This article has been updated from a previous version to clarify comments by
Ambati regarding Opko's Molecular Vision study.
Ambati regarding Opko's Molecular Vision study.
The senior author of a recent paper describing how siRNAs suppress neovascularization regardless of their sequences or targets fired back at a critic last week, defending his finding that Opko Health’s siRNA-based treatment for wet age-related macular degeneration works as a result of a cellular immune response rather than an RNAi effect.
The researcher, the University of Kentucky’s Jayakrishna Ambati, also raised questions about the data Opko has published regarding the ability of siRNAs to inhibit choroidal neovascularization, which includes a paper published in Retina in 2004 but later retracted by its authors.
As reported by RNAi News, Ambati and colleagues published in last week’s Nature data showing that any dsRNA at least 21 nucleotides in length activated toll-like receptor 3, which in turn suppressed angiogenesis, in the retinas of a mouse in which laser burns were used to trigger CNV (see RNAi News, 3/27/2008).
Among the siRNAs examined were Opko’s AMD drug candidate bevasiranib, a naked siRNA targeting vascular endothelial growth factor that recently entered phase III testing (see RNAi News, 6/27/2007). Opko gained the rights to bevasiranib as part of its acquisition of Acuity Pharmaceuticals (see RNAi News, 3/29/2007).
In response to the paper, Sam Reich, Opko’s executive vice president of ophthalmologics, told RNAi News that Ambati’s findings were inaccurate and possibly the result of the use of a “technically difficult” mouse model known to yield “variable outcomes.”
“Everyone who works with this model knows that you must take the results with great caution,” Reich said last week. “It is a first model that is tested and followed up by many other models to look for confirmation … because it is very common for different labs to get different findings … using the laser-induced mouse model of CNV.”
Reich said that Ambati did not provide such confirmation, while Opko’s data showing the RNAi-mediated silencing of VEGF, meantime, has been validated by the company in multiple models, as well as by independent labs.
Ambati, in an interview with RNAi News this week, disagreed. “We tested … several dozen different siRNAs to reach the conclusion that angiogenesis suppression is sequence-independent. Also, the eye experiments were reproduced both in my lab” and by collaborators in Japan.
Additionally, Ambati noted that experiments demonstrating the generic anti-angiogenic property of siRNAs in the skin of a mouse model were conducted by colleagues at the University of Illinois, Chicago.
“I would submit that the replication of our observation in multiple models in multiple laboratories establishes the credibility of our study. Moreover, our work survived … Nature's rigorous peer-review process,” Ambati told RNAi News in an e-mail.
He also pointed to data published in the Nature paper showing that both bevasiranib and AGN211745, Allergan’s siRNA-based AMD drug candidate targeting VEGF receptor-1, did not suppress neovascularization in mice lacking TLR3 or downstream molecules.
“These findings definitively establish that both these drugs function solely via TLR3 activation because genetic knockouts are the gold standard for establishing molecular mechanisms in animal models,” Ambati wrote.
“We tested … several dozen different siRNAs to reach the conclusion that angiogenesis suppression is sequence-independent. Also, the eye experiments were reproduced both in my lab” and by collaborators in Japan.
Ambati added that his lab has conducted additional experiments in a corneal model that confirm the Nature data. He said that these findings are currently under review for publication.
In responding to Ambati’s paper last week, Reich cited a paper he and colleagues published in Molecular Vision in 2003 as evidence that siRNAs targeting VEGF can inhibit ocular neovascularization via RNAi. But Ambati pointed out that in that paper, the investigators used the same laser-injury mouse model used in the Nature paper — a fact Ambati called “ironic” since Reich had suggested the model can yield inaccurate results.
Furthermore, in the Molecular Vision paper the investigators indicate that they conjugated their siRNAs with a transfection reagent, which enabled the oligos to penetrate cells, Ambati said. “In contrast … bevasiranib is naked and not administered with any such permeating agent. Therefore, it is not surprising that with the aid of [a] transfection reagent the … siRNA enters cells and does indeed execute RNA interference.”
Still, Reich defended his statements on the mouse model, telling RNAi News this week that Opko maintains “one of the few labs that can run [the] model … and we know a lot about a lot about the variable outcomes.”
In a telephone interview, Ambati told RNAi News that he found Reich's suggestion that his work had not been independently validated jarring because he is unaware of any independent verification of the siRNAs evaluated in the Molecular Vision paper and their ability to inhibit neovascularization.
“Is there any paper out there by anybody?” he asked. “No, there is nothing.”
But Reich pointed to a 2006 Gene Therapy paper from Peter Campochiaro and colleagues at the Johns Hopkins School of Medicine showing that AGN211745, which was originally developed by Sirna Therapeutics as Sirna-027 (see RNAi News, 10/7/2005), could suppress neovascularization compared with control sequences in the laser-injury mouse model.
“It’s a less-than-perfect model,” Reich said this week. “But people [such as Campochiaro] who are very experienced running it have had results” showing the ability of siRNAs to suppress neovascularization through an RNAi effect.
Ultimately, Ambati said that Reich’s criticism of his Nature paper was “particularly egregious” in light of the retraction of a 2004 Retina paper, authored by Reich, Acuity co-founder Michael Tolentino, and researchers from the University of Pennsylvania, that claimed to demonstrate the ability of intravitreally injected siRNAs to inhibit VEGF and suppress vascular growth and permeability in a primate model of CNV.
According to the retraction, a dataset sent to a biostatistician for analysis “had an error in linking the codes for each animal with the gradings for angiographic leakage. Therefore, the results reported for the effect of treatment on leakage were not correct.”
After the angiograms were regraded, it was found that there was “no statistically significant effect of dose,” the authors wrote in their retraction.
In a separate letter published with the retraction, the Acuity authors said that while they agree with the retraction, a separate statistical analysis showed that “there is an effect in reducing leakage demonstrated by” their retracted data.
Yet a few months later, two researchers who were affiliated with centers where the angiograms were regraded and one who analyzed the data for the original Retina paper sent a letter to the editor of the journal stating that “the results of the experiment [published by Reich and colleagues] do not support the assertion of the original article that VEGF siRNA inhibits vascular permeability in a dose-dependent manner.
”Given the few animals in the experiments … we believe that there is not sufficient precision in the data to determine whether the observed differences in leakage among the … treatment groups are attributable to random variation in response to laser burns or to true effects of siRNA on leakage,” they concluded.
But this week Reich said the retraction stemmed from “communication issues” between the teams at Acuity and the University of Pennsylvania regarding blood vessel leakage. “It was really only one figure [in the paper] that required correction,” he said, noting that the CNV and safety results “were unaffected.”
Because the paper was authored by multiple investigators with different affiliations, “the easiest thing to do was retract the paper,” he said. In the end, “we actually confirmed [the retracted data] in our phase II trial” of bevasiranib, he said.