Introduction

Replication-selective oncolytic viruses hold promise for the treatment of advanced cancers that are refractory to standard treatment approaches¹. These viruses are engineered to preferentially replicate in cancerous tissues. Adenoviruses (Ads) have been extensively studied as oncolytic agents². Engineered selectivity mechanisms have included two main approaches: (1) the viral gene deletion approach and (2) the use of tissue- and/or tumor-specific promoters to drive critical early viral gene expression. In the former approach a gene that is necessary for replication in normal cells but expendable in cancer cells is deleted. Examples include the E1A-CR2 gene region mutant dl922/947³ and the E1B–55-kD gene–deleted Ad dl1520 (also known as Onyx-015)⁴, which has entered a phase III clinical trial in head and neck patients. Tissue-specific transcriptional regulatory elements (TREs) have included prostate-specific promoters and enhancers^5,6,40, and bladder urothelium-specific TRE⁷. These TREs have been used to control expression of E1A alone or E1A in combination with E1B or E4 gene regions. Each of these approaches is able to target replication and cytopathic effects to cancer cells and/or specific tissues selectively.

The paradigm for oncolytic virus clinical development was established for the first such engineered oncolytic virus to enter clinical trial testing, Ad dl1520 (also known as Onyx-015)⁸. The initial trials in patients with recurrent head and neck cancer explored intratumoral injection to maximize safety^9,10. Following demonstration of safety and intratumoral viral replication in these trials, subsequent trials explored intracavitary injection (into the peritoneal cavity of patients with ovarian carcinoma)¹¹, then hepatic arterial infusion (in patients with liver metastases)^12,13,24, and finally intravenous dosing¹⁴. In a phase I and subsequent phase II trials of intravenous administration in patients with metastatic carcinomas, doses as high as 2 10¹² vp were generally well tolerated following injections every 1 to 2 weeks; individual patients received single doses of approximately 6 10¹² vp and 2 10¹³ vp each without significant toxicity^14,15. Flulike symptoms were the most common adverse events, and transient liver function test abnormalities were noted at the highest dose level. Notably, however, detailed laboratory analyses to detect disseminated intravascular coagulation (DIC) were not consistently reported. Liver toxicity assessment was complicated by underlying liver metastases. These data are of critical importance given the DIC and fatal complications reported from a previous trial with a replication-incompetent Ad in a patient with ornithine transcarbamylase deficiency²⁹. Likewise, detailed analyses of changes in the genome concentration in the blood over time were not reported; these data can potentially be used to indirectly assess viral replication and shedding into the bloodstream in patients following treatment^10,26. In addition, viral shedding to the environment was not adequately assessed. Finally, doses higher than 2 10¹² vp were only minimally evaluated¹⁴. No objective responses were reported from these trials.

Only recently has chemotherapy for metastatic, hormone-refractory prostate cancer been shown to prolong life significantly, with docetaxel-based chemotherapy demonstrating an approximately 2- to 3-month increase in median overall survival versus the previous standard mitoxantrone and prednisone^{17,25,34,38,39}. Given the limited current treatment options for patients with metastatic hormone-refractory prostate cancer, the investigation of novel therapeutics in this setting is warranted. Demonstration of systemic antitumor activity will be necessary for oncolytic Ads to become clinically relevant for the treatment of advanced prostate cancer.

Three oncolytic Ads have been tested in localized prostate cancer phase I clinical trials. CG7060 (previously designated CN706 and CV706; Cell Genesys, South San Francisco, CA, USA) used the PSA promoter-enhancer element to control E1A expression and thus limit replication to prostate tissue (cancer and normal); the virus was well tolerated following intraprostatic injection in 20 patients and resulted in dose-dependent reductions in PSA in some patients⁵. Similar results were obtained in a 20-patient phase I trial of CG7870 (previously designated CV787; Cell Genesys, South San Francisco, CA, USA; unpublished data) injected in similar fashion. CG7870 expresses E1A under control of the rat probasin promoter and E1B under control of the PSA promoter-enhancer; in contrast to CG7060, this virus expresses the E3 region genes⁶. Finally, an E1B–55-kD gene–deleted Ad expressing a tk/CD fusion gene¹⁸ was injected into the prostate in a phase I trial with similar tolerability and anti-PSA effects in the context of prodrug administration¹⁹.

In this trial CG7870 was administered by a single intravenous injection to patients with metastatic, hormone-refractory prostate cancer. We report novel information about pharmacokinetics, biodistribution, immunologic, and clinical safety and efficacy endpoints following intravenous administration of this oncolytic Ad.

Patient monitoring

After treatment, patients were monitored in the hospital for 24 hours after infusion of CG7870. After discharge from the hospital, patients were seen on days 2 and 4, then weekly for 1 month, and then on months 2, 3, 4, and 6 during the study. Physical examination was performed at each follow-up visit, as was an interval medical history. Safety monitoring included adverse event monitoring using the National Cancer Institute Common Toxicity Criteria (Version 2) and standard laboratory toxicity grading. Laboratory evaluation included: hematology, liver and renal function, coagulation studies, serum chemistry, urinalysis, analysis for virus in blood and urine samples, serum PSA, and testing for anti-Ad5 neutralizing antibody titers. At the final off-study visit, a physical examination and routine clinical laboratory tests were performed.

Neutralizing antibody titer determination

Neutralizing antibodies to Ad5 were assessed at baseline and on days 8, 15, 22, 29, and 6 months (or time of early termination in trial participation). Patient and control serum samples were incubated at 56°C for 20 minutes to inactivate complement. A clinical serum sample with a previously determined anti-Ad neutralization antibody titer was used as an assay serum control. Clinical serum samples were twofold serially diluted in RPMI 1640 (modified) containing 10% fetal bovine serum (FBS) and 10 or more total dilutions tested per time point. For baseline samples neutralization antibody titers were determined using a primary dilution of 1:2; initial dilutions of 1:2 to 1:100 were used for advanced time points depending on baseline values. Diluted serum samples were individually mixed with an equivalent volume of Ad5 stock diluted to 2 10⁵ pfu/ml in RPMI 1640 (modified) containing 10% FBS. After incubation at 37°C for 1 hour, virus-containing diluted serum samples (100 l) were used to infect separate wells, in duplicate, of a 96-well tissue culture plate containing 70% to 80% confluent 293 HEK cells (1 10⁴ cells/well). After incubation at 37°C, 5% CO₂ for 1 hour, RPMI 1640 medium (modified) containing 10% FBS (200 l) was directly added to each well without inoculum removal and plates reincubated for an additional 6 days. Individual wells on plates were scored for the presence or absence of virus-induced cytopathic effects (CPE) by microscopic examination. Serum sample endpoint titers were reported as the reciprocal of the highest antibody dilution failing to display evidence of virus-induced CPE in duplicate test wells.

Quantitative PCR for CG7870 and wild-type Ad

Patient plasma samples were tested for CG7870 and wild-type Ad genomes by quantitative polymerase chain reaction (qPCR) on days 1 (5, 10, 15, 30, 60, 90, 180 minutes; 6, 9 and 12 hours), 2, 4, 8, 15, 22, 29, and 6 months (or time of early termination in trial participation). Samples were tested by qPCR methods designed to detect and distinguish between vector and wild-type Ads. Quantification of wild-type Ad was performed using a qPCR method designed to detect a revertant wild-type E1b promoter/gene junction. Assay specificity for wild type was determined by a fluorescent-labeled probe oligomer that hybridizes at the junction between the E1b promoter and coding sequences. CG7870 virus sequences are not quantitated in the wild-type qPCR assay because of insertion of the prostate-specific element (PSE) between the E1b promoter and gene sequence in the engineered virus construct. Determination of circulating CG7870 genomes was performed, in parallel, using an independent qPCR assay designed to amplify sequences unique to the virus construct. Assay specificity for CG7870 quantitation was derived by a reverse primer with sequence complementarity to the rat probasin promoter, and a fluorescent oligonucleotide probe with sequence homology spanning the junction between E1a and rat probasin promoters. Wild-type Ad fails to be quantified in the CG7870-specific assay because of the normal absence of the rat probasin promoter in the wild-type human Ad genome.

DNA was isolated from aliquots of patient plasma samples. Individual samples (1 ml total) were centrifuged at 50,000 g and pelletable material resuspended in 4 M guanidine thiocyanate, 0.14 M 2-mercaptoethanol by vortex. Following sequential phenol–chloroform, chloroform-alone extractions, sample nucleic acid was precipitated with ammonium acetate (1.6 M final concentration) and one volume of isopropyl alcohol. Precipitated pellets were washed with 70% ethanol, dried under vacuum, resuspended in a 50-l volume of 10 mM Tris, 1 mM ethylenediaminetetraacetate (EDTA) (TE), and stored at –80°C before analysis.

CG7870 levels were determined in all patients' plasma before treatment (that is, true negatives) to set a threshold for positivity for the assay. Following determination of the mean and two standard deviations, a value of less than 100 genomes/ml was considered background and was not considered positive.

Circulating CG7870 Ad was quantitated from duplicate 5-l aliquots tested in the CG7870-specific qPCR assay. Dilutions of plasmid CP257 DNA (containing PSA/E1a and rat probasin/E1b sequences) ranging from 5 10^–1 to 5 10⁵ copies, diluted in 10 mM Tris, 1 mM EDTA containing 25 g/ml transfer RNA, were used to generate a reference standard curve. Assay amplification specificity for CG7870 virus sequences alone was previously qualified to the limit tested, 1 10⁷ wild-type Ad genomes per milliliter reaction. The lower limits of CG7870 sequence detection and quantitation was 5 and 50 copies, respectively, per 50-l reaction. Mean CG7870 titer values (expressed in units of genome copies per milliliter plasma) were calculated between replicate samples.

Plaque assay for infectious Ad

Urine and saliva samples collected at baseline and on days 1, 2, 4, 8, 15, 29, and 6 months (or time of early termination in trial participation) were tested for infectious Ad by plaque assay. Patient urine samples were diluted 1:100 in RPMI 1640 (modified) containing 10% FBS to minimize sample matrix-associated cytoxicity, and one dilution was tested per time point. Patient saliva samples were diluted 1:100 in RPMI 1640 (modified) containing 10% FBS, 1% penicillin/Sstreptomycin (100 units/100 g/ml, respectively) and filtered through a 0.2-m syringe filter. Saliva samples were tested at 1:100, 1:200, and 1:500 final dilution, following sterile filtration, for each time point. Diluted clinical samples (1 ml) were inoculated into duplicate wells of a six-well tissue culture plate containing at least 90% confluent 293 HEK cells and incubated undisturbed at 37°C, 5% CO₂ for 24 hours. Following 37°C incubation, sample inoculum was aspirated from wells, cell monolayers overlaid with an agarose–medium solution containing 0.8% agarose, 1 RPMI 1640 (modified), 10% FBS, 0.375% NaHCO₃, 4 mM L-glutamine (3 ml), and plates reincubated at 37°C, 5% CO₂ for 7 days. After 7 days' incubation, cell monolayers were again overlaid with an additional 3 ml of agarose–medium solution and reincubated for 5 days at 37°C, 5% CO₂. Cell monolayers were stained by adding a neutral red staining solution (0.025% in RPMI 1640–10% FBS–4 mM L-glutamine) to each well, and plates were incubated at 37°C, 5% CO₂ for an additional 16 to 18 hours. Plaques were visualized by transillumination and counts determined from dilutions containing mean replicate values of between 5 and 50 plaques per well. Final plaque titers, expressed in pfu per mL, were calculated as the arithmetic mean of summed plaque counts from replicate wells/dilutions corrected for the dilution factor.

Cytokine assays

Cytokine enzyme-linked immunosorbent assays (ELISAs; R+D Quantikine kits, Minneapolis, MN) were performed on patients' serum as previously described¹⁴. The following cytokines were assessed: IL-1, IL-6, IL-10, tumor necrosis factor- (TNF-). In brief, patients' serum samples were extracted from clotted blood samples and stored at –80°C. Multiple serum samples collected during each patient's treatment course were analyzed simultaneously (in triplicates), using cytokine-specific immunoassay reagents according to manufacturer's protocols. The colorimetric reaction was quantified as a function of optical density (OD) absorbance at 540 nm (SpectraMax 340, Molecular Devices, Sunnyvale, CA, USA). Cytokine concentration was calculated according to a reference standard curve generated with four-parameter logistic (4-PL) curve fit and OD values of known, graded concentrations of recombinant cytokine.

Results

Patient characteristics

Twenty-three patients were enrolled and treated with a single intravenous injection of CG7870. Patient demographic and baseline PSA data are listed in Table 1. The median age was 71 years (range 45–87 years). The median PSA was 42.5 ng/ml (range 5.0–1495 ng/ml). ECOG performance status was 0 to 2 in all patients (21/23 were 0–1). Of the 23 patients, 5 had soft-tissue disease; all other patients had bone-only disease.

Table 1 - Demographics.

Full table

Treatment-related toxicity

Reported adverse events

Table 2 lists adverse events (AEs) that were reported in 10% or more of patients by grade of toxicity, regardless of relationship to CG7870 treatment. The most common AEs were mild to moderate flulike symptoms (fever, fatigue, rigors, nausea, and vomiting) and asymptomatic, transient grade I–II hypotension (that is, not requiring IV fluid support). The hypotension was dose related (occurring at doses greater than 6 10¹¹ vp) and was delayed (occurring approximately 10–12 hours after infusion of CG7870). All serious AEs, as well as grade 3 or 4 AEs considered to be possibly, probably, or definitely related to treatment, are listed in Table 3. Only a single serious adverse event was categorized as related (definitely) to treatment; this grade 3 fatigue lasted for 1 day and resolved completely. In addition, two cases of grade 3 fever were categorized as related to CG7870 but were not serious AEs. No deaths or irreversible AEs were experienced.

Table 2 - Adverse events.

Full table

Table 3 - Related grade 3 events and serious adverse events.

Full table

Laboratory data

Asymptomatic transient laboratory changes included transaminitis, lymphopenia, anemia, thrombocytopenia, and isolated D-dimer elevations. The frequency of transaminitis was dose related, although grade 1 transaminitis was seen even at the lowest dose levels in approximately half of all patients; transaminitis was not associated with hyperbilirubinemia or liver synthetic function abnormalities. At doses greater than 10¹² vp, all five patients had grade 1–2 transaminitis. Transaminases typically peaked between days 4 and 8, although elevations occurred as early as day 2 and as late as day 15. The duration of transaminitis varied; some patients still had grade 1 transaminitis as late as day 22. No grade 3 transaminitis was observed.

Lymphopenia (grade 2–3) was seen at all dose levels and typically reached a nadir early (day 2); duration varied greatly (1–29+ days). Thrombocytopenia (grade 1) or decreased platelet counts within the normal range were noted at doses of 10¹¹ vp or higher in approximately one-fourth (4/17) of patients. The nadir generally occurred by day 2, and counts returned to the normal range within 1 to 14 days. D-dimer elevations were common at baseline, presumably as a result of subclinical low-grade DIC in this patient population associated with their tumor burden. Six patients had a significant increase in D-dimer levels after treatment (defined as an absolute increase from baseline of >1.0 g/ ml and >20% increase from baseline); D-dimer elevations were seen over the entire dose range studied. Both patients at the 6 10¹² vp dose had significant D-dimer elevations. D-dimers peaked generally between day 2 and 7. Five out of six patients, including the two at the top dose level, had no other clinical or laboratory evidence for DIC (that is, D-dimer elevations were isolated). One patient developed clinical DIC in the context of PSA progression. Notably, this patient had an extremely high baseline D-dimer level (19.7 g/ml) that remained stable until day 22 after treatment. On day 22, however, the D-dimer level increased to 27.6 g/ml and was associated with a grade 2 fibrinogenemia, an International Normalized Ratio of 1.5, grade 1 thrombocytopenia; clinically, epistaxis and easy bruising were noted. The PSA increased from 31 to 77 ng/ml during development of DIC. Given the apparent tumor progression, baseline marked elevation of the D-dimer level and the timing of the onset, this DIC seemed to be related to tumor progression rather than CG7870 treatment and was defined as not related to treatment with CG7870 by the site investigator.

At the highest dose level (6 10¹² vp) both patients had significant D-dimer elevations, transaminitis, and a decrease in platelet count (Fig. 1). Although the events did not constitute a DLT as defined by the protocol, dose escalation was halted because of the observed constellation of findings. The D-dimer elevations were significant (baseline <1.0 mg/dl, peak 3.06–3.85 mg/dl at 24 hours) and relatively prolonged (Fig. 1A); resolution to less than 1.5 mg/dl occurred by day 20. Prothrombin time, partial thromboplastin time, and fibrinogen remained within normal limits (Fig. 1C). The grade 1 to 2 transaminitis peaked between days 4 and 7, and nearly completely resolved within 8 days (Fig. 1B); bilirubin levels remained normal. In addition, both patients experienced decreases in platelet counts of approximately 50%, although only one qualified as grade 1 (day 2–4) (Fig. 1D). Lymphopenia was grade 3 in both patients (Fig. 1E). Marked increases in IL-6 and IL-10 were notable compared with lower dose levels (see below).

Figure 1.

Representative laboratory data on the two patients treated with the highest dose of CG7870 (6 10¹² vp). (A) D-dimer; (B) prothrombin time (PT), partial thromboplastin time (PTT); (C) aspartate transaminase, alanine transaminase; (D) platelets; (E) lymphocytes.

Full figure and legend (185K)

PSA response

No patients had a 50% or more decline in PSA. Five patients had a decline in PSA of 25% to 49%, which appeared to be dose dependent. In 4 of 11 (36%) patients a PSA decline of 25% to 49% occurred at doses of 6 10¹¹ vp or higher, whereas only 1 of 12 (8%) patients had such a decline at lower doses. The median time to PSA progression was 60 days, and 27% of patients were free from progression at 6 months.

Neutralizing antibodies

All patients had anti-Ad5 antibody titers less than 1:20 at baseline by ELISA; all but three patients were negative by neutralizing antibody assay. All patients developed detectable neutralizing antibodies after treatment (data not shown). By day 8, 65% of patients were positive for neutralizing antibody (titer >1:20); 91% were positive at day 21, and by day 28 all patients were positive with a median titer of 1:300 and a range of 1:50 to 1:12,800. No correlation was noted between the rate of antibody rise or absolute titer and toxicity or PSA measurements.

CG7870 in blood, saliva and urine

All patients were found to have circulating CG7870 genomes in the blood as determined by Q-PCR after intravenous administration that persisted through 90 minutes after treatment in all patients, day 8 in 13/23 patients, and through day 29 in 3/23 patients (Fig. 2A).

Figure 2.

CG7870 genome positivity over time following CG7870 intravenous infusion. The percentage of patients in each of the three dose ranges is shown, with detectable genomes in plasma at each of the time points following intravenous infusion of the virus: (A–C) low (10¹⁰, 3 10¹⁰, 10¹¹), white bar; (D–F) intermediate (3 10¹¹, 6 10¹¹, 10¹²), black bar; (G, H) high (3 10¹² and 6 10¹²), gray bar.

Full figure and legend (70K)

The peak genome concentration was noted 5 to 15 minutes after infusion and was dose related. The three low-dose groups (A–C) had a mean concentration of 1 10⁵ genomes /ml, whereas the three mid-dose groups (D–F) had 6 10⁵ genomes/ml and the top two groups (G, H) had 3 10⁷ genomes/ml. The alpha half-life was approximately 15 minutes. The genome nadir typically occurred between 9 and 12 hours. Thereafter we considered a significant increase to be more than 10 times the nadir concentration; this "secondary" or "delayed" peak was suggestive of genome replication and shedding into the bloodstream. Secondary peaks were demonstrated in 70% (16/23) of patients overall and generally occurred between day 2 and day 8 (Fig. 2B). Patients in the higher dose groups had more frequent and higher secondary peaks. By day 4, 100% of the five highest dose patients had secondary peaks detected as compared with approximately 45% of the lowest dose patients. The height of the peak was also dose related; on day 4 the mean was 1 10⁵ genomes/ml in the high-dose group and 50 genomes/ml in the low-dose group. The duration of genome positivity in blood was also dose related (Fig. 2A). The two patients treated at the highest dose level (6 10¹² vp) showed a profound secondary peak and had detectable viral genomes in the blood at 29 days after infusion. Notably, this analysis reflects only the number of genomes in the plasma and does not reflect viral genomes that were (1) produced but not shed from tumor, or (2) genomes in the cellular fraction of the blood (e.g., red blood cells). The true genome replication may have been significantly higher.

All urine samples tested for evidence of viral shedding via plaque assay remained negative at testing on days 1, 2, 4, 8, 15, and 29. Saliva samples were positive (for infectious units) in three patients. Two of five patients treated with doses higher than 10¹² vp were positive (both on day 4), whereas only 1 of 18 patients treated with doses of 10¹² vp or less was positive (day 8). All samples from days 1, 2, 15, and 29 were negative. Wild-type adenovirus was not detected via PCR in either the urine or blood.

Cytokines blood levels

IL-1, IL-6, IL-10, and TNF- levels were measured after administration of CG7870. IL-6 and IL-10 blood concentrations increased progressively with higher dose cohorts of CG7870. Peak IL-6 responses occurred at 6 hours, and peak IL-10 levels were evident between 12 and 24 hours after treatment (Figs. 3A, B). Notably, the peak blood concentration time point for IL-6 was associated with a transient decrease in blood pressure (Fig. 4). IL-1 levels peaked higher and earlier in the high-dose patient groups; in the top-dose cohort the mean IL-1 peak was approximately 4 pg/ml (between 3 and 6 hours after treatment) (data not shown). TNF- increases were only evident at the highest dose levels. Cohort G (3 10¹² vp) patients had a doubling of TNF- levels, on average, within approximately 6 to 9 hours after treatment. Patients in cohort H (6 10¹² vp) had a nearly 10-fold increase over baseline, peaking at approximately 6 hours (data not shown).

Figure 3.

Acute cytokine induction following CG7870 intravenous (IV) infusion. Concentrations of interleukin 6 (IL-6; pg/ml) 6 hours after IV infusion of the virus are represented for each patient per dose cohort (dose levels shown on x-axis); concentrations shown on log scale on y-axis (A). IL-6 concentrations peaked at 6 hours after dosing. Concentrations of interleukin-10 (IL-10; pg/ml) 12–24 hours after IV infusion of the virus are represented for each patient per dose cohort (dose levels shown on x-axis); concentrations shown on linear scale on y-axis (B). IL-10 concentrations peaked between 12 and 24 hours after dosing.

Full figure and legend (75K)

Figure 4.

Dose-related decreases in diastolic and systolic blood pressure after CG7870 dosing. The maximum percentage change in patient blood pressure from baseline to 8–12 hours after treatment is represented for each patient in gray bars (diastolic top panel; systolic, lower panel). The mean change for each dose group is represented by the black bar (absolute number for mean change shown). Bars directed upward (above the zero line) represent an increase in blood pressure, whereas a bar directed downward represents a decrease in blood pressure.

Full figure and legend (143K)

Discussion

This dose escalation phase I trial is the first detailed report of viral, immunologic, and clinical correlates following a single intravenous dose of a replication-selective oncolytic Ad. Intravenous CG7870 was tolerable overall, without irreversible toxicity at doses from 1 10¹⁰ to 6 10¹² vp. Mild, transient flulike symptoms, including fever, nausea, chills and lymphopenia, as in other reported Ad clinical trials, were the most common adverse events. Despite the lack of clinically significant adverse events, given the constellation of D-dimer elevation, transaminitis, intranormal decrease in platelet counts from baseline, and marked IL-6 induction in the two highest dose patients, the decision was made to halt dose escalation on this trial.

The toxicity data from this trial have generated insights that should be useful in guiding future trials of intravenous Ad for several reasons. First, low-grade transaminitis was observed at all dose levels on this trial; the severity and duration seemed to be higher at the top-dose level of 6 10¹² vp, but this must be confirmed in future studies. Liver toxicity is one of the most serious potential Ad-related toxicities. Previous clinical trials may have underestimated the frequency of low-grade transaminitis, because cancer-related transaminitis was present at baseline and increased with tumor progression. Liver function was not detectably compromised at these doses. Similar findings were reported from a phase I trial of Addl1520 (also known as Onyx-015)²⁰; however, no transaminitis was reported from a phase II trial (dose 2 10¹² vp)¹⁵. This discrepancy seems to be related to trial design issues (less intensive liver enzyme assessments) and to the liver involvement with cancer (making assessment more difficult). Nevertheless, although liver toxicity is notable in mice, no clinically significant treatment-related hepatotoxicity has been reported on an oncolytic Ad trial to date¹⁶. Similar findings were reported from a trial of Ad-p53 (replication-incompetent) administered by hepatic artery infusion¹⁶. Murine liver has higher relative coxsackie-Ad receptor (CAR) expression than does human liver³²; further cautious dose escalation will be necessary to determine whether hepatotoxicity will be dose-limiting. The frequency of repeat dosing intravenously may be partially dictated by the recovery time required for these liver abnormalities to resolve. It is clear from a phase II trial of Addl1520 (also known as Onyx-015) that doses of 2 10¹² vp can be administered weekly¹⁵.

As was the case for liver toxicity monitoring, the intensity of blood testing for coagulopathy (e.g., D-dimers, fibrinogen) was greater on this trial than on previous trials^15,16. DIC was noted with replication-deficient adenoviral vectors in primate toxicity studies and in a patient with ornithine transcarbamylase deficiency²⁹; DIC was not previously reported on oncolytic Ad studies, although detailed laboratory testing was not done^8,9. Several novel findings were noted on this study. First, many of the patients in this study had elevated levels of D-dimers at baseline; low-grade DIC is common in patients with metastatic prostate carcinoma. However, during the course of the study, significant increases in D-dimers occurred (defined as an increase of 1 ng/ml or more and 20% or more from baseline) in approximately one-fourth of all patients. These increases were not dose related at the doses tested on this trial, were asymptomatic, and were not associated with clinically significant coagulopathy. It is impossible to determine if this increased level of D-dimers was caused by CG7870 or progression of prostate cancer. Future studies of oncolytic Ads should monitor changes in laboratory and clinical coagulation parameters carefully, especially as higher dose therapy is investigated. Moreover, given the relatively nonspecific nature of the D-dimer elevation alone, such changes should be assessed in the context of the overall coagulation status of the patient.

With respect to pharmacokinetics and biodistribution after intravenous treatment, CG7870 genomes were detectable in the plasma for as long as 90 minutes in all patients and for as long ass 4 weeks at the top dose levels. A delayed secondary peak in viral genomes (increase from nadir >10-fold) occurred in 70% of the treated patients, and in all patients in the top two dose cohorts; these delayed genome spikes were highly suggestive of genome replication and shedding^12,23 and are not seen with replication-incompetent Ads^16,21,22. Ethical approval was not obtained for biopsy of bone metastases present in these patients to prove replication in situ. Shedding of CG7870 to the environment via urine was not detectable, whereas three patients had positive saliva samples; 40% of patients treated with more than 10¹² vp were positive as compared with 5% with less than 10¹² vp. In high-dose patient cohorts the secondary peak in circulating viral genomes (day 4) was associated with the presence of infectious virus in saliva at matched time points.

Acute, dose-related cytokine inductions were noted. In particular, IL-6 levels peaked at approximately 6 hours and correlated closely with dose and with transient grade 1 to 2 hypotension. IL-6 is a potent cytokine induced by significant physiological stresses such as infection, sepsis, surgery, and hypotension²⁷. IL-6 can be induced by TNF²⁸, for example, and is able to induce IL-10. IL-6 and IL-10 levels were only significantly induced at the top two dose levels. Notably, IL-6 levels in high-dose patients on this trial were similar to those reported from a hepatic artery infusion trial of another oncolytic Ad dl1520 (Onyx-015)^12,13, with the exception of one patient on the hepatic artery trial; this patient had a severe systemic inflammatory syndrome associated with markedly higher levels of IL-6 (250 vs. 1–7 ng/ml) and a paradoxical decrease in IL-10¹³. IL-6 was elevated in a patient experiencing significant toxicity due to a nonreplicating adenoviral vector in a gene therapy trial studying hepatic arterial administration^29,30; reported IL-6 levels were similar to that reported in high-dose patients on this trial (range 1–6 ng/ml). Notably, it was unclear whether IL-6 had a causative role in the patient's toxicity or whether IL-6 induction was a secondary phenomenon. Additional study will be necessary to determine the role of IL-6 in Ad-induced toxicity. TNF infusion can cause a laboratory profile resembling that seen on this trial, including transaminitis and thrombocytopenia²⁸. In preclinical models with Ad gene delivery vectors, TNF is induced after particle uptake by reticuloendothelial cells. Whether TNF plays a primary causative role in Ad-mediated toxicity is also unclear. Investigators have proposed pretreatment with anti-TNF therapies such as etanercept (Enbrel; Amgen-Immunex, Thousand Oaks, CA, USA) and infliximab (Remicade; Centocor, Horsham, PA, USA) in an attempt to reduce Ad-mediated toxicity (J. Nemunaitis, personal communication), but clinical data are lacking to date.

This phase I dose escalation trial was designed to assess safety rather than efficacy. Nevertheless, the PSA decreases demonstrated on this trial are the first reported evidence of PSA modulation after intravenous administration of an oncolytic Ad. Future trials under consideration will focus on maximizing antitumoral efficacy including repeat dosing. Because induction of neutralizing antibodies occurs in nearly all patients treated with adenoviral vectors within 3 to 4 weeks³¹, as was observed in this study, it may be desirable to inhibit antibody induction and/or to treat aggressively over the first several weeks. This approach is similar to "induction" chemotherapy approaches. Transient suppression of antibody development may delay induction of neutralizing antibodies, thereby increasing the window of time for intravenous viral treatment³³. Additionally, numerous approaches are being explored to increase the intravenous delivery of Ads to distant tumors. Native CAR- and integrin-binding sites on the fiber and penton proteins of the virus, respectively, have been ablated to increase the viral half-life within the bloodstream and to decrease hepatic uptake in mice³⁵. In addition, fiber shaft mutations have been shown to alter biodistribution by some investigators. Finally, polymer coating of the virus can be used to shield the virus from antibody-binding and natural uptake mechanisms, thereby increasing the circulating half-life of the active virus³⁷; tumor-specific ligands can be incorporated into the coat to direct tumor-specific uptake^35,36. It will probably be necessary to use a constellation of these approaches to maximize intravenous delivery to tumors and enhance antitumoral efficacy.

Materials and methods

Study approvals

The study protocol was submitted to and approved by the Recombinant DNA Advisory Committee, the U.S. Food and Drug Administration, and the individual hospital Institutional Review and Biosafety Committees.

Study design

The primary objective of this study was to determine the safety and maximum tolerated dose of CG7870 after intravenous administration in patients with hormone-refractory, metastatic prostate cancer. Patients were enrolled at three institutions (University of California, San Francisco, CA; Kimmel Cancer Center at Johns Hopkins University, Baltimore, MD; University of Wisconsin, Madison, WI) and received treatment at one of eight dose levels in a group-sequential dose escalation design. Cohorts were treated sequentially at dose levels of 1 10¹⁰, 3 10¹⁰, 10¹¹ (Cohorts A, B, and C), 3 10¹¹, 6 10¹¹, 1 10¹² (Cohorts D, E, and F), 3 10¹², and 6 10¹² (cohorts G and H) viral particles (vp), with three patients at each dose level except the final dose level (only two patients were treated at this dose level because of toxicity). Additional endpoints included efficacy (PSA regression), studies of immunological response (neutralizing antibodies and cytokines), viral pharmacokinetics, viral concentrations in plasma over time, and viral shedding into the environment (urine and saliva).

Patient selection

To be eligible for enrollment patients had to have hormone-refractory adenocarcinoma of the prostate with evidence of metastatic disease on bone scan or computed tomography scan. Patients continued therapy with luteinizing hormone–releasing hormone agonists if not previously treated with orchiectomy. Antiandrogen therapy was discontinued at least 6 weeks before study entry, and therapy with estrogen, megesterol acetate, ketoconazole, chemotherapy, and/or PC SPES (a combination herbal therapy) had to be discontinued at least 4 weeks before enrollment. Treatment with corticosteroids was not allowed within 2 weeks of enrollment, and prior radiopharmaceuticals had to have been administered at least 8 weeks before study entry. All patients were required to test negative for Ad serotype 5 (Ad5) antibody (titer <12.5 g/ml)^5,41. Other inclusion criteria included: serum testosterone of less than 50 ng/dl, expected survival of at least 3 months, and an Eastern Cooperative Oncology Group (ECOG) Performance Status of 2 or less. Exclusion criteria included more than one prior chemotherapy regimen, treatment with chemotherapy or other experimental drugs within 4 weeks of enrollment, brain metastasis, active acute infection, human immunodeficiency virus infection or other immunodeficiency, historical or laboratory evidence of coagulation defect (not to include isolated D-dimer elevation), uncontrolled medical condition, and a history of malignancy (except curatively treated nonmelanomatous skin cancer or stage Ia bladder cancer) within 5 years of enrollment. Additionally all patients were required to have adequate liver (bilirubin 1.5 times the institutional upper limits of normal ( ULN), aspartate transaminase/alanine transaminase (AST/ALT) 2 ULN), renal (creatinine 2 mg/dl), and bone marrow function (white blood cells 2,000/mm³, absolute neutrophil count 1500, platelets 100,000/mm³, hemoglobin 9 g/dl) to participate in this trial. Signed informed consent was attained from all patients.

Manufacturing and preparation of CG7870

CG7870 is a replication-selective oncolytic Ad that targets prostate cells because of insertion of prostate-specific TREs (Fig. 5). CG7870 was provided by Calydon, Inc., now a part of Cell Genesys, Inc. (South San Francisco, CA, USA), as previously described⁵. CG7870 was supplied to clinical trial sites as a sterile, clear to opalescent, frozen liquid in 2.0-ml vials containing virus at a concentration ranging from approximately 2.5 10¹¹ to approximately 4.0 10¹² vp/ml. CG7870 was formulated in ARCA buffer (5% sucrose, 0.05% polysorbate-80, 1% glycine, 10 mM Tris, and 1 mM magnesium chloride, pH 7.8). After thawing CG7870 was diluted in 0.9% normal saline to the appropriate final concentration.

Figure 5.

CG7870 genetic construct. PB, rat probasin promoter (controls expression of E1A); PSE, prostate-specific element promoter-enhancer region (controlling expression of E1B region genes). ITR, inverted terminal repeat region of Ad5.

Full figure and legend (50K)

Treatment procedure

The virus was diluted with normal saline to a total volume of 10 to 25 ml in a sterile class II biosafety cabinet. A single infusion of 10 to 25 ml of CG7870 in normal saline was administered over approximately 10 minutes through an intravenous line. Biocompatibility studies were performed on the syringes and intravenous infusion lines used in the clinic; significant virus inactivation was not reported.

References

1. Kirn, D., Martuza, R. L. and Zwiebel, J. (2001). Replication-selective virotherapy for cancer: biological principles, risk management and future directions. Nat Med. 7: 781–787. | Article | PubMed | ISI | ChemPort |
2. Kirn, D. (2000). Replication-selective oncolytic adenoviruses: virotherapy aimed at genetic targets in cancer. Oncogene 19: 6660–6668. | Article | PubMed | ISI | ChemPort |
3. Heise, C., et al. (2000). An adenovirus E1A mutant that demonstrates potent and selective antitumoral efficacy. Nat. Med. 6: 1134–1139. | Article | PubMed | ISI | ChemPort |
4. Heise, C., Sampson-Johannes, A., Williams, A., McCormick, F., Von Hoff, D. D. and Kirn, D. H. (1997). ONYX-015, an E1B gene-attenuated adenovirus, causes tumor-specific cytolysis and antitumoral efficacy that can be augmented by standard chemotherapeutic agents [see Comments] Nat. Med. 3: 639–645.  | Article | PubMed | ISI | ChemPort |
5. DeWeese, T., et al. (2001). A phase I trial of CV706, a replication-competent, PSA selective oncolytic adenovirus, for the treatment of locally recurrent prostate cancer following radiation therapy. Cancer Res. 61: 7464–7472. | PubMed | ISI | ChemPort |
6. Yu, D., Chen, Y., Seng, M., Dilley, J. and Henderson, D. R. (1999). The addition of adenovirus region E3 enables calydon virus 787 to eliminate distant prostate tumor xenografts. Cancer Res. 59: 4200–4203. | PubMed | ISI | ChemPort |
7. Zhang, J., et al. (2002). Identification of human uroplakin II promoter and its use in the construction of CG8840, a urothelium-specific adenovirus variant that eliminates established bladder tumors in combination with docetaxel. Cancer Res. 62: 3743–3750. | PubMed | ISI | ChemPort |
8. Kirn, D. (2001). Clinical research results with dl1520 (Onyx-015), a replication-selective adenovirus for the treatment of cancer: what have we learned?. Gene Ther. 8: 89–98. | Article | PubMed | ISI | ChemPort |
9. Nemunaitis, J., et al. (2000). Selective replication and oncolysis in p53 mutant tumors with Onyx-015, an E1B-55kD gene-deleted adenovirus, in patients with advanced head and neck cancer: a phase II trial. Cancer Res. 60: 6359–6366. | PubMed | ISI | ChemPort |
10. Khuri, F. R., et al. (2000). A controlled trial of intratumoral ONYX-015, a selectively-replicating adenovirus, in combination with cisplatin and 5-fluorouracil in patients with recurrent head and neck cancer. [In Process Citation] Nat. Med. 6: 879–885.  | Article | PubMed | ISI | ChemPort |
11. Vasey, P., et al. (2002). Phase I trial of intraperitoneal injection of the E1B-55kD-deleted adenovirus ONYX-015 adenovirus in patients with recurrent/refractory epithelial ovarian cancer. J. Clin. Oncol. 20: 1562–1569. | Article | PubMed | ISI | ChemPort |
12. Reid, T., et al. (2001). Intra-arterial administration of a replication-selective adenovirus (dl1520) in patients with colorectal carcinoma metastatic to the liver: a phase I trial. Gene Ther. 8: 1618–1626. | Article | PubMed | ISI | ChemPort |
13. Reid, T., et al. (2002). Hepatic arterial infusion of a replication-selective oncolytic adenovirus (dl1520): phase II viral, immunologic, and clinical endpoints. Cancer Res. 62: 6070–6079. | PubMed | ISI | ChemPort |
14. Nemunaitis, J., et al. (2002). Intravenous infusion of a replication-selective adenovirus (ONYX-015) in cancer patients: safety, feasibility and biological activity. Gene Ther. 8: 746–759.
15. Hamid, O., et al. (2003). Phase II trial of intravenous CI-1042 in patients with metastatic colorectal cancer. J. Clin. Oncol. 21: 1498–1504. | Article | PubMed | ISI | ChemPort |
16. Reid, T., Warren, R. and Kirn, D. (2002). Intravascular adenoviral agents in cancer patients: lessons from clinical trials. Cancer Gene Ther. 9: 979–986. | Article | PubMed | ISI | ChemPort |
17. Tannock, I., et al. (1996). Chemotherapy with mitoxantrone plus prednisone or prednisone alone for symptomatic hormone-resistant prostate cancer: a Canadian randomized trial with palliative end points. J. Clin. Oncol. 14: 1756–1764. | PubMed | ISI | ChemPort |
18. Freytag, S. O., Rogulski, K. R., Paielli, D. L., Gilbert, J. D. and Kim, J. H. (1998). A novel three-pronged approach to kill cancer cells selectively: concomitant viral, double suicide gene, and radiotherapy [see Comments]. Hum. Gene Ther. 9: 1323–1333.  | PubMed | ISI | ChemPort |
19. Freytag, S., et al. (2002). Phase I study of replication-competent adenovirus-mediated double suicide gene therapy for the treatment of locally recurrent prostate cancer. Cancer Res. 62: 4968–4976. | PubMed | ISI | ChemPort |
20. Nemunaitis, J., Cunningham, C., Buchanan, A., Blackburn, A., Edelman, G., Maples, P., Netto, G., Tong, A., Randlev, B., Olson, S. and Kirn, D. (2001 (May)). Intravenous infusion of a replication-selective adenovirus (ONYX-015) in cancer patients: safety, feasibility and biological activity. Gene Ther. 8 (10): 746–759.
21. Nemunaitis, J., et al. (2000). Adenovirus-mediated p53 gene transfer in sequence with cisplatin to tumors of patients with non-small-cell lung cancer. J. Clin. Oncol. 18: 609–622. | PubMed | ISI | ChemPort |
22. Senzer, N., et al. (2004). TNFerade biologic, an adenovector with a radiation-inducible promoter controlling expression of TNF-alpha. J. Clin. Oncol. 22: 592–601. | Article | PubMed | ISI | ChemPort |
23. Wein, L. M., Wu, J. T. and Kirn, D. H. (2003). Validation and analysis of a mathematical model of a replication-competent oncolytic virus for cancer treatment: implications for virus design and delivery. Cancer Res. 63: 1317–1324. | PubMed | ISI | ChemPort |
24. Reid, A., Galanis, E., Abbruzzese, J., Romel, L., Rubin, J. and Kirn, D. (1999). A phase I/II trial of ONYX-015 administered by hepatic artery infusion to patients with colorectal carcinoma. EORTC-NCI-AACR Meeting on Molecular Therapeutics of Cancer.
25. Rosenberg, J. and Small, E. J. (2003). Prostate cancer update. Curr. Opin. Oncol. 15: 217–221. | Article | PubMed | ChemPort |
26. Blasberg, R. G. and Tjuvajev, J. G. (2003). Molecular-genetic imaging: current and future perspectives. J. Clin. Invest. 111: 1620–1629. | Article | PubMed | ISI | ChemPort |
27. Mistchenko, A. S., et al. (1994). Cytokines in adenoviral disease in children: association of interleukin-6, interleukin-8, and tumor necrosis factor alpha levels with clinical outcome. J. Pediatr. 124: 714–720. | PubMed | ChemPort |
28. Selby, P., et al. (1987). Tumor necrosis factor in man: clinical and biological observations. Br J Cancer 56: 803–808. | PubMed | ISI | ChemPort |
29. Raper, S. E., et al. (2003). Fatal systemic inflammatory response syndrome in a ornithine transcarbamylase deficient patient following adenoviral gene transfer. Mol. Genet. Metab. 80: 148–158. | Article | PubMed | ISI | ChemPort |
30. Raper, S. E., et al. (2002). A pilot study of in vivo liver-directed gene transfer with an adenoviral vector in partial ornithine transcarbamylase deficiency. Hum. Gene Ther. 13: 163–175. | Article | PubMed | ISI | ChemPort |
31. Heise, C. and Kirn, D. H. (2000). Replication-selective adenoviruses as oncolytic agents. J. Clin. Invest. 105: 847–851. | PubMed | ISI | ChemPort |
32. Tomko, R., Xu, R. and Philipson, L. (1997). HCAR and MCAR: the human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses. Proc. Natl. Acad. Sci. USA 94: 3352–3356. | Article | PubMed | ChemPort |
33. Yu, D., et al. (2001). Antitumor synergy of CN787, a prostate cancer-specific adenovirus, and paclitaxel and docetaxel. Cancer Res. 61: 517–525. | PubMed | ISI | ChemPort |
34. Small, E. J. (2001). Docetaxel in prostate cancer. Anticancer Drugs 12 (Suppl 1): S17–S20. | PubMed | ChemPort |
35. Akiyama, M., et al. (2004). Ablating CAR and integrin binding in adenovirus vectors reduces nontarget organ transduction and permits sustained bloodstream persistence following intraperitoneal administration. Mol. Ther. 9: 218–230. | Article | PubMed | ISI | ChemPort |
36. Suzuki, K., Fueyo, J., Krasnykh, V., Reynolds, P. N., Curiel, D. T. and Alemany, R. (2001). A conditionally replicative adenovirus with enhanced infectivity shows improved oncolytic potency. Clin. Cancer Res. 7: 120–126. | PubMed | ISI | ChemPort |
37. Fisher, K. D., Stallwood, Y., Green, N. K., Ulbrich, K., Mautner, V. and Seymour, L. W. (2001). Polymer-coated adenovirus permits efficient retargeting and evades neutralizing antibodies. Gene Ther. 8: 341–348. | Article | PubMed | ISI | ChemPort |
38. Tannock, I. F. TAX 327 Investigators et al. (2004). Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N. Engl. J. Med. 7 Oct; 351 (15): 1502–1512. | Article | PubMed | ISI |
39. Petrylak, D. P., et al. (2004 (7 Oct)). Docetaxel and estramustine compared with mitoxantrone and prednisone for advanced refractory prostate cancer. N. Engl. J. Med. 351 (15): 1513–1520.
40. Yu, D. C., Sakamoto, G. T. and Henderson, D. R. (1999 (1 Apr)). Identification of the transcriptional regulatory sequences of human kallikrein 2 and their use in the construction of calydon virus 764, an attenuated replication competent adenovirus for prostate cancer therapy. Cancer Res. 59 (7): 1498–1504.
41. Chen, Y., Yu, D. C., Charlton, D. and Henderson, D. (2000). Pre-existent adenovirus antibody inhibits systemic toxicity and antitumor activity of CN706 in the nude mouse LNCaP xenograft model: implications and proposals for human therapy. Hum. Gene Ther. 11: 1553–1567. | Article | PubMed | ISI | ChemPort |

Acknowledgements

This work was supported by National Cancer Institute grant NCI-CAP50-58236 (JHU).