Medicinal Chemistry Training Is in Big Trouble - ACS Publications

Subscriber access provided by Heriot-Watt | University Library

Perspective

No Denying It: Medicinal Chemistry Training is in Big Trouble Michael F. Rafferty J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00741 • Publication Date (Web): 26 Sep 2016 Downloaded from http://pubs.acs.org on September 27, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Medicinal Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Medicinal Chemistry

No Denying It: Medicinal Chemistry Training is in Big Trouble

Michael F. Rafferty*

*Correspondence to: Department of Medicinal Chemistry 4070 Malott Hall University of Kansas, Lawrence, KS 66045

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 24

Abstract There has been little consensus between the pharmaceutical industry and academic communities concerning the best approach to train medicinal chemists for drug discovery. For decades the pharmaceutical industry has shown preference for synthetic organic graduates over candidates with degrees from medicinal chemistry programs on the assumption that medicinal chemistry expertise will be acquired on the job. However, ongoing changes to pharmaceutical drug discovery organizations and practices threaten to undermine this training model. There is a compelling argument to be made for establishment of a strong industry academic partnership to train new candidates with sophisticated knowledge of contemporary drug design concepts and techniques to ensure that the future needs of both industry and academic drug discovery research can be served. Introduction The training of medicinal chemists for pharmaceutical industry research has been discussed and debated in the literature for decades.1-6 Every formal definition of medicinal chemistry acknowledges that it is much more than simply a subheading under organic or pharmaceutical chemistry, and that success in our field requires a fundamental understanding and application of a continually evolving portfolio of concepts and disciplines. In recent years, the strategic importance of medicinal chemistry in the pharmaceutical industry has been challenged by the impressive growth and success of biologics. However, biologics with all of their promise are also limited in ways that may never be overcome. The pharmaceutical industry will therefore continue to depend on


2

Page 3 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


medicinal chemists for perhaps the majority of new product candidates well into the foreseeable future.

Academic Medicinal Chemistry Training for Careers in Industry Despite its central importance to drug discovery, there has never been a firm consensus on the best approach to academic training of medicinal chemists for drug discovery roles in the pharmaceutical industry. It is worth noting that, for every scientific discipline important to drug discovery, hiring managers in industry recruit graduates of academic programs for the same discipline with one exception. For medicinal chemistry positions, hiring managers in industry have preferred candidates with a degree in synthetic organic over medicinal chemistry graduates. For many companies, this preference is openly acknowledged. In a comprehensive survey of hiring managers in the pharmaceutical industry by the Medicinal Chemistry Section of IUPAC in 1992 and 1993 concerning academic training in medicinal chemistry, the prevailing opinions were that any academic training of medicinal chemistry recruits other than advanced organic chemistry was simply unnecessary.1 Even among those survey responders that expressed an interest in additional formal training of subjects beyond organic chemistry, there was little agreement on which ones should be taught.2 The prevailing opinion was that the necessary additional skills of a medicinal chemist would be acquired through time spent “on the job”. Traditionally, new recruits were paired with seasoned medicinal chemists who were trusted to mentor and teach them what they needed to know to become successful drug discoverers. Indeed, many academic faculty positions in medicinal chemistry are also being filled by organic chemistry graduates, presumably with the same


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 24

expectations that they will mature as medicinal chemists perhaps with some mentoring by more experienced faculty. “On the job training” or experiential learning obviously depends on the nature of the experiences from which learning is derived. This point is important, because no two drug design projects are alike. Every lead, molecular target, therapeutic target, and therapeutic endpoint all present different challenges for drug designers. For example, medicinal chemists who work primarily on CNS diseases will likely face different problems than oncology specialists. So the accrued expertise of a chemist over time will depend mainly on the nature of the projects to which she or he is exposed. Experiential learning is also a slow, incremental, and uneven process. A senior industrial manager who was one of the authors of a policy report from a 2006 symposium on the future of medicinal chemistry education estimated that for industry medicinal chemists “the learning curve was about 40% complete after 5 years of employment.”6 Experiential learning theory also recognizes that the learning process itself is highly individualized. Learning experts have defined four distinct experiential learning styles that correlate with personality profiles, which in turn are definable by standard personality profiling tools such as the Myers-Briggs test.7 It is therefore possible or even likely that individuals may each learn something different from the same experience. In other words, each chemist will accrue a unique portfolio of biases and cues that will form the basis upon which they make decisions derived from their individual experiences as influenced by their individual personalities and learning styles. Furthermore, an individual experience could also be misleading; “it happened to me once, therefore it will always happen” events can lead to a lasting impression and personal bias that might not


4

Page 5 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


even be scientifically valid. Individual experiences lead to individual biases, which form the basis for judgment criteria or “cues” for making decisions. In a 1974 paper on the definition of an expert, Einhorn proposed that the level of expertise within a group is related the degree to which “experts” organize information relevant to decisions; “experts should weigh and combine information in similar ways.” By this definition, if “expertise” is not being instilled with uniform effectiveness among industry medicinal chemists, then experience alone does not make “experts”. 8 Clues can be found in the literature that may provide some insights into just how effective experience-based training has been for medicinal chemists in industry. Two separate studies explored the extent to which experienced industry chemists agreed on the selection of a new program lead and whether years of experience had any effect on their choices. Lead selection is arguably one of the most important decisions that a medicinal chemist will make and so was identified as a good way to evaluate the consistency of decisions by chemists and the nature of the cues that they relied on to make them.9 The chances of success of a project depend a great deal on the choice of a high quality lead without serious deficiencies that would require extensive optimization effort. In 2004, Lajiness et al.10 conducted a simple experiment in which a group of 13 chemist volunteers from drug discovery at Pharmacia were asked to label any structure that they judged as unacceptable from a list of 2000 candidate structures. Each list was unique with the exception of 250 compounds that were included in all of the lasts in order to measure the level of consistency in agreement among the participants. All of the 250 embedded compounds had been previously rejected as lead candidates. Nine of the 13 chemists involved in this study were asked to review more than one list containing the same 250


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 24

compounds, which allowed for assessment of within-subject (self vs. self) consistency. The experience of these chemists ranged from 3 years to over 25 years. Surprisingly, only one of the embedded compounds was rejected by all 13 chemists; the other 249 structures were all acceptable to at least one of the participants. The within-subject results showed that the likelihood of a chemist repeating the same judgment of a compound was roughly 50%. Perhaps most disturbing of all was the observation that “experience had little to do with consistency of opinion” in that the level of agreement between two chemists with 25+ years experience was only 23%. The authors concluded from the results of this study that chemists make quality decisions about structures based on “their own personal set of guidelines”, and even in cases where the same parameters (such as lipophilicity) were identified as decision cues there was no consistency in the acceptable value of the parameter. Since 2004 there have been a number of publications by industry researchers on property value criteria and the utility of property-based tools and metrics for lead selection.11,12 It might therefore be expected that a repeat of the 2004 study would show greater consistency among chemists making the same sort of decisions. However, in a second investigation by Kutchukian et al.13 published in 2012, the outcome was similar to the earlier study in virtually every aspect. The Pharmacia study was focused on the question of whether a candidate structure should be rejected from consideration. This more recent study out of Novartis asked volunteer chemists to select attractive fragment leads from a list of approximately 4000 candidate structures. This study was also designed to identify the conscious and sub-conscious cues and biases relied upon by the volunteers in making their selections. Even though the question being posed was different


6

Page 7 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


for the two studies (rejection vs. acceptance) the results were very similar in that the degree of consistency among chemists was low. Statistical analyses to identify the cues that drove the decisions of individual chemists indicated that even where chemists reported using the same compound property or properties, the cutoff values used for these properties varied significantly. This study also revealed that chemists were not always aware of the personal biases that were most influential in their choices. And once again, there was no relationship between consistency of decisions and years of service. The level of agreement in compound selections among the most experienced participants was no better than for a group of randomly selected chemists. It needs to be acknowledged that selecting a lead from a list of hits is at least partly subjective and it would be unreasonable to expect that chemists would all make exactly the same choices all of the time. However, what these two studies showed is that even experienced chemists differ significantly on the features and cues they use in considering structures, and the extent to which unrecognized personal bias can affect their decisions. A surprising aspect of the Kutchukian study was the lack of consistency in the consideration of critically important physical properties of molecules, notably lipophilicity. The results indicated that the degree to which chemists considered lipophilicity to be important varied significantly and even where chemists agreed on the importance of this term there was no consistency with respect to what degree of lipophilicity was acceptable. A more consistent understanding and use of molecular property criteria for making key strategic decisions won’t necessarily lead every chemist to select the same exact lead from a hit list, but would certainly mean greater consistency in avoiding poor quality leads. It is important to also keep in mind that medicinal


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 24

chemists are expected to make decisions daily, and that these decisions are all influenced by their backgrounds. Making good decisions at a critical time increases efficiency, productivity, and quality, so it is logical that companies would be interested in developing researchers who make good decisions. The problem is, that at least the studies summarized here suggest that the current model isn’t very efficient and there is room for improvement in the training of medicinal chemists who can more consistently make good decisions. Review of the drug discovery literature reveals another perhaps related issue, which is that many chemists engaged in drug discovery don’t appear to attend to physical properties during optimization. A recent study included 261 small molecule lead optimization papers published in leading journals during the first half of 2014 to determine whether property-based drug design concepts and tools were being used, and if so by whom. 14 Their findings showed that overall only 33% of the papers reported interest in controlling for lipophilicity. Additionally, a comparison of industry vs. academic projects revealed a significant difference in the consideration of molecular properties along with potency, finding that approximately 40% of industry papers controlled for lipophilicity compared with 15% of the academic papers. Not surprisingly, projects that controlled for lipophilicity during optimization tended to yield better quality development leads. As has been discussed in several recent reviews, failure to attend to physical properties while relying on potency as the sole measurement of progress generally leads to highly potent compounds with poor drug-like properties.24,25 In particular, dissolution and solubility, ADME, off-target promiscuity, and other behaviors such as efflux transporter susceptibility have all been shown to be influenced by physical


8

Page 9 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


properties including lipophilicity.24-28 Molecules which possess an optimum drug-like property profile are less likely to encounter problems in development and are more likely to make it to market, regardless of therapeutic application. A similar investigation was carried out for this Perspective, which included a review of the small molecule optimization papers published in Volume 58 of this journal (2015). This analysis included only those papers that described an SAR investigation with the stated or inferred intent to identify a potential therapeutic candidate. Only small molecule projects that aimed for an orally administered agent were considered, and so optimization of peptides by sequence modification is an example of projects that were excluded. This left a total of 455 papers to be reviewed, of which 131 were from industry, 259 were from academic institutions, and 65 came from public/private research nondegree institutions such as the NIH. A total of 269 unique affiliations were among the reviewed papers. Each paper was searched for the following property keywords: lipophilicity, hydrophobicity, logP (or log P), logD (or log D), PSA, MW, polar surface area, or pKa, Lipinski, or rule of 5. A single mention of any of these terms within the body of the text or tables was counted as evidence of consideration of that property in compound design. Papers were also searched for evidence of the mention or use of calculated property/efficiency metrics such as ligand efficiency (LE),15 lipophilic ligand efficiency (LLE or LipE),16 LE/logP (LELP),17 Fsp3,18 aromatic ring count, or other property analysis and scoring tools such as CNS-MPO19 or QED.20 The papers were then grouped as follows: 1. no mention of properties or property metrics; 2. one or more property terms were identified; and 3. property metrics were used. The results are presented in Table 1. Overall, 35% of the papers mentioned or considered one or more


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 24

compound physical property. And as in the case of the earlier study, the differences between industry and academia were stark- 61% (80/131) of the industry publications mentioned one or more properties, whereas only 24% (79/324) of the non-industry publications did so. Reference to the use of one or more property metric was found in 10% of the papers overall, but only 3% of the non-industry affiliated papers cited a calculated property metric. Large pharma researchers were just slightly more likely to consider properties and to use property metrics than small/medium pharmaceutical company groups. Even though ligand efficiency was first described in the literature in 2004 with reports of several other efficiency metrics in subsequent years, it appears that either the academic community remains unaware of these tools or doesn’t see a need to adopt them. There is also some evidence that chemists who acknowledge the need to manage molecular properties may not understand how to do so. An analysis by Leeson and Young of recent patent applications revealed that many compounds with poor physicochemical property profiles were still being made.14 Among the factors suggested by the authors that could be enabling poor design decisions was the inappropriate reliance on the Lipinski “rules-of-5” to determine acceptability of a compound based on calculated properties. Reward systems may also be a factor in organizations that rely on the number of compound submissions as a measure of performance by valuing quantity over quality. The widespread use of automated synthesis technologies to rapidly expand a lead series may also encourage synthesis of some low quality compounds simply because the capacity and building blocks is available. Synthesis of compounds with poor properties has multiple negative consequences.


10

Page 11 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For example, synthesis and testing of low quality compounds likely results in the need for still more compounds and more chemistry before the development candidate is discovered. Evidence for this was reported a recent study out of AstraZeneca.21 And every compound submission imposes overhead costs for analysis, registration, storage, distribution, and testing. So burdening programs and compound libraries with low quality compounds can add up to a significant drain on research budgets. All of this suggests that industry would do well to consider the potential benefits of a better, more consistent approach to training chemists to make better, more consistent decisions as a way to improve R&D efficiency. Impact of Industry’s Organizational Changes on Medicinal Chemistry Training The ongoing consolidations across the entire pharmaceutical industry have dramatically altered the roles and responsibilities of medicinal chemists who remain after staffing cuts. These chemists increasingly are finding that their primary responsibility is to design new targets for synthesis by either dedicated internal or external synthesis groups. These “designers” may do little or no bench work themselves, and instead are committed full time to the application of highly sophisticated cheminformatics and drug design software tools. Nussbaumer has actually proposed a “two expert paradigm” pairing a designer with a highly skilled synthesis expert to make the targets, which is operationally similar to the outsourced chemistry model used by most biotechs and a growing number of large pharmaceutical research organizations.22 Chemistry CRO’s worldwide have realized explosive growth as the demand grows for low cost, budgetflexible synthesis support. And large pharmas have undoubtedly realized significant operating cost savings as a result of chemistry staff reductions and elimination of


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 24

expensive synthesis lab space. However, a potentially important and far-reaching consequence of this new operating model for industry could be the disappearance of experiential training as a way of training new medicinal chemists. Drug designers must be expert users of a growing suite of computational and data analysis tools to identify SAR trends and inspire analogue design. Successful designers must possess a deep understanding of the molecular properties of drug-like space and intimate knowledge of biological processes that influence drugs, the nature of ligandtarget interactions, and protein structure and function. General knowledge of basic physiology, biochemistry, drug metabolism, transporters, pharmacokinetics, and in vivo drug behaviors across species are critically important. Designers must be able to properly interpret all types of data and understand statistics and principles of experimental design. Understanding of the medical need, target market profile, and clinical challenges of the therapeutic target are also important design considerations over the course of a project that best suit the intended clinical use. What designers don’t require is sophisticated inlab training as a synthetic chemist. Obviously, designers must understand synthesis well enough so as to be capable of designing makeable targets; but synthetic prowess is less important. As this new operating model continues to evolve, it will make less and less sense for large pharmaceutical companies to recruit highly trained synthetic chemists for a position which may not require much in the way of synthesis, rather than recruits with more relevant training and experience and a much shorter learning curve. Biotech firms that rely mostly or entirely on contract synthesis have been successful in recruiting designers from among the experienced chemists whose positions were eliminated. Eventually, however, the entire pharmaceutical industry may encounter


12

Page 13 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a serious shortage of trained recruits to replace the current designers as they retire, because the traditional way of grooming the next generation may not be affordable. On the job training of the next generation of designers by hiring in recruits with skills that don’t fit the roles can’t meet the needs of this evolving operating model. If industry is to avoid a critical staff shortage in the not-so-distant future, it will need to come up with a new way of training the next generation of designers or develop a new source of recruits with the requisite skills. Issues with Current Academic Medicinal Chemistry Training The studies discussed herein suggest that industry’s historical reliance on on-thejob training strategy doesn’t work as well as is generally believed. And if the scenario above plays out, training new medicinal chemists by experiential learning may well become unaffordable. A formalized graduate training program that emphasizes the core principles and practices of drug design both in formal lectures and graduate research could potentially enable industry to sustain their outsourcing strategy for the long term. Such programs have been training students for as long as 60 years in medicinal chemistry training programs at major universities. Although these programs were dismissed by respondents to the IUPAC survey as “synthesis light”,2 this criticism may no longer be relevant in the new age of designers. In principle, academic training programs should be well positioned to serve the future needs of industry. However, there may well be another reason why industry might continue to be concerned about medicinal chemistry graduate training. In a 2004 editorial, Simon Franz opined that the differences in research philosophies and practices between academia and pharmaceutical industry were greater in the medicinal chemistry field than perhaps for any other scientific discipline. 3 Over the


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 24

past several years, medicinal chemistry departments have diversified in an attempt to better represent the full scope of drug discovery, with the result that many of the research programs within their departments don’t actually train many students to be medicinal chemists in the classical sense. Many medicinal chemistry departments now recruit faculty whose skills and interests are useful and important in the context of drug discovery, but the training that they may receive in their graduate program may not prepare them for a designer role. Career motivations for academic researchers to publish and succeed in attracting grants and research support are quite different career development motivators compared with the product-oriented priorities of industry. But even where academic research programs have been focused on novel therapeutics discovery, where one might expect to see a greater level of agreement between academics and industry, Franz noted that their research philosophies were generally quite different. The data presented in Table 1 are perhaps consistent with this conclusion. To underscore the point being made, only one out of the 68 publications from U.S. based laboratories included in the Table 1 data reported using property metrics in their optimization.23 Academic researchers tend to emphasize target potency as their only measure of progress during optimization, which as noted earlier has been shown to lead to compounds with poor physicochemical profiles in multiple studies from industry scientists. 24-29 These results are also consistent with my personal observations as a frequent reviewer of small molecule NIH translational grants over the past 10 years, where research plans were often based entirely on maximizing target potency. It is possible, therefore, that industry’s low regard of academic medicinal chemistry training may be more nuanced than just the


14

Page 15 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


“synthesis light” argument. Industry may also be concerned that current academic training and research practices of drug design aren’t keeping up with the times. Industry and Academic Medicinal Chemistry Programs must find solutions together. Without question, the drug discovery community includes many outstanding medicinal chemists, and thanks to these true experts the pharmaceutical industry continues to launch important new products every year. But it is also important to acknowledge from the data discussed here that the number of “true experts” in drug discovery behind these innovations may be lower than assumed. Industry, for many of the reasons already discussed, should recognize that inconsistency and generally poor decision making by discovery chemists can be wasteful and expensive, and then do something about it. And the medicinal chemistry academic community, for many of these same reasons, should recognize that they are failing to train graduates who will be attractive to industry. It also needs to be emphasized that the pressures on academic research by university administrations to produce opportunities for licensing and royalty revenue is not likely ever to diminish, so academic researchers need to focus not just on grants and publications but also on producing development candidates as a source of research funding. For many reasons, both research communities have reason to find common solutions before the situation becomes critical. Among the most obvious approaches would be to establish industry-academic partnerships to jointly train new medicinal chemists. Industry already commits tens of millions toward industry-academic drug discovery efforts, in many cases involving physical placement of industry scientists on university campuses. Examples include


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 24

Pfizer’s university-based Centers for Therapeutic Innovations and Lilly’s Open Innovation Drug Discovery program. It may be possible to incorporate medicinal chemistry graduate training co-directed by industry and academic mentors for relatively little added cost. As another possibility could be to establish Industrial Ph.D. programs, which have been in place in Japan for many years. Universities should also take a hard look at course content to ensure that students are being exposed to the innovations coming from industry, and recognize that their own research program would benefit from these techniques by improving the chances of discovering an attractive commercialization opportunity. It is possible to develop formal courses that not only teach contemporary drug discovery tools, but create situations where students can put them to practice. A special topics course that I have been teaching at the University of Kansas for over 10 years was designed specifically for this purpose.30 Funding agencies could also play a significant role; for example, the Innovation Fund Denmark has already funded thousands of industrial Ph.D. research programs since it’s creation in 2007.31 Novo Nordisk, NeuroSearch, and Lundbeck are among the companies who have taken advantage of this program. A similar program within the EU is Marie Curie Actions.32 In both cases, Ph.D. candidates receive financial support from the funding agency and conduct their research at the industry partner laboratories with joint mentorship by industry scientists and academic faculty. These agencies also fund short term sabbaticals for faculty to conduct research in company laboratories, where they are exposed to industry tools and practices. Another EU program, the Innovative Medicines Initiative, also provides training grants for industry-academic collaborations.33 And in the US, the National Science Foundation’s Grant Opportunities for Academic Liaison with Industry (GOALI) provides funding for


16

Page 17 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


academic faculty and trainees to work and learn in an industry setting, and for industry scientists to engage with educational institutions to develop and improve curricula and training programs to better suit the needs of industry.34 All of these programs could serve as models for a more comprehensive medicinal chemistry graduate training partnership. Academic departments should also look to recruit from the pool of experienced industry chemists affected by the ongoing reorganizations, or who may simply be interested in a career change. Already many industry chemists have been hired as research faculty to lead university-based drug discovery initiatives, although it is not clear to what extent they participate in graduate training. The pharmaceutical industry has been, and will likely always be, the largest employer of medicinal chemists, and academic departments need to be more responsive to the needs of the discipline’s largest employer if both communities are to thrive. I am among the many who experienced first hand the evolution of drug discovery in industry, and among the very few who had the opportunity to reengage the academic community and observe both communities up close. These perspectives have given me an opportunity to consider the opportunities open to both communities to radically change the medicinal chemistry training paradigm if they are willing to act. Perhaps, under the circumstances, the time to act is now.


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 24

Table 1. Investigation of small molecule discovery publications in Volume 58 (2015) of the Journal of Medicinal Chemistry for evidence of property considerations during optimization Affiliation Grouping

# institutions

#papers

#papers per institution 2.47 4.2 1.06 1.47

#papers mentioning any property (%)1,2 80 (61.0) 57 (62.0) 16 (51.6) 63 (24.3)

#papers citing use of property metric (%)2,3 34 (25.9) 26 (28.3) 7 (21.2) 8 (3.1)

Industry Large Pharma Small/Medium Academic Degree Institutions U.S. Universities Research institutes

53 22 31 176

131 92 33 259

58 40

89 65

1.53 1.65

1(1.1) 16 (24.6)

1(1.1) 3 (4.6)

Totals

269

455

1.69

159 (34.9)

45 (9.9)

1. Any mention of a physical property of importance to drug behavior was counted. The term must be in the body of the paper; terms found only in the title of references were not included. The terms searched were logP (or log P), logD (or LogD), PSA, Lipinski, rules of 5, # rotatable bonds, and Fsp3. 2. Comparison of industry vs. non-industry percentages by two-tailed Statistical Comparison of Proportions resulted in P values

Medicinal Chemistry Training Is in Big Trouble - ACS Publications

Recommend Documents