Inflammation, Apoptosis, and Necrosis Induced by Neoadjuvant Fas Ligand Gene Therapy Improves Survival of Dogs With Spontaneous Bone Cancer

Abstract

Fas ligand (FasL) gene therapy for cancer has shown promise in rodents; however, its efficacy in higher mammals remains unknown. Here, we used intratumoral FasL gene therapy delivered in an adenovirus vector (Ad-FasL) as neoadjuvant to standard of care in 56 dogs with osteosarcoma. Tumors from treated dogs had greater inflammation, necrosis, apoptosis, and fibrosis at day 10 (amputation) compared to pretreatment biopsies or to tumors from dogs that did not receive Ad-FasL. Survival improvement was apparent in dogs with inflammation or lymphocyte-infiltration scores >1 (in a 3-point scale), as well as in dogs that had apoptosis scores in the top 50th percentile (determined by cleaved caspase-3). Survival was no different than that expected from standard of care alone in dogs with inflammation scores ≤1 or apoptosis scores in the bottom 50th percentile. Reduced Fas expression by tumor cells was associated with prognostically advantageous inflammation, and this was seen only in dogs that received Ad-FasL. Together, the data suggest that Ad-FasL gene therapy improves survival in a subset of large animals with naturally occurring tumors, and that at least in some tumor types like osteosarcoma, it is most effective when tumor cells fail to express Fas.

Introduction

Every cell in the body has the potential to commit suicide under the appropriate physiological conditions. Cell suicide typically occurs as a result of activation of a cell death pathway that leads to apoptosis.¹ While apoptotic death pathways are present in all normal cells, the Fas “death receptor” (CD95 or APO-1) is often acutely “sensitized” in cells that are chronically activated and/or rapidly dividing. These can include cancer cells and activated leukocytes, including autoreactive lymphocytes and regulatory T cells. In tumors where signaling by Fas is intact, treatment with anti-Fas antibodies induces apoptosis in vitro.^2,3 However, major drawbacks of using anti-Fas antibodies in vivo are that Fas-resistant tumors would be spared and that, when administered systemically to destroy metastatic tumors, engagement of Fas in hepatocytes by anti-Fas antibodies can be lethal.⁴

The role of targeting Fas on tumor cells is controversial for other reasons. There is evidence that desensitization of tumor cells to Fas-mediated apoptosis by downregulation or inactivation of the Fas receptor or its associated signaling molecules can enhance metastatic potential to the lungs.^5,6,7 On the other hand, recent data showed that human (xenografted) and murine (genetically predisposed) ovarian tumors, and liver tumors induced by partial hepatectomy, required Fas-signaling for survival and growth.⁸ This raises multiple dilemmas about the potential efficacy of tumor therapy directed at the Fas receptor, especially using anti-Fas antibodies.

The physiological ligand for Fas is called Fas ligand (FasL). FasL is a type II transmembrane (and secreted) protein expressed by cytotoxic lymphocytes, monocyte/macrophages and some tumor cells. FasL-binding promotes oligomerization of Fas receptors, and in turn induces recruitment of Fas-associated death domain and procaspase-8 to the Fas cytoplasmic tail. Efficacy of ectopically administered FasL to treat tumors has been documented repeatedly in a variety of preclinical models using myriad approaches.^{2,9,10,11,12,13,14,15,16,17} Interestingly, the therapeutic response in these models does not require functional Fas expression by tumor cells, but instead, its common thread is localization of FasL to the tumor microenvironment where it induces robust inflammation. The inflammatory response is absolutely dependent on expression of functional Fas by host cells, especially neutrophils,^15,18 which upon interaction with the ectopically expressed FasL undergo apoptosis and promote further infiltration by professional antigen-presenting cells and, in immunocompetent hosts, eventually by lymphocytes.^14,18,19 Innate inflammation and rejection of the primary tumor occurs efficiently in athymic nude mice^16,20 and in lymphocyte-deficient SCID mice.¹⁷ However, infiltrating lymphocytes mount specific, durable, and transplantable antitumor responses only in immunocompetent hosts.^{12,16,17,19,20,21,22,23} Thus, FasL targeting of normal tissue promotes an innate immune system-mediated inflammatory response that destroys the primary tumor and leads to presentation of dead tumor antigens to cells of the adaptive immune system, in turn inducing a response where tumor specific T lymphocytes are generated.

These data support the concept of using Ad-FasL to induce tumor specific, re-circulating T lymphocytes to treat highly metastatic tumors that are resistant or inaccessible to conventional therapies. To translate this therapeutic approach to a clinically relevant model,²⁴ we designed a protocol to treat pet dogs diagnosed with spontaneous osteosarcoma using intratumoral administration of Ad-FasL in the neoadjuvant setting. The study followed an open-label design with the primary intent to address safety in large companion animals that meet similar clinical criteria to human cancers. Secondary endpoints were efficacy and assessment of mechanisms underlying antitumor activity, as a previous study using intratumoral administration of FasL (rDNA plasmid) to dogs with spontaneous melanoma suggested that direct tumor apoptosis might promote efficient immune priming.^10,25 In this study, administration of Ad-FasL was followed by implementation of standard of care after a 10-day delay. The delay (10-day window) was meant to provide time for the Ad-FasL to induce inflammation and T-cell priming, with peripheral migration of effector T cells that could target distant metastases.^10,17 Even though our approach was limited to only one dose of FasL before primary tumor removal, we hypothesized that FasL-mediated inflammation would lead to improved systemic outcomes defined either by time to progression (disease-free interval) or by an overall survival benefit.

The data show a good safety profile and a positive correlation between FasL-induced inflammation and local tumor destruction. In addition, the magnitude of the inflammatory response, as well as the disease-free interval and overall survival were significantly increased in dogs whose tumors had reduced or absent Fas expression.

Results

Demographics of participating dogs and tumor characteristics

Fifty-six dogs were enrolled in the study and received Ad-FasL therapy; 62 dogs with appendicular osteosarcoma treated by limb amputation with adjuvant carboplatin chemotherapy during the study enrollment period were used as the comparison population. The age, breed, and weight distributions for the study cohort and the comparison cohort were typical for dogs with osteosarcoma, as were clinical tumor stage and location (Supplementary Table S1). Exclusions for dogs that could not be evaluated for toxicologic assessment (N = 2), histopathologic assessment (N = 3), or efficacy (N = 8) and treatment variances are shown in Figure 1.

Figure 1.

Enrollment, exclusions, and outcomes. Flow chart provides details on dogs enrolled in the study and exclusions from each of the three measured endpoints. Two dogs were excluded from all the analyses. Treatment variances and censoring are indicated in the efficacy assessments box.

Intratumoral Ad-FasL therapy is safe in the outpatient setting

We monitored environmental shedding of adenovirus as suggested by the Institutional Biosafety and Recombinant DNA Advisory Committees. Buccal, nasal, and rectal swab samples collected in the hospital over 48 hours from the first five dogs in the study showed no cytopathic effects when cultured with HEK-293-crmA cells. Given the sensitivity of this cytopathicity assay, it is reasonable to conclude that very little or no infectious virus was shed from either the saliva or rectally from any of the treated dogs. Therefore, subsequent dogs were treated with Ad-FasL on an outpatient basis and released to their owners.

The toxicity profile of Ad-FasL therapy is acceptable

The characteristics of the study population and the dose schedule are described in Supplementary Tables S1 and S2. Intratumoral injections and Ad-FasL therapy were well tolerated. Dose escalation in the dog cohorts started at dose 4 (N = 3) and progressed uneventfully through dose #5 (N = 3) and dose #6, with this final cohort expanded to include 50 dogs. Twenty-two of the evaluable 54 dogs (41%, Figure 1) had no reportable toxicity according to VCOG criteria²⁶ in the 10 days before amputation, and 26 of 54 (48%) had grade-1 or grade-2 toxicity as the worst event. Thus, therapy-related toxicity was unremarkable in 48/54 dogs (89%). Of the remaining 6 dogs, 4 had grade-3 toxicity (7%) and 2 had grade-4 toxicity (4%) as the worst event during the evaluation period.

Recorded toxicities are listed in Table 1. In general, biochemical toxicity consisted of transient increases in aspartate transaminase and creatine phosphokinase with no apparent clinical impact. In one dog with grade-4 toxicity (hypotension), septicemia developed on day 8 and in spite of aggressive supportive care the dog died on day 10. Necropsy showed changes associated with generalized sepsis with the source undetermined; however large vegetative lesions were found on the valves of the heart indicating endocarditis as the probable source. The other dog with grade-4 toxicity (renal) developed azotemia on day 3, possibly secondary to nonsteroidal anti-inflammatory drugs, and the owners elected euthanasia in lieu of supportive care.

Table 1. Adverse events associated with administration of intratumoral Ad-FasL in dogs with osteosarcoma.

Fifty-four dog owners returned quality of life surveys for evaluation. Of those, 34 (63%) had no change (zero score) or a mild perceived increase in quality of life (positive score) at the end of the evaluation period. Fifteen (28%) had a mild perceived decrease (negative score) at the end of the evaluation period. Overall, owners did not perceive significant changes in quality of life parameters after the treatment was administered and during the 10-day delay of therapy period.

Intratumoral injection of Ad-FasL induces tumor-associated inflammation and apoptosis

Previous data from tumor-bearing laboratory animals show intratumoral administration of Ad-FasL is associated with inflammation and apoptosis.^2,10 Conversely, osteosarcomas are characterized by a striking paucity of inflammation.^27,28 Tissues from 53 dogs treated with Ad-FasL were used for histopathological evaluation. Two dogs with a final diagnosis that was different from osteosarcoma were included in assessments of toxicity and pathology, but not outcome (Figure 1). Pretreatment osteosarcoma biopsy samples from two dogs in the study had sufficient material for histological and immunohistochemical evaluation and were used as controls to establish a baseline for histological measures, along with biopsy samples from nine additional dogs that did not receive Ad-FasL. Supplementary Figure S1 shows photomicrographs of osteosarcomas from dogs after treatment with Ad-FasL to illustrate the appearance of tumors with mild (score = 1), moderate (score = 2), and severe (score = 3) fibrosis and necrosis.

Figure 2a shows the distribution of inflammatory infiltrates, necrosis, and fibrosis in control dogs that did not receive Ad-FasL (N = 9) or in dogs before Ad-FasL therapy (N = 2) and in the evaluable dogs treated with Ad-FasL (N = 53). Figure 2b shows representative photomicrographs of tumors from Ad-FasL treated dogs with mild (score = 1), moderate (score = 2), and severe (score = 3) inflammation. Seven of the 10 dogs with inflammatory infiltrates in the control population had mild inflammation, and the 3 dogs with moderate to severe inflammation in these untreated/pretreatment controls predominantly had acute, neutrophilic inflammation that was responsible for the high inflammatory scores. Neutrophils were typically within the peripheral areas of tumor necrosis or surrounding reactive fibrosis, but not within areas of viable tumor. Occasional admixed plasma cells were present in a few control samples, and lymphocytic inflammation, when present, was observed in areas of necrosis and fibrosis as well as within viable areas of tumor.

Figure 2.

Fas ligand therapy promotes inflammation, lymphocytic infiltrates, necrosis, and fibrosis in spontaneous canine osteosarcoma. Dogs with appendicular osteosarcoma received 1.12 × 10⁸ to 4.48 × 10⁸ pfu of Ad-FasL intratumorally during the initial diagnostic biopsy procedure. After 10 days, the affected limb was amputated and tumors were evaluated for inflammation, necrosis, or apoptosis using histomorphometry, and assigned scores of 0–3 (a), where 0 represents no evidence of inflammation, necrosis, or fibrosis in the tumor and 3 represents extensive, severe inflammation. The control population represents nine dogs biopsied with similar delay of therapy that did not receive Ad-FasL, and two dogs from the Ad-FasL group for which there was sufficient material for assessment at the time of initial biopsy (pretreatment). The top row in panel b shows photomicrographs of tumors with inflammation scores of 1, 2, and 3 (hematoxylin and eosin (H&E), 200X magnification). The middle and bottom rows in panel b show photomicrographs with examples of immunostaining for CD3 and cleaved caspase-3 in tumors from Ad-FasL-treated dogs (modified ABC staining, 200× magnification). FasL, Fas ligand.

Each of the dogs treated with Ad-FasL had measurable inflammation (Figure 2a); 26/53 had moderate to severe mononuclear (lymphocytic) inflammation, and only 8 had moderate to severe neutrophilic or mixed inflammation. Significantly greater (P < 0.05) areas of necrosis (90% compared to 35% with scores of 2 or 3) and fibrosis (80% compared to 30% with scores of 2 or 3) also were observable in tumors from Ad-FasL-treated dogs than in tumors from untreated dogs (Figure 2 and Supplementary Figure S1).

Consistent with the observations from hematoxylin and eosin-stained slides, tumors from dogs treated with Ad-FasL had an increased proportion of T-cells as compared to tumors from untreated dogs. Specifically, the box plot in Figure 3a illustrates the relationship between histomorphometric measures of lymphocyte infiltration and the number of CD3 T-cells in tumors from dogs that received Ad-FasL. The median fraction of infiltrating T-cells as a proportion of the total nucleated cell population was 17% in Ad-FasL-treated dogs but only 8% in untreated dogs (Figure 3b). B-cells made up only a minority of lymphocytic infiltrates in samples from untreated and from treated dogs (data not shown).

Figure 3.

Fasaret-mediated T-cell infiltrates and apoptosis. (a) Box plots showing the correlation between CD3 immunostaining and lymphocyte infiltrate scores. The white line in each box represents the median, the top and bottom of the box represent the 75% confidence intervals, and the whiskers represent outliers for each group (Pearson correlation score = 0.47). (b) Empirical cumulative distribution (%) of control dogs and treated dogs as a function of the proportion of labeled CD3⁺ cells in tissue sections. Median percentage of CD3⁺ cells in tumors from control dogs was 8%, whereas it was 17% in tumors from dogs treated with Ad-FasL (i.e., 8% of nucleated cells in control tumors were T-cells and 17% of nucleated cells in treated tumors were T-cells). (c) Empirical cumulative distribution (%) of control dogs and treated dogs as a function of the proportion of cells expressing cleaved caspase-3 in tissue sections. Median percentage of cleaved caspase-3⁺ cells in tumors from control dogs was 14%, whereas it was 41% in tumors from dogs treated with Ad-FasL (i.e., 14% of nucleated cells in control tumors were apoptotic and 41% of nucleated cells in treated tumors were apoptotic). FasL, Fas ligand.

There also was greater tumor-associated apoptosis in tumors from dogs treated with Ad-FasL than in tumors from untreated dogs (Figures 2b and 3c). Specifically, the median number of cells expressing cleaved (activated) caspase-3 as a proportion of the total nucleated cell population in treated dogs was 41% vs. 14% in untreated dogs (P < 0.05, Figure 3c). There were no significant differences in overall tumor cellularity between treated and untreated dogs; the inflammatory infiltrates appeared to create a trend toward greater cellularity in tumors from treated dogs, which was probably balanced by loss of cells due to apoptosis. The tumors from treated dogs also had increased matrix (data not shown).

The extent of tumor necrosis was not directly correlated with expression of cleaved caspase-3 (data not shown). This was predictable from the evaluation of hematoxylin and eosin-stained slides, which suggested that coagulative necrosis was the dominant type seen in the tumor samples. Likewise, many samples had large regions of replacement fibrosis where presumably tumor had been present originally.

Ad-FasL-mediated tumor inflammation is associated with improved survival

In laboratory animal models, FasL-mediated inflammation promotes antitumor immunity with rejection of tumor challenge, and under some circumstances improved survival.^2,10,17 Previous work from our lab suggested that FasL-mediated tumor apoptosis also might enhance antitumor immunity.²⁵ Thus, we examined correlations between inflammation and survival and between apoptosis and survival. The box plot in Figure 4a shows that dogs treated with Ad-FasL that developed moderate to severe lymphocyte infiltration at the tumor site had longer overall median survival than dogs whose tumors had mild lymphocyte infiltration (Pearson χ² = 0.049). Similar data were observed when total inflammation was used as a measure for efficacy. Figure 4b shows the proportion of dogs with no detectable metastasis and the proportion of dogs surviving at 200 days. Data from the comparison group are shown with the data from Ad-FasL-treated dogs that had mild (score = 1), moderate (score = 2), and severe (score = 3) tumor-associated inflammation. It is worth noting that dividing dogs into groups with moderate and severe inflammation resulted in small samples; so, even though it appears that greater inflammation might confer greater clinical benefit, these two groups were combined for additional survival analyses. Figure 4c shows Kaplan–Meier plots for disease (event)-free survival and Figure 4d shows Kaplan–Meier plots for overall survival in study dogs with mild (score = 1) or moderate to severe (scores = 2 or 3) inflammation, as well as for the comparison group. The table shows the median survival time and the 95% confidence intervals for each group. Overall survival for Ad-FasL-treated dogs with moderate to severe inflammation was significantly improved compared to dogs with mild inflammation or dogs that received only standard of care (log rank P ≤ 0.05), and a similar trend was present for disease-free survival (log rank P = 0.08). There was a trend for improved survival in dogs that had tumors with more extensive apoptosis (log rank P = 0.057), although there was no direct correlation between tumor-associated inflammation (or T-cell infiltrates) and the extent of apoptosis in the tumors of dogs after Ad-FasL therapy (r = 0.20 and r = –0.08, respectively).

Figure 4.

Inflammation induced by Ad-FasL neoadjuvant therapy is associated with improved survival of dogs with osteosarcoma. (a) Survival data as a function of inflammatory scores. Dogs with inflammation scores of 2 or 3 show longer median overall survival than dogs with inflammatory scores of 1 (P < 0.05). (b) Percent of dogs treated with standard of care or with standard of care plus neoadjuvant Ad-FasL stratified according to inflammation scores that showed no evidence of metastasis (disease-free) or that were alive at 200 days. (c) Kaplan–Meier analysis of disease free interval of dogs treated with standard of care (amputation plus adjuvant carboplatin, dashed line) or with the addition of neoadjuvant Ad-FasL that achieved inflammatory scores of 1 (solid line) or inflammatory scores of 2 or 3 (dotted line). Crosshatches represent censored dogs. (d) Kaplan–Meier analysis of overall survival of dogs treated with standard of care (dashed line) or with the addition of neoadjuvant Ad-FasL that achieved inflammatory scores of 1 (solid line) or inflammatory scores of 2 or 3 (dotted line). Crosshatches represent censored dogs. The table below the graphs shows the median survival and the 95% confidence intervals for each group. Overall survival between treated dogs with inflammation scores of 2/3 vs. standard of care alone or vs. dogs with inflammation scores of 1 are statistically significant as determined by log rank analysis (P ≤ 0.05). FasL, Fas ligand.

Tumor inflammation is associated with insensitivity to FasL

In laboratory animals, FasL-mediated inflammation is observed primarily in tumors that lack Fas receptors or otherwise have disengaged the Fas-mediated apoptosis response.¹⁰ In dogs, ~50% of melanoma and osteosarcoma samples tested express Fas receptor mRNA, and expression of this receptor generally predicted an apoptotic response upon exposure to Ad-FasL.^25,29 Thus, we examined whether Fas receptor expression by tumor cells was correlated with the magnitude of the observed inflammatory response in vivo. We quantified Fas mRNA expression in cultures enriched for tumor cells from pre-Ad-FasL treatment study samples using quantitative reverse transcriptase PCR. The fractional value for Fas expression normalized to GAPDH ranged from 4 × 10^–8 to 2 × 10^–4, with a median value of 4.9 × 10^–5. Data were median-centered with the median sample assigned a value of 1.00 and samples divided into 2 groups called “Fas-lo” (N = 7 samples with Fas mRNA expression below the median) and “Fas-hi” (N = 8 samples with Fas mRNA expression equal to or greater than the median, Supplementary Figure S2a). Eight pre-Ad-FasL treatment samples had sufficient material for immunohistochemical evaluation (Supplementary Table S4) and were assessed for Fas expression by immunohistochemistry. Examples of both “Fas-lo” (dogs 47 and 55) and “Fas-hi” (dogs 29 and 56) are shown in Supplementary Figure S2b. A correlation was present between Fas mRNA expression on cells cultured ex vivo and Fas protein expression on tissues (Supplementary Figure S2 and Supplementary Table S5). The 4 samples with the lowest Fas mRNA (dogs 47, 50, 52, 55) had no detectable Fas staining either in the whole tumor or in part of the tumor, and 3 of 4 samples with higher Fas mRNA (dogs 29, 37, 56) had clearly detectable staining across the whole tumor (Supplementary Table S5).

The magnitude of inflammation observed in vivo at day-10 after Ad-FasL therapy was inversely proportional to the levels of Fas mRNA in the cultured cells (Figure 5a). Furthermore, similar inverse correlations were observed between Fas mRNA expression and disease-free interval and between Fas mRNA expression and overall survival (Figure 5b). These differences were statistically significant (P = 0.05 and P = 0.01, respectively). There was no correlation between Fas mRNA expression and survival in dogs that did not receive neoadjuvant Ad-FasL therapy. Furthermore, while an abbreviated prognostic signature we defined previously for osteosarcoma (inversely related “cell cycle” and “microenvironment” gene clusters³⁰) was apparent in samples from dogs treated with neoadjuvant Ad-FasL, the pattern was not predictive for outcome (data not shown).

Figure 5.

Inflammation induced by Ad-FasL and outcomes are inversely proportional to Fas expression in tumor cells. (a) Box plots showing the correlation between Fas expression (measured by quantitative reverse transcriptase PCR (RT-qPCR)) and histomorphometric measures of inflammation in 15 study dogs tested. The white line in each box represents the median, the top and bottom of the box represent the 75% confidence intervals, and the whiskers represent outliers for each group (Pearson correlation score = –0.89). (b) Kaplan–Meier plots showing disease-free interval (left) and survival (right) for “Fas-lo” and “Fas-hi” groups from dogs treated with Ad-FasL. “Fas-lo” tumors are represented by solid lines and “Fas-hi” tumors are represented by dashed lines. Both disease-free interval and overall survival between dogs with lo Fas expression in their tumors vs. dogs with hi Fas expression in their tumors were statistically significant as determined by log rank analysis (P = 0.05 and P = 0.01, respectively).

Discussion

This study was designed to examine the effect of neoadjuvant Ad-FasL given to large animals living in the community (pet dogs) that develop cancer spontaneously. The choice of Ad-FasL was based on studies in rodents showing clinical benefit in experimental animals using established tumor cell lines. While the translational fidelity of these rodent models to human disease remains to be fully established, naturally occurring cancers of dogs serve a robust purpose in developmental therapeutics, both because of their inherent molecular and clinical similarities to human diseases and because of the potential to apply such therapies in the veterinary setting.²⁴ Canine osteosarcoma was especially well suited to test the safety of neoadjuvant Ad-FasL in a spontaneous cancer setting. The disease in dogs and in humans shares extensive molecular similarities.^29,30 Moreover, disease progression, even with standard of care, is sufficiently rapid so that assessments of improvement can be made in a relatively short period of time.²⁴

The sum of our data indicates that a single dose of neoadjuvant Ad-FasL delivered into the tumor was able to engage cells in the microenvironment of tumors, as predicted by the laboratory animal models, and induce a potent inflammatory response. A survival advantage was apparent in dogs that achieved higher inflammation and apoptosis scores. Since osteosarcoma is a heterogeneous disease where primary tumors in human patients (and dogs) are about as likely to express Fas receptors as not,^7,29 we examined expression of mRNA for Fas, as well as for 10 genes that form a minimally prognostic signature in canine and human osteosarcoma in a representative subset of samples from dogs treated with Ad-FasL in this study. We conclude that reduced Fas expression by tumor cells was the only variable associated with a survival advantage when Ad-FasL was used in the neoadjuvant setting.

The study design was conservative, starting at a dose of Ad-FasL that had reasonable probability to show a biological effect, and with room for escalation and de-escalation. A previous study that administered a replication-deficient adenovirus vector into the prostate gland of normal dogs at a similar dose to the one we used here showed no local (urinary system) or systemic toxicity associated with the adenovirus.³¹ Expression of the transduced reporter was seen exclusively in prostatic epithelial cells, and the dogs had no episodes of fever, suggesting minimal distribution of the adenovirus outside the target tissue.³¹ We similarly saw no toxicity that was attributable to the adenovirus vector, and overall, the data show that intratumoral Ad-FasL therapy was extremely safe. This is especially noteworthy given the route of administration into the medullary cavity of a long bone tumor using a simple catheter without micro-precision instruments. The bone marrow is intimately connected to the systemic vasculature, so the absence of systemic toxicity suggests that the adenovirus has sufficiently high affinity for stromal and/or tumor cells so as not to cause unacceptable systemic toxicity. It also is possible that reduced blood flow in osteosarcoma tumors³² might have contributed to maintain the adenovirus vector and cargo in the local microenvironment. Transduction with or intratumoral administration of empty adenovirus vectors or adenoviruses carrying reporter genes (e.g., Ad-GFP) do not elicit either tumor cell killing or protective antitumor responses in rodent models,^10,17 indicating that the inflammatory response that promotes antitumor immunity is initiated by gene transfer of FasL.¹⁷ The adenovirus-mediated expression of FasL also is abrogated in a relatively short time (~14 days) because the transduced cells are killed as a consequence of the inflammatory response.³³ Therefore, both mechanisms (Fas-mediated apoptosis and inflammation leading to tumor cell death) minimize persistence of FasL in the system.

The most commonly observed toxicities were transient increases in aspartate transaminase and creatine phosphokinase. While the specific cause for these changes is unclear, they became less common in dogs enrolled later in the study. We also observed elevations in serum alkaline phosphatase, an enzyme which is frequently elevated in dogs with osteosarcoma (bone source).³⁴ We did not establish the source of origin for the alkaline phosphatase in these dogs.

Two severe adverse events were seen. The dog that developed septicemia showed signs only 8 days after administration of the Ad-FasL. The source of the septic event could not be identified, but no other dogs showed persistent signs of systemic inflammation or any transient or persistent signs of infection. The dog that developed azotemia, and eventually renal failure, similarly did not show signs until 3 days after the therapy. Although these events were limited to one dog each, the development of FasL or similar products might require special attention to these events, especially in a setting where further escalation or repeat dosing would be applied.

A question that remained unanswered due to the limitations of the murine models was whether the inflammation and the consequent systemic antitumor immune response elicited by Ad-FasL would be seen equally in tumors that undergo Fas-mediated apoptosis, which dogmatically is a “silent” death, and in tumors that resist Fas-mediated apoptosis and where FasL can act principally to attract neutrophils and other leukocytes. Stratification of the treated population in our study revealed a subset of dogs that showed significant benefit, reflected by parallel improvements in disease-free interval and overall survival (in dogs, both tend to be closely linked due to the high frequency with which owners choose humane euthanasia upon the diagnosis of metastatic disease). This stratification was based on the presence of moderate to severe tumor-associated inflammation, and especially on the expression level of Fas receptors in tumor cells in a subset of dogs. The differences in inflammation were not due to other components of the therapeutic management. In fact, inflammation was consistently seen in tumors of the Ad-FasL-treated dogs in spite of the concurrent use of a nonsteroidal anti-inflammatory drug for pain control in all of the dogs enrolled in this study.

Our observation that inflammation induced by Ad-FasL and outcome were inversely proportional to Fas expression in tumor cells might be considered counterintuitive, as one could imagine that tumors expressing the highest levels of Fas would also be those most susceptible to FasL-mediated destruction. The paradoxical findings regarding the role of Fas expression to alternatively increase “fitness”⁸ or metastatic potential⁶ in tumors may be context-dependent and could be explained by two alternative mechanisms. One possible mechanism that we favor, and that is supported by our data on overall survival and Fas sensitivity, is that Fas-negative (or “Fas-lo”) cells are more resistant to the proapoptotic effects of the expressed FasL, and so allow for greater effects of this molecule on the microenvironment. We provide evidence that the low expressing Fas tumors are present in an environment where Fas expression in leukocytes is normal and no different from samples where Fas expression by the tumor was detectable at moderate to high levels. Thus our data support the interpretation that the “Fas-lo” tumor cells are only killed when inflammation ensues, generating a suitable environment for immune stimulation and systemic antitumor immunity, whereas “Fas-hi” tumor cells are likely to be more sensitive to FasL and die through “silent” apoptosis,²⁹ thus inducing a subdued immune response.

An alternative mechanism is that those tumors most susceptible to destruction by FasL are those that express the least amount of Fas-dependent survival signaling. If this were the case, it might be expected that FasL therapy would have a negative impact on survival in patients with “Fas-hi” tumors. Our data did not show a statistically significant survival advantage or disadvantage of “Fas-hi” tumor-bearing dogs treated with FasL as compared to what is observed using standard of care alone.

Thus, the sum of our data indicates that a single dose of neoadjuvant Ad-FasL delivered into the tumor was able to engage cells in the microenvironment of “Fas-lo” tumors, as predicted by the laboratory animal models, and induce a potent inflammatory response that was beneficial for the patient (in this case, the dog) that was independent of other molecular factors that influence tumor behavior. Despite extensive preclinical data in laboratory animals suggesting this therapy is likely to be effective against a large variety of solid tumors, its clinical application to tumors other than osteosarcoma may require direct testing to establish a link between efficacy signals and Fas receptor status for other potential histologies. Alternatively it may be that multiple dosing of Ad-FasL will permit the therapy to provide efficacy even for Fas-hi tumors.

Materials and Methods

Clinical study design. The design was a standard dose escalation with an expansion cohort. Details about the study design and the manufacturing of Ad-FasL are provided in Supplementary Materials and Methods (Supplementary Figure S3 and Supplementary Tables S1–S3). The study was conducted with approval from the institutional animal care and use committees and the Institutional Biosafety Committees from each participating institution (Colorado State University and the University of Minnesota). Fifty-six dogs were enrolled in the study (Supplementary Table S1 and Figure 1). Data from 54 dogs were evaluable for assessment of safety (toxicological monitoring); the causes for excluding 2 dogs from this assessment are shown in Figure 1. Tissue samples from 53 dogs were evaluable for histopathologic assessment; one biopsy sample did not yield sufficient tissue. Data from 48 dogs were evaluable for efficacy; the causes for excluding dogs from this assessment are shown in Figure 1.

Principle of therapy. The principle of treatment was based on a 10-day delay of therapy, based on preclinical data that showed statistically significant immunologic protection in animals treated 10 days before challenge with transplantable (syngeneic) tumors,¹⁰ as well as on the practicality of any delay of therapy approach. The experimental therapy was performed on day 1, and dogs were followed to assess local and/or systemic toxicity for 10 days. At the end of the experimental period, dogs were treated using standard-of-care for appendicular osteosarcoma (limb amputation plus neoadjuvant carboplatin chemotherapy³⁵). All dogs in the study received nonsteroidal anti-inflammatory drugs during the delay period and after amputation as part of the standard pain control regimen. Treatment variances are listed in Figure 1. The treatment sequence used is illustrated in Supplementary Table S3.

Cell culture. Processing of tumor specimens and culture of cells derived from these tumors was done as previously described.^36,37 A detailed description of the methods for derivation of these cells is provided in the Supplementary Materials and Methods section. Cells were maintained in culture in Dulbecco's modified Eagle's medium containing 5% glucose, supplemented with 10% fetal bovine serum, 0.1 % Primocin (Invivogen, San Diego, CA) and 10 mmol/l HEPES in a humidified atmosphere of 5% CO₂ at 37 °C.

RNA isolation and real-time quantitative reverse transcriptase PCR. RNA was isolated from cultured cells, quantified, and assessed for quality as described.³⁸ Elimination of genomic DNA and reverse transcription were done using QuantiTect Reverse Transcription Kit (Qiagen, Valencia, CA). Real-time quantitative reverse transcriptase PCR was performed on an Eppendorf Mastercycler ep realplex with FastStart SYBR Green Master Mix Protocol (Roche, Indianapolis, IN). Primer sequences are provided in Supplementary Table S6. GAPDH was used as the reference standard for normalization and relative levels of steady state mRNA were established using the comparative ΔΔCt method.³⁹ The transcript levels for FAS, GAPDH, AURKA, BUB1B, CCNA2, CCNB2, CTSS, FGF7, IGFBP2, RECK, SMARCA1, and TOP2A, were median-centered for reference. Additional validation included verification of a linear relationship between previously published microarray data and quantitative reverse transcriptase PCR values for transcripts of interest in an independent cohort.³⁰

Statistics. Descriptive statistics were established for variables of breed, gender, age at diagnosis, histological subtype and each of the histopathologic criteria examined. Categorical data are expressed as percentages, while continuous data are expressed as means or medians (ranges). Histological score comparisons were done using the chi square test or Fisher's exact test for data when the expected in a subgroup was <5 samples. Event-free survival was defined as the interval from diagnosis until there was radiographic evidence of metastatic disease; overall survival was defined as the interval from diagnosis until death (all causes) or termination of the study. Kaplan–Meier analysis was done using an online tool from the Walter and Elisa Hall Institute Department of Bioinformatics. Statistical significance was calculated using the log rank test and a P < 0.05 was considered significant.

SUPPLEMENTARY MATERIAL Figure S1. Fibrosis and necrosis in tumors from dogs treated with Ad-FasL neoadjuvant therapy. Figure S2. Fas mRNA expression in cultured osteosarcoma cells established from study dog tumors. Figure S3. Imaging of tumors and administration of Ad-FasL. Table S1. Demographics of 56 participating dogs and comparison cohort with tumor characteristics. Table S2. Dose schedule for intratumoral Ad-FasL¹. Table S3. Treatment sequence for dogs enrolled in Ad-FasL study. Table S4. Demographic characteristics of Fas analysis subset. Table S5. Correlation between expression of Fas mRNA and immunohistochemical detection of Fas protein. Table S6. Primers for RT-qPCR. Materials and Methods.

Supplementary Material

Figure S1.

Fibrosis and necrosis in tumors from dogs treated with Ad-FasL neoadjuvant therapy.

Figure S2.

Fas mRNA expression in cultured osteosarcoma cells established from study dog tumors.

Figure S3.

Imaging of tumors and administration of Ad-FasL.

Table S1.

Demographics of 56 participating dogs and comparison cohort with tumor characteristics.

Table S2.

Dose schedule for intratumoral Ad-FasL¹.

Table S3.

Treatment sequence for dogs enrolled in Ad-FasL study.

Table S4.

Demographic characteristics of Fas analysis subset.

Table S5.

Correlation between expression of Fas mRNA and immunohistochemical detection of Fas protein.

Table S6.

Primers for RT-qPCR.