(2001) 8, 89–98  2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00 www.nature.com/gt REVIEW Clinical research results with dl1520 (Onyx-015), a replication-selective adenovirus for the treatment of cancer: what have we learned?

D Kirn Viral and Genetic Therapy Programme, Imperial College School of Medicine, Imperial Cancer Research Fund, Molecular Oncology Unit, Hammersmith Hospital, London W12 0NN, UK

Replication-selective microbial agents hold promise as a cation was generally transient (Ͻ10 days), however, and novel cancer treatment platform. dl1520 (Onyx-015), an was variable depending on tumor histology. Single agent E1B-55 kD gene-deleted adenovirus, was the first such gen- efficacy has been limited to date (0–14% local tumor etically engineered agent to be tested in humans. Over 200 regression rates). In combination with , how- cancer patients have been treated to date on over 10 clinical ever, encouraging antitumoral activity has been demon- trials (phases I-III). The virus was generally well-tolerated at strated. These clinical research results demonstrate the doses of up to 2 × 1012 particles by intratumoral, intraperi- potential of this novel treatment platform, as well as the toneal, hepatic arterial and intravenous administration; no hurdles to be overcome. Novel replication-selective agents maximally tolerated doses were identified by any route of with improved potency are needed. Gene Therapy (2001) 8, administration. Viral replication was tumor-selective, and 89–98. was documented after administration by all routes; repli-

Keywords: clinical trials; adenovirus; cancer; dl1520; oncolytic

Introduction against cancer for nearly a century.7 However, up until the last decade investigators had used only wild-type or Cancer gene therapy, the genetic modification of cells for tumor-tropic isolates from a variety of viral species.8–10 therapeutic benefit against cancer, holds great promise. Following the development of recombinant DNA tech- Encouraging antitumoral efficacy and safety have been nology and molecular biology advances, viruses could be demonstrated in preclinical animal models using varied engineered to enhance the desired anti-cancer phenotype. approaches, including replace- Martuza et al11 first demonstrated the potential utility of ment, prodrug-activating expression and immu- genetically engineered replication-selective viruses for nomodulatory strategies. However, clinical trial results cancer treatment almost 10 years ago. Since that time have been disappointing to date. Although encouraging tumor selectivity has been genetically engineered or safety has been demonstrated, in general, antitumoral enhanced in a number of replication-selective microbes, 1,2 efficacy has been minimal or non-existent. The reasons including herpesvirus,12,13 adenovirus14,15 and Salmonella for this are certainly varied, but clearly two major limi- typhimurium.3,5 In addition, inherently replication-selec- tations have been the inability to achieve (1) sufficiently tive agents have been developed for cancer treatment3 high levels of gene expression in (2) sufficiently large including Newcastle disease virus,16 autonomous parvo- numbers of target cells to result in clinical benefit. These viruses,17 reovirus,18 vaccinia,19 poliovirus20 and VSV.21 limitations need to be overcome if cancer gene therapy Although preclinical data reported with these agents is ever to fulfil its promise. have been encouraging, many critical questions have One novel and potentially fruitful approach to achiev- awaited results from clinical trials. Microbial agents such ing higher level gene expression and more widespread as adenovirus have complex biologies, potentially includ- within tumors is the use of replication-selective ing species-specific interactions with host machinery 3,4 agents. Alternatively referred to as ‘oncolytic’ agents or and/or immune response effectors.22,23 Antitumoral effi- as ‘cancer biotherapy’, this approach capitalizes on the cacy and safety studies have been performed in rodent fact that microbial agents such as viruses and bacteria can or primate models, and all published animal tumor replicate within human tissues to levels that are many model data with replication-selective adenoviruses have logs higher than the input ‘dose’, kill the infected cell and come from immunodeficient mouse–human tumor xeno- 5,6 subsequently spread to adjacent cells. Replication- graft models.6,24,25 Therefore, data from cancer patients selective viruses have been studied for their utility have been eagerly awaited. After over 4 years of clinical development with dl1520, roughly 15 clinical trials have been completed and analyzed involving approximately Correspondence: D Kirn 250 patients. This review is the first attempt to assess the Clinical trial results with dl1520 (Onyx-015) D Kirn 90 data from these studies in aggregate to determine what intraperitoneal), intra-arterial infusion (initially hepatic we have learned about: (1) the clinical utility of dl1520 artery) and eventually intravenous administration. In (Onyx-015), specifically; and (2) about the biology and addition, only patients with advanced and incurable can- potential utility of replication-selective adenoviruses, in cers were initially enrolled on trials. Only after safety had general. Future research directions are also discussed. been demonstrated in terminal cancer patients were trials initiated for patients with premalignant conditions. Background: dl1520 (Onyx-015) Finally, clinical trials of combinations with chemotherapy were initiated only after the safety of dl1520 as a single One approach to engineering replication selectivity is to agent had been documented by the relevant route of delete viral genes that are necessary for efficient repli- administration. cation in normal cells but are expendable in tumor cells. The deletion approach was first described with herpesvi- rus. Martuza et al26 showed the therapeutic potential of Intratumoral indications the thymidine kinase gene-deleted herpesvirus dlsptk Cancer patients can benefit from the effective local ther- initially. Subsequently, this group constructed G207; this apy of an established tumor mass if the target tumor virus has the lacZ gene inserted into the ribonucleotide 27 mass causes morbidity or death before other masses do. reductase gene and has in both copies of the For example, patients with recurrent glioblastoma multi- ICP34.5 gene. dl1520 (Onyx-015) was the first adenovirus forme or head and neck cancer frequently die from local described to mirror this approach. McCormick hypothes- tumor progression without evidence of distant metast- ized that an adenovirus with deletion of a gene encoding ases. In contrast, eradication of a localized skin lesion in a -inhibitory , E1B-55kD, would be selective 14 a patient with widespread pulmonary or CNS metastases for tumors that had already lost p53 function. P53 func- is unlikely to be of benefit. Patients with recurrent head tion is lost in the majority of human cancers through and neck carcinomas were enrolled into the initial clinical mechanisms including gene , overexpression of trials because most suffer severe morbidity, and even p53-binding inhibitors (eg mdm2, human papillomavirus mortality, from the local/regional progression of treat- E6) or loss of that indirectly modulate p53 func- ARF 28–30 ment-refractory tumors; therefore, intratumoral adminis- tion such as p14 . E1B-55kD gene deletion from tration had the potential to cause substantial palliation adenovirus was associated with decreased replication + and even survival prolongation. This population was also and cytopathogenicity in p53( ) tumor cells versus chosen because of the accessibility of superficial tumors matched p53(−) tumor cells, relative to wild-type aden- 14,31 for direct injection and biopsy in the outpatient clinic set- ovirus, in RKO and H1299 cells. In addition, normal ting. Finally, patients with tumors in superficial neck and epithelial and endothelial cells were shown to be rela- oral locations would presumably better tolerate peritu- tively resistant to dl1520-induced cytopathic effects and 15 moral inflammation and swelling than patients with replication. Based on these data, clinical trials were intraparenchymal tumors (eg intracranial, intrapulmon- initiated. Subsequently, our group and others demon- ary or intrahepatic). Once the safety of intratumoral injec- strated that the mechanism of selectivity of this virus was tion was demonstrated in the superficial neck and oral far more complex in vitro than was originally proposed regions, trials of intratumoral injection in solid organs (see section labeled ‘Controversy . . .’ in this manuscript). (pancreas, liver) were carried out. Nevertheless, the definitive data on the clinical utility of this virus have to come from clinical trials. Intracavitary indications A staged approach to the clinical Tumor types that spread and/or cause complications development of replication-selective agents primarily within specific body cavities are potentially amenable to intracavitary administration of therapeutic dl1520 (Onyx-015, now CI-1042, Pfizer Corporation, Gro- agents. Examples include mesothelioma (pleural cavity), ton, CT, USA) is a novel agent with a novel mechanism ovarian carcinoma (peritoneal cavity) and recurrent of action. This virus was the first genetically engineered, superficial bladder carcinoma (bladder). In addition, sev- replication-selective virus to be used in humans. We pre- eral premalignant conditions are also amenable to super- dicted that both toxicity and efficacy might be dependent ficial intracavitary administration, including Barrett’s on multiple factors including: (1) the inherent ability of esophagus and oral dysplasias (eg oral leukoplakia). a given tumor to replicate and shed the virus; (2) the Intraperitoneal administration to patients with advanced, location of the tumor to be treated (eg intracranial versus refractory ovarian carcinoma was followed by intra-eso- peripheral); and (3) the route of administration of the phageal instillation in patients with Barrett’s esophagus. virus. In addition, we felt it would be critical to obtain The virus was sequestered within the affected region of biological data on viral replication, antiviral immune the esophagus following instillation through a Wilson– responses and their relationship to antitumoral efficacy Cook catheter by occlusive proximal and distal balloons. in the earliest phases of clinical development. Finally, oral dysplasias were targeted through adminis- We therefore designed and implemented a novel tration as a mouthwash. staged clinical research and development approach (Figure 1). The goal of this approach was to increase sys- temic exposure to the virus sequentially only after safety Vascular delivery: intra-arterial and with more localized delivery had been demonstrated. intravenous administration Following demonstration of safety and biological activity by the intratumoral route, trials were sequentially Although patients with the indications listed above can initiated to study intracavitary instillation (initially potentially benenfit from local–regional therapy, systemic

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 91

Figure 1 A staged clinical research and development approach for a replication-selective agent in cancer patients. Following demonstration of safety and biological activity by the intratumoral route, trials were sequentially initiated to study intracavitary instillation (initially intraperitoneal), intra- arterial infusion (initially hepatic artery) and eventually intravenous administration. In addition, only patients with advanced and incurable cancers were initially enrolled on trials. Only after safety had been demonstrated in terminal cancer patients were trials initiated for patients with premalignant conditions. Finally, clinical trials of combinations with chemotherapy were initiated only after the safety of dl1520 as a single agent had been documented by the relevant route of administration. antitumoral efficacy can have a much greater impact on moral injections with dl1520 (Onyx-015) are outlined in overall cancer-related mortality. Preclinical studies Table 1. Dosing on clinical trials was done on the basis proved that intravenous adenovirus could infect, repli- of infectious units (plaque-forming units, p.f.u.), but it is cate within and inhibit the growth of established meta- now standard to dose on the basis of viral particles. static tumors.6 However, nude mouse–human tumor xen- Therefore, all doses used in this review are p.f.u. con- ograft models were of unkown relevance to human verted to particles (particle: p.f.u. ratio 20:1). cancer patients from both safety and efficacy standpoints. No maximally tolerated dose or dose-limiting toxicities Targeted intra-arterial infusions were studied in the first were identified following intratumoral injection doses of intravascular trial.32 Colorectal carcinoma metastases to up to 2 × 1012 particles.34 This safety is true not only for the liver cause morbidity and death in a large proportion tumors injected in superficial neck and oral sites, but also of these patients, and these metastases receive у90% of for intrahepatic and intra-pancreatic tumor masses, as their blood flow from the hepatic artery. Hepatic artery well.35 No clinically significant hepatitis or pancreatitis infusions had therefore been used previously to target was demonstrated; transient low-grade episodes were colorectal liver metastases with a variety of agents.33 rarely documented. Flu-like symptoms were the most Once safety by this route of administration had been common associated toxicities. No clear association demonstrated, intravenous trials were initiated in between flu-like symptoms and viral dose or treatment patients with lung metastases. cycle was demonstrable. Phase I/II and phase II trials reported a similar lack of clinically significant tox- Results from clinical trials with dl1520 icities.36,37,63 This safety is remarkable given the daily or (Onyx-015) even twice-daily dosing that was repeated every 1–3 weeks in the head and neck region or pancreas. Local Toxicity complications of intratumoral injections in the pancreas The toxicity findings from trials of dl1520 (Onyx-015) are appeared to be related to the endoscopic ultrasound- outlined in Tables 1–3. Data from phase I trials of intratu- guided injection procedure rather than to the agent itself;

Table 1 Toxicity data from phase I trials of dl1520 (ONYX-015): intratumoral injection

Tumor type n Location of injected Dose/Cycle (particles) Regimen/Cycle Maximally Most frequent related tumor frequency tolerated adverse events (% of dose/Dose- patients with event) limiting toxicity

Head and neck- 32 cervical, 2 × 108–2 × 1012 Single dose/q4 wk none Fever (35) squamous supraclavicular, 1011–1012 Daily × 5/q4 wk none Chills (35) oral, tongue Injection site pain (25) Gastrointestinal- 19 liver 2 × 109–2 × 1012 Single dose/q4 wk none Fever (75) colorectal, gastric, Asthenia (40) pancreatic Chills (35) Pancreatic 22 pancreas 2 × 109–2 × 1012 Single dose/q4 wk none Fever (78) Asthenia (57) Nausea (43)

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 92 Table 2 Toxicity data from phase I/II or phase II trials of dl1520 (ONYX-015): intratumoral injection

Tumor type n Location of injected Dose/Cycle Regimen/Cycle frequency Most frequent related adverse tumor (particles) events (% of patients with event)

Head and neck-squamous 30 cervical, 1012 Daily × 5/q3 wk Fever (73) (standard) supraclavicular, Asthenia (50) oral, tongue Injection site pain (47) Head and neck-squamous 10 same 4 × 1012 Twice daily (×5 days)/2 wk on, Injection site pain (80) hyperfractionated 1 wk off (hyperfractionated) Fever (60) Asthenia (60) Pancreatic 21 pancreas 2 × 1011 Single dose/day 1, 5, 8, 15a Fever (52) (first 3 pts (endoscopic ultrasound Nausea (48) 2 × 1010) administration) Asthenia (38)

aSubsequent cycles administered with gemcitabine chemotherapy.

Table 3 Toxicity data from phase I and I/II trials of dl1520 (ONYX-015): intraperitoneal, intra-arterial or intravenous injection

Route of n Tumor type Dose/cycle Regimen/Cycle Maximally tolerated dose/Dose- Most frequent adverse events administration (particles) frequency limiting toxicity (% of patients with event)

Intraperitoneal 16 ovarian 1011–1013 Daily × 5/q Bulky tumors (у2 cm) – Fever (75) phase I 3 wk DLT Ȱ 1012: abdominal pain, Abdominal pain (60) diarrhea, nausea/vomiting Nausea/vomiting (60) Non-bulky tumors (Ͻ2 cm) – no DLT Ȱ 1013 Intra-arterial 33 gastrointestinal – 2 × 109– Single Fever (90) (hepatic primarily colorectal 2 × 1012 dose/q 1 wk/ none Chills (68) artery) day 1, 8, 22a,50a Nausea (42) phase I/II Intravenous 10 metastatic 2 × 108– Single none Fever (100) phase I carcinoma 2 × 1012 dose/q 1 wk Chills (100) (3 wk on, 1 off) Transaminitis (67) grade 1–2 (mild–moderate) transient (Ͻ10 days)

aCycles given in combination with chemotherapy.

these included bacteremia, cyst formation and a tear of of normal for the assay; although the maximally tolerated the doudenal wall.36 Each of these complications were intravenous dose has not yet been defined, it may there- avoided once procedural changes were made and fore be close to this dose. Further dose escalation was prophylactic antibiotic treatment was mandated. limited by virus supplies. Intraperitoneal administration was also well-tolerated, in general (Table 3).38 Intraperitoneal administration was Viral replication feasible at doses up to 1013 particles divided over 5 days. Viral replication was assessed by two methods during The most common toxicities included fever, abdominal clinical trials with dl1520. The first was in situ hybridiz- pain, nausea/vomiting and bowel motility changes ation for adenoviral DNA in tumor biopsy samples; repli- (diarrhea, constipation). The severity of the symptoms cation was demonstrated by nuclear predominance of appeared to correlate with tumor burden. Patients with adenoviral DNA staining and associated cytopathic heavy tumor burdens reached a maximally tolerated dose effects. The second method was to test blood samples for at 1012 particles (dose-limiting toxicities were abdominal adenoviral genomes by quantitative PCR; since the initial pain and diarrhea), whereas patients with a low tumor input of viral genomes is cleared from the blood in 6– burden tolerated 1013 without significant toxicity. 24 h, genomes in the blood on day 3 or after are indica- No dose-limiting toxicities were reported at doses up tive of viral replication. These two methods gave nearly to 2 × 1012 particles (hepatic artery)32 or 2 × 1013 particles identical results in patients with head and neck cancers. (intravenous).39 Fever, chills and asthenia following intra- Histologic analysis of tumor biopsies is attractive vascular injection were more common and more severe because the nature and distribution of the infected cells than after intratumoral injections (grade 2–3 fever and can be evaluated. However, the evaluation of biopsy chills versus grade 1). Dose-related transaminitis was samples to assess replication has severe limitations. First, reported infrequently. The transaminitis was typically false negatives are possible given the small amount of transient (Ͻ10 days) and low-grade (grade 1–2) and was tissue obtained. Post-treatment biopsies can also be inev- not clinically relevant at doses up to 2 × 1012 particles. aluable if necrotic tissue is obtained. Ethical and practical However, at the highest intravenous doses administered considerations also limit the number of samples that can (6 × 1012 particles to 2 × 1013 particles), AST/ALT levels be obtained over time, particularly in tissues that are not reached approximately three to five times the upper limit superficial and therefore require invasive procedures to

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 93 access them. Quantitative PCR is far more practical. Immune response Blood samples can be conveniently obtained and ana- Neutralizing titers to the coat (Ad5) of dl1520 lyzed at multiple time-points. False negative results are were positive but relatively low in roughly 50–60% of all possible, however, since the lower limit of detection is clinical trial patients at baseline.34,40 Antibody titers 104 genomes per milliliter and viral shedding into the increased uniformly following administration of dl1520 bloodstream is required for detection. It was encouraging by any of the routes tested, in some cases to levels that the frequency of replication detected on several head Ͼ1:80 000.34 Antibody increases occurred regardless of and neck cancer trials by biopsy staining was nearly evidence for replication or shedding into the blood- identical to that determined by plasma sample PCR test- stream. Flu-like symptoms (fevers, rigors) were signifi- ing. At this time, it appears that these two approaches cantly more frequent and severe with intravascular each have merits and that they are complementary. administration than with intratumoral injections.41 The Viral replication has been documented at early time- acute inflammatory cytokine response to hepatic arterial points after intratumoral injection in head and neck can- infusion was evaluated using RT-PCR for specific cyto- cer patients by both tests.37,63 Approximately 70% of kine mRNAs from buffy coat leukocyte samples.32 The patients with either biopsy analysis or plasma testing by levels of the following were determined before treatment, PCR had evidence of replication on days 1–3 after the last 3 h post- and 18 h post-treatment: IL-1, IL-6, IL-10, inter- treatment. In contrast, days 14–17 samples were uni- feron-gamma, tumor necrosis factor. Significant increases formly negative. This time-course for replication mirrors were demonstrated within 3 h for IL-1, IL-6, tumor closely the clinical evidence for biological activity (eg necrosis factor and to a lesser extent interferon-gamma; local inflammation and necrosis). Intratumoral injection all cytokines were back down to pretreatment levels by of liver metastases (primarily colorectal) led to similar 18 h. In contrast, IL-10 did not increase until 18 h. Future PCR results at the highest doses of a phase I trial analyses will attempt to correlate clinical outcomes with (unpublished data); high quality biopsy samples could cytokine levels. not be collected given the location of these tumors. Patients with elevated neutralizing antibody titers Patients with injected pancreatic tumors, in contrast, before treatment were less likely to have evidence of viral 32 showed no evidence of viral replication by plasma PCR DNA in the blood 3–5 days after treatment; intra- or fine needle aspiration (Gene Therapy, in press). Simi- arterial administration was more sensitive to antibody = larly, intraperitoneal dl1520 could not be shown reprodu- inhibition (P 0.01) than intratumoral administration = 32 cibly to infect ovarian carcinoma cells within the perito- (P 0.09) (and unpublished data). This effect cannot neum.38 None of the plasma PCR samples was positive simply be ascribed to enhanced clearance of the virus and only one of 12 peritoneal fluid samples was faintly from the plasma by ; viral pharmacokinetics were unchanged between cycle 1 (low antibody titers) positive in 3–5 cells. Therefore, different tumor types can 32 vary dramatically in their permissiveness for viral infec- and cycle 3 (high antibody titers). Whether plasma PCR tion and replication. negativity reflects decreased viral replication, decreased Proof of concept for tumor infection following intra- shedding or both is not clear at this time. arterial or intravenous administration with human aden- Efficacy with dl1520 (Onyx-015) as a single agent ovirus has been achieved. Approximately half of the The single agent efficacy of ONYX-015 is outlined in roughly 25 patients receiving hepatic artery infusions of Table 4. Two phase II trials enrolled a total of 40 patients 2 × 1012 particles were positive by PCR 3–5 days follow- with recurrent head and neck cancer.42,63 Tumors were ing treatment.32 Three of four patients with metastatic treated aggressively with six to eight daily needle passes carcinoma to the lung treated intravenously with у × 12 for 5 consecutive days (30–40 needle passes per 5 day 2 10 particles were positive for replication by PCR cycle; n = 30) and 10–15 per day on a second trial (50–75 ± 39 on day 3 ( 1). Therefore, it is feasible to infect distant needle passes per cycle; n = 10). The median tumor vol- tumor nodules following intravenous or intra-arterial ume on these studies was approximately 25 cm3; an aver- administration. age cm3 of tumor therefore received an estimated four to Perhaps the optimal method to determine the exact five needle passes per cycle. Despite the intensity of this extent and duration of intratumoral replication is to treatment, the unconfirmed response rate (у50% shrink- resect the entire tumor mass at predetermined time- age at a single point in time) was only 13%. Therefore, points after injection. This strategy was used in patients even in a tumor that can be extensively and repeatedly with surgically resectable head and neck tumors (S Mor- injected, the majority of injected tumors did not objec- ley and S Kaye, Beatson Institute, Glasgow). Half of the tively respond. Interestingly, there was no correlation tumor mass was injected; the contralateral tumor half between evidence of antitumoral activity and neutraliz- was used as an internal control during evaluation of rep- ing antibody levels at baseline or after treatment.37,63 lication, necrosis and immune cell infiltration. Viral repli- No objective responses were demonstrated in patients cation results were very similar to those obtained with tumor types that could not be directly and aggress- through biopsies and/or blood PCR; replication was ively injected (due to their deep locations). Although common within р72 h after injection (particularly in p53 some evidence of minor shrinkage or necrosis was tumors), but it was short-lived. Reproducible obtained, no objective responses were documented with treatment-associated immune cell infiltrations were not intratumoral injection of either pancreatic cancer (phase seen in these tumors. No replication or necrosis induction I and II trials; n = 43 patients)36 or gastrointestinal carci- was demonstrable following direct injection of normal nomas (phase I trial, primarily colorectal; n = 19 patients) buccal mucosa. Transient tumor-selective replication was (unpublished data). Similarly, no responses were seen therefore confirmed on this study (S Morley, personal following intraperitoneal administration (phase I; n = 16 communication). ovarian cancer patients)38 or following intravenous

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 94 Table 4 Antitumoral efficacy data of dl1520 (ONYX-015 as a single agent: intratumoral, intraperitoneal, or intravenous injection

Route of administration Tumor type Phase Dose/Cycle Regimen/Cycle frequency у50% tumor regressiona (particles) No. responders/total (%)

Intratumoral head and neck I 2 × 108–2 × 1012 Single dose/q 4 wk Intratumoral head and neck II 1012 Daily × 5/q 3 wk unconfirmed, intent-to- treat: 4/30 (13) confirmed, intent-to-treatb: 2/30 (7) unconfirmed, evaluable: 4/19 (20) Intratumoral gastrointestinal liver I 2 × 109–2 × 1012 Single dose/q 4 wk 0/19 (0) metastases – colorectal, gastric, pancreatic Intratumoral pancreatic (CT-guided) I 2 × 108–2 × 1012 Single dose/q 4 wk 0/22 (0) Intratumoral pancreatic (endoscopic US) I 2 × 1010 (n = 3) Single dose/days 0/3 (0) II 2 × 1011 (n = 18) 1,5,8,15 c0/18 (0) Intraperitoneal ovarian I 1011–1013 Daily × 5/q 3 wk 0/16 (0) Intravenous carcinoma metastatic to lung I 2 × 108–2 × 1012 Single dose/q 1 wk 0/9 (0) (3 wk on, 1 off)

aNon-necrotic cross-sectional area used for regression assessment (ie necrotic area subtracted from total cross-sectional area). All regressions refer to shrinkage of the injected tumor mass only (ie distant, non-injected tumors not included). All regressions were in tumors with a p53 gene mutation. bEvaluable patients defined as those receiving Ͼ1 cycle of therapy and measurable tumor at baseline and at least one occasion Ͼ6 weeks after treatment initiation (ie patients without follow-up tumor measurements after 1+ cycles of treatment were excluded). Intent-to-treat analysis includes all patients receiving at least one dose of ONYX-015. The confirmed responses reflect those that were confirmed to be durable for у4 weeks on an intent-to-treat basis. cResponses of single agent ONYX-015 determined after 4 cycles (on day 35) on the pancreatic EUS phase I/II trial. Subsequent cycles given with chemotherapy.

administration in nine patients with metastatic carci- tiple trials (Table 5). Encouraging clinical data have been nomas (phase I; n = 9).39 No objective responses were obtained in patients with recurrent head and neck cancer demonstrated during the virus only treatment stage of a treated with intratumoral dl1520 in combination with phase I/ II hepatic artery infusion trial (n = 33 patients, intravenous and 5-fluorouracil.44 Thirty-seven predominantly metastases); responses patients were treated and 19 responded (54%, intent-to- were only seen on this trial in combination with chemo- treat; 63%, evaluable); this compares favorably with therapy (see below).32 Of note, many of the patients response rates to chemotherapy alone in previous trials described above were treated on the phase I portions of (30–40%, generally). The time-to-tumor progression was these trials in which tumor response was not a primary also superior to previously reported studies. However, endpoint. In addition, many of these patients had tumors comparisons with historical controls are unreliable. We that were refractory to standard therapies and/or were therefore used patients as their own controls whenever highly fibrotic, potentially making responses less likely. possible (n = 11 patients). Patients with more than one In summary, single agent responses across all studies tumor mass had a single tumor injected with dl1520 while were rare, and therefore combinations with chemo- the other mass(es) was left uninjected. Since both masses therapy were explored. were exposed to chemotherapy, the effect of the addition of viral therapy to chemotherapy could be assessed. The Efficacy in combination with chemotherapy: – dl1520-injected tumors were significantly more likely to chemotherapy augmentation respond (P = 0.017) and less likely to progress (P = 0.06) Cancer treatment failure results when tumors become than were non-injected tumors. Non-injected control resistant to standard therapies. Therefore, novel treat- tumors that progressed on chemotherapy alone were sub- ments that are not cross-resistant with standard chemo- sequently treated with Onyx-015 in some cases; two of therapies (ie work by different mechanisms) are needed. the four injected tumors underwent complete regressions. Combination therapy with agents that act by different These data illustrate the potential of viral and chemo- mechanisms should make the emergence of resistant dis- therapy combinations. The clinical utility of dl1520 in this ease less likely. Ideally, the toxicities associated with indication will be definitively determined in a ran- these agents would be non-overlapping, thus allowing domized phase III trial. safe combination treatment. Adenoviruses are therefore A phase I/II trial of dl1520 administered by hepatic well-suited for use in combination with chemo- artery infusion in combination with intravenous 5-fluor- therapies.4,43 Preclinical mouse tumor model studies have ouracil and leukovorin was carried out (n = 33 total).32 demonstrated that these agents can be safely and effec- Following phase I dose escalation, 15 patients with col- tively combined, and that efficient viral replication can orectal carcinoma who had previously failed the same still occur despite concomitant chemotherapy with cispla- chemotherapy were treated with combination therapy tin and 5-fluorouracil.64 after failing to respond to dl1520 alone; one patient Evidence for a favorable interaction between adenovi- underwent a partial response and 10 had stable disease ral therapy and chemotherapy has been obtained on mul- (2–7+ months). Chemosensitization of colorectal liver

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 95 Table 5 Evidence for potential synergy* between dl1520 ONYX-015) and chemotherapy from clinical trials

Route of Tumor type Phase Dose/Cycle Regimen/Cycle frequency Evidence for potential synergya administration (particles)

Intratumoral head and neck II 1012 ONYX-015 daily × 5 ONYX-015 injected tumors significantly + more likely to respond than matched, cisplatin day 1 i.v.b. non-injected control tumors (P = 0.017; 5-FU days 1–5 c.i. McNemar’s test) ONYX-015 injected tumors less likely to q3 wk progress than matched, non-injected control tumors (P = 0.006; log rank test) 2 of 4 tumors progressing on chemotherapy responded to same chemotherapy plus ONYX-015 Uncontrolled: response rate 63% versus historical 30–40% with chemotherapy and 14% with ONYX-015 alone Intratumoral pancreatic (endoscopic I 2 × 1010 (n = 3) ONYX-015 single dose US) II 2 × 1011 (n = 18) + gemcitabine i.v.b. None – 2 of 21 patients responded to q1 wk × 3, 1 wk off combination Intraperitoneal ovarian I 1011–1013 ONYX-015 daily × 5/ One patient had tumor response (Ͼ50% q3 wk reduction in CA-125) on platinum-based chemotherapy following ONYX-015, despite previous tumor progression on platinum-based chemotherapy alone and on ONYX-015 alone Intra-arterial gastrointestinal liver I 2 × 109–6 × 1011 ONYX-015 single dose One partial response, 10 stable disease (hepatic artery) metastases – primarily II 2 × 1012 + (2–7+ months) to combination ONYX-015 colorectal 5-FU/leucovorin i.v.b. plus 5-FU/leucovorin in patients with q4 wk tumor progression on both single agent ONYX-015 and on 5-FU/leucovorin alone 3/3 chemotherapy-naive patients responded to combination therapy Intravenous Metastatic carcinoma I 2 × 1010–2 × 1013 Single dose/q 1 wk q1wk (3 wk on, 1 off), then Stable disease for 6.5+ months in a with weekly patient with carboplatin/paclitaxel- carboplatin/paclitaxel refractory disease. aAlthough synergy cannot be definitively proven in phase II clinical trials, these clinical trial results are consistent with synergy and/or positive interaction between ONYX-015 and chemotherapy with cisplatin and/or 5-FU. i.v.b., intravenous bolus; c.i., continuous infusion; 5-FU, 5-fluorouracil; US, ultrasound. metastases is therefore possible via hepatic artery ing intravascular treatment. Acute inflammatory cyto- infusions, although the magnitude and frequency of this kines (especially IL-1 and IL-6) increased within 3 h fol- effect remains to be determined. In contrast, data from a lowing intra-arterial infusion. Neutralizing antibodies phase I/II trial studying the combination of dl1520 and increased in all patients, regardless of dose, route or gemcitabine chemotherapy (n = 21 patients) were not tumor type. Viral replication was documented in head encouraging; the combination resulted in only two and neck and colorectal tumors following intratumoral or responses, and these patients had not received prior gem- intra-arterial administration. Neutralizing antibodies did citabine. Therefore, potential synergy was demonstrated not block antitumoral activity in head and neck cancer with dl1520 and chemotherapy in two tumor types that trials of intratumoral injection. However, viral supported viral replication (head and neck, colorectal), replication/shedding into the blood was inhibited by but not in a tumor type that was resistant to viral repli- neutralizing antibodies; intra-arterial virus was more cation (pancreatic). sensitive to antibody inhibition than was intratumorally injected virus. Single agent antitumoral activity was lim- Results from clinical trials with dl1520 ited in head and neck cancers that could be repeatedly (Onyx-015): summary injected directly (13% unconfirmed regression rate). No objective responses were documented with single agent dl1520 has been extremely well-tolerated at the highest therapy in phase I or I/II trials in patients with pancre- × 12 practical doses that could be administered (2 10 – atic, colorectal or ovarian carcinomas. A favorable and × 13 2 10 ) by intratumoral, intraperitoneal, intra-arterial potentially synergistic interaction with chemotherapy and intravenous routes. The lack of clinically significant was discovered in multiple tumor types and by multiple toxicity in the liver or other organs was remarkable. Flu- routes of administration. like symptoms (fever, rigors, asthenia) were the most common toxicities and were increased in patients receiv-

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 96 Controversy regarding the mechanism of inadequate viral receptor expression (eg CAR, ) action of dl1520 and other mechanisms that are as yet unidentified. The original experiments demonstrated that E1B-55kD Future directions: improving the efficacy of gene deletion from adenovirus was associated with decreased replication and cytopathogenicity in p53(+) replication-selective agents tumor cells versus matched p53(−) tumor cells, relative to Given the high degree of safety but inadequate single wild-type adenovirus, in RKO cells.14 In addition, normal agent efficacy of dl1520 against advanced solid tumors, epithelial and endothelial cells were shown to be rela- second generation viruses will clearly be engineered for tively resistant to dl1520-induced cytopathic effects and greater potency. For example, a promising adenoviral replication.15 Based on these data clinical trials were E1A mutant has been described that demonstrates not initiated. Subsequently, however, our group and others only tumor selectivity but also significantly greater anti- demonstrated that the mechanism of selectivity of this tumoral efficacy in vivo compared with dl1520 (all models virus was far more complex in vitro than was originally tested) and even wild-type adenovirus (in a breast cancer proposed. First, p53 function can be lost in many cancers metastasis model).50,54 Similar results have been reported through mechanisms besides gene mutation; therefore, a with another E1A mutant.55 Potency can also be lack of correlation with the p53 gene status of tumors was improved by arming viruses with therapeutic genes (eg not unexpected in retrospect.43,45,46 The precise role of p53 prodrug-activating ),56–58 or even utilizing pro- in the inhibition of adenoviral replication has not been drug-activating enzyme genes already present in the clearly defined to date. In addition, other adenoviral pro- virus;59 however, prodrug conversion can inhibit viral teins also have p53 inhibitory effects (eg E4ORF6).47 replication in some models.57,59 Viral coat modifications Finally, E1B-55kD itself has important viral functions that may be beneficial if inadequate CAR expression plays a are unrelated to p53 inhibition (eg viral mRNA transport, role in the resistance of particular tumor types.60 host cell protein synthesis shut-off).48,52 Therefore, the Improved systemic delivery may require novel formu- replication selectivity of this E1B-55kD deletion mutant lations or coat modifications, as well as suppression of may be complicated by numerous factors.15,31,45,49 dl1520 the humoral immune response. Finally, we must identify (Onyx-015) has been extensively studied in vitro by many the mechanisms leading to the potential synergy between groups, and conflicting data on the role of p53 in modul- replicating adenoviral therapy and chemotherapy. This ating dl1520 replication and/or c.p.e. have come from dif- understanding may then allow us to bolster this interac- ferent cell systems; no p53 effect was demonstrated in tion. In addition to adenovirus, other viral species are matched U2OS cells, for example.46 In the HCT116 cell being developed.18,26,61 Since intratumoral spread also system, viral replication was inhibited by expression of appears to be a substantial hurdle for viral agents, p14ARF in a p53 gene wild-type tumor line but not in the inherently motile agents such as bacteria may hold great matched p53(−) tumor line, suggesting that loss of p14ARF promise for this field.62 expression may prevent p53-mediated blockage of viral The clinical development of the first-generation aden- replication; this effect was relatively minor, however ovirus dl1520 (Onyx-015) has taught us a great deal about (one-half log).65 Clinical trials were ultimately necessary the hurdles to be overcome with the replication-selective to determine the clinical utility of dl1520. adenovirus approach. However, this novel therapeutic platform clearly has great potential to improve and pro- Future directions: Why has dl1520 (Onyx- long the lives of cancer patients. 015) failed as a single agent for refractory Acknowledgements solid tumors to date? The following individuals have been instrumental in Future improvements with this approach will be possible making this manuscript possible: John Nemunaitis, Stan if the reasons for dl1520 failure as a single agent, and rela- Kaye, Tony Reid, Fadlo Khuri, James Abruzzesse, Eva tive success in combination with chemotherapy, are Galanis, Joseph Rubin, Antonio Grillo-Lopez, Carla uncovered. Factors that are specific to this adenoviral Heise, Larry Romel, Chris Maack, Sherry Toney, Nick mutant as well as factors that may be generalizable to LeMoine, Britta Randlev, Patrick Trown, Fran Kahane other viruses and/or bacteria should be considered. and Margaret Uprichard. Regarding this particular adenoviral mutant, it is important to remember that this virus is attenuated rela- tive to wild-type adenovirus in most tumor cell lines in References vitro and in vivo, including even p53 mutant 1 Merritt J et al. Clinical phase I experience with INGN 201 (Ad- tumors.31,45,46,50 This is not an unexpected phenotype p53) in non-small cell lung cancer and head and neck cancer. since this virus has lost critical E1B-55kD functions that Cancer Gene Ther 1997; 4: S12 (Abstr.). are unrelated to p53, including viral mRNA transport.51,52 2 Nemunaitis J et al. Three phase II trials of intratumoral injection Innate and acquired immune responses to the virus may with a replication-deficient adenovirus carrying the p53 gene be critical. Of note, a second deletion in the E3B gene (AdCMV-p53) in patients with recurrent/refractory head and region of dl1520 may make this virus more sensitive to neck cancer. Proc Am Soc Clin Oncol 1999; 18: 1661. 51,53 3 Kirn D. Replication-selective micro-organisms: fighting cancer innate antiviral immune effectors (eg TNF-alpha); an with targeted germ warfare. J Clin Invest 2000; 105: 836–838. immunocompetent animal model will need to be ident- 4 Heise C, Kirn D. Replication-selective adenviruses as oncolytic ified in order to explore the role of the immune response. agents. J Clin Invest 2000; 105: 847–851. Factors likely to be an issue with any virus include bar- 5 Pawelek J, Low K, Bermudes D. Tumor-targeted Salmonella as riers to intratumoral spread, antiviral immune responses, a novel anticancer vector. Cancer Res 1997; 57: 4537–4544.

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 97 6 Heise C et al. Intravenous administration of ONYX-015, a selec- 30 Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 tively-replicating adenovirus, induces antitumoral efficacy. Can- mutations in human cancers. Science 1991; 253: 49–53. cer Res 1999; 59: 2623–2628. 31 Harada J, Berk A. p53-independent and -dependent require- 7 Kirn D. Selectively-replicating viruses as therapeutic agents ments for E1B-55kD in adenovirus type 5 replication. J Virol against cancer. In: Gerschenson LA (ed.). Cancer Gene Therapy, 1999; 73: 5333–5344. Vol. 1. Academic Press: San Diego, 1998, pp 235–248. 32 Reid T et al. Hepatic arterial infusion of a replication-selective 8 Smith R et al. Studies on the use of viruses in the treatment of adenovirus, Onyx-015: a phase I/II clinical trial. Proc Am Soc carcinoma of the cervix. Cancer 1956; 9: 1211–1218. Clin Oncol 2000; 19: 953 (Abstr.). 9 Webb HE, Smith CE. Viruses in the treatment of cancer. Lancet 33 Kemeny N et al. Hepatic arterial infusion of chemotherapy fol- 1970; 1: 1206–1208. lowing resection of hepatic metastases from colorectal cancer. N 10 Asada T. Treatment of human cancer with mumps virus. Cancer Engl J Med 1999; 341: 2039–2048. 1974; 34: 1907–1928. 34 Ganly I et al. A phase I study of Onyx-015, an E1B attenuated 11 Martuza RL et al. Experimental therapy of human glioma by adenovirus, administered intratumorally to patients with recur- means of a genetically engineered virus mutant. Science 1991; rent head and neck cancer. Clin Cancer Res 2000; 6: 798–806. 252: 854–856. 35 Hecht JR et al. A phase I study of multiple direct injections of 12 Mineta T, Rabkin SD, Martuza RL. Treatment of malignant ONYX-015 adenovirus under endoscopic ultrasound guidance. gliomas using ganciclovir-hypersensitive, ribonucleotide Proc Am Soc Clin Oncol 1999; 18: 186 (Abstr.). reductase-deficient herpes simplex viral mutant. Cancer Res 36 Hecht R et al. Endoscopic ultrasound-guided intratumoral injec- 1994; 54: 3963–3966. tion of pancreatic carcinomas with a replication-selective aden- 13 Mineta T et al. Attenuated multi-mutated herpes simplex virus- ovirus: a phase I/II clinical trial. Proc Am Soc Clin Oncol 2000; 1 for the treatment of malignant gliomas. Nat Med 1995; 1: 19: 1039 (Abstr.). 938–943. 37 Kirn DH et al. A phase II trial of ONYX-015, a selectively-rep- 14 Bischoff JR et al. An adenovirus mutant that replicates selec- licating adenovirus, in combination with cisplatin and 5-fluor- tively in p53-deficient human tumor cells (see comments). ouracil in patients with recurrent head and neck cancer. Proc Science 1996; 274: 373–376. Am Soc Clin Oncol 1999; 18: 1505 (Abstr.). 15 Heise C et al. ONYX-015, an E1B gene-attenuated adenovirus, 38 Vasey P et al. A phase I trial of an E1B-55kD gene-deleted aden- causes tumor-specific cytolysis and antitumoral efficacy that can ovirus administered by intraperitoneal injection into patients be augmented by standard chemotherapeutic agents (see with advanced, refractory ovarian carcinoma. Proc Am Soc Clin comments). Nat Med 1997; 3: 639–645. Oncol 2000; 19: 1512 (Abstr.). 16 Lorence RM et al. Complete regression of human fibrosarcoma 39 Nemunaitis J, Cunningham C, Randlev B, Kirn D. A phase I xenografts after local Newcastle disease virus therapy. Cancer trial of intravenous administration with a replication-selective Res 1994; 54: 6017–6021. adenovirus, dl 1520. Proc Am Soc Clin Oncol 2000; 19: 724 17 Van Pachterbeke C et al. Parvovirus H-1 inhibits growth of (Abstr.). short-term tumor-derived but not normal mammary tissue cul- 40 Khuri F et al. A controlled trial of Onyx-015, an E1B gene-deleted tures. Int J Cancer 1993; 55: 672–677. adenovirus, in combination with chemotherapy in patients with 18 Coffey M, Strong J, Forsyth P, Lee P. Reovirus therapy of tumors recurrent head and neck cancer. Nat Med 2000; 6: 879–885. Science 282 with activated ras pathway. 1998; : 1332–1334. 41 Reid A et al. A phase I/II trial of ONYX-015 administered by 19 Mastrangelo M, Eisenlohr L, Gomella L, Lattime E. Poxvirus hepatic artery infusion to patients with colorectal carcinoma. vectors: orphaned and underappreciated. J Clin Invest 2000; 105: EORTC-NCI-AACR Meeting on Molecular Therapeutics of Can- 1031–1034. cer, 1999. 20 Gromeier M et al. Intergeneric poliovirus recombinants for the 42 Kirn D et al. A phase II trial of intratumoral injection with an treatment of malignant glioma (see comments). Proc Natl Acad E1B-deleted adenovirus, ONYX-015, in patients with recurrent, Sci USA 2000; 97: 6803–6808. refractory head and neck cancer. Proc Am Soc Clin Oncol 1998; 21 Stojdl DF et al. Exploiting tumor-specific defects in the interferon 17: 391a. pathway with a previously unknown . Nat Med et al 2000; 6: 821–825. 43 Heise C . ONYX-015, an E1B gene-attenuated adenovirus, 22 Wold WS, Hermiston TW, Tollefson AE. Adenovirus proteins causes tumor-specific cytolysis and antitumoral efficacy that can that subvert host defenses. Trends Microbiol 1994; 2: 437–443. be augmented by standard chemotherapeutic agents (see 23 Sparer TE et al. The role of human adenovirus early region 3 comments). Nat Med 1997; 3: 639–645. proteins (gp19K, 10.4K, 14.5K, and 14.7K) in a murine pneu- 44 Khuri FR et al. A controlled trial of intratumoral ONYX-015, a monia model. J Virol 1996; 70: 2431–2439. selectively-replicating adenovirus, in combination with cisplatin 24 Rodriguez R et al. Prostate attenuated replication competent and 5-fluorouracil in patients with recurrent head and neck can- adenovirus (ARCA) CN706: a selective cytotoxic for prostate- cer (see comments). Nat Med 2000; 6: 879–885. specific -positive cells. Cancer Res 1997; 45 Goodrum FD, Ornelles DA. p53 status does not determine out- 57: 2559–2563. come of E1B 55-kilodalton mutant adenovirus lytic infection. J 25 Heise C, Williams A, Olesch J, Kirn D. Efficacy of a replication- Virol 1998; 72: 9479–9490. competent adenovirus (ONYX-015) following intratumoral 46 Rothmann T et al. Replication of ONYX-015, a potential injection: intratumoral spread and distribution effects. Cancer anticancer adenovirus, is independent of p53 status in tumor Gene Ther 1999; 6: 499–504. cells. J Virol 1998; 72: 9470–9478. 26 Martuza RL et al. Experimental therapy of human glioma by 47 Dobner T, Horikoshi N, Rubenwolf S, Shenk T. Blockage by means of a genetically engineered virus mutant. Science 1991; adenovirus E4orf6 of transcriptional activation by the p53 tumor 252: 854–856. suppressor. Science 1996; 272: 1470–1473. 27 Mineta T et al. Attenuated multi-mutated herpes simplex virus- 48 Yew PR, Liu X, Berk AJ. Adenovirus E1B oncoprotein tethers 1 for the treatment of malignant gliomas. Nat Med 1995; 1: a transcriptional repression domain to p53. Genes Dev 1994; 8: 938–943. 190–202. 28 Ganly I, Soutar D, Brown R, Kaye S. p53 alterations in recurrent 49 Goodrum FD, Ornelles DA. The early region 1B 55-kilodalton squamous cell cancer of the head and neck refractory to radio- oncoprotein of adenovirus relieves growth restrictions imposed therapy. Br Cancer 2000; 82: 392–398. on viral replication by the . J Virol 1997; 71: 548–561. 29 Scheffner M, Munger K, Byrne JC, Howley PM. The state of the 50 Heise C et al. An E1A mutant adenovirus with potent and selec- p53 and retinoblastoma genes in human cervical carcinoma cell tive systemic antitumoral efficacy. Nat Med 2000; 6: 1134–1139. lines. Proc Natl Acad Sci USA 1991; 88: 5523–5527. 51 Barker DD, Berk AJ. Adenovirus proteins from both E1B reading

Gene Therapy Clinical trial results with dl1520 (Onyx-015) D Kirn 98 frames are required for transformation of rodent cells by that retains an intact thymidine kinase gene. Cancer Res infection and DNA transfection. Virology 1987; 156: 107–121. 1994; 54: 5745–5751. 52 Dobbelstein M et al. Nuclear export of the E1B 55-kDa and E4 60 Roelvink P et al. Identification of a conserved receptor-binding 34-kDa adenoviral oncoproteins mediated by a rev-like signal site on the fiber proteins of CAR-recognizing . sequence. EMBO J 1997; 16: 4276–4284. Science 1999; 286: 1568–1571. 53 Gooding LR. Regulation of TNF-mediated cell death and 61 Lattime EC, Lee SS, Eisenlohr LC, Mastrangelo MJ. In situ cyto- inflammation by human adenoviruses. Infect Agents Dis 1994; 3: kine gene transfection using vaccinia virus vectors. Semin Oncol 106–115. 1996; 23: 88–100. 54 Kirn D et al. Adenovirus E1A CR2 as selectively-rep- 62 Low K et al. Lipid A mutant salmonella with suppressed viru- licating agents for cancer. Cancer Gene Ther 1998. lence and TNF-alpha induction retain tumor-targeting in vivo. 55 Fueyo J et al. A mutant targeting the Rb Nat Biotechnol 1999; 17: 37–41. pathway produces anti-glioma effect in vivo. Oncogene 2000; 19: References added in proof 2–12. 63 Nemunaitis J et al. Selective replication and oncolysis in p53 56 Hawkins L et al. Replicating adenoviral gene therapy. Proc Am mutant tumors with Onyx-015, an E1B-55kD gene-deleted aden- Assoc Cancer Res 1999; 40: 476 (Abstr.). ovirus, in patients with head and neck cancer. Cancer Res 2000; 57 Freytag SO et al. A novel three-pronged approach to kill cancer 60: 6359–6366. cells selectively: concomitant viral, double suicide gene, and 64 Heisse C, Kirn D. Combination therapy with a replication-selec- radiotherapy (see comments). Hum Gene Ther 1998; 9: 1323–1333. tive adenovirus and cisplatin chemotherapy; dependence on 58 Wildner O, Blaese RM, Morris JM. Therapy of colon cancer with sequencing but independence from p53 gene status or route of oncolytic adenovirus is enhanced by the addition of herpes sim- administration. Clin Cancer Res (in press). plex virus-thymidine kinase. Cancer Res 1999; 59: 410–413. 65 Ries S et al. Loss of p14ARF in tumor cells facilitates replication 59 Boviatsis EJ et al. Long-term survival of rats harboring brain of the adenovirus mutant dl1520 (Onyx-015). Nature Med 2000; treated with ganciclovir and a herpes simplex virus 6: 1128–1133.

Gene Therapy