A Phase I Study of Intravenous Oncolytic Reovirus Type 3 Dearing in Patients with Advanced Cancer

There is accumulating evidence for the potential clinical utility of oncolytic viruses capable of tumor-selective replication and cytolysis. Reovirus type 3 Dearing (Reolysin; Oncolytics Biotech), a wild-type member of the Reoviridae family, is nonpathogenic in humans, and infections are usually asymptomatic or associated with only minor respiratory/enteric symptoms (1). Reovirus is ubiquitous as witnessed by the high seropositivity rate in normal adult populations (2–4). Reovirus was initially shown to be oncolytic (tumor-selective) by virtue of its ability to replicate selectively in transformed, but not normal, cells (5). NIH-3T3 cells are naturally resistant to reovirus infection but become highly permissive when transformed with the epidermal growth factor receptor gene, v-erbB, sos, or ras, which are all activators of Ras signaling (6, 7). In normal cells, reovirus infection causes phosphorylation of cellular PKR (dsRNA-activated protein kinase) and arrest of viral protein translation and replication. In contrast, in Ras-activated cells, PKR remains inactive (unphosphorylated) and is incapable of aborting viral translation, replication, and cytolysis (7, 8). Ras-activating mutations promote tumor growth, angiogenesis, and metastasis and are found in ∼30% of human cancers. However, two-thirds of cancers may be susceptible to reovirus therapy through Ras pathway targeting by virtue of activation or overexpression of elements upstream or downstream of Ras, such as epidermal growth factor receptor or platelet-derived growth factor receptor (9).

Reovirus therapy has yielded complete tumor regressions in murine models after intratumoral and systemic delivery (8, 10, 11). Reovirus was initially evaluated in humans via the intratumoral route (12). Toxicities were mild and consisted of transient flu-like symptoms and headache. As in the murine models, tumor regressions were observed in these studies in both injected and adjacent uninjected lesions.

This phase I, open-label, dose-escalation study of i.v. reovirus was conducted across two centers in the United Kingdom. The primary objective was to determine the safety and tolerability of i.v. reovirus and thereby to define the maximum tolerated dose (MTD). Secondary objectives included analysis of viral biodistribution, antiviral immune responses, and antitumor activity.

Eligible patients with histologically or cytologically confirmed advanced-stage primary or metastatic solid tumors that were refractory to standard therapy were enrolled. Acute toxic effects of prior radiotherapy, chemotherapy, or surgical procedures had to have resolved to Common Terminology Criteria for Adverse Events (version 3.0) grade ≤1, with any surgery occurring ≥28 days before study enrollment. Patients were ages ≥18 years; using adequate birth control; had a life expectancy of >3 months; had measurable or assessable disease; an Eastern Cooperative Oncology Group performance status of 0 to 2; cardiac ejection fraction of ≥50% by multiple-gated acquisition scan; and adequate hepatic, renal, and bone marrow function (aspartate transaminase/alanine transaminase ≤2.5 × institutional upper limit of normal; total bilirubin ≤1.5 × institutional upper limit of normal; serum creatinine ≤1.5 × institutional upper limit of normal; hemoglobin ≥9.0 mg/dL; absolute neutrophil count ≥1,500/μL; platelet count ≥100,000/μL). Exclusion criteria included known brain metastases, pregnancy or breast-feeding, concurrent immunosuppressive therapy, known HIV and hepatitis B or C infections, clinically significant cardiac disease (New York Hearth Association class III or IV), and inability to give written informed consent. The study protocol was approved by the local ethics committees.

Dose escalation. In the first (or lowest dose) cohort, virus was administered at 1 × 10⁸ TCID₅₀ as a single dose every 4 weeks. In the second cohort, virus was given at 1 × 10⁸ TCID₅₀ for 3 consecutive days every 4 weeks. In the third cohort, virus was given at 1 × 10⁸ TCID₅₀ for 5 consecutive days every 4 weeks, with 28 days defined as 1 cycle. Subsequent cohorts received 5 consecutive injections of escalating doses of virus every 4 weeks with the following dose-escalation scheme: cohort 4, 3 × 10⁸ TCID₅₀: cohort 5, 1 × 10⁹ TCID₅₀; cohort 6, 3 × 10⁹ TCID₅₀; cohort 7, 1 × 10¹⁰ TCID₅₀; and cohort 8, 3 × 10¹⁰ TCID₅₀ (the manufacturing limit for dose escalation; Table 1). Cohorts of three were individually assessed for safety and dose-limiting toxicity (DLT). Subjects were “evaluable for dose-escalation decisions” if they received at least 1 cycle or withdrew from the study because of drug-related toxicity. Subjects withdrawn from the study without meeting these criteria were replaced. If one of three subjects in a cohort experienced a DLT during the first cycle, three more subjects were added to that dose group. If two or more subjects in a dose group experienced a DLT during the first cycle, the previous lower dose would be defined as the MTD. Intrasubject dose escalations were not permitted. Patients continued reovirus treatment as long as it was tolerated and there was no evidence of disease progression.

Virus administration. Reolysin was supplied by Oncolytics Biotech in single-use 1 mL glass vials containing a frozen viral suspension in PBS. Stock was stored at -70°C and thawed rapidly over 2 min, and the appropriate TCID₅₀ dose was diluted to 250 mL in 0.9% sodium chloride and infused over 60 min through a peripheral line. Patients were observed closely (including blood pressure, temperature, and heart rate monitoring) during and for at least 6 h after infusion. For the first five cohorts, patients remained as in-patients until it was confirmed that they were not shedding reovirus. For cohorts 6 to 8, hospital discharge occurred after the fifth dose of virus had been administered without analysis for viral shedding.

Dose-limiting toxicity. Toxicities were graded according to Common Terminology Criteria for Adverse Events version 3.0. DLT was defined during the first cycle of treatment and included absolute neutrophil count <0.5 × 10⁹ with sepsis, platelet count <25 × 10⁹/L, grade 2 neurotoxicity or cardiotoxicity, inability to tolerate one course of therapy due to toxicity, and any other drug-related nonhematologic grade 3 or 4 toxicity (except flu-like symptoms and nausea and vomiting if appropriate prophylactic or therapeutic measures had not been administered).

Pretreatment and follow-up studies. Safety was assessed by evaluating the type, frequency, and severity of adverse events; changes in clinical laboratory tests (including hematology, clinical chemistry, and urinalysis); immunogenicity; and physical examination. Multiple-gated acquisition scan (or echocardiography) and radiologic studies of assessable disease were done at baseline and after alternate cycles. Patients continued treatment in the absence of progressive disease or intolerable toxicity.

Response evaluation. Response was measured radiologically by Response Evaluation Criteria in Solid Tumors and by densitometric analysis of computed tomography imaging (13). For this latter analysis, the computed tomography attenuation coefficient (density) of the target lesion in Hounsfield units (HU) was measured on a commercially available workstation (AGFA 4.0) by drawing a region of interest circumscribing the margin of the tumor. The tumor HU measurements of all lesions in a patient were combined to calculate an average HU for each patient. The percentage of change in tumor density from the pretreatment evaluation to the post-treatment evaluation was calculated for each lesion, and the average percentage of change was then estimated for the patient. The percentage of change in tumor enhancement was graded on a scale of 1 to 4 as described previously (13). The tumor density was graded as follows: grade 4, decrease of >30%; grade 3, decrease of 11% to 30%; grade 2, decrease of ≤10% or increase of ≤10%; and grade 1, increase of >10%. Tumor response was also followed, where possible, by measurement of carcinoembryonic antigen (CEA), prostate-specific antigen (PSA), and CA19.9 tumor markers.

Analysis of viral shedding by reverse transcription-PCR. Initial evaluation of the detection limit of reovirus RNA by 25 or 35 cycles of reverse transcription-PCR (RT-PCR) was done. Viral RNA was extracted from 140 μL stock using the QIAamp viral RNA mini-kit (Qiagen) and serially diluted and 5 μL was assayed directly by RT-PCR using the One-Step RT-PCR Enzyme Mix Kit (Qiagen). Reovirus s3 cDNA targeted primers used were forward 5′-GGGCTGCACATTACCACTGA and reverse 5′-CTCCTCGCAATACAACTCGT. PCR conditions were 50°C for 30 min (for reverse transcription); 95°C for 15 min; and 25 (or 35) cycles of 95°C for 30 s, 62°C for 45 s, and 72°C for 45 s followed by 72°C for 7 min (Supplementary Fig. S1). For the clinical assays, blood samples were taken at baseline, post-treatment dose and weekly during the first cycle. Blood from the contralateral arm was collected into EDTA tubes, centrifuged at 1,200 × g for 10 min at 4°C, and stored at -70°C. Urine, sputum, and fecal swab (after PBS elution) samples were also stored at -70°C. Samples were analyzed after the last treatment dose in each cycle and weekly using the 25- and 35-cycle RT-PCR. Reovirus RNA (300-bp PCR product) and water were included in all experiments as positive and negative controls, respectively.

Detection of neutralizing antireoviral antibodies. A modified neutralizing antibody assay was used to detect antibody titers at baseline and weekly during the first 2 cycles of treatment by measuring the effect of patient serum samples on the ability of reovirus to kill a monolayer of target mouse L929 cells (14). The neutralizing antireoviral antibodies (NARA) titer of serum samples was expressed as the last dilution causing <80% cell killing as described previously (15).

Estimation of virus titer from tumor samples. Paired pretreatment and post-treatment tumor biopsy samples were obtained in 3 patients and stored at -80°C until the time of analysis. At this time, the biopsies were thawed and macerated in 1 mL DMEM and centrifuged at 3,600 rpm for 5 min. Supernatant was taken, serially diluted (1:10), and placed on to L929 cells in quadruplicates in a 96-well plate. Titer was calculated using the Kärber statistical method for a standard TCID₅₀ assay. In addition, photomicrographs were taken of the plates at the time of calculating the viral titer.

Patients. Thirty-three patients with a variety of underlying malignant diagnoses were enrolled and treated over 8 dose levels with a total of 76 cycles (median, 2; range, 1-6). Patients received the following numbers of treatment cycles: 1 cycle, RL02, RL03, RL08, RL13, RL16, RL23, RL29, and RL32; 2 cycles, RL01, RL04, RL07, RL10, RL11, RL14, RL18, RL21, RL22, RL24, RL25, RL26, RL28, RL30, and RL33; 3 cycles, RL06, RL09, RL17, RL19, and RL31; 4 cycles, RL05, RL15, and RL20; 5 cycles, RL27; and 6 cycles, RL12. Patient demographics are displayed in Table 2. Tumor diagnoses and exposure to prior therapies are detailed in Table 3.

Safety. Treatment was well tolerated in all cohorts with no grade 4 toxicities. Grade 3 toxicity was restricted to flu-like symptoms (Table 4A) and uncomplicated lymphopenia and neutropenia, which were more prominent in cohort 8 (Table 4B). The most common toxicities observed were grade 2 (or lower) flu-like symptoms (fever, chills, fatigue, nausea/vomiting, and headache) that primarily occurred 2 to 6 h after reovirus administration and resolved fully with acetaminophen and nonsteroidal anti-inflammatory therapy. These symptoms were more frequent and pronounced on cycle 1. There was no relationship between the reovirus dose level and the incidence and grade of these symptoms. Patients who developed grade 2 flu-like symptoms on day 1 received prophylactic nonsteroidal anti-inflammatory drugs on other treatment days. One patient with pancreatic cancer presented with disease-related pulmonary embolism during cycle 1 at the 1 × 10⁸ TCID₅₀ dose level. Following anticoagulation treatment, he was rechallenged uneventfully with the same dose. During the first cycle, the second patient treated at 1 × 10¹⁰ TCID₅₀ had an asymptomatic increase in liver transaminases (grade 2), CK-MB (grade 3), and troponin-I levels (grade 1 at baseline to grade 3 on day 3). Electrocardiogram and echocardiogram were normal. These biochemical toxicities resolved to baseline on day 15 and the patient was rechallenged with a lower dose (3 × 10⁹) with no further sequelae. To characterize further the safety at the 1 × 10¹⁰ TCID₅₀ dose level, 4 more patients were included. No similar or other significant toxicities were observed and dose escalation continued to the 3 × 10¹⁰ dose level where 8 patients were treated. No DLT was observed at this level and this dose was recommended for phase II evaluation.

Viral biodistribution. All pretreatment and post-treatment blood, urine, saliva, and rectal samples were negative for reovirus detection using RT-PCR screening based on 25 cycles of amplification. This assay was used for decision-making for discharge from the hospital. As a result, all patients were discharged on schedule and no patient was kept in hospital because of viral persistence in screened samples. However, in recognition of the fact that the initial 25-cycle assay for viral excretion may have missed relatively low levels of viral RNA in the samples, we conducted further RT-PCR analysis based on 35 cycles of amplification (where the detection limit was 200 TCID₅₀; Supplementary Fig. S1). In these assays, positive signal was detected in cycle 1 in a small number of patients at days 5 and 15 after viral administration (Table 5; Fig. 1A and B).

NARA response. Data were available for the fold increase in neutralizing antibody titers for 29 patients. After treatment, 28 patients showed an increase in NARA titers to a maximum at 4 weeks (median increase, 250-fold; range, 9-6,437; Fig. 1C). This NARA titer remained constant during subsequent cycles (15).

Response assessment. No objective radiologic responses were observed in terms of Response Evaluation Criteria in Solid Tumors. Eight of the patients who were radiologically evaluable showed disease stabilization or minor response. Data on the sum of the diameters of target lesions before and after reovirus therapy are provided for 10 patients who had clinical evidence of disease stability (Table 6). There was no clear correlation between clinical stabilization and dose or duration of reovirus therapy. In patient RL19, a computed tomography scan done after 2 cycles of therapy showed bilateral necrosis of involved iliac lymph nodes and stable disease by Response Evaluation Criteria in Solid Tumors for 3 months (Fig. 2A-C; Supplementary Fig. S2). Response Evaluation Criteria in Solid Tumors analysis recorded a 16.4% reduction in the sum of the longest diameters of predefined target lesions at 84 days. Densitometric evaluation of the change in contrast enhancement in tumor tissue revealed a grade 3 (12%) reduction from a baseline level of 75.5 to 66 HU. The iliac lymph nodes were biopsied and histologic evaluation revealed extensive necrosis in keeping with the imaging appearances (Fig. 3). Patient RL27 had a 20% reduction in the sum of the longest diameters of predefined target lesions at 143 days (Fig. 4A-C; Supplementary Fig. S3). Densitometric evaluation of the change in contrast enhancement in tumor tissue revealed a grade 2 (<10%) reduction from a baseline level of 65 to 61 HU. Sequential data were not available for similar lesions without any treatment or after chemotherapy treatment.

Reductions in tumor markers (CEA, CA19.9, and PSA) were detected in 3 patients (Fig. 5A-D). Two patients (RL12 and RL17) with colorectal cancer had reductions in CEA levels of 68% and 27% (Fig. 5A and B) and stable disease for 3 and 6 months, respectively. Patient RL17 also had a detectable level of CA19.9 and this fell by 18% (Fig. 5C). Patient RL19 showed a 51% fall in PSA from 100.9 to 49.2 ng/mL over 4 weeks after completion of cycle 2 (Fig. 5D).

Viral replication in tumor biopsies. Biopsy analysis confirmed the presence of viable virus in post-treatment (but not pretreatment) tumor samples in 3 patients (RL03, RL04, and RL11). Supernatant from freeze-thawed and macerated tumor biopsies that were taken after the initiation of reovirus therapy caused cytopathic effect in L929 target cells (Fig. 6A, vi-x). Pretreatment samples did not cause cytopathic effect (Fig. 6A, i-v). For the 3 patients analyzed, the titer of reovirus recovered correlated directly with the dose of reovirus administered (Fig. 6B).

Oncolytic reovirus offers significant advantages over recently described virotherapies in that it has activity against a wide range of tumor types with activated Ras signaling pathways (16–18). The lack of a requirement for genetic modification retains the antitumor activity of wild-type reovirus (in contrast to other genetically modified viruses that sacrifice potency in achieving selectivity). Intratumoral delivery of reovirus has been shown to be safe in humans, with antitumor activity reported in patients with high-grade gliomas (12). This study has extended this approach to systemic delivery of reovirus to target disseminated disease.

The purpose of this trial was to determine the safety, DLT, and MTD of i.v. reovirus in patients with refractory solid malignancies, with an aim of defining the recommended phase II dose. Importantly, no DLT was observed and 3 × 10¹⁰ TCID₅₀ was defined as the MTD by virtue of the fact that this was the highest dose available for administration. The activation status of the ras pathway was not measured in this study, but we do not believe that this variable would have affected the MTD or recommended phase II dose. We have shown previously that viral replication in tumors in immunocompetent mice is not associated with systemic toxicity. Therefore, in a clinical situation in which the immune system is operating, we believe that the MTD or recommended phase II dose will be independent of the Ras pathway status in the tumor. It is also significant that there was no evidence of viral shedding in any patient using the RT-PCR assay designed for safety analysis. This finding suggests that this agent is potentially suitable for office-based administration. Further viral biodistribution studies using a more sensitive assay based on 35 cycles of RT-PCR showed detectable virus at 5 or 15 days in a small minority of patients. Although these data offer reassurance about patient and environmental safety, they may also represent evidence of relatively rapid clearance of virus from the systemic circulation.

As expected, a significant number of patients had measurable NARA titers pretreatment. All patients showed an early increase in NARA titers that remained essentially constant during subsequent treatment cycles. This may account for the relatively lower toxicity scores despite much higher doses of reovirus in later treatment cycles. Clearly, this NARA response may limit the effect of repeated therapy by rapid neutralization of virus before it reaches tumor sites. However, despite this, we found evidence of viable virus in biopsies taken after 2 cycles of treatment from 3 patients by quantitative assay. This suggests that reovirus is able to access tumor tissue in patients and persists there even in the face of a rapidly evolving NARA response.

Although evaluation of antitumor activity was not a primary endpoint of the study, patients were monitored for evidence of antitumor efficacy. There were no objective responses by Response Evaluation Criteria in Solid Tumors, but there were several indications of activity. Eight patients had radiologic evidence of stable disease, including 2 patients (RL19 and RL27) with 16.4% and 20% reductions in the sum of the longest diameters of target lesions. Densitometric analysis of the radiologic data revealed a grade 3 reduction in patient RL19 and a grade 2 reduction in patient RL27. These data must be considered in light of the fact that we do not have data from similar untreated or drug-treated lesions. However, they are of sufficient interest to lead us to include densitometric analyses in ongoing studies in which we are combining reovirus and chemotherapy. In addition, there were reductions in tumor marker levels (CEA, CA19.9, and PSA) in 3 patients and biopsy-confirmed evidence of tumor necrosis in metastatic lymph nodes in 1 patient. Clearly, in the absence of data on the Ras pathway activation status in the individual tumors, it is not possible to attribute the changes in tumor sizes and marker levels to reovirus treatment. Future studies that include molecular analyses of the Ras pathway may clarify this area.

Development of reovirus as a cancer therapeutic is likely to depend on an improved understanding of the mechanisms of reovirus-induced cell death, circumvention of humoral and cellular immune responses, and concomitant use with other anticancer agents. Reovirus may be a relevant therapeutic approach in both tumors with oncogenic Ras mutations and those with upstream or downstream Ras pathway activation. This opens up the prospect of using reovirus in a wide range of common tumor types. In certain tumor models, reoviral cytotoxicity may be mediated through activation of apoptosis rather than simply as a result of viral replication and lysis (19). Patients were not selected for this study based on the Ras status of their tumors, as this was primarily a safety evaluation. Future phase II studies, using the recommended phase II dose, will be enriched for patients with tumors that have known activation of the Ras pathway. Moreover, preclinical studies have confirmed that modulation of the immune response by concomitant administration of either cyclosporine A or anti-CD4/CD8 antibodies results in improved antitumor activity of i.v. administered reovirus in immunocompetent hosts (11, 20). Such immune modulation, by agents such as cyclophosphamide (21) or rituximab, may maximize antitumor activity and is now being pursued for this oncolytic virus in clinical trials. Furthermore, combinations of reovirus and standard anticancer agents, such as radiotherapy and cytotoxic drugs, are being pursued (22–24).

In conclusion, this trial confirms the feasibility of i.v. delivery of repeated high doses of reovirus. The recommended phase II dose is 3 × 10¹⁰ for 5 consecutive days every 4 weeks. This study has shown successful intratumoral localization of reovirus after systemic administration. These data provide the essential background for subsequent studies of reovirus either as a single agent or in combination with standard anticancer therapies and immune modulating agents.

M. Coffey is employed by Oncolytics Biotech and has an ownership interest in that company; K. Harrington, A. Melcher, and R. Vite have received research grants from Oncolytics Biotech; J. de Bono has received research support from Oncolytics Biotech.