Being under ever-increasing pressure to compete in a challenging economic and technological environment, pharmaceutical and biotech companies must continually innovate in their R&D programmes to stay ahead of the game.
External innovations come in different forms and originate in different places -- from university labs, to privately held venture capital-backed startups and contract research organizations (CROs). Let’s get to reviewing some of the most influential research trends which will be “hot” in 2018 and beyond, and summarize some of the key players driving innovations.
Last year BioPharmaTrend summarized several important trends affecting biopharmaceutical industry, namely: an advancement of various aspects of gene editing technologies (mainly, CRISPR/Cas9); a fascinating growth in the area of immuno-oncology (CAR-T cells); an increasing focus on microbiome research; a deepening interest in precision medicine; some important advances in antibiotics discovery; a growing excitement about artificial intelligence (AI) for drug discovery/development; a controversial but rapid growth in the area of medical cannabis; and continuous focus of pharma on engaging in R&D outsourcing models to access innovations and expertise.
Below is a continuation of this review with several more active areas of research added to the list, and some extended commentaries on the trends outlined above -- where relevant.
1. Adoption of Artificial Intelligence (AI) by pharma and biotech
With all the hype around AI nowadays, it is hard to surprise anybody with this trend in pharmaceutical research. However, it should be noted that AI-driven companies really start getting traction with big pharma and other leading life science players, with lots of research partnerships and collaborative programs -- here is a list of key deals so far, and here is a brief review of some notable activity in the “AI for drug discovery” space over the last several months.
A potential of AI-based tools is now explored at all stages of drug discovery and development -- from research data mining and assisting in target identification and validation, to helping come up with novel lead compounds and drug candidates, and predicting their properties and risks. And finally, AI-based software is now able to assist in planning chemical synthesis to obtain compounds of interest. AI is also applied to planning pre-clinical and clinical trials and analyzing biomedical and clinical data.
Beyond target-based drug discovery, AI is applied in other research areas, for instance, in phenotypic drug discovery programs -- analyzing data from high content screening methods.
With a major focus of AI-driven startups on small molecule drug discovery, there is also an interest in applying such technologies for biologics discovery and development.
2. Expanding chemical space for drug discovery explorations
A vital part of any small molecule drug discovery program is hit exploration -- identification of those starting point molecules which would embark on a journey towards successful medications (rarely they survive this journey, though) -- via numerous optimization, validation and testing stages.
The key element of hit exploration is the access to an expanded and chemically diverse space of drug like molecules to choose candidates from, especially, for probing novel target biology. Given that existing compound collections at the hands of pharma were built in part based on the small molecule designs targeting known biological targets, new biological targets require new designs and new ideas, instead of recycling excessively the same chemistry.
Following this need, academic labs and private companies create databases of chemical compounds far beyond what is available in typical pharmaceutical company compound collections. Examples include GDB-17 database of virtual molecules containing 166,4 billion molecules and FDB-17 of 10 million fragment-like molecules with up to 17 heavy atoms; ZINK -- a free database of commercially-available compounds for virtual screening, containing 750 million molecules, including 230 million in 3D formats ready for docking; and a recent development of synthetically accessible REadily AvailabLe (REAL) chemical space by Enamine -- 650 million molecules searchable via REAL Space Navigator software, and 337 million molecules searchable (by similarity) at EnamineStore.
An alternative approach to access new drug-like chemical space for hit exploration is using DNA-encoded library technology (DELT). Owing to the “split-and-pool” nature of DELT synthesis, it becomes possible to make huge numbers of compounds in a cost- and time-efficient manner (millions to billions of compounds). Here is an insightful report on the historic background, concepts, successes, limitations, and the future of DNA-encoded library technology.
3. Targeting RNA with small molecules
This is a hot trend in drug discovery space with a continuously growing excitement: academics, biotech startups and pharmaceutical companies are increasingly active about RNA targeting, although uncertainty is also high.
In the living organism, DNA stores the information for protein synthesis and RNA carries out the instructions encoded in DNA leading to protein synthesis in ribosomes. While a majority of drugs is directed at targeting proteins responsible for a disease, sometimes it is not enough to suppress pathogenic processes. It seems like a smart strategy to start earlier in the process and influence RNA before proteins were even synthesized, therefore substantially influencing the translation process of genotype to unwanted phenotype (disease manifestation).
The problem is, RNAs are notoriously terrible targets for small molecules -- they are linear, but able to clumsily twist, fold, or stick to itself, poorly lending its shape to suitable binding pockets for drugs. Besides, in contrast to proteins, they compose of just four nucleotide building blocks making them all look very similar and difficult for selective targeting by small molecules.
However, a number of recent advances suggest that it is actually possible to develop drug-like, biologically active small molecules that target RNA. Novel scientific insights prompted a golden rush for RNA -- at least a dozen companies have programs dedicated to it, including big pharma (Biogen, Merck, Novartis, and Pfizer), and biotech startups like Arrakis Therapeutics with a $38M Series A round in 2017, and Expansion Therapeutics -- $55M Series A early in 2018.
4. New antibiotics discovery
There is a growing concern about the rise of antibiotic-resistant bacteria — superbugs. They are responsible for about 700,000 deaths worldwide each year, and according to a U.K. government review this number can dramatically increase -- up to 10 million by 2050. Bacteria evolve and develop resistance to the antibiotics which were traditionally used with great success, and then become useless with time.
Irresponsible prescription of antibiotics to treat simple cases in patients and a widespread use of antibiotics in livestock farming jeopardize the situation by accelerating the rate of bacterial mutations, rendering them resistant to drugs with alarming speed.
On the other hand, antibiotics discovery has been an unattractive area for pharmaceutical research, compared to developing more ‘economically feasible’ drugs. It is probably the key reason behind a drying up of the pipeline of novel antibiotic classes, with the last one introduced more than thirty years ago.
Nowadays the antibiotics discovery is becoming a more attractive area due to some beneficial changes in regulatory legislature, stimulating pharma to pour money into antibiotics discovery programs, and venture investors -- into biotech startups developing promising antibacterial medicines. In 2016, one of us (AB) reviewed the state of antibiotics drug discovery and summarized some of the promising startups in the space, including Macrolide Pharmaceuticals, Iterum Therapeutics, Spero Therapeutics, Cidara Therapeutics, and Entasis Therapeutics.
Notably, one of the more exciting recent breakthroughs in the antibiotics space is the discovery of Teixobactin and its analogs in 2015 by a group of scientists led by Dr. Kim Lewis, Director of the Antimicrobial Discovery Center at Northeastern University. This powerful new antibiotics class is believed to be able to withstand the development of bacterial resistance against it. Last year, researchers from the University of Lincoln successfully developed a synthesized version of teixobactin, making an important step forward.
Now researchers from the Singapore Eye Research Institute have shown the synthetic version of the drug can successfully cure Staphylococcus aureus keratitis in live mouse models; before the activity of teixobactin was only demonstrated in vitro. With these new findings, teixobactin will need another 6-10 years of development to become a drug that doctors can use.
Since the discovery of teixobactin in 2015, another new family of antibiotics called malacidins were revealed in early 2018. This discovery is still in its early stages, and not nearly as developed as the latest research on teixobactin
5. Phenotypic screening
Image credit: SciLifeLab
In 2011 authors David Swinney and Jason Anthony published results of their findings about how new medicines had been discovered between 1999 and 2008 unveiling the fact that considerably more of the first-in-class small molecule drugs had actually been discovered using phenotypic screening than target-based approaches (28 approved drugs vs 17, respectively) -- and it is even more striking taking into account that it was target based approach that had been a major focus over the period stated.
This influential analysis triggered a renaissance of the phenotypic drug discovery paradigm since 2011 -- both in pharmaceutical industry and in academia. Recently, scientists at Novartis conducted a review of the current state of this trend and came to a conclusion that, while pharma research organizations encountered considerable challenges with phenotypic approach, there is a decreasing number of target-based screens and an increase of phenotypic approaches in the past 5 years. Most probably, this trend will continue far beyond 2018.
Importantly, beyond just comparing phenotypic and target based approaches, there is a clear trend toward more complex cellular assays, like going from immortal cell lines to primary cells, patient cells, co-cultures, and 3D cultures. The experimental setup is also becoming increasingly sophisticated, going far beyond univariate readouts toward observing changes in subcellular compartments, single-cell analysis and even cell imaging.
6. Organs (body)-on-a-chip
Microchips lined by living human cells could revolutionize drug development, disease modeling and personalized medicine. These microchips, called ‘organs-on-chips’, offer a potential alternative to traditional animal testing. Ultimately, connecting the systems altogether is a way to have the whole “body-on-a-chip” system ideal for drug discovery and drug candidate testing and validation.
This trend is now a big deal in drug discovery and development space and we have already covered the current status and context of the “organ-on-a-chip” paradigm in a recent mini-review.
While a lot of skepticism existed some 6-7 years ago, when perspectives on the field were articulated by enthusiastic adopters. Today, however, the critics appear to be in full retreat. Not only have regulatory and funding agencies embraced the concept, but it is now increasingly adopted as a drug research platform by both pharma and academia. Over two dozen organ systems are represented in on-chip systems. Read more about it here.
The area of bioprinting human tissues and organs is rapidly developing and it is, undoubtedly, the future of medicine. Founded in early 2016, Cellink is one of the first companies in the world to offer 3D printable bioink – a liquid that enables life and growth of human cells. Now the company bioprints parts of the body -- noses and ears, mainly for testing drugs and cosmetics. It also prints cubes enabling researchers to “play” with cells from human organs such as livers.
Cellink recently partnered with CTI Biotech, a French medtech company specializing in producing cancer tissues, in order to substantially advance the area of cancer research and drug discovery.
The young biotech startup will essentially help CTI to 3D print replicas of cancer tumors, by mixing the Cellink’s bioink with a sample of the patient's cancer cells. This will help researchers in identifying novel treatments against specific cancer types.
Another biotech startup developing 3D printing technology for printing biological materials -- an Oxford University spinout company, OxSyBio, which just secured £10m in Series A financing.
While 3D bioprinting is an extremely useful technology, it is static and inanimate because it considers only the initial state of the printed object. A more advanced approach is to incorporate “time” as the fourth dimension in the printed bio-objects (so called “4D bioprinting”), rendering them capable of changing their shapes or functionalities with time when an external stimulus is imposed. Here is an insightful review on 4D bioprinting.
Even without a deep dive into each of the top trends just described, it should become apparent that AI will be taking an ever increasing part of the action. All these new areas of biopharma innovation have become big data centric. This circumstance in itself presages a pre-eminent role for AI, noting also, as a postscript to this coverage of the topic, that AI comprises multiple, analytical and numerical tools undergoing continuously evolution. The applications of AI in drug discovery and early stage development are for the most part targeted at uncovering hidden patterns and inferences connecting causes and effects otherwise not identifiable or comprehensible.
Thus, the subset of AI tools that are employed in pharmaceutical research fall more appropriately under the moniker of “machine intelligence” or “machine learning”. These can be both supervised by human guidance, as in classifiers and statistical learning methods, or unsupervised in their inner workings as in the implementation of various types of artificial neural networks. Language and semantic processing and probabilistic methods for uncertain (or fuzzy) reasoning also play a useful role.
Understanding how these different functions can be integrated into the broad discipline of “AI” is a daunting task that all interested parties should undertake. One of the best places to look for explanations and clarifications is the Data Science Central portal and especially the blog posts by Vincent Granville, who regularly elucidates the differences between AI, machine leaning, deep learning, and statistics. Becoming conversant on the ins and outs of AI as a whole is an indispensable component of keeping abreast or ahead of any biopharma trends.
Please, share if you liked the post, and leave your comments below, especially, if something important was omitted from the list and you belive it should belong there.