King cancer: The top 10 therapeutic areas in biopharma R&D

July 23, 2017

It’s not going to come as a surprise to anyone who’s been paying attention to drug R&D trends that cancer is the number 1 disease in terms of new drug development projects. But it is amazing to see exactly how much oncology dominates the industry as never before.

At a time the first CAR-T looks to be on the threshold of a pioneering approval and the first wave of PD-(L)1 drugs are spurring hundreds of combination studies, cancer accounted for 8,651 of the total number of pipeline projects counted by the Analysis Group, crunching the numbers in a new report commissioned by PhRMA. That’s more than a third of the 24,389 preclinical through Phase III programs tracked by EvaluatePharma, which provided the database for this review.

That’s also more than the next 5 disease fields combined, starting with number 2, neurology — a field that includes Parkinson’s and Alzheimer’s. Psychiatry, once a major focus for pharma R&D, didn’t even make the top 10, with 468 projects.

Moving downstream, cancer studies are overwhelmingly in the lead. Singling out Phase I projects, cancer accounted for 1,757 out of a total of 3,723 initiatives, close to half. In Phase II it’s the focus of 1,920 of 4,424 projects. Only in late-stage studies does cancer start to lose its overwhelming dominance, falling to 329 of 1,257 projects.

PhRMA commissioned this report to underscore just how much the industry is committed to R&D and significant new drug development, a subject that routinely comes into question as analysts evaluate how much money is devoted to developing new drugs instead of, say, marketing or share buybacks.

The report makes a few other points to underscore the nature of the work these days.

— Three out of four projects in the clinic were angling for first-in-class status, spotlighting the emphasis on advancing new medicines that can make a difference for patients. Me-too drugs are completely out of fashion, unlikely to command much weight with payers.

— Of all the projects in clinical development, 822 were for orphan drugs looking to serve a market of 200,000 or less. Orphan drugs have performed well, able to command high prices and benefiting from incentives under federal law.

— There were 731 cell and gene therapy projects in the clinic, with biopharma looking at pioneering approvals in CAR-T, with Novartis and Kite, as well as the first US OK for a gene therapy, with the first application accepted this week for a priority review of a new therapy from Spark Therapeutics.

Distribution of products and projects by therapeutic area and phase

Source: Analysis Group, using EvaluatePharma data

Unique NMEs in development by stage (August 2016)

New AI-Based Search Engines are a “Game Changer” for Science Research

November 14, 2016

ee203bd1-b7e0-4864-a75641c2719b53a8By Nicola Jones, Nature magazine

A free AI-based scholarly search engine that aims to outdo Google Scholar is expanding its corpus of papers to cover some 10 million research articles in computer science and neuroscience, its creators announced on 11 November. Since its launch last year, it has been joined by several other AI-based academic search engines, most notably a relaunched effort from computing giant Microsoft.

Semantic Scholar, from the non-profit Allen Institute for Artificial Intelligence (AI2) in Seattle, Washington, unveiled its new format at the Society for Neuroscience annual meeting in San Diego. Some scientists who were given an early view of the site are impressed. “This is a game changer,” says Andrew Huberman, a neurobiologist at Stanford University, California. “It leads you through what is otherwise a pretty dense jungle of information.”

The search engine first launched in November 2015, promising to sort and rank academic papers using a more sophisticated understanding of their content and context. The popular Google Scholar has access to about 200 million documents and can scan articles that are behind paywalls, but it searches merely by keywords. By contrast, Semantic Scholar can, for example, assess which citations to a paper are most meaningful, and rank papers by how quickly citations are rising—a measure of how ‘hot’ they are.

When first launched, Semantic Scholar was restricted to 3 million papers in the field of computer science. Thanks in part to a collaboration with AI2’s sister organization, the Allen Institute for Brain Science, the site has now added millions more papers and new filters catering specifically for neurology and medicine; these filters enable searches based, for example, on which part of the brain part of the brain or cell type a paper investigates, which model organisms were studied and what methodologies were used. Next year, AI2 aims to index all of PubMed and expand to all the medical sciences, says chief executive Oren Etzioni.

“The one I still use the most is Google Scholar,” says Jose Manuel Gómez-Pérez, who works on semantic searching for the software company Expert System in Madrid. “But there is a lot of potential here.”

Microsoft’s revival

Semantic Scholar is not the only AI-based search engine around, however. Computing giant Microsoft quietly released its own AI scholarly search tool, Microsoft Academic, to the public this May, replacing its predecessor, Microsoft Academic Search, which the company stopped adding to in 2012.

Microsoft’s academic search algorithms and data are available for researchers through an application programming interface (API) and the Open Academic Society, a partnership between Microsoft Research, AI2 and others. “The more people working on this the better,” says Kuansan Wang, who is in charge of Microsoft’s effort. He says that Semantic Scholar is going deeper into natural-language processing—that is, understanding the meaning of full sentences in papers and queries—but that Microsoft’s tool, which is powered by the semantic search capabilities of the firm’s web-search engine Bing, covers more ground, with 160 million publications.

Like Semantic Scholar, Microsoft Academic provides useful (if less extensive) filters, including by author, journal or field of study. And it compiles a leaderboard of most-influential scientists in each subdiscipline. These are the people with the most ‘important’ publications in the field, judged by a recursive algorithm (freely available) that judges papers as important if they are cited by other important papers. The top neuroscientist for the past six months, according to Microsoft Academic, is Clifford Jack of the Mayo Clinic, in Rochester, Minnesota.

Other scholars say that they are impressed by Microsoft’s effort. The search engine is getting close to combining the advantages of Google Scholar’s massive scope with the more-structured results of subscription bibliometric databases such as Scopus and the Web of Science, says Anne-Wil Harzing, who studies science metrics at Middlesex University, UK, and has analysed the new product. “The Microsoft Academic phoenix is undeniably growing wings,” she says. Microsoft Research says it is working on a personalizable version—where users can sign in so that Microsoft can bring applicable new papers to their attention or notify them of citations to their own work—by early next year.

Other companies and academic institutions are also developing AI-driven software to delve more deeply into content found online. The Max Planck Institute for Informatics, based in Saarbrücken, Germany, for example, is developing an engine called DeepLife specifically for the health and life sciences. “These are research prototypes rather than sustainable long-term efforts,” says Etzioni.

In the long term, AI2 aims to create a system that will answer science questions, propose new experimental designs or throw up useful hypotheses. “In 20 years’ time, AI will be able to read—and more importantly, understand—scientific text,” Etzioni says.

This article is reproduced with permission and was first published on November 11, 2016.