Research and Development of Photodynamic Therapy for AMD

While PDT had a long history as a potential treatment for cancer, there were no approved drugs in the 1990s. By combining a light-activated photosensitizer delivered to target tissue with activation from a light source at a specific wavelength, PDT has a direct toxicity to tumor cells. Since PDT also causes vascular injury, a number of investigators pursued various photosensi-tizers and PDT in animal models of subfoveal choroi-dal neovascularization.

During my first year as a faculty member at Mass Eye and Ear, PDT using verteporfin, a liposomal formulation of benzoporphyrin derivative, was in clinical trial for skin metastases. If verteporfin PDT could safely treat experimental CNV, there would be a shorter, potential path for clinical application in nvAMD. To investigate this, I began conducting experiments using a non-human primate laser injury model in the macula, recapitulating many of the features of human CNV. For PDT, we needed a laser that could provide visualization and deliver laser light to the macula at the back of the eye with the right power and spot size. The optical bench laser used at Massachusetts General Hospital to treat experimental ocular tumors produced too large a spot size and had no ability to focus on the macula. We required a laser with slit lamp delivery typical of ophthalmic lasers, but able to produce a larger spot size and lower power. The group at Coherent, including David Dewey and George Marcelino, worked with me to build the first device, and they built early and improved prototype lasers with the necessary specifications that were used in our early studies. Once we moved to clinical trials, Coherent was joined by Zeiss to develop lasers for clinical use. QLT Phototherapeutics provided the photosensitizer, benzoporphyrin derivative, although in a form that had to be dissolved in dimethyl sulfox-ide rather than the liposomal form verteporfin, being used in dermatologic clinical trials.

The first experiments were performed using an irradiance of 150 mW/cmReference Miller ² — the same irradiance that was used in the skin cancer clinical trials and in experimental choroidal melanoma. With this irradiance, treatment times were very long at 16 minutes, and it was difficult to focus using the low luminance setting of the slit lamp and the dim aiming beam. While these parameters worked experimentally, such a long treatment time for PDT with the difficulties in maintaining focus and localization, was not going to be practical for clinicians treating patients. Knowing that the choroid is a heat sink that could ameliorate any increased temperatures, we decided to try higher irradiance — all the way up to 1800 mW/cmReference Miller ² . The experiments were a success, with no complications, and we were still able to achieve CNV closure as determined by fluorescein angiography and histopathology.Reference Miller ² Importantly, we found that when treating normal retina, we could use the same parameters and avoid significant damage to the outer retina. From these experiments, we selected 600 mW/cmReference Miller ² for the clinical application, and this was the irradiance that was used in the nvAMD and pathological myopia clinical trials, and approved for clinical use. ³ The clinical trials proceeded from 1995 to 2000 and ultimately led to FDA approval for verteporfin, the first pharmacologic treatment for nvAMD. ⁴

Research and Development of Anti-VEGF Therapy

During the same time period that we were developing PDT, we were also investigating the molecular drivers of neovascularization in CNV and ischemic retinopathies. The search for “Factor X” began with Michael-son’s observations in the 1940s that there must be a secreted factor released from ischemic retina to drive neovascularization of the retina, optic nerve, and iris.Reference Michaelson ⁵ In the 1970s, Judah Folkman’s laboratory identified tumor angiogenesis factor,Reference Gimbrone ⁶ and Dr. Folkman published his seminal theory of tumor angiogenesis: that angiogenesis, or the recruitment and growth of new blood vessels, was required for tumor growth.Reference Folkman ⁷ While Dr. Folkman’s theory was initially met with skepticism, he is now recognized for pioneering the concept of tumor angiogenesis, successfully elucidating tumor biology, and developing new cancer treatments.Reference Klagsbrun and Moses ⁸

Napoleone Ferrara at Genentech was also initially seeking new treatments for cancer and cloned, sequenced, and characterized vascular endothelial growth factor (VEGF), recognizing its role in angiogenesis.Reference Leung ⁹ VEGF turned out to be the same molecule as vascular permeability factor, a very potent soluble permeability factor — 50,000 times more potent than histamineReference Keck ¹⁰ — that was previously identified and described by Harold Dvorak at Harvard Medical School.Reference Senger ¹¹ As VEGF was a secreted endothelial cell mitogen and an angiogenesis factor that was regulated by hypoxia, a condition known to stimulate neovascu-larization in certain retinopathies, it met the characteristics of the elusive Factor X. Our two Boston groups — composed of ophthalmology clinician scientists and vision scientists — set out to explore the role of VEGF in ocular disease.

Anthony Adamis, Patricia D’Amore, David Shima, Evan Gragoudas, and our team demonstrated that VEGF protein levels correlated with neovascularization in vivo.Reference Miller ¹² Lloyd Paul Aiello, George King, and their team showed that elevated VEGF protein levels also correlated with active retinal neovascularization in patients.Reference Aiello ¹³ Next, we demonstrated that VEGF inhibition with an anti-VEGF antibody — the precursor of bevacizumab (Avastin) — provided by Napoleon Fer-rara at Genentech, prevented neovascularization in non-human primates.Reference Adamis ¹⁴ Finally, we were able to show that injection of VEGF into normal monkey eyes led to iris neovascularization and neovascular glaucoma, recapitulating the pathology seen clinically with retinal ischemia.Reference Tolentino ¹⁵ VEGF was both sufficient and necessary for intraocular neovascularization — we had solved for Factor X.

Yet, despite the success of our VEGF studies, we struggled to obtain funding for our work. Genentech funded the VEGF discovery work of Napoleon Ferrara, while the National Institutes of Health (NIH) funded grants to support some of the initial studies, such as those performed in laboratories of Harold Dvorak, Patricia D’Amore, and George King. Additionally, while some of the more basic science studies in Drs. D’Amore’s and King’s laboratories were funded by NIH grants, the majority of our translational work was funded by small foundational grants and philanthropic funds. We applied multiple times for NIH funding for our nonhuman primate studies but were unsuccessful, primarily because of the use of nonhuman primates, even though they were the best available model and very predictive of human drug and light dose parameters for both efficacy and safety. Additionally, given the translational nature of these studies, it was generally felt that they should be industry supported.

Although the two groups had collectively demonstrated the key role of VEGF in ocular disease, and Genentech clearly had the best potential therapeutic, progress to clinical application foundered. Genentech remained primarily focused on developing a treatment for cancer. Genentech may also have been negatively influenced by their part owner Roche (Basel, Switzerland), who had supported unsuccessful clinical trials in nvAMD using α-interferon.Reference Engler ¹⁶ Genentech considered licensing the intellectual property to another sponsor or trying to develop an ophthalmology program internally.

Two former Genentech employees, Marty Glick and John McLaughlin, saw the potential for anti-VEGF development in ophthalmology and left Genentech to pursue this path. Joined by three ophthalmologists — David Guyer, Samir Patel, and Anthony Adamis — they formed Eyetech Pharmaceuticals, licensing anti-VEGF aptamer technology from Gilead Sciences. Eyetech developed pegaptanib (Macugen) for AMD, and entered into clinical trials, progressing to Phase III trials with data demonstrating efficacy in slowing vision loss in all forms of nvAMD.Reference Gragoudas ¹⁷ This was groundbreaking as the first demonstration of efficacy of an antiangiogenic therapy in ophthalmology. The vision outcomes with pegaptanib were similar to those achieved with verteporfin,Reference Miller ¹⁸ but achieved across a broader range of nvAMD subtypes.

The development and early promise of pegaptanib spurred interest at Genentech to enter the development path for their anti-VEGF agent ranibizumab (Lucentis), a recombinant, humanized monoclonal antibody fragment designed for intraocular use. Phase III results using ranibizumab for nvAMD were dramatic. Results from the 1-year MARINA trialReference Rosenfeld ¹⁹ presented publicly in 2005, showed for the first time, that sustained vision improvement was possible in patients given intravitreal injections with anti-VEGF therapy (0.3 mg or 0.5 mg ranibizumab). Similar results were obtained at 2 years and after 1 year of treatment in the ANCHOR study,Reference Brown ²⁰ comparing ranibizumab to verteporfin.

In addition to ranibizumab, Genentech had also developed bevicizumab, a full-length monoclonal antibody administered intravenously to treat cancer. In 2004, the US Food and Drug Administration (FDA) approved bevicizumab for the systemic treatment of metastatic colon cancer. Ranibizumab was created from the same parent mouse antibody as bev-acizumab and designed specifically for injection in the eye to block blood vessel growth in AMD. Since ranibizumab was going to be administered locally in the eye via intravitreal injection, we recommended that it be smaller in size than a full-length antibody, as we were concerned that a full-length antibody would not be able to cross the blood-retinal barrier. The same year that the ranibizumab results were presented (in fact, at the same meeting), Philip Rosenfeld presented a single case of a patient with nvAMD treated with intravitreal bevacizumab showing improvement in optical coherence tomography findings.Reference Rosenfeld, Moshfeghi and Puliafito ²¹ This was dramatic and unexpected, suggesting that the full-length antibody could penetrate the retina and be effective. This finding, combined with the extremely positive data from the large Phase III trials presented at the same meeting, provided a strong basis for anti-VEGF antibodies for AMD, and led to widespread off-label use of bevacizumab for nvAMD both before and after ranibizumab was approved by regulatory agencies. Genentech was placed in an awkward position, as ophthalmologists were keen to try bevacizumab, but Genentech was prohibited from promoting an off-label product. Moreover, Genentech had concerns about systemic safety of the full-length antibody based on their experience with systemic delivery — and, of course, Genentech had invested heavily in the development of ranibizumab for AMD.

With support from the National Eye Institute, Daniel Martin organized and led a multicenter trial called the Comparison of AMD Treatments Trials (CATT) to compare ranibizumab and bevacizuamb for the treatment of nvAMD. The CATT 1-year data showed that ranibizumab and bevacizumab had nearly identical effects on visual acuity, and that as-needed dosing did not lessen these effects. ²² The CATT 2-year results confirmed that ranibizumab and bevacizumab had similar vision outcomes over the 2-year period, although most of the vision change occurred in year 1. ²³ Importantly, CATT (both 1-year and 2-year data) also showed similar rates in cardiac serious adverse events between the 2 drugs, which had been more of a concern for bevacizumab. ²⁴ Similar findings comparing the safety and efficacy of ranibizumab and bevacizumab intravitreal injections to treat nvAMD were reported in the IVAN 1-year and 2-year studies. ²⁵

We now have four FDA-approved anti-VEGF agents — brolucizumab (Beovu), aflibercept (Eylea), ranibizumab (Lucentis), and pegaptanib (Macugen), with additional agents in development. Bevacizumab is approved by for the treatment of colorectal and other cancers, and continues to be used off-label as a therapy for nvAMD and other conditions. These anti-VEGF therapies have revolutionized care for patients with AMD and other neovascular ocular conditions, and are administered millions of people worldwide.

Contributions of Industry, Academia, Nonprofits, and Government in Translational R&D for AMD

Partnership between the pharmaceutical industry and academia is essential for the development of approval-level clinical trials and deployment of new therapies to patients. Expertise on both ends — research and patient care — is important for proper study design and a critical factor in patient selection, dosage, and route of administration in early phase clinical trials. In the case of anti-VEGF therapy, the method of drug delivery was initially uncertain. Intravenous administration was considered, but experience with the Roche clinical trials of α-interferon for macular degeneration made our Boston group skeptical about systemic treatments for patients with AMD. The a-interferon doses studied in the clinical trials for AMD were similar to those used with acceptable side effects for HIV-infected patients. However, AMD patients are a much older population, and systemic side effects were a major problem in the interferon clinical trials for AMD. Local administration, or an intravitreal injection directly into the eye, seemed to be a better approach than systemic drug delivery. While there was considerable skepticism among retina specialists about this route of administration, specifically regarding obtaining an adequate dose into the eye and the practicality of repeated intravitreal injections, clinical trials were designed using monthly intravitreal injections. As described above, these trials were dramatically successful — accepted by clinicians, and importantly, by patients fearful of blindness without treatment.

In addition to the partnership between industry and academia, another important aspect in research and development is the role of clinician scientists, as they are able to bridge the gap between research and patient care. In our PDT work, as a clinician scientist, I was able to understand the practical requirements for an effective treatment. Long treatment times at a low irradiance (150 mW/cmReference Miller ² ) worked at an experimental level, but not as a practical treatment for patients. Recognizing this, we were able to modify our approach and use a higher irradiance (600 mW/cmReference Miller ² ) to lessen treatment times and still obtain CNV closure without damage to the outer retina. Ultimately, this higher irradiance was used in AMD clinical trials and approved for clinical use. ²⁶

It is also essential to recognize and value the contributions of different individuals and groups early in the process, especially in regards to intellectual property for development of technologies or therapies. Communicating with collaborators and making arrangements regarding invention disclosures, licensing, authorship, and other important details will help reduce the number of disagreements among interested parties. In the case of verteporfin, a lack of reasonable agreements at the outset led to more than a decade-long legal battle between Mass Eye and Ear and QLT/Novartis. Litigation and appeals over PDT ultimately resulted in a $126 million judgment for Mass Eye and Ear, and an ongoing royalty of about 3% of all sales worldwide as long as the drug is used. ²⁷

While the academic setting is supportive for transla-tional research, obtaining funding for these studies is difficult. Funding from federal sources, such as NIH, is often directed at earlier stage basic science studies or later stage clinical trials, leaving translational work underfunded. Our group was never able to obtain NIH funding for our PDT work and only obtained federal funding for our VEGF work at the basic science level. Moreover, even when researchers and clinician scientists have potential therapeutic treatments and approaches with good pre-clinical data, the pharmaceutical industry is hesitant to invest, asking for additional proof-of-concept studies in patient populations. Unfortunately, this space is difficult for academic researchers and clinician scientists to navigate, and those who do tend to seek out support from start-up companies and individual investors. The end result of this approach is that a number of promising therapeutic approaches end up “dying on the vine” or are studied in an incomplete and haphazard manner, never reaching clinical trial. Thus, there are likely many opportunities and potential treatments on which we are missing out.

However, once established positive proof-of-concept studies in patients have been obtained for a potential therapy, the pharmaceutical industry is necessary for conducing large clinical trials, and if successful, product distribution. Successful Phase I and II studies often partner the pharmaceutical industry and academia, as clinician scientists are a valuable resource for helping to guide study design as mentioned above. However, on reaching Phase III trials, the pharmaceutical companies now routinely conduct these studies in community-based practices as they benefit from the more streamlined processes offered there. Exceptions for this would be in the case of gene-based therapies for inherited retinal disorders, as these are more complex, require a unique patient population, and specialized instrumentation that tends to be housed within academic medical centers.

As for the government, its role should be to provide support for research at the basic and translational level. Federal funding is a somewhat limited resource — directing it towards later stage research and development limits the number of basic and translational projects that can be funded. As these studies are vital in feeding the research pipeline, it makes sense to focus federal resources here and then establish partnerships with the pharmaceutical industry to drive large-scale clinical trials and product distribution. In 2008, the UK government boosted translational research when the National Institute for Health Research (NIHR) announced 12 new hospital-based biomedical research units, funding them with up to £1 million per year.Reference Adams ²⁸ To further transform health research, the UK also aligned two previously separate funding entities — the Medical Research Council (MRC) and the NIHR — to coordinate basic discovery and clinical evaluation. As part of this initiative, the MRC is working to expand partnership opportunities with industry, while the NIHR is setting up large networks for clinical trials. ²⁹

Despite supporting initial research and development of potential therapeutics, the government should not try to keep an “equity” position in these inventions or therapies. Their role should be to support research than can end debilitating and/or lethal conditions. Instead, the U.S. government, could work at the other end of the pipeline, and negotiate drug prices as other countries do. This may help avoid a problematic debate on which disease areas are the most important and deserve funding and research effort. Additionally, adoption of a “one size fits all” approach is unnecessary, as there will be some therapeutic areas, such as antibiotics, where government mission-oriented investing may make sense.Reference Mazzucato ³⁰