Development of a gene transfer-based treatment for radiation-induced salivary hypofunction

Abstract

A significant long-term side effect of radiation therapy for head and neck cancers is xerostomia, a dry mouth, due to salivary gland damage. Despite continuing efforts to eliminate this problem, many patients continue to suffer. This brief review describes our efforts to develop a gene transfer approach, employing the aquaporin-1 cDNA, to treat patients with existing radiation-induced salivary hypofunction. A Phase I/II clinical trial, using a recombinant adenoviral vector to mediate gene transfer, is currently underway.

Introduction

Each year in the United States there are 35,000–40,000 new cases of oral cavity and pharyngeal (oral) cancers¹. The five-year relative survival rate for these cancers has changed modestly, but significantly, over the last three decades, i.e., from 53% in 1975–77 to 60% for 1996–2004, the most recent period examined¹. The treatment of most patients with oral cancers includes radiation therapy (RT). It has been long recognized that salivary glands in the radiation field can suffer irreversible damage leading to a marked reduction in salivary flow and, as a consequence, xerostomia, dysphagia and oral infections ^2,3. This RT-induced salivary hypofunction results in a significant diminution in quality of life for a large number of surviving oral cancer patients^4,5. As a result, there has been a substantial effort to minimize or eliminate this major side effect of RT for oral cancers. Among the strategies that have been employed are the following: use of hyperbaric oxygen⁶, chemoprevention with Amifostine⁷, surgical gland transfer⁸, and intensity-modulated RT^9–11. Despite these efforts, there remain many patients who survive oral cancer but experience significant xerostomia, i.e., patients in RTOG categories 2 and 3¹². It is for these individuals that we have tried to develop a novel, gene transfer-based corrective treatment for RT-induced salivary hypofunction¹³.

Specifically, for the past ~15 years, our laboratory has studied the value of transferring the human aquaporin-1¹⁴ (hAQP1) cDNA to restore salivary flow in patients with RT-damaged salivary glands¹³ (Table 1). hAQP1 encodes a water channel, a plasma membrane protein that facilitates rapid transmembrane water movement in response to an osmotic gradient. While the idea for these studies was conceived in late 1991, the actual progression of research from the bench into the clinic began in 1995 (Table 1). The following brief description (see also Figure 1) outlines the premise behind our studies^13,15. RT damages most of the fluid secreting acinar cells in salivary glands, leaving patients with relatively water impermeable duct cells as the principle surviving epithelial cells in the glands. Normally, salivary ducts are an absorptive epithelia, i.e., they re-absorb almost all of the NaCl secreted by acinar cells in the isotonic primary salivary fluid¹⁶. However, in the absence of substantial numbers of acinar cells and a significant volume of primary fluid, such a role is absent. We hypothesized that under such conditions duct cells could generate an osmotic gradient (lumen > interstitium) that water could follow, however, they lacked a facilitated water permeability pathway in their luminal membrane^13,15,17. Transfer of the hAQP1 cDNA into the surviving duct epithelial cells would provide a pathway for water to follow if an osmotic gradient was produced and, consequently, increased fluid secretion from the irradiated gland would result^13,15,17.

Table 1.

Key steps in taking AdhAQP1 from the bench to the clinic

Initial construction of the vector (1995–1996)15 Extensive functional testing in vitro (1995–1997)38 Cross-sectional testing of efficacy in irradiated rats (1995–1997)15 Longitudinal testing of safety and efficacy in irradiated rhesus macaques (1998–1999)21 Longitudinal testing of safety and efficacy in irradiated miniature pigs (2004–2006)23 GLP* level biodistribution and toxicology study in rats (2005–2006)24 Development of the clinical protocol (2005–2006) Approval of protocol (2006–2007) Production of clinical grade (GMP*) vector (2006–2007) Establishment of infrastructure and start of patient enrollment (2007–2008)

^*GLP and GMP are abbreviations, respectively, for the United States Food and Drug Administration's Good Laboratory Practice and Good Manufacturing Practice standards.

Figure 1.

Schematic depiction of the general mechanism hypothesized to occur following AdhAQP1-mediated gene transfer to irradiation-induced hypofunctional parotid glands. Shown to the left is a parotid gland that has been damaged as a result of radiation therapy. The acinar cells shown in gray are either damaged or dysfunctional (acinar cells could also be lost), while some surviving and functional acinar cells remain (in white with black nuclei). Acinar cells are normally the only water-permeable epithelial cells in these glands and are considered a secretory epithelium¹⁶. Duct cells (in white with black nuclei), either not or minimally damaged by radiation therapy, are normally water-impermeable and are an absorptive epithelium. The overall result is little to no saliva is secreted from this gland. Following administration of AdhAQP1 (large arrow) the acinar cells are unchanged, but the duct cells now are water permeable, i.e., capable of secreting fluid in an exocrine direction, and thus more “saliva” is secreted into the mouth. The presence of the aquaporin-1 transgenic protein is indicated as small circles in the luminal and basal plasma membranes of the duct cells. The secreted “saliva” is placed in quotation marks because it would not be of exactly the same composition as the normal saliva secreted by parotid glands. While it would contain exocrine proteins and various electrolytes, its specific composition would be different from that produced by a parotid gland before radiation therapy. This hypothesized mechanistic scheme is based on experimental results obtained in pre-clinical studies with miniature pigs²³. See that study, the study with irradiated rats by Delporte et al ¹⁵ and the text herein, for additional details.

In vivo animal model studies

We originally tested this hypothesis in irradiated rats, using a recombinant serotype 5, adenoviral vector to transfer the hAQP1 cDNA¹⁵ (AdhAQP1, Table 2). Rats have long been used as a model of RT-induced salivary hypofunction (e.g., see references18, 19). Our initial studies were cross-sectional in design, i.e., comparing different groups of rats subjected to a single relatively high radiation dose (data for 21 Gy shown) and a relatively high AdhAQP1 dose (5×10⁹ plaque forming units; pfu), given their size. Controls employed included sham-irradiation and treatment with a control adenoviral vector¹⁵. As shown in Table 1, AdhAQP1 was effective in restoring salivary flow to near-normal levels (~84%) in this rat model.

Table 2.

Comparison of the efficacy of AdhAQP1 in two animal models of salivary gland irradiation damage ^*

Species and gland used	Animal size (Kg)	Study design	Radiation dose (Gy)	AdhAQP1 dose (pfu)	Change>control	Percent of “normal flow”
Rat submandibular	~0.3	Cross-sectional	1×21	5×109	2.3-fold	~84
Miniature pig parotid	~30	Longitudinal	1×20	1×109	2.7-fold	~81

^*These results are based on data presented in Delporte et al¹⁵ (rat) and Shan et al²³ (miniature pig). The data for changes>control and percent of normal salivary flow are based on a summary presented in Table 1 of Baum et al³⁷. Key data from the miniature pig study²³ are shown in Figure 1. Kg-kilogram; Gy-Gray; pfu-plaque forming units. See text for additional details.

A critical step in the development of a clinically useful gene transfer treatment is the extension of results, with reasonable dose scaling, to a large animal model²⁰. We initially conducted a small longitudinal study in five-rhesus macaques²¹. The results, however, were equivocal, in part we suspect because of the small number of animals studied (one control; two animals in each of two dosage groups). We needed to utilize a large animal that was more affordable in order to increase the number of animals studied. Miniature pigs provide an excellent and affordable large animal model of salivary gland irradiation damage²² and we next tested the AdhAQP1 vector in irradiated parotid glands in this species²³ (Tables 1, 2; Figure 2). The data shown in the figure are representative of results from three animal cohorts studied longitudinally. The radiation dose used was 20Gy and the AdhAQP1 vector dose in the experiments shown in Figure 1 was 1×10⁹ pfu, i.e., both doses lower than used in the rat experiments. As can be clearly seen from Figure 1 and Table 1, the gene transfer maneuver in miniature pigs was also able to restore salivary flow to near-normal levels in the targeted parotid gland, albeit transiently.

Figure 2.

Pattern of parotid salivary flow in miniature pigs following irradiation (20 Gy) and adenoviral vector administration (1×10⁹ pfu). Parotid saliva was collected twice prior to irradiation and normalized to 100%. All other results shown are relative to that initial normalized salivary output (mean ±SEM). Animals were administered (arrow) either a control vector (AdCMVluc; encodes luciferase; filled circles) or the experimental vector (AdhAQP1; open circles) 17 weeks following irradiation. This figure is reproduced from Shan et al²³.

Given the strongly positive results of these efficacy studies, we next conducted a detailed vector biodistribution and toxicology study, under the US Food and Drug Administration's (FDA's) Good Laboratory Practice (GLP) standards, to assess the safety of AdhAQP1²⁴. This large study (100 adult rats of each gender) supported the earlier non-GLP assessments made in macaques²¹ and miniature pigs²³, i.e., that administration of AdhAQP1 to a single salivary gland was safe.

The clinical trial

Based on those safety findings, plus the pre-clinical efficacy studies conducted in rats and miniature pigs, we submitted a clinical protocol for testing this vector in patients (Tables 1,3). The protocol, “Open-label, dose-escalation study evaluating the safety of a single administration of an adenoviral vector encoding human aquaporin-1 to one parotid salivary gland in individuals with irradiation-induced parotid salivary hypofunction”, received all required US approvals (NIDCR Institutional Review Board, NIH Institutional Biosafety Committee, Recombinant DNA Advisory Committee, and the FDA), and is overseen by an independent Data Safety and Monitoring Board. The NIH protocol number for this trial is 06-D-0206 and the FDA's Investigational New Drug number for AdhAQP1 is 13102. Production of the clinical grade (FDA's Good Manufacturing Practice standards) AdhAQP1 vector occurred at the Belfer Gene Therapy Core Facility at Cornell University Medical College (New York, NY).

Table 3.

Summary of key characteristics of NIH protocol 06-D-0206

Phase I/II Single site (NIH Clinical Center) Open label Dose escalation Single dose/patient Three patients in 5 dose cohorts

This trial is currently recruiting subjects and is registered at the clinical trials.gov website (http://www.clinicaltrials.gov/ct/show/NCT00372320?order). There is also a patient oriented website (http://www.drymouthstudy.com/) that provides general information about the study and study personnel, as well as a contact link for more information. Key eligibility criteria for the study are listed in Table 4, but the detailed inclusion and exclusion criteria can be found at the clinicaltrials.gov website. The trial is designed as a single site, open label, dose escalation study (Table 3), i.e. Phase I/II. Safety is being evaluated using conventional clinical and immunological parameters. The primary outcome measures for biological efficacy are parotid gland salivary output and subjective responses to a standardized questionnaire and visual analogue scale^25,26. As a Phase I/II trial, the major purpose of this study is to determine vector safety in humans. However, as noted above in the miniature pig pre-clinical study, the effects of AdhAQP1 treatment are transient (~2–4 weeks). Such a result is expected with any first generation adenoviral vector²⁷, given the typical immune response to their administration in salivary glands²⁸. There is no intent by us to treat patients with additional doses of AdhAQP1.

Table 4.

Key enrollment criteria for NIH protocol 06-D-0206

Adults ≥18 years old Previous diagnosis of a head and neck cancer Experiencing a dry mouth due to radiation therapy Cancer-free for 5 years Not pregnant or breast feeding Non-smoker

Thus, this study also is testing an important proof of concept: does hAQP1 gene transfer increase salivary flow and improve dry mouth in humans with RT-induced parotid hypofunction? If the treatment increases salivary flow, even transiently, we have developed a serotype 2 adenoassociated viral (AAV2) vector (rAAVhAQP1)²⁹ and AAV2 vectors are capable of mediating long lived transgene expression in the salivary glands of rodents and non-human primates^30,31. In the non-human primate study³¹, AAV2-mediated reporter transgene (rhesus erythropoietin) expression was stable for ~6 months, the longest time studied. Also, biodistribution assays at animal sacrifice demonstrated that the vast majority of the AAV2 vector was found in the targeted parotid gland (>99%), with little vector detected elsewhere³¹. Furthermore, we found no consistent adverse changes in serum chemistry or hematology parameters in the AAV2-vector treated animals. Thus, using an AAV2 vector to mediate extended hAQP1 expression should be beneficial for the long-term correction of salivary gland radiation damage. Accordingly, we have begun extensive testing of the rAAVhAQP1 vector in radiation damaged miniature pig parotid glands. Patients enrolled in the current AdhAQP1 clinical trial would be eligible to enroll in future studies with rAAVhAQP1.

The current study takes one year to complete and requires 10–12 visits to the NIH Clinical Center in Bethesda, MD. There is 1 required inpatient visit that lasts 3–4 days, and the remaining outpatient visits typically last 2 days. There is no expense to a patient for their participation. The first two visits involve multiple clinical and imaging evaluations and are designed to ensure an individual's eligibility. The AdhAQP1 vector is administered on the third visit, and all remaining visits are designed to learn more about the safety and efficacy of the hAQP1 gene transfer. After establishing all necessary infrastructures for the study, we began treating patients with AdhAQP1 in 2008 and have completed the first dose cohort (4.8×10⁷ vector genomes, vg, to a single parotid gland; see Table 5). We are currently treating subjects enrolled in the second dose cohort.

Table 5.

Approved AdhAQP1 doses for NIH protocol 06-D-0206^*

Dose Cohort	Dose in vector genomes
1	4.8 × 107
2	2.9 × 108
3	1.3 × 109
4	5.8 × 109
5	3.5 × 1010

^*Three subjects are planned for each dose cohort. The first cohort has been completed. Dose level 4 corresponds approximately to that shown to be effective in miniature pigs when normalized for infusate volume. See text for additional details.

It is important to recognize that the dose units (vg) employed in the clinical trial are different from those used in our pre-clinical studies (pfu). This is because of a change recommended by the FDA. The pfu designation indicates the number of infectious units administered, while vg defines the number of vector particles delivered (both infectious and non-infectious, i.e., including “defective” ones). The FDA reasoned that the latter is more reproducible to assess in different laboratories, as it is measured by quantitative PCR. The former is the result of a functional assay and shows more inter-laboratory variability. Thus, the dosage units used in clinical studies are vg, with an FDA-specified maximum number of vg tolerated per measurable infectious unit. Comparing the vg/pfu ratio of the laboratory grade AdhAQP1 vector used in the miniature pig studies with similar values from the clinical vector preparation, administration of 5.8 × 10⁹ vg/gland (the fourth dose cohort; Table 5) provides a similar multiplicity of infection (MOI; vector dose; defined as infectious units/mL infusate; assuming a 500 μL infusate volume) to that successfully used pre-clinically (Figure 1). It also is important to recognize that comparable (scaled) transgene expression in murine submandibular glands and miniature pig parotid glands is seen with this same effective MOI³².

The highest proposed dose of 3.5 × 10¹⁰ vg/gland for this clinical study has not been associated with any reported major adenoviral vector-adverse events when administered to other tissues and is one well tolerated by patients (e.g., see references33,34). Furthermore, the highest dose is about 1000-times less than the adenoviral vector dose that was associated with a death in a liver gene therapy study at the University of Pennsylvania³⁵. However, it is essential to recognize that a clinical trial employing a novel biological reagent such as AdhAQP1 poses several theoretical major risks for participants (Table 6). The parotid glands, however, have several distinct advantages as a gene transfer target site³⁶, including: (i) being well-encapsulated to limit undesirable vector spread, (ii) ductal access uses a limited fluid volume, not diluted on administration and allowing use of low vector doses, and (iii) not being critical for life in case of a local severe adverse event. Thus, while the theoretical risks listed in Table 6 have been recognized during the review process of this protocol, the anticipated likelihood of the occurrence of one is considered to be extremely low.

Table 6.

Theoretical major risks associated with salivary gland gene therapy

Death of a participant Severe, non-fatal idiosyncratic reaction to vector not previously observed in pre-clinical animal studies, including worsening of existing gland function Error in vector delivery during infusion resulting in duct trauma and vector spread beyond gland and transduction of major organs, e.g., liver and spleen Recombination with a wild type DNA virus within gland tissue, present as a latent infection, resulting in a replication competent, mutated virus Recurrence of head and neck cancer Unknown effects of the long-term presence in gland tissue of vector DNA, i.e., viral genome, promoter and other vector cassette sequences

Concluding remarks

This brief review has described the development of a gene transfer-based treatment for RT-induced salivary hypofunction. The ongoing clinical trial represents the first use of gene therapy in human salivary glands. Pre-clinical animal model results with the AdhAQP1 vector were quite encouraging, although relatively short-lived because a first-generation adenoviral vector was used. If the AdhAQP1 clinical trial shows positive results, an alternative strategy using an AAV2 vector, which should lead to extended transgene expression in humans, has also been developed.