The contribution of clinical pharmacology surrogates and models to drug
`development—a critical appraisal
`
`Paul Rolan
`Medeval Ltd and Neuraxis Ltd, University of Manchester, Manchester Science Park, Lloyd St North, Manchester M15 6SH, UK
`
`Keywords: surrogates, models, validation, drug development
`
`Br J Clin Pharmacol 1997; 44: 219–225
`
`Why do we need surrogates?
`
`Separating the wheat from the chaff at the earliest possible stage
`
`Over the last decade or so, the objectives and process of
`clinical drug development have been changing in response
`to an altered regulatory, medical and business environment.
`Due to the increasing costs of clinical drug development
`and increased market competitiveness, companies can no
`longer afford to continue to late Phase III with drugs which
`are unlikely to be therapeutically effective, or to market
`new products which lack superiority over existing treatments.
`Hence companies are increasingly willing to take novel
`compounds into man, but with the expectation of an early
`answer to the likely clinical and commercial success with
`abandonment of the compound if the target profile is not
`likely to be met.
`Such an approach has changed the traditional view of
`clinical drug development. Typically, this has been divided
`into three phases [1]. In Phase I, small numbers of volunteers
`or patients are exposed to the drug, usually starting with
`single doses, to examine kinetics, tolerability and ‘safety’.
`Frequently, the ‘maximum tolerated dose’
`is sought. In
`Phase II, dose-response relationships are sought in highly
`selected and studied patient groups, with confirmatory
`efficacy and ‘safety’ in Phase III. However, there is increasing
`pressure, for the reasons listed above, to determine the likely
`probability of therapeutic and commercial success of a new
`drug as early as possible, ideally after the first human study,
`and to optimise the dosing regimen early.
`Because the traditional approach required considerable
`resources
`to estimate likely therapeutic success and has
`resulted in overdosing, many companies are viewing clinical
`drug development differently. Instead of comprising three
`phases, it is divided into two phases, ‘exploratory’ and ‘full’.
`‘Exploratory’ development consists of all clinical work
`required to demonstrate the likely chance of therapeutic
`success
`[3]. Sometimes this
`is referred to as
`‘proof-of-
`principle’. This may require as
`little as one study but
`typically it covers Phase I and Phase IIa in the traditional
`scheme. ‘Full’ development consists of completion of the
`registration dossier. The objectives of exploratory develop-
`ment are to demonstrate dose- and concentration-response
`relationships, perhaps for both desired and undesired effects.
`This approach is particularly important to small pharmaceut-
`ical companies such as many in the biotechnology sector.
`Such companies frequently do not have the resources to
`take a new drug to market, and need to attract a development
`partner. The probability of attracting a partner, and the
`value of the partnership to the initial company, will depend
`heavily on whether the ‘proof-of-principle’ point has been
`reached. In this paper I will give some views on how I
`
`© 1997 Blackwell Science Ltd
`
`believe that appropriate use of surrogates and models may
`be useful in expediting exploratory development.
`
`Getting the dose right
`
`industry
`the pharmaceutical
`the major errors
`One of
`continues to make is to attempt to register a dose which is
`too high. This may be partly due to the development
`approach used. In the traditional development approach, the
`Phase I studies are often little more than human toxicity
`studies sometimes examining the effects of doses far in
`excess of those ultimately clinically useful. The concept of
`the ‘maximum tolerated dose’ is a flawed one. Most adverse
`effects are dose- and concentration-related. An important
`adverse effect occurring at a frequency of 1 in 20 at a given
`high dose level
`is likely to be unacceptable in clinical
`practice but is unlikely to be detected or correctly attributed
`to the study drug in a typical volunteer Phase I study. With
`increasingly potent drugs, a new molecule may appear to
`be well-tolerated in Phase I at very large multiples of a
`pharmacologically effective dose. However, clinical trials at
`too high a dose may attribute an unacceptable safety profile
`to an otherwise good drug.
`The following are examples of drugs which may have
`been labelled as having a tolerability problem due to
`excessive doses. Captopril was initially licensed at doses of
`approximately 500 mg day−1,
`and with the resulting
`proteinuria and rash it was regarded as a specialist-only drug.
`At doses of 25 mg day−1, it is sufficiently well-tolerated to
`be a first-line antihypertensive. Benoxaprofen was withdrawn
`from the market, at least partly because elderly patients were
`overdosed. We do not need to go so far back to see drugs
`for which the individual patient dose may be much less than
`that initially marketed. Sumatriptan was marketed orally
`initially at 100 mg. However 50 and 25 mg have been
`shown to be equivalently effective as starting doses [2].
`For some of these drugs, the doses selected for clinical
`trials were based on the ‘maximum tolerated dose’ approach
`or fractions thereof. However, as a result of the overdosing
`of patients using this approach, most regulatory authorities
`are asking for demonstration of the minimum, maximally
`effective dose. In this paper I will describe the concepts
`behind and the use of surrogates and models to assist in dose
`selection.
`
`What is a surrogate?
`Ideally, decisions about likely efficacy and dose selection
`can be based on careful measurement of the therapeutic
`response in the earliest human studies. For some drugs, the
`desired pharmacological effects can be observed directly,
`even in non-therapeutic situations, e.g. antihypertensives,
`Genentech 2134
`Hospira v. Genentech
`IPR2017-00737
`
`219
`
`

`

`P. Rolan
`
`anticoagulants, antiplatelet drugs. However, for many drugs,
`these effects cannot be directly observed, and a surrogate may
`be needed.
`In this context, a surrogate may be defined as a clinical
`measurement, known to be statistically associated with and
`believed to be pathophysiologically related to a clinical outcome
`[4, 5]. In its literal sense, a surrogate can be used to replace
`the ultimate clinical measurement. However, for reasons
`which will be discussed later, such ‘surrogates’ fulfilling the
`literal definition of the word, are rare and particularly rare
`when novel therapies are being evaluated.
`
`What is a model?
`
`A model is an experimental system or paradigm, used in drug
`development to simulate some aspects of the disease of
`interest in which the effects of the drug are examined. For
`a model to be useful it must share two of the three main
`criteria defined for a surrogate above,
`i.e.
`it should be
`known to be statistically associated with and believed to be
`pathophysiologically related to the clinical outcome. Hence
`although models and surrogates are different operationally
`(one is an experimental system and the other is a clinical
`measurement), it will be argued that they can fulfil similar
`functions in expediting exploratory clinical development
`because of their potential to predict therapeutic efficacy.
`Sometimes the distinction between surrogates and models is
`not clear. For example,
`in asthma, urinary excretion of
`leukotriene E4 is a surrogate; allergen challenge is a model,
`but measurement of histamine sensitivity as an index of
`non-specific bronchial reactivity could be considered as
`the difference between surrogates and
`either. However,
`models is operational and not conceptual and in this article
`I will use the two terms interchangeably when referring to
`their strategic function in drug development.
`
`The properties of surrogates—their ‘dimensions’
`
`In order to critically review the utility of potential surrogates
`and models, I will describe some of their properties which
`I will call the ‘dimensions’.
`
`Validation
`
`Although the power of the use of surrogates is attractive,
`the major obstacle to their more widespread use is whether
`they adequately fulfil the criteria stated in the definition
`given earlier, i.e. the extent of validation of the potential
`surrogate. Surrogates which are truly sufficiently validated
`to replace the clinical endpoint and hence fulfil the literal
`definition of the word ‘surrogate’ are rare. Blood pressure
`
`Table 1 Examples of well-validated surrogates.
`
`as a surrogate for cardiovascular risk, as mentioned above, is
`one [6], but with the recent controversy about short-acting
`calcium antagonists [7], even this is in doubt.
`The first component of validation comprises the analysis
`of data on the statistical properties of the surrogate, relating
`to reproducibility, accuracy and bias of measurement and
`may also include observer error and variability due to assay
`techniques in addition to biological variation. Some consider-
`ations of these statistical aspects are discussed by Prentice [8]
`and Boissel et al. [9]. Not only must the surrogate show a
`close correlation with the clinical variable in a stable
`environment, but to be useful as a guide to making decisions
`about drugs it must also be sensitive to interventions. This
`aspect of validation is also referred to as criterion validity [10]
`and is established by previous clinical data. The more difficult
`aspect of validation relates to the information required to
`support the assumption (required in the definition) that the
`surrogate shares a causal mechanism with an ultimate clinical
`outcome i.e. it has construct validity [10]. Although a good
`statistical correlation between the surrogate and the clinical
`outcome is necessary for this assumption, it is not sufficient
`as the two may have a common root cause but not share a
`common mechanism. In addition, there must be a plausible
`mechanistic connection between the two and an adequate
`quantity of experimental data to demonstrate that changes in
`the surrogate quantitatively predict changes in the clinical
`outcome after several interventions of different types. A good
`illustration of the need for the latter point comes from the
`results of recent studies of the effect of dietary supplemen-
`tation with vitamin A to reduce the risk of cancer. The
`studies confirmed the inverse relationship between plasma
`vitamin A levels and cancer risk, but disappointingly showed
`that increasing plasma vitamin A levels by supplementation
`was associated with an increased, not decreased, incidence of
`cancer [11]. Hence validation of a potential surrogate is
`usually not a binomial variable (valid/invalid) but
`is a
`continuous variable, with surrogates of varying validity.
`Tables 1–3, respectively, list examples of surrogates which
`are well-validated; examples which are of incomplete validity
`but which are considered useful; and examples of invalidated
`surrogates. At the beginning of validation of a potential
`surrogate when little data exists, an estimate can be made on
`a theoretical basis. As new experimental data become available
`this probability estimate is progressively updated (Figure 1).
`
`Innovation
`
`in Table 1 have been sufficiently well
`The surrogates
`validated that one can have high confidence in designing a
`therapeutic dosage regimen from the data using these
`surrogates alone. However, developing a new compound in
`
`Drug/class
`
`Antihypertensive
`Glaucoma
`H1-receptor antagonist
`
`H2-receptor antagonist
`
`220
`
`Clinical endpoint
`
`Cardiovascular disease
`Visual acuity
`Allergic rhinitis, dermatitis
`
`Peptic ulcer
`
`Surrogate
`
`Blood pressure [6]
`Intraocular pressure
`Suppression of histamine-induced
`cutaneous weal and flare [33]
`24 h suppression of gastric acid [13]
`
`© 1997 Blackwell Science Ltd Br J Clin Pharmacol, 44, 219–225
`
`

`

`The contribution of clinical pharmacology surrogates and models to drug development—a critical appraisal
`
`Table 2 Examples of incompletely-validated surrogates and models which may be useful in drug development.
`
`Drug/class
`
`5HT3-receptor antagonists
`Lipoxygenase inhibitors
`
`Leukotriene antagonists
`MAO-B inhibitor antiparkinsonian
`
`Anxiety
`Panic disorder
`Dementia
`
`Antiinfectives
`Analgesic
`
`Clinical endpoint
`
`Chemotherapy emesis
`Asthma
`
`Asthma
`Depression
`
`Symptoms
`Symptoms
`Social function
`
`Infection
`Pain
`
`Surrogate/Model
`
`Ipecacuanha emesis [34]
`Lipoxygenase inhibition in neutrophils [12];
`allergen challenge [12]
`Bronchial challenge with leukotriene [35]
`MAO-B inhibition in platelets [24];
`MAO-B receptor occupancy in vivo [24]
`Conditioned aversive anxiety [25]
`Simulated public speaking [25]
`Psychometric tests [36]; scopolamine
`model in volunteers [37]
`Infective challenge, e.g. malaria [38]
`Pain models [39]
`
`Table 3 Examples of invalidated
`surrogates.
`
`Drug/class
`
`Clinical endpoint
`
`Surrogate
`
`Type 1 antiarrhythmics
`Vitamin A supplements
`
`Cardiac arrest
`Cancer
`
`Supression of ventricular ectopic beats [40]
`Plasma vitamin A levels [11]
`
`(cid:20)(cid:19)(cid:19)(cid:8)
`
`(cid:57)(cid:68)(cid:79)(cid:76)(cid:71)(cid:68)(cid:87)(cid:76)(cid:82)(cid:81)
`
`(cid:19)(cid:8)
`
`(cid:36)(cid:80)(cid:82)(cid:88)(cid:81)(cid:87)(cid:3)(cid:82)(cid:73)(cid:3)(cid:71)(cid:68)(cid:87)(cid:68)(cid:3)(cid:82)(cid:81)(cid:3)(cid:86)(cid:88)(cid:85)(cid:85)(cid:82)(cid:74)(cid:68)(cid:87)(cid:72)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:82)(cid:88)(cid:87)(cid:70)(cid:82)(cid:80)(cid:72)
`
`Figure 1 Schematic representation of the relationship between
`the amount of clinical data in support of a potential surrogate and
`an estimate of its validity.
`
`one of these classes is most unexciting as there are already
`more than adequate numbers of therapeutic alternatives.
`This brings us to another dimension of
`surrogates—the
`degree of innovation. In general, the degree of innovation
`varies inversely to validity because of the lack of prior data,
`by definition. This ‘Catch-22’ situation, where the least
`well-validated surrogates and models exist for diseases with
`the greatest unmet clinical need and hence highest develop-
`ment risk, is the single greatest point limiting the utility of
`surrogates and models in exploratory development of novel
`therapies.
`
`Proximity
`
`In a chain of events from drug-receptor interaction to
`therapeutic response, there may be many events, each of
`which may be measurable as a potential surrogate. Some
`surrogates, however, will be closer, pathophysiologically, to
`the desired endpoint and hence can be regarded as more
`proximate. An example of such a chain of surrogates can be
`seen with inhibitors of 5-lipoxygenase. The initial, but distal
`surrogate (from the clinical endpoint) could be inhibition
`of 5-lipoxygenase in peripheral blood neutrophils, a measure
`of the intended pharmacological effect. A more proximal
`
`step could be inhibition of allergen challenge [12]. Both are
`potential surrogates but differ in proximity to the clinical
`endpoint, i.e. change in spirometry and symptoms.
`Plasma (or even tissue) drug concentrations could be
`regarded as distal surrogates and this concept underlies the
`utility of therapeutic drug monitoring.
`The major use of a distal surrogate is confirmation of
`pharmacological activity in man, and establishing a dose-
`and concentration-response relationship. It is unlikely that a
`drug could be registered on such data, but the utility is in
`making a decision during the drug development process.
`Such distal
`surrogates may be of particular use when
`evaluating follow-up compounds
`in a series, where the
`forerunner has validated this early marker with the ultimate
`clinical effect.
`A well-validated proximal surrogate could be used as an
`outcome measure for drug registration—examples include
`CD4 count
`for HIV infection, plasma cholesterol
`for
`cardiovascular disease due to hypercholesterolaemia, and
`some of the surrogates listed in Table 1.
`
`Specificity
`
`A surrogate may be well-validated for one type of
`intervention but not to be useful for another. For example,
`as mentioned above, 24 h gastric acid suppression is a well-
`validated surrogate for the healing effect of H2-receptor
`antagonists on peptic ulcers [13], but would probably be
`unhelpful for predicting the effect of antimicrobial therapy
`directed at H. pylori for the same disease. In a similar way
`to the usual
`inverse relationship between validation and
`novelty, proximity and specificity may represent the same
`axis.
`
`Practicality
`
`Although a potential surrogate might seem appropriate in
`terms of
`the conceptual dimensions
`listed above,
`its
`
`© 1997 Blackwell Science Ltd Br J Clin Pharmacol, 44, 219–225
`
`221
`
`

`

`P. Rolan
`
`robustness and ease of use will be additional
`determining its utility.
`A diagrammatic representation of the five dimensions of
`surrogates is presented in Figure 2.
`
`factors in
`
`The use of surrogates and models in decision making
`in early drug development
`
`As discussed above, a well-validated surrogate or model can
`substantially shorten clinical development time or time to
`reach a critical decision point in exploratory development.
`Does this mean that only well-validated surrogates are useful
`or
`is there a role for imperfectly validated markers of
`drug effect?
`Decision making in drug development, clinical medicine
`or other scientific disciplines is rarely based on a single piece
`of data. We are required in clinical medicine and drug
`development to make decisions based on several pieces of
`data, often of
`imperfectly understood predictive power
`and sometimes conflicting. Decision analysis formalises the
`process of making rational decisions on such data sets,
`starting with an initial probability estimate and revising this
`in light of subsequent data, taking into account the variability
`of the new data and the uncertainty about its predictive
`power [14]. Surrogates and models for which the predictive
`power is incompletely understood may be similarly useful in
`decision making, but often a larger amount of data is
`required to offset the uncertainty in predictive ability. I will
`illustrate this with some examples.
`
`Example 1. Use of existing models with a novel compound
`
`interested in examining
`The Wellcome Foundation was
`whether inhibition of non-adrenergic, non-cholinergic nerve
`activity in the lung by a peripherally-acting m-opiate agonist,
`443C81, with inhibition of release of inflammatory neuro-
`peptides, would be a useful therapeutic strategy in asthma.
`No well-validated method for assessing such activity existed
`to our knowledge, and we assumed that an adequately-
`powered clinical trial of at least 1 month duration would be
`required to test
`this hypothesis. During the time while
`
`(cid:54)(cid:83)(cid:72)(cid:70)(cid:76)(cid:73)(cid:76)(cid:70)
`(cid:51)(cid:85)(cid:82)(cid:91)(cid:76)(cid:80)(cid:68)(cid:79)
`
`(cid:51)(cid:85)(cid:68)(cid:70)(cid:87)(cid:76)(cid:70)(cid:68)(cid:79)
`
`(cid:51)(cid:82)(cid:82)(cid:85)(cid:79)(cid:92)(cid:3)(cid:89)(cid:68)(cid:79)(cid:76)(cid:71)(cid:68)(cid:87)(cid:72)(cid:71)
`(cid:44)(cid:81)(cid:81)(cid:82)(cid:89)(cid:68)(cid:87)(cid:76)(cid:89)(cid:72)
`
`(cid:44)(cid:80)(cid:83)(cid:85)(cid:68)(cid:70)(cid:87)(cid:76)(cid:70)(cid:68)(cid:79)
`
`chronic toxicity studies were performed to support such a
`study, we embarked upon a series of studies attempting to
`demonstrate an effect of
`the compound in bronchial
`challenge models. The drug was given intravenously and by
`nebuliser, to normals and asthmatics, using the bronchial
`challenge stimuli of cold dry air and inhaled metabisulphite
`[15–18]. All studies showed no effect of the drug. Taken
`separately, no study could be regarded as pivotal, but
`together the results undermined confidence in the com-
`pound. A small clinical trial showed no trend for benefit and
`further development was stopped. This example shows that
`imperfectly validated surrogates and models can be useful in
`decision making in drug development.
`
`Example 2. Development of a new surrogate
`
`In the previous example, existing models were used to assess
`the action of a novel compound. Frequently, no existing
`model or surrogate seems appropriate and a new surrogate
`or model needs to be developed. Such an example comes
`from work at the Wellcome Research Laboratories on a
`potential prophylactic treatment
`for
`sickle cell disease,
`tucaresol. This drug was a product of rational drug design.
`It was
`intended to left-shift
`the haemoglobin-oxygen
`saturation curve, and hence reduce the proportion in
`insoluble deoxy sickle haemoglobin at tissue oxygen tensions.
`A novel surrogate, the proportion of haemoglobin molecules
`modified to the high-affinity form, or
`‘%MOD’ was
`developed to guide exploratory development
`[19]. This
`surrogate was soundly based on a theoretical view with the
`expectation from two independent sources that between
`15–30% MOD would be likely to inhibit
`the clinical
`consequences of sickling. The surrogate was used in all the
`preclinical pharmacology and toxicity studies. Due to
`markedly different pharmacokinetics in the animal species
`used,
`the dose-response relations were not consistent
`between species but the relationships between %MOD and
`pharmacology and toxicity were consistent
`[20]. This
`substantially facilitated the design of the first human studies
`[21, 22]. We were also able to make accurate predictions
`about
`the likely clinical dose based on the use of
`this
`surrogate, as in sickle cell disease patients, %MOD levels
`predicted to be likely to be therapeutic were associated with
`substantial reductions in several haematological
`indices of
`haemolysis [23]. These tests however are also surrogates, but
`more closely related to the basic pathology and hence can
`be regarded as more proximate to the disease than %MOD.
`The success of this approach due to the planning ahead of
`the preclinical pharmacologists, by developing a method for
`measuring the effects of this novel compound which could
`be used in all test species including man.
`
`(cid:57)(cid:68)(cid:79)(cid:76)(cid:71)(cid:68)(cid:87)(cid:72)(cid:71)
`(cid:49)(cid:82)(cid:87)(cid:3)(cid:76)(cid:81)(cid:81)(cid:82)(cid:89)(cid:68)(cid:87)(cid:76)(cid:89)(cid:72)
`
`(cid:49)(cid:82)(cid:81)(cid:86)(cid:83)(cid:72)(cid:70)(cid:76)(cid:73)(cid:76)(cid:70)
`(cid:39)(cid:76)(cid:86)(cid:87)(cid:68)(cid:79)
`
`Example 3. Missed opportunity
`
`Figure 2 The dimensions of surrogates plotted as three
`orthogonal axes. Due to the likely inverse relationship between
`validity and innovation, these have been collapsed to one axis.
`Similarly, because of the likely covariance of specificity and
`proximity, these have also been collapsed to one axis. Any
`potential surrogate can be plotted in this space to estimate its
`likely utility.
`
`When no well validated surrogate exists for a given type of
`compound, the time to explore the development of a new
`surrogate is during preclinical development, not at the time
`of first administration to man. In the example of %MOD
`given above,
`the use of
`this
`surrogate linked all
`the
`preclinical pharmacology and toxicology with the design
`and interpretation of the early human results. Another more
`
`222
`
`© 1997 Blackwell Science Ltd Br J Clin Pharmacol, 44, 219–225
`
`

`

`The contribution of clinical pharmacology surrogates and models to drug development—a critical appraisal
`
`complex example is illustrated with a study by Bench et al.
`[24] describing a study in the development of an MAO-B
`inhibitor. This study was performed because early human
`studies had indicated that the maximum well-tolerated dose
`was below that predicted on a body weight basis from the
`animal data. The difficult decision for the company is
`whether to abandon a compound which might be useful or
`to proceed to major clinical trials with a dose which may
`be subtherapeutic. The paper describes an elegant PET study
`which showed that receptor occupancy likely to be associated
`with therapeutic activity was obtained with well tolerated
`doses. The real point of the paper, though, was the excellent
`correlation between easily measurable receptor occupancy
`on platelets and cerebral receptor occupancy. A systematic
`investigation of this possible relationship in the preclinical
`studies may have avoided the dilemma using the maximum
`well-tolerated dose approach and could have avoided the
`loss of significant clinical development time.
`
`Continuing validation during development
`Once a compound has moved in to clinical trials of efficacy,
`should the imperfectly validated surrogate be abandoned? It
`is rare for only one compound of a therapeutic class to be
`in development. By continuing to use the surrogate in the
`efficacy trials, retrospective validation of the surrogate may
`be possible. This could lead to major time saving in the
`development of subsequent compounds, with perhaps the
`ability to dose range over a narrower interval. Hence
`validation is a continuous cycle, not a one-off process. This
`cycle of continuous development is schematically presented
`in Figure 3.
`
`The therapeutic effect is difficult to measure
`trials are difficult and
`For many CNS disorders clinical
`sometimes apparently adequately powered clinical trials may
`fail to show the known effect of a drug. When developing
`truly innovative therapies for such conditions, a model, even
`though imperfectly validated, can be used to support the
`decision to commit to a major clinical trial program and to
`assist with dose selection. Examples of such models are those
`for generalised anxiety disorder [25] and panic disorder [25].
`For other conditions, there is no single reliable measure of
`efficacy and surrogates may be helpful in measuring disease
`activity, e.g. rheumatoid arthritis [26] and HIV infection
`[27, 28].
`
`There is a long delay between drug exposure and effect
`
`to confirm pharmacological activity may be
`Surrogates
`useful when developing drugs given for prophylaxis of
`uncommon events, e.g. seizures, cardiac arrest.
`
`Novel proposed therapeutic action
`
`Many novel drugs act on a pathway for which the role in
`disease is not well understood. Without a surrogate or model
`to confirm pharmacological activity, clinical failure could be
`attributed either to insufficient drug activity at
`the site
`(where a higher dose, more frequent dosing, alternative
`route of administration or more potent analogue might be
`effective) from the proposed mechanism not being relevant
`to the disease. For example, the demonstration that PAF
`antagonists blocked exogenous PAF challenge but were
`ineffective in asthma showed that the development of more
`potent antagonists was unlikely to be clinically useful
`[29–31].
`
`In which therapeutic areas do we need good
`surrogates and models?
`
`The clinical trial needs large sample size
`
`A surrogate or model is likely to be of the greatest use in
`the following situations:
`
`large groups sizes may be
`In conditions such as stroke,
`required to demonstrate efficacy. In such conditions a
`
`(cid:54)(cid:72)(cid:79)(cid:72)(cid:70)(cid:87)
`
`(cid:86)(cid:88)(cid:85)(cid:85)(cid:82)(cid:74)(cid:68)(cid:87)(cid:72)
`
`(cid:40)(cid:86)(cid:87)(cid:76)(cid:80)(cid:68)(cid:87)(cid:72)
`(cid:89)(cid:68)(cid:79)(cid:76)(cid:71)(cid:76)(cid:87)(cid:92)
`
`(cid:50)(cid:69)(cid:87)(cid:68)(cid:76)(cid:81)
`(cid:71)(cid:68)(cid:87)(cid:68)(cid:3)(cid:82)(cid:81)
`(cid:86)(cid:88)(cid:85)(cid:85)(cid:82)(cid:74)(cid:68)(cid:87)(cid:72)
`
`(cid:50)(cid:69)(cid:87)(cid:68)(cid:76)(cid:81)
`(cid:70)(cid:79)(cid:76)(cid:81)(cid:76)(cid:70)(cid:68)(cid:79)
`(cid:71)(cid:68)(cid:87)(cid:68)
`
`Figure 3 Schematic representation of a continuous cycle of review of the validity of a surrogate in light of new clinical data.
`
`© 1997 Blackwell Science Ltd Br J Clin Pharmacol, 44, 219–225
`
`223
`
`

`

`P. Rolan
`
`well-validated proximal surrogate could substantially reduce
`sample size and trial duration. For example, Wittes et al.
`[32] demonstrated that using coronary artery patency as a
`surrogate for the effect of thromolytic therapy, sample size
`is reduced to 200 compared with 4000 where death is the
`clinical endpoint, and that a result for a patient is determined
`within 90 min rather than at 5 years.
`In other conditions where a large sample size is required
`(e.g. stroke) but where a novel mechanism of action is used
`(e.g. NMDA antagonists), a distal surrogate, measuring the
`pharmacological effect of the compound would be valuable
`in deciding whether to proceed to full development and
`perhaps to assist with dose selection.
`
`The role of the academic and industrial clinical
`pharmacologist in development of surrogates
`
`In therapeutic areas of greatest clinical need where there are
`few or no existing therapies it is most likely that the least
`well-validated surrogates exist. Development of a surrogate
`may come from advances in the understanding of basic
`mechanisms of a disease but such surrogates need to be
`prospectively evaluated by people experienced in designing
`and testing methods of measuring drug action in man.
`Hence academic clinical pharmacologists should be able to
`play a major role is devising ways to measure the effects of
`truly novel compounds. The industrial clinical pharmacol-
`ogist may also undertake this role. However it is essential
`that this development process starts during the exploratory
`preclinical development of
`the compound, with close
`interaction between the biochemists, preclinical and clinical
`pharmacologist and continual updating the estimate of
`the validity of
`the surrogate during later phase clinical
`development.
`The increasing pressure for rapid decision making in
`exploratory clinical development gives clinical pharmacol-
`ogists a major opportunity to assume a pivotal role in drug
`development. I believe that taking responsibility for the
`development and validation of new surrogates and models
`represents
`the greatest opportunity for our profession
`to thrive.
`
`References
`
`1 Toon S. Relevance of pharmacokinetic studies in volunteers.
`Clin Pharmacokinet 1993; 25: 259–262.
`2 Cutler N, Mushet GR, Davis R, et al. Oral sumatriptan for
`the acute treatment of migraine: evaluation of three dosage
`strengths. Neurol 1995; 45 Suppl 7: S5–9.
`3 Posner J. Exploratory development. In The Textbook of
`Pharmaceutical Medicine, Second Edition, eds Griffin JP,
`O’Grady J, Wells FO. Belfast: Queens University of Belfast,
`1994: 291–314.
`4 Lipicky RJ, Packer M. Role of surrogate end points in the
`evaluation of drugs for heart failure. J Am Coll Cardiol 1993;
`22 (Suppl A): 179A–184A.
`5 Rolan PE. General characteristics of clinical surrogates. In
`Advances in Headache Research Vol 5: Experimental headache
`models, eds Olesen J, Moskowitz M, New York: Lippincott-
`Raven, 1995: 27–32.
`6 Wittes J, Lakatos E, Probstfield J. Surrogate endpoints in
`clinical trials: cardiovascular diseases. Stat Med 1989; 8:
`415–425.
`
`7 Fagan TC. Calcium antagonists and mortality. Another case of
`the need for clinical judgement. Arch Int Med 1995; 155: 20.
`8 Prentice R. Surrogate endpoints in clinical trials: definition
`and operational criteria. Stat Med 1989; 8: 431–440.
`9 Boissel J-P, Collet J-P, Moleur P, et al. Surrogate endpoints: a
`basis for a rational approach. Eur J Clin Pharmacol 1992; 43:
`235–244.
`10 Sim J, Arnell P. Measurement validity in physical therapy
`research. Phys Ther 1993; 73: 102–115.
`11 Marwick C. Trials reveal no benefit, possible harm of beta
`carotene and vitamin A for lung cancer. JAMA 1996; 275: 6.
`12 Nasser SM, Bell GS, Hawksworth RJ, et al. Efficacy of the
`5-lipoxygenase inhibitor ZD 2138 on allergen-induced early
`and late responses. Thorax 1994; 49: 743–748.
`13 Howden CW, Burget DW, Hunt RH. Appropriate acid
`suppression for optimal healing of duodenal ulcer and gastro-
`oesophageal reflux disease. Scan J Gastroenterol— Supplement
`1994; 201: 79–82.
`14 Cantor SB. Decision analysis: theory and application to
`medicine. Prim Care 1995; 22: 261–270.
`15 Gray SJ, Rolan PE, Posner J. Effect of i.v. BW443C in cold
`air-induced broncoconstriction in healthy volunteers. Abstract
`for 4th World Conference on Clinical Pharmacology and
`Therapeutics, July 1989.
`16 Gray SJ, Rolan PE, Phillips SC, Posner J. Acute tolerance,
`effects on spirometry and systemic absorption of nebulised
`443C81 in healthy volunteers and asthmatic patients. Abstract
`for British Thoracic Society, July 1990.
`17 Gray SJ, Rolan PE, Posner J. Effect of inhalation of a novel
`enkephalin analogue (443C81) on the response to cold air and
`metabisulphite challenge in healthy volunteers. Br J Clin
`Pharmacol 1991; 32: 662–663.
`18 Phillips SC, Rolan PE, Gray SJ, Posner J. Effect of i.v.
`infusion of a novel enkephalin analogue (443C81) on
`metabisulphite induced broncoconstriction and intradermal
`histamine flare in healthy volunteers. Br J Clin Pharmacol
`1991; 32: 663.
`19 Wootton R. Compounds designed to bind to haemoglobin.
`In The design of drugs to macromolecular targets, ed Beddell CR,
`John Wiley and Sons Ltd, London, 1992.
`20 Rolan PE. The exploratory clinical development of tucaresol, an
`antisickling agent, using a novel surrogate marker. MD thesis,
`University of Adelaide, Adelaide, 1995.
`21 Rolan PE, Parker JE, Gray