developed this finding within the first 3 months of protocol therapy. Only 2 had symptoms consistent with clinical adrenal insufficiency: both these patients had amelioration of symptoms with corticosteroid therapy.

Fourteen patients underwent baseline PET scans with FDG and O-15 isotopes to assess tumor metabolism and perfusion, respectively. However, only 5 patients were able to obtain a posttreatment PET scan after at least two cycles of protocol therapy. In this limited cohort, 1 patient with radiographic SD (by CT scanning) had decreased metabolism and perfusion by PET (Fig. 2) in response to treatment. Another patient with PD by CT had increased tumor metabolism and perfusion by PET posttherapy (Fig. 2,B). The sole patient with a minor response had minimal change in the metabolic and perfusion images pre- and posttherapy. Paired PET scan results are summarized in Table 4.

Plasma samples were drawn from participating patients before initiation of SU5416 therapy and at the end of each cycle. Eighty-two plasma specimens from 25 patients were investigated for serial VEGF levels, whereas 76 were additionally analyzed for serial PAI-1levels.

Sixteen patients had a minimum of three serial blood draws in the following categories: a pretreatment draw; a draw from the first cycle (∼21 days after initiation of treatment); and a draw from the second or third cycle (40–65 days after initiation of treatment), and these were analyzed. Baseline (pretreatment) plasma concentrations of PAI-1 ranged from 4 to 278 ng/ml, with a median value of 83 ng/ml. A comparison of pre- and posttreatment specimens demonstrated a common trend among individual patients: a drop in PAI-1 levels that was usually maintained in subsequent cycles (Fig. 3). A repeated measures ANOVA test showed that, as a group, PAI-1 levels were significantly reduced at each posttreatment time point compared with pretreatment levels.

To determine whether pretreatment levels of PAI-1 could be used to predict patient outcome, baseline levels were divided into two groups, those with SD and those with PD. Patients that were not evaluable for response were excluded from this analysis. Results of this analysis are illustrated in Fig. 5. Patients with SD had a significant decline in PAI-1 levels compared with baseline by cycle 2 or 3 of protocol therapy. Although patients with PD had generally lower baseline levels of PAI-1, a statistical difference was not demonstrated.

Individually, patient PAI-1 levels were analyzed to identify patterns of response to treatment. Of the 10 patients who had SD during treatment, 8 showed a drop in PAI-1 levels after the first cycle of treatment. Two of the 10 SD patients had a different pattern of response, where PAI-1 plasma levels increased during the course of treatment. There was no correlation between PAI-1 response patterns and EFS. Although only 4 PD patients were available for analysis, the PAI-1 response patterns of this group generally showed less of a reduction in PAI-1 levels than SD patients.

Baseline (pretreatment) plasma concentrations of VEGF ranged from 3 to 340 pg/ml, with a median value of 77 pg/ml. After treatment, a drop in VEGF levels was noted in a majority of patients that was usually maintained in subsequent cycles (Fig. 4). A repeated measures ANOVA test showed that, collectively, patient posttreatment VEGF levels were significantly reduced compared with pretreatment levels. A comparison of patients with SD versus patients with PD revealed that patients with PD generally had lower baseline levels of VEGF, although this did not achieve statistical significance (Fig. 5).

Individually, patient VEGF levels were analyzed to identify patterns of response to treatment. Of the 10 patients who had SD during treatment, 7 patients showed a drop in VEGF levels after the first cycle that was maintained or further reduced in subsequent cycles. Three of the 10 patients with SD had a different pattern of response, where VEGF plasma levels increased during the course of treatment. No correlation was observed between VEGF response patterns and EFS. The general VEGF response of all patients with SD demonstrates marked reduction of VEGF levels after initiation of treatment. The VEGF response patterns of patients who continued to progress during treatment were somewhat different from that of the SD patients. Of the 4 informative patients, 2 showed dramatic increases in VEGF levels, 1 was modulated only slightly, and only 1 showed a reduction of VEGF after treatment. In general, VEGF levels were not decreased in patients with PD in contrast to the patients with SD.

A comparison of PAI-1 and VEGF plasma levels in individual patients revealed that in almost all cases, both proteins had highly similar patterns of response, showing comparable relative baseline concentrations (i.e., if high for VEGF than high for PAI-1) and mirroring the posttreatment trends. This correlation was highly statistically significant (P < 0.001).

Levels of the procoagulant proteins factor VIII, IX, fibrinogen, as well as vWF activity were measured immediately before and within 4 h after the first infusion of SU5416. There were no significant changes in the levels of these factors with the infusion of SU5416. However, it should be noted that the baseline level of these proteins were elevated in these patients, reflecting their nature as acute phase reactants that are often elevated in patients with malignancies. Similarly, there was no significant change in the levels of the anticoagulant proteins protein C, S, and antithrombin (data not shown). We also determined whether platelets might become acutely activated with infusions of SU5461. Using three different determinations for platelet activation, namely platelet surface expression of CD62, the activated conformation of platelet glycoprotein gpIIb/IIIa and platelet microparticle formation, there was no difference pre- and postinfusion of SU5416 (data not shown). Thus, there was no evidence for acute platelet activation as the result of SU5416 infusion. In fact, there was actually a decrease in the ex vivo expression of activation markers in response to epinephrine and ADP after infusion of the study drug, suggesting inhibition of platelet activation.

Antiangiogenic therapy was once predicted to revolutionize cancer care in the early 21st century, therefore unleashing a flurry of clinical investigations in a variety of malignancies, all seeking to identify an appropriate target or active agent (21). VEGF signaling appeared to be a one such target for angiogenesis inhibition, after several in vivo studies investigating its role in tumor angiogenesis. Flk-1 receptors lacking the intracellular kinase domain blocked the activation of the endogenous Flk-1 receptor activity in cultured cells (22), leading to inhibition of tumor xenografts in nude mice. Furthermore, antisense constructs that prevented VEGF expression inhibited growth of glioma cells in nude mice; this was associated with reduced intratumoral microvessel density (23). Finally, reduction of VEGF levels with neutralizing antibodies resulted in clear inhibition of growth of a variety of solid tumor xenografts models in nude mice (24, 25, 26).

One of the earliest candidate compounds was SU5416, a novel small molecule inhibitor of the intracellular tyrosine kinase domain of the VEGF receptor Flk-1. In preclinical studies, SU5416 inhibited VEGF-dependent mitogenesis of human endothelial cells in vitro, whereas systemic administration in mice resulted in inhibition of s.c. growth of a variety of tumor xenografts (9). Despite a short half-life, SU5416 was also found to also have long-lasting inhibitory activity against VEGF-signaling in vitro, hence supporting a twice-weekly administration schedule (27).

The rationale for combining two well-characterized antiangiogenic agents such as SU5416 and IFN in the treatment of RCC stems from emerging observations that a critical series of events are required in the development of an intratumoral network of blood vessels and that simultaneously targeting two of these events with antiangiogenic agents possessing dissimilar mechanisms of action may result in enhanced antitumor activity (28). Because of the unique targeted mechanism of action of these agents and the potential for sparing normal tissues from their toxic effects, it was initially presumed that such therapy would be relatively more tolerable than conventional cytotoxic therapy. This has not been the case for SU5416 nor for low-dose IFN. For example, early phase trials with SU5416 reported several thrombotic events, including deep venous thrombosis, pulmonary embolism, and myocardial infarction, likely related to the agent’s antivascular activity (18). [Vascular events, principally pulmonary hemorrhage, were also seen in a phase II trial of the monoclonal antibody against the VEGF receptor in non-small cell lung cancer (29).] Although only 2 patients in this current study had documented thromboembolic events related to protocol therapy, there were a substantial occurrence of grade III and IV nonhematological toxicities, resulting in dose modifications and treatment delays. Expected toxicities of protracted low-dose IFN dosing were also noted, principally fatigue, depressed mood, and flu-like symptoms. Furthermore, there were 3 on-study deaths, 2 of which were felt to be associated with protocol therapy. One of these was a patient with predominantly nodal disease and a baseline Zubrod performance status of 0 who developed fulminant hepatorenal failure and sepsis during the initial treatment course.

The toxicities of SU5416 can be explained, in part, by formulation issues surrounding the compound. Its insolubility in saline required the use of Cremophor as the primary diluent, hence necessitating steroid premedications to prevent known allergic reactions to this compound. When given at its recommended twice-weekly schedule, patients were therefore frequently subjected to pulsed corticosteroids, likely contributing to this regimen’s toxicity profile. It is also conceivable that the on-study deaths may have been influenced by immunosuppression from chronic steroid therapy. An ideal antiangiogenic agent must possess not only antiendothelial cell activity and subsequent antitumor activity but must be suitably formulated to allow chronic dosing, preferably through an oral route. Furthermore, because of the potential need for long-term use, the toxicity profile must necessarily be optimal because even moderate side effects when occurring over a long period of time will eventually interfere with quality of life and symptom control. As currently formulated, SU5416 appears to be inadequate in this regard: not only is it an impractical water-insoluble i.v. agent, it is also clearly associated with enhanced toxicities, some of which are apparently life threatening. It is also possible that the combination with IFN decreased tolerability to SU5416. In addition, the use of prophylactic dexamethasone could have theoretically blunted the antiangiogenic and immunological effects attributed to IFN. In at least one preclinical model, corticosteroids have been shown to decrease IFN-mediated cell death (30). Dexamethasone has also been shown to suppress the expression of IFN-inducible protein 10, a naturally occurring angiostatic compound, in an in vitro model (31).

Another important issue is that of designing appropriate phase II trial end points for novel antiangiogenic agents such as SU5416. The tumor response rate, where measurable regressions in radiographically or clinically evident disease, is often used as a surrogate for patient benefit for traditional cytotoxic agents. In contrast, antiangiogenic agents are expected to have a more cytostatic effect; therefore, measurable tumor shrinkage as a trial end point might be an overly ambitious end point. We selected the 1-year EFS rate as our primary outcome measure for this disease and expected that in most cases, this end point would be defined by progression. It is certainly a relatively more time-consuming end point in the Phase II context for a disease such as RCC because it requires a lead-in accrual period (in our study, ∼15 months) and a follow-up time of at least a year for the majority of enrolled patients. Nevertheless, our study established that using such an end point is appropriate and feasible for testing novel cytostatic agents for metastatic RCC in the phase II setting and that efficacy results can be obtained in a relatively rapid fashion. Therefore, the clinical trial design and correlative studies used here can potentially be used in future trials evaluating other novel agents with antiangiogenic activity. Additional studies of such agents in earlier stage disease (e.g., in the adjuvant setting after a radical nephrectomy), where biological or antiangiogenic agents related to SU5416 are likely to have maximal benefit, are currently being explored.

A study objective was to identify informative markers of drug activity and develop strategies to better predict patient response or outcome to the combination of SU5416 and low-dose IFN. Analysis of PAI-1 and VEGF plasma concentrations revealed several trends that warrant further study in trials with larger cohorts. First, both PAI-1 and VEGF showed remarkably similar patterns of response in most patients. Secondly, both PAI-1 and VEGF pretreatment concentrations were generally higher in patients with SD during treatment than those who progressed. The current trial, however, was not sufficiently powered to verify this finding statistically. Thirdly, over the course of treatment, plasma concentrations of both proteins tended to decrease, in some cases, quite considerably. Overall, the drop in plasma levels of both PAI-1 and VEGF was highly statistically significant. This trend was particularly evident in patients with SD but was less evident in patients with PD.

In our previous experience with these markers in a phase I trial of the hypoxic cytotoxin tirapazamine, we observed decreases in PAI-1 and VEGF in patients that responded to therapy (32). This allows us to speculate that perhaps a similar biological response is occurring, where the hypoxic cells (which theoretically produce high levels of PAI-1 and VEGF) are being preferentially inhibited. This observation was evident in one patient with SD by CT scanning but had reductions in PET-FDG and O-15 uptake postprotocol therapy. Future trials of angiogenic inhibitors and hypoxic cytotoxins by the California Cancer Consortium will incorporate these markers to determine the significance of PAI-1 and VEGF response patterns with regard to patient response, toxicity, and outcome.

The results of the coagulation and platelet activation studies do not support an effect of SU5416 on these parameters as a reason for the increased thrombosis observed in patients receiving this drug. In fact, the ex vivo platelet activation data would suggest an inhibitor effect on platelets. However, coagulation studies were done immediately after the first infusion only, and a cumulative effect of SU5416 on coagulation or anticoagulation proteins cannot be ruled out by this preliminary study.

In summary, SU5416 plus low dose IFN-α 2B exhibits biological activity in RCC as evidenced by significant declines in serial VEGF and PAI-1 plasma levels. However, the 1-year EFS of 6% and adverse toxicity profile diminish enthusiasm for additional studies with this combination in advanced RCC. Despite this disappointing clinical outcome, angiogenesis inhibition remains a rational therapeutic target in this disease. Ongoing investigations may yet identify other antiangiogenic agents with optimal pharmaceutical formulation, increased potency, substantial clinical activity, and enhanced tolerability in RCC. A phase III trial of IFN with or without an antibody directed against VEGF (RhuMab-VEGF) in RCC has already been initiated by the Cancer and Leukemia Group B to further explore this issue.

In conclusion, this phase II trial (and its associated correlative studies) provides a suitable framework on which future efficacy trials evaluating novel agents with antiangiogenic or similar mechanisms can be pursued.

We thank Tracey Baratta for data management support, Christopher Ruel for biostatistical assistance, and Mark Jacobs (Sugen, Inc.) for his manuscript review. We also thank the support of the Cancer Therapy Evaluation Program (Drs. James Pluda and Percy Ivy) of the National Cancer Institute in the development and conduct of this trial.