REALIZE HIDDEN VALUE THRDDGH TIMELY PDRTFOLID DECISIDNS

REALIZE HIDDEN VALUE THRDDGH TIMELY PDRTFOLID DECISIDNS

Dynamic portfolio simulation with aligned management processes can improve pharmaceutical research productivity.

Roland Mohr, Harald Pad and Marcus Hartmann

OVERVIEW: Sophisticated portfolio management tools have been developed and implemented to maximize the value of existing projects in the later phases of drug development. However, the almost equally cost-intensive early research phase with a large number of relatively small and risky projects has twt heen examined appro- priately. By taking into account key drivers of early phar- maceutical research, namely leveraging internal and external expertise identically across the organization, and recognizing the temporary validity of respective portfolio decisions while using utility value-based dynamic portfolio management techniques, it is possible to derive a consistent organizational model that should significantly enhance research productivity and pipeline value.

KEY CONCEPTS: dynamic simulation, portfolio opti- mization, research productivity.

The development and launch of an innovative new drug may take 10 to 15 years, and cost estimates vary from around $500 million to more than $2.000 million, depending on the therapy and the developing fimi (/).

Roland Mohr is managing director oflnfraserv Höchst, Frankfurt. Germany. Prior to this he was head of drug innovation and approval-site operations, at Aventis Germany. He holds a master in chemistry from the Uni- versity of Wuerzhurg, a Ph.D. from the University of Muenster, and a Controlling degree from the IHK in Frankfurt, roland.mohr@infraserv.coin

Harald Pad has many years ‘ experience in the pharma- ceutical industry. He holds a masters in chemistry, a masters in business administration and a Ph.D. in organic chemistry.

Marcus Hartmann has been working in the pharmaceu- tical industry since 2003. He has a Ph.D. in chemical economics from Technical University Berlin, and a masters degree in both biomédical sciences jrom Univer- sity^ of Marburg and finance from Frankfurt School of Finance and Management.

R&D budgets of international pharmaceutical com- panies trended around 15 percent of sales in 2006 (2). Additionally, the pharmaceutical R&D process is burdened with extremely high attrition rates, resulting in ever-fewer innovative dnigs coming to the market.

Despite scientific breakthroughs and rapid technological advances in miniaturization and automation, potential drug candidates face a high degree of project termination during the rigorous testing and selection process. Faced with a dire need to close the innovation gap, a reorienta- tion toward value maximization has transformed the R&D departments of the large, muhinational pharma- ceutical companies (3).

There have been any number of reports on value-driven portfolio management. However, current approaches {4) show an intriguing gap in addressing the specific aspect of early drug discovery, e.g., management of portfolios with projects exhibiting high uncertainty and a high number and frequency of decision points. This paper presents for the first time a dynamic model of an inte- grated portfolio management process for the research phase. We specifically discuss the minimum necessary decision bodies involved, key processes, number and temporary validity of the decisions, and address the general issue of portfolio-wide resource allocation.

Organization of Pharmaceutical R&D The pharmaceutical industry provides innovative therapies to patients through new proprietary products. Each company defines its therapeutic areas of interest in the context of formulating the respective enterprise strategy. For simplicity. Figure I divides the complex R&D process into three different segments based on common defmitions in the industry and characteristic project and risk profiles. Based on the chosen therapeutic areas, specific disease knowledge for various indications is created internally, a mandatory prerequisite for suc- cessful partnering through in- or out-licensing.

We believe that an open enterprise structure is needed in all phases without any discrimination between internal and external project ideas. The pharmaceutical industry

D Research • Technology Management

has made significant progress toward this goal, particu- larly in the early and late development phases. Because these final stages of the pharmaceutical development process (clinical development) are highly regulated, information for a project in the clinical stage is extensive and standardized and can be readily transferred between companies. The large number of licensing deals is proof of this (5). However, for the research stage this still seems an elusive goal. In particular, unresolved intellectual property issues make companies reluctant to share data openly. Even in long-established successful strategic partnerships, data created by one partner do not always fit the infonnation needs of the other, thereby resulting in additional experiments and difficult project evaluation.

One way to address these issues is with the model of Innocentive, Inc.. which Eli Lilly founded. Companies wishing to solve a particular problem (seekers) are anonymously connected with thousands of scientists around the world (solvers); all intellectual rights reside with the company whereas the solver receives a cash award for an accepted solution. This approach has yielded impressive results but seems limited in applica- tion (6). In any case, we see the ability to treat internal and external projects identically and integrate them into the same decision now as an important pillar of future pharmaceutical innovation without which above- average company success will not be possible.

Given the rapid ascent (and descent) of hot, breakthrough technologies, which no single company can comprehen-

We need an explicit teciineiegy strategy that defines in-heuse core competencies

and areas fer partnering.

sively perform in-house, there is a clear need for an explicit technology strategy that defmes in-house core competencies and areas for partnering. This strategy then has a strong impact on project organization and future investments. An organizational structure and culture with clearly defined internal roles and integration of external opportunities is most likely to ensure a continu- ous flow of innovative drug candidates ( 7).

Characteristics of Portfolio Management In general, management prioritizes and selects active projects based on appropriate value maximization criteria. The same criteria are applied to decisions

R&D strategy implementation

Research Early

development Late

development

Alliance portfolio (external) &

Project portfolio (internal)

Ranking, selection

Processing of individual projects

Resource allocation

Resource pool (staff, financial, capacity)

c 0)

î ö u 0)

Figure L—The pharmaceutical research and development process has three stages, bnplementation of research and technology strategies drive resource allocation and partnering.

November—December 2008

regarding adequate finances, capacity, headcount, and resource allocation. A lot of effort has been invested in the two later stages of pharmaceutical R&D. The devel- opment stage especially, with individual project budgets of $100 million or more and extensive clinical trials running over years, has received a lot of attention. Despite scientific differences in preclinical and clinical phase I work as compared to the development phase, the overall characteristics of the early development phase like duration, number of projects in the portfolio and indi- vidual project budgets can also be adequately addressed by existing portfolio management techniques (4).

In contrast, the research phase is characterized by many comparably small, usually short-lived projects with high attrition rates. In this area, formal individual project budgeting and organization would be cumbersome and is done only toward the end of this stage in preparation for a transfer to the next development stage. Nevertheless, within the research phase an unusually high number of portfolio decisions and resource allocation have to take piace, which clearly should also be based on value creation criteria.

This crucial aspect—stringent execution of portfolio selection—is for praetical reasons often neglected or

viewed as a static problem. We believe that this repre- sents an inadequate approach to early drug discovery and therefore present an alternative. This seems all the more important because, while individual project budgets/ investments in the research phase are relatively small compared to late development projects, the overall budget spent on research is nearly comparable (8). On an operational level, a resource decision in this phase is often assigned to individual labs, e.g. assignment of a complete medicinal chemistry lab to work on a given project for three months. This is a decision that may easily cost more than $250,000. Equal diligence should therefore be exercised in “small,” day-to-day resource allocation decisions.

Dynamic Management ofthe Research Portfolio

Qualitative valuation

In the research phase of pharmaceutical drug develop- ment, negative results from critical experiments bring about immediate project termination resulting in unallo- cated capacity with a need to respond to a changed situation. Therefore, for an efficient R&D process, a dynamic way of reallocating free resources is desirable.

Potential criteria for scoring model:

Project candidates

External options

Option to expand/ add target profile \ Option to license out \ Option to deiay \

Number of back ups / Degree of innovation / Doability / Medical need / Patent situation / Epidenniology / Peak sales, margin/ Market situation /

Adjustment of critical resources

Portfolio with

project values

Ranking, selectio

Project xecutio

Resource allocation

Transfer to next stage

• Termination Divest

Figure 2.—The resource allocation process for the pharmaceutical research stage in its simplest form consists of portfolio evaluation, selection/ranking ofprojfcts. resource allocation, and project execution.

Research • Technology IVIanat;eiiient

Figure 2 illustrates a potential process for resource real- location. This process should be done within the poiitblio conte.xt and also include decisions on new entries into the portfolio. Given incomplete project infor- mation, any ranking and selection of projects should be based upon utility values derived from an appropriate scoring model (Figure 2, left).

The utility value comprises all relevant qualitative aspects, ranging from biomédical assessments and chemical feasibility to real-option criteria and intellec- tual property considerations (9). After summation, the qiiasi-quantification of each individual criterion using appropriate weighting factors yields a utility value that is specific for each company’s project and directly reflects core competencies and strategy. The actual methodology for deriving a utility value for a project in the portfolio is considered secondary to the fact of having one for each project.

It seems sensible to revisit these utility values on a yearly basis. Between reviews, the calculated utility values should generally not be challenged. Any project newly entering the portfolio has to be evaluated accordingly to ensure consistency before proceeding with the dynamic portfolio management approach. It has to be stressed that quantitative economic aspects for individual projects in the phamiaceutical research stage cannot be evaluated with any degree of certainty. Too many questions regarding the potential drug profile, its eompetitive labeling, etc., remain unanswered.

Ranking, selection and resourcing The results of the utility value determination constitute the basis for the following steps in the decision process of

a Research Management Board. Using all potential project ideas, which have been assigned individual project utility values, a preliminary ranking of projects is peifonned. In addition to the consideration of the indi- vidual scoring model results, comprehensive portfolio aspects have to be taken into account for the final ranking, e.g., market criteria like marketed product portfolio strategic outlook, or qualitative biomédical criteria like number of projects in one indication/disease area.

Figure 3 illustrates the results of the resource allocation process. For practical ranking reasons, three buckets (high, medium, low) are often considered sufficient. From past experience, existing capacity is normally deemed sufficient to process high and medium project candidates. Thus, in a first step, only projects with either high or medium utility values are selected and tentatively resourced. If resources are still undeiutilized additional projects with low utility value are added to capacity. All other projects not resourced are put on hold and reas- sessed in the next period (Figure 2), terminated or licensed out to third parties. The stim of all utility values of resourced projects in the selected active portfolio con- stitutes the corresponding portfolio utility value, which is then used as an indicator for value maximization.

So far, this procedure may sound familiar, almost trivial, to the expert. It is an accepted method for aligning projects and resources with strategy; it makes a lot of sense for many portfolio and is widely used in various fonns. Even in phannaceutical R&D it is the method of choice in the later stages, namely preclinical and clinical development.

O Bubble size indicates expected peak sales

Projects fully resourced

Minimal utility value for project continuation after capacity analysis

Therapeutic Therapeutic Therapeutic area 1 area 2 area 3

Figure 3.—Capacity analysis yields a minimum utility value for project continuation. Projects with utility values below the threshold are not resourced.

November—December 2008

However, consider the case of a research portfolio of approximately 100 projects with around five critical experiments to perform per project {go/no-go decisions) in the course of one year. You complete the above ranking and selection exercise and start processing projects. Given the number of decision points (approxi- mately 10 a week) you wouid shortly have to reconsider the resuits of the resource aliocation to stay on the path of pursuing an optimai portfoiio vaiue. The originai projects are either moved to the next phase or temiinatcd. Resources can be reailocated to projects that were not resourced initially and can now be added to the active portfolio.

Additionally, new project ideas come up and have to be compared to existing projects. How long is this portfolio shift and gradual erosion of the portfoiio vaiue accept- able? When is it necessary to reassess aii project values? How does one ensure that a realiocation of resources under new conditions yieids optimai portfolio utility value? What frequency of review does management need in order to maintain portfolio progress?

Surprisingiy, in many cases companies do not react very dynamicaliy at this point. They often retain the originai “core” resource aliocation and add new projects only following resource availability between project utiiity value reviews. Review frequencies either simply evolve or are fixed without directly examining the link between portfolio size and change rate.

In some cases, a portfolio reassessment is done every six months. In the above example this correlates to a change in portfolio activities of approximately 50 percent. Utility values of projects are often reviewed not more often than once a year, if used at all. Whiie this is a workable and deceptively simple answer to the problem, it is obvious that most of the time it results in the pursuit of sub-optimal portfolio utility values with high oppor- tunity cost to the enterprise.

Realizing hidden value

One can fmd striking similarities between planning parallel pharmaceutical research projects and batch scheduling in manufacturing. Figure 4 summarizes their most important characteristics. While the two situations appear at first giance to be completely different, a closer

A dynamic resource allecatien of

research prejects is possible witii

operatiens researcii metiiods.

iook reveals distinct parallels regarding process charac- teristics, complexity, objectives, etc. Since batch sched- Liiing methods are well establisiied and widely used in multi-purpose batch manufacturing plants, there is no reason to assume a properly modified model shouid not yield significant improvements in the pharmaceutical research arena. Obviously, there are constraints that cannot be ignored, e.g., different time segments (weeks instead of minutes) or scientists who cannot be reallo- cated to a new project with the same ease as an assembly line or a vessel. i

Another issue often in evidence is the idea that pharma- ceutical research cannot be pianned but is inherently ser- endipitous. Although we certainly do not recommend deterministic planning of research activities, we do believe that the research process—not the results of research—can be modeled with benefit. In case of changes or scientific findings that require a portfolio change, this new situation can then be simulated accord- ingly. Especiaiiy in pharmaceutical R&D. serendipity is highly valued and has likely hampered a rational approach to this problem in some cases. Whiie aware of the difficulties, we nevertheless conclude tiiat a dynamic resource allocation of research projects is possible using operations research (OR) methods

Other groups have also built upon the similarities between batch processes in manuiacturing and the phar-

Munufacturint; Pharmaceutical Rescarcli

Problem Process Uniqueness Complexity Resources Events Objectives

Scheduling Batch process Repetitive Complex, large number of constraints People, assets, money Breakdowns, delays Schedule. NPV, R O l . . . .

Resource allocation on projects Batch process Scientifically unique, workwise repetitive Defined milestones, less complex People, money Serendipity, delays, scientific issues Schedule, milestones, c o s t , . . .

Figure 4.—Mamtfactiiring proces.ses and the pluirnuiccutical research process share a number of cownum eluiracterisfies.

Research * Iechnoldíjv

maceutical research process (7/). In all these cases, the application of OR methods to specific aspects of drug discovery and development has resulted in value being added to the enterprise. Examples range from time com- pression of development programs helped by multi-class queuing network models (12) to value-enhanced portfolio decisions by discrete event simulation (13), to improved resource allocation using dynamic investment policies (14).

In our example, resource needs for individual projects are modeled using normalized standard intensities. All possible projects, irrespective of the individual project utility value or its ranking, can be considered. Subse- quently, the overall portfolio utility value can be optimized by appropriate variation of those intensities. Evidently, the results of this exercise differ from the tra- ditional “bucket” approach, usually leading to a project utility value threshold lower than the traditional one. Since all available resources are used as best they can be, potential bottlenecks are either avoided or at least iden- tified. Figure 5 illustrates the non-intuitive results of OR methods in resource allocation. Note that two projects in therapeutic area 1 and 2 are not resourced despite having a medium project utility value, probably because some necessary resources are already used to maximum capacity. However, projects with a lower project utility value not using these particular resources can be fully supported.

Results like this are typical when applying OR methods but are not easily understood. Generally, more projects can be resourced efficiently using OR methods, with value improvements typically in the double-digit range. Obviousiy. the derivation of an enterprise-specific model and appropriate boundary conditions necessary for a fruitful application of these OR methods is not trivial but the costs are easily outweighed by the benefits (14). It has to be stressed that all simulations can be performed on a standard personal computer within seconds and thus allow examination of a wide variety of scenarios and comparison of numerous options for a realistic portfolio (>IOO projects).

The decision to support and resource a project also includes related decisions on technology investments and external partnering needs in the research stage. All selected active projects within the portfolio would be processed using the deñned number of given resources. Ifthe research stage is successfully completed, projects are then transferred to the next stage, early development, using a standard stage-gate process. Projects may be ter- minated at any point within the research stage depending on the results of critical experiments.

The utilization of resources and potential capacity adjustment of critical resources is assessed yearly. Again, OR methods can support the identification of critical resources as well as the utility value contribution

It is crucial to evaluate externai preject optious iu the sanie way as internai prejeots.

of investments in additional resource capacity (10). Naturally, from a more strategic enterprise perspective, the key financial figures for the company like net debt, R&D expenses as percent of sales, etc.. also have to be considered to avoid disturbing financial market percep- tions. If a decision to increase capacity is taken, a subse- quent standard make-or-buy business case will have to be assessed depending on the portfolio situation and the particular resource and technology strategy. In cases of under-utilization a divestment deeision may be appropri- ate (Figure 2).

Summary Conclusions

The integrated management process we have presented is based upon recognizing that a truly dynamic portfolio

Therapeutic Therapeutic Therapeutic area 1 area 2 area 3

Projects resourced

Figure 5.—Selection and resourcing of projects using an operations research approach results in non-intuitive solutions. No minimum utility value for project continuation exists: instead, overall portfolio value has been optimzed. Bubble size indicates expected peak sales.

November—December 2008

Decision bodies

Strategy board

Development management

board

Research management

board

Execution

l W J M ] Q1 Q2 Q3 Q4

Calculation of project values Figure ó.Stringent hierarchical set-up and meeting frequency of decision bodies result in an appropriate organizational model for R&D decisions.

management should make use of dynamic simulation tools to maximize portfolio value and ensure smooth and optimal resource reallocation. Frequent changes within the portfolio require an adequate number of management reviews. This number is enterprise-specific but generally high.

In our example of approximately 100 projects in parallel for a pharmaceutical research unit, an appropriate meeting interval should be at least bi-weekly. This reflects the need to respond quickly and efficiently to the ubiquitous changes within the portfolio. It also strength- ens the innovation ability of the respective research unit by optimally allocating limited resources to projects where value is created best. Based on this align- ment, similar considerations allow the development of an organizational model for all R&D-relevant decisions and all decision-making bodies as shown in Figure 6.

Finally, it is crucial to evaluate external project options in the same way as internal projects despite the discussed difficulties, management issues and potential infonna- tion gap. Only by fully integrating these three aspects— use of dynamic portfolio simulation tools based upon utility values, total alignment of management processes with optimal frequency, and identical treatment of all projects irrespective of their origin—can portfolio man- agement become truly dynamic and value maximizing after all.

PLACE THE ORDER WITH US TODAY AND GET A PERFECT SCORE!!!