safer medicines report · such studies may support the safe development of novel therapies that may...

1

Safety Pharmacology working group report

Safer Medicines Report


November 2005

Safety Pharmacology-02.qxd 11/4/05 20:38

2


Contents

Page

Summary of key findings and recommendation 3

Chapter one: Introduction 5

Chapter two: General/secondary pharmacology studies 8

Chapter three: Cardiovascular and renal safety pharmacology 10

Chapter four: Respiratory safety pharmacology 12

Chapter five: Central nervous system pharmacology 14

Chapter six: Hepatic function 16

Chapter seven: Gastrointestinal safety pharmacology 18

Chapter eight Sensory safety pharmacology 20

Chapter nine: Immuno-safety pharmacology 22

Chapter ten: Paediatric indications 24

Chapter eleven: Biotechnology compounds 26

Chapter twelve: Compounds selective for human targets 28

Chapter thirteen: Education and training 29

References 32


3


A review of human safety information concluded thefirst administration to man of NCEs is very safe.

This suggests that the current paradigm for safetytesting for untoward effects related to the major lifesustaining physiological systems (respiratory,cardiovascular and major CNS effects) successfullyprevents compounds being administered to man atconcentrations that may cause these effects. Theestablishment of dose response in these models clearlyfits with the successful identification of hazard andcareful risk assessment prior to early clinicalevaluation.

Ninety-five medicines withdrawn from the US marketbetween 1960 and 1999 were reviewed to try tounderstand the main causes for ADR-related drugwithdrawals. The principal interesting clusters ofreasons were: arrhythmias – 12%; neuropsychiatriceffects / abuse liability / dependency – 12%;hepatotoxicity – 9%; bone marrow toxicity- 7%;allergy- 6%.

Understanding proarrhythmia:

Considerable emphasis has been placed on studyingdrug effects on the QT interval. However, the real riskto human safety is the risk of causing Torsade dePointes (TdeP), for which QT prolongation is abiomarker. The relationship between QTprolongation and TdeP is poorly defined and shouldbe the subject of further non-clinical and clinicalresearch. Such studies may support the safedevelopment of novel therapies that may otherwise bediscarded during non-clinical testing.

Abuse potential

The potential for centrally active drugs to have abusepotential and drug withdrawal reactions is an area ofincreasing regulatory interest. Although non-humanprimates are frequently used for these studies, literaturedata suggest that the rat has similar predictive value toman. A systematic comparison between the rat andnon-human primate is required to identify thepreferred species to predict this liability in man.

CNS pharmacology

Neuropsychiatric adverse events (such as somnolence,dizziness and asthenia) have been shown to be dose-limiting in a number of Phase I/II clinical trials as wellas accounting for 12% of drug withdrawals from themarket. In addition to these effects there is increasingconcern related to the potential for CNS active drugsto cause adverse affective disorders (anxiogenic effectsand suicidal tendencies). Animal models are used todetect drugs with the potential to have anxiolytic andantidepressant activity, although the value of thesemodels to detect untoward affective effects isunknown. This should be an area of further research todevelop animal models that can be used to test foranxiogenic/depressant activity.

Sensory function

The incidence of adverse drug reactions on vision andhearing is relatively low compared with the incidenceof headache, nausea & vomiting, diarrhoea anddizziness, although the consequence to volunteers andpatients may be greater. However, it is probable thatsymptoms (changes in vision and tinnitus) would bedetected prior to the risk of long-term damage.Nevertheless, investments should be made to increasethe understanding between drug effects on visualfunction in animals and man to further refine non-clinical testing paradigms. Furthermore, studiesshould be undertaken to develop predictive in vitrotests of ototoxicity.

Immune mediated adverse events of lowfrequency

There is confidence that currently available non-clinical models are able to detect immunosuppressionand impaired host resistance despite the validation ofthese methods being incomplete. Methods to identifyallergenicity have been widely used and for contactsensitization several approaches are validated. Thereremains a need to gain full acceptance for the methodsavailable to detect respiratory and gastrointestinalsensitisers. Approaches for the identification ofautoimmune responses or idiosyncratic immune

Summary of key findings and recommendation


4


response have proven to be elusive and there is a needfor more intensive study. This should includeconsideration of genetic susceptibility factors andenvironmental factors (diet, overall health status, lifestyle etc). Furthermore, it is pertinent to ensure that allinformation relevant to a consideration of immunefunction from conventional toxicity studies, and anappreciation of cytokine expression patterns areincluded in an overall evaluation of impacts onimmune status.

Training and education for safetypharmacology and clinical pharmacology

Formal training in life sciences within the UK has beensteadily declining over the past 10 years and inparticular under graduate and post graduate traininginvolving the use of animals, in particular in vivo, inresearch has seem a marked decline. A clear need toinvest in formal training within the UK exists. Thenumber of pharmacology departments providingundergraduate training must not be allowed to reducefurther. Post-graduate training needs to be supportedmore consistently with funds available to be targetedspecifically at pharmacology and toxicology thusallowing development of safety pharmacologists withthe appropriate experience. Initiatives within industrytogether with the BPS and BTS are welcomed but areinsufficient to meet the future needs for industry,academia and often overlooked the regulatory bodieswho will face dossier submissions containing data withincreased complexity. Continual professionaldevelopment and a formal accreditation system tosupport the development and maintenance ofindividuals involved in the non-clinical riskassessment of new medicines should be considered.

First in human paradigms

Early and more efficient evaluation of new moleculestogether with more rapid validation of novel targets inhumans will lead to a major advance in the

identification of more efficacious medicines. Thecurrent requirements for non-clinical studies prior tofirst dose in man needs to be re-evaluated with theobjective of permitting limited, highly controlledhuman studies to be conducted safely. Experiencewith Safety Pharmacology studies together withconventional toxicology studies has given reassurancethat highly toxic compounds are readily detected.Furthermore, safety pharmacology studies have greatpotential to support first in human studies using ‘novel’non-clinical safety paradigms.

Improving non-clinical testing paradigmsthrough data sharing

Developing better non-clinical testing paradigms is inthe interest of the patients, doctors, regulatoryauthorities and the pharmaceutical industry.Experience with groups such as ILSI/HESI hasshown that considerable advancements in ourknowledge of the predictivity of non-clinical tests canbe made through the sharing of anonymous data. Suchactivities can effectively review current test paradigms,make recommendations for change and then assessthe value of these changes stimulating a cycle ofpositive change.

In order to achieve the above proposals werecommend the creation of a UK centre,perhaps virtual, for development of DrugSafety Sciences. Such a centre wouldfacilitate the sharing of data withoutcompromise to commercial interest andenable structured assessment of theconcordance between safety pharmacologystudies in animal models against clinicaloutcome to be prospectively assessed andimproved. Such a centre would also providethe academic focus to support the training ofnon-clinical and clinical pharmacologistsand toxicologists.


5


1.1 Studies conducted to investigate thepharmacology of new chemical entities (NCEs)are frequently classified as:

� primary pharmacodynamics (studies toinvestigate the designed mode of actionexpected to provide the desired clinicalbenefit);

� secondary pharmacodynamics (studiesdesigned to explore the broaderpharmacology of a compound e.g. actions notexpected from its primary mode of action thatmay arise from additional actions of thecompound);

� safety pharmacology (studies designed toinvestigate the potential undesirablepharmacodynamic effects of a substance onphysiological functions in relation to exposureswithin the therapeutic range and above).

1.2 Adverse events are defined as any untowardmedical occurrence in a patient or clinicalinvestigation subject administered apharmaceutical product and which does notnecessarily have a causal relationship with thistreatment. Adverse Drug reactions are a subsetof Adverse Events that are thought to be causallyrelated to the use of a medicine. Adverse Eventsand Adverse Drug Reactions can further bedefined as seriousness criteria or by severityratings (such as severe, moderate or mild).Serious Adverse Events are those that s result indeath, require hospitalisation or extension ofhospital stay, are life-threatening, result inpersistent or significant disability or are a causeof congenital abnormality. In general, severeadverse events prevent regular activities and arenot relieved with symptomatic treatment.Moderate ADR’s are bothersome, interfere withactivities and are only partially relieved withsymptomatic treatment with mild ADR’s beingslightly bothersome and are relieved withsymptomatic treatment.

1.3 Non-clinical Safety Pharmacology testing (incombination with other non-clinical sciences) isdirected towards prevention of serious ADR’s inearly clinical testing. This is approached by

defining the concentration/dose-responserelationship for effects in non-clinical modelsthat are used to predict the likelypharmacologically mediated adverse effectsassociated with either the pharmacology of theprimary therapeutic target or with a secondarytarget that happens to be a property of the drug.Information from safety pharmacology studiesis used to guide the starting dose and set likelystopping criteria in initial clinical studies andalso provides guidance on potential adverseevents for which monitoring in further clinicaltrials is appropriate. Safety Pharmacology alsoplays a role in understanding drug toxicitiesidentified in humans during clinical evaluation.

1.4 The design and conduct of non-clinical safetypharmacology studies has been defined in ICHS7A (ICH S7A Safety Pharmacology Studies forHuman Pharmaceuticals, CPMP/ICH/539/00)that was introduced in 2001. Thus, prior to thefirst administration to humans, core battery(cardiovascular, central nervous and respiratorysystem assessment) studies will be completed toinvestigate the safety pharmacology of the NCE.A further non-clinical guidance document onthe conduct of Safety Pharmacology Studies forAssessing the Potential for Delayed VentricularRepolarisation (QT Interval Prolongation) byHuman Pharmaceuticals (ICHS7B) has alsorecently been finalised. Understanding thesafety pharmacology associated with a newdisease target has significant potential to reducefailure in development. Furthermore the designof Safety Pharmacology packages should beinfluenced by an understanding of the target, itsdistribution and function and not simplythrough fulfillment of ICH S7A. Mechanisticunderstanding will lead to better understandingof risk/benefit in man.

1.5 A literature review (Sibille et al., 1998) and ananalysis of Phase I clinical trials from 4 majorpharmaceutical companies (the working groupwas blinded to the therapeutic class underevaluation) has demonstrated that the FIH trialsare very safe with few, if any, serious ADRs beingreported. This suggests that safety testing for

Chapter one - Introduction


untoward effects related to the major lifesustaining physiological systems (respiration,cardiovascular and major CNS effects e.g.convulsions) are successfully preventingcompounds being administered to humans atdoses that may cause these effects. It should alsobe noted that these initial clinical studies areconducted extremely diligently with intensivemonitoring for the emergence of potentialADR’s. Such careful entry into man hasprevented serious ADR’s occurring frequentlyeven in drug classes commonly associated withhigh toxicity e.g. cytotoxic anticancer agents.Adverse events do occur in these early studiesbut are generally more frequently related to theprocedure rather than the drug under study.

1.6 The above analysis from 4 pharmaceuticalcompanies revealed with remarkableconsistency that common ADRs (10–30% ofvolunteers) were: headache, nausea &vomiting, diarrhoea and dizziness. A largenumber of other lower frequency ADRs wasreported – these tended to be specific to themolecule under investigation and often relatedto the mode of action. A review of the dose-limiting toxicities for 25 randomly selectedmolecules tested in single and multiple doses inhealthy volunteers revealed that sevenmolecules had no dose-limiting ADRs.Interestingly, nausea and vomiting, diarrhoeaand neuropsychiatric effects were dose-limitingin four cases but these ADR’s were notreadily identified from the non-clinicalsafety pharmacology data. Conversely, safetypharmacology testing clearly predicted effectsupon the QT interval of the electrocardiogram,blood pressure effects, respiratory effects anddiuresis. These findings are in agreement withthe published literature that single dose non-clinical safety studies both under and overpredict the effects seen in man (Olson et al2000; Greaves et al 2004). Effects thatfrequently emerge during Phase 1 clinicalevaluation and appear to be poorly predictedor extrapolated from the animal studies andwarrant further development of predictive testsappear to be in the areas of GI disturbance andneuropsychiatric effects.

1.7 Approximately one in six drugs in clinicaldevelopment is discontinued for clinical safety

6


reasons (Centre for Medicines Research 2003).Non-serious ADR’s are often mechanism, drugclass or disease related and may limit the utilityof the medicine but usually do not pose asignificant safety issue. Serious ADR’s in Phase2b or Phase 3 tend to occur at low frequencyand may be related to the pharmacology or tothe chemical properties of the drug. The formershould be predictable from safetypharmacology assessment whereas the lattermay induce toxicity via direct chemical toxicityor via hypersensitivity or immunologicalmechanisms. The impact of such toxicity isalways very high requiring warnings,precautions and contraindications or evenwithdrawal from development or if detectedpost approval drug withdrawal from marketing.

1.8 The actual incidence of serious ADR’sproduced by marketed medicines is difficult toestimate accurately. ADR’s are thought to beresponsible for 4–6% of hospital admissions(Stephens 2004) with up to 106,000 deaths peryear in the USA attributed to ADR’s (Lazaraouet al 1998). The latter figure may be an overestimate but there is no doubt that ADR’s causesignificant patient harm and are a significantburden to health care. Although the frequencymay be very low the establishment of theconnection between pharmacologicalmechanism and the event may yieldopportunities to reduce risk. For example therelationship between hERG potassium channelblockade, prolongation of the QT interval andinduction of Torsade de Pointes (TdeP) has ledto the rapid development of non-clinicalscreening tests that will help identify moleculeswith a reduced risk for this ADR.

1.9 Ninety-five medicines withdrawn from the USmarket were reviewed to try to understand themain causes for ADR-related drug withdrawals.The principal interesting clusters of reasonswere: arrhythmias – 12%; neuropsychiatriceffects / abuse liability / dependency – 12%;hepatotoxicity – 9%; bone marrow toxicity- 7%; allergy- 6% (Stephens 2004). An earlierreview of 121 medicines withdrawn from theworldwide market between 1960 and 1999showed a similar spectrum with hepatic toxicity26%, haematological 10%, cardiovascular 9%,skin effects 6% and carcinogenicity issues 6%.


7


pharmacology. Is the compound bioavailableand tolerated by the species underconsideration? This may influence the choice ofroute of administration and therefore, the use ofeither conscious or anaesthetised animals.

1.12 Advances in technology and in particular theminiaturisation of telemetric devices will giveopportunities to improve the non-clinicalmodels and potentially positively influence 3Rinitiatives. There is a need to agree internationalacceptance criteria for the validation of new orimproved non-clinical models. In addition, withthe continuous recording of physiologicalparameters and concomitant measurements ofdrug plasma levels, these new technologies have great potential to be used to develop a better understanding of the pharmacokinetic:pharmacodynamic relationship in safetypharmacology studies. Such an approach may better inform clinical colleagues of the likely adverse effect concentration in man and potentially improve our confidence inthe safety margins defined in non-clinicalstudies.

1.13 The changing demographics of the populationwith increasing numbers of elderly peoplerequiring multiple-drug therapy is increasingthe concerns around polypharmacy. There is aneed to develop better testing paradigms.

1.10 With the above background it is clear thatnon–clinical safety assessment and safetypharmacology in particular is providinginformation that is preventing harmfulcandidate drugs entering clinical development,is identifying some key and important adverseevents but is clearly failing to predict all adverseevents. In considering areas for future researchit is clear that much is effort has been devoted tounderstanding drug induced QT intervalprolongation. There clearly remains a need tounderstand drug induced pro-arrhythmia andits true relationship to altered repolarisation.Hepatotoxicity, bone marrow toxicity andimmune mediated toxicity and GI toxicity areareas warranting further development. It isstriking that neuropsychiatric effects account fora large group of drug withdrawals and they arealso prominent as dose limiting toxicities inearly volunteer patient studies that are not wellpredicted from the non-clinical models.

1.11 The study design and choice of species to beused in safety pharmacology studies needs totake into account a number of factors. Firstly,the known pharmacology of the compound, isthe target under investigation expressed in thenon-clinical species and/or does the compoundinteract with the animal target? This is ofparticular importance when studying thepotential safety implications of the primary


8


2.1 Actions of a potential drug at molecular targets(receptors, ion channel and enzymes) other thanthe desired therapeutic target may translate intodeleterious, beneficial or no functionaleffects. Increased knowledge of the ‘secondary’target interactions can be expected to leadto a better understanding of potential adversepharmacology.

Current situation

2.2 Secondary pharmacodynamic studies have beendefined as those that investigate the mode ofaction and/or effects of a substance not related toits therapeutic target (ICH S7A, 2000).

2.3 Secondary pharmacology has traditionally beenassessed in vivo and effects observed followed upusing a range of tissue preparations in vitro.Effects are quantified and a mechanism ofaction elucidated by application of in vitrostudies which may be radio ligand bindingassays, enzyme assays or cell/tissue basedfunctional assays to determine which moleculartarget the drug has affinity for. The in vivostudies are performed where possible inconscious animals and involve investigation ofthe effects of the drug on physiological systems.This in vivo approach reveals unexpected effectsin an integrated system, determines thetherapeutic margin and so aids risk assessmentwhen entering clinical development.

2.4 More recently the above approach has beenreversed since it has become possible to screenfor binding against a wide range of moleculartargets (receptors, ion channels, transportersand enzymes), often derived from human tissueor human cell lines. The methods employed arepredominantly radioligand binding or enzymeactivity assays. For activity that occurs withinthe therapeutic concentration range additionalstudies to determine functional consequence ofthe secondary pharmacological activity iswarranted. The methods used are cell/tissueassays together with specifically designed in vivostudies. This approach minimizes the use of

animals and facilitates the screening of largernumbers of compounds during candidate drugselection.

Challenges

2.5 The in vivo approach may not be sensitive tosubtle effects and inter species variation needs tobe considered when extrapolating from animalmodels to man. Should the animal target havehigher affinity for the drug false positives willoccur, potentially resulting in loss of valuablecandidate drugs. These studies require the use ofanimals and so prompt ethical considerations;

2.6 The in vitro approach permits generation oflarge databases with defined ‘hits’ at specifictargets. For many targets the correlationbetween binding affinity and functionalconsequence remains unclear. Often the targetis a recombinant human target and anassumption is made that the affinity is identicalto the native human target. A major question iswhether in vitro radioligand binding assay datais predictive of secondary pharmacology effectsin man. There is a clear need to establishconfidence in such predictions;

Recommendations

2.7 Establishment of databases that bridge the non-clinical data with observations from clinicalstudies will clearly aid the predictivity of thenon-clinical data. Access to data that summarisethe non-clinical secondary pharmacologicaltargets for marketed drugs would be the first steptowards enhancement. Commercially availabledatabases are being developed (DrugMatix™,BioPrint™ and Receptor selectivity mappingdatabase are examples) but to fully exploit thisknowledge access to data from compounds indevelopment (including those that fail) as well asmarketed drugs will greatly facilitate thisunderstanding. Such a database could beestablished within a confidential frameworkshould a centre for the safety assessment of drugsbe set up.

Chapter two - General/secondary pharmacology studies


9


2.8 The emergence of large databases and theirconsiderable potential in this area highlightthe need for bio-informatic expertise. Thereis clearly a need to focus training anddevelopment of this key skill. Furthermore,the availability of skills to integrate themolecular information with in vivopharmacology has diminished. Focus on

developing infrastructure to produce highlyskilled in-vivo pharmacologists is essential.

2.9 The development of databases that facilitate abetter understanding of the functionalconsequence of secondary targets will aiddevelopment of structure activity tools leadingto better drug design.


10


3.1 The evaluation of drug effects on thecardiovascular system can be effectivelyconducted using a range of techniques and non-clinical species.

Current situation

3.2 The development of telemetry techniques inconscious animals has had a major impact onthe conduct of cardiovascular safetypharmacology studies. These techniquessupport the continuous collection of a range ofphysiological data, including heart rate, bloodpressure, left ventricular pressure, ECG andbody temperature over 24 hours. Thus, drugeffects can be studied under ‘normal’physiological conditions often using the clinicalroute of administration (e.g. oral or intravenous).The use of these technologies also supportusing a cross-over study design (each animalreceives active treatment and vehicle) thatprovides a high statistical power to detect drugeffects which, in turn, reduces the number ofanimals used in each study. In addition, usingtelemetered techniques drug effects can bestudied at plasma Cmax and Cmin levels withthe potential to define the pharmacokinetic:pharmacodynamic relationships. Furthermore,effects of metabolites may be investigated.Echocardiography applied to animal studies isa valuable technique that is probably underutilised. It is also important to consider theneed to measure regional blood flow on a case-by-case basis.

3.3 Anaesthetised animals can be used when, forexample, the compound is poorly tolerated inthe chosen species (for example due to emesis)or relatively little is known about the compoundat the time of evaluation. This raises thecomplication of the unknown effect ofanesthesia on responses to the compound understudy, although it is the view of this workinggroup that, at least the anaesthetised dogresponds to cardiovascular effects of drugs in aqualitatively similar extent as the conscious dog.

3.4 The concordance between cardiovascular effectsof drugs in animals (in particular dogs and

primates) and in humans is good. This isconsistent with the low incidence ofcardiovascular ADRs in FIH trials and the useof similar animal models in Drug Discovery todetect novel cardiovascular products as newtherapeutics.

3.5 Of great scientific, medical and regulatoryinterest is in better understanding adverse drugeffects on cardiac repolarisation (QT interval ofthe ECG) and its associated risk of causing thelife threatening arrhythmia, TdeP. Severalstudies have established that a vast majority ofcompounds that prolong the QT interval inman do so via inhibition of the potassiumcurrent, IKr. Conversely, not all IKr blockersprolong the QT interval in vivo. During the last2–5 years, significant advances have been madein our ability to test for drug effects on the IKrcurrent and a range of other cardiac ionchannels. These in vitro assays, in conjunctionwith the above in vivo cardiovascular/cardiacstudies, are able to detect a vast majority ofcompounds known to prolong the QT intervalin man. The validity of these assays has beenestablished through ILSI/HESI and also JPMAinitiatives. Both of these groups studied a rangeof compounds with known torsadogenic risks inman in comparison with non-torsadogeniccompounds and were able to differentiatebetween these two classes of compounds.

3.6 Thus, the risk of drug-induced changes incardiac repolarisation in man has been greatlyreduced. Despite these advances a major area ofchallenge in the cardiac/cardiovascular fieldremains the understanding of the relationshipbetween drug effects on cardiac repolarisationand the risk of causing cardiac arrhythmias. Thisis an area that this working group recommendsfurther investment in both non-clinical andclinical research (see section 1).

3.7 The effects of NCEs on renal function can be studiedin conscious animals, frequently the rat, following oraldosing. The primary end points are urine volume andelectrolyte concentrations. Although the methodsused have not changed greatly in the last decade or so,the low incidence of renal adverse effects would

Chapter three - Cardiovascular and renal safety pharmacology


11


suggest that they are effective at detecting risks forhuman safety.

Challenges and opportunities

3.8 Greater understanding of structure-activityrelationships will enhance drug candidateselection. However, there is no crystal structureof the hERG channel known at present due to itsmembrane bound nature. Site directed mutationand homology models based on the templatebacterial KcsA channel have been used to inferimportant amino acid residues likely to beinvolved in the drug/channel interaction. Suchmodels may support in silico screening tominimise drug interactions with the hERGchannel. Studies to investigate drug effects oncardiac ion channels should of course includeother ion channels such as sodium and calcium.Furthermore, understanding pharmacogeneticsand the importance of ‘outliers’ is likely to givesignificant benefit in assessing human risk ofcardiovascular drug toxicity.

3.9 The potential to expand in silico models such as,“CardioprismTM”, using the IC50 profile fordifferent ion channels and a library of validatedcomputer models to simulate the effects ofcompounds on in vitro preparations should beexplored further. The library contains in silicomodels of the sinus node, Purkinje fibers, atrialand ventricular myocytes including epicardial,midmyocardial and endocardial cells may havefuture value to reduce adverse cardiac effects ofnew potential drugs. Furthermore, these modelsare customized to reflect different speciescommon in non-clinical testing, namely rabbits,dogs, and guinea pigs.

3.10 The main area of challenge in the cardiovascularfield is the improvement of the understandingof the relationship between drug effects oncardiac repolarisation (QT interval) and therisk of causing TdeP arrhythmias, since theformer is only a biomarker of the latter. Thecharacteristics of subpopulations such asdiseased, young and old individuals as well asmale/female differences have not sufficientlybeen investigated. TdeP arrhythmia is an area

that this working group recommends furtherinvestment in both non-clinical and clinicalresearch.

3.11 The challenges for the future include thefurther development of telemetry methods tosupport the recording of more physiologicalend points in a single animal. Companies aredeveloping methods to record pulmonarymechanics, so the ideal would be to study drugeffects on cardiovascular, cardiac andpulmonary function in a single non-rodentanimal model. This would provide an excellentevaluation of the main physiological systems ofconcern to the safety pharmacologist andreduce and refine the use of animals in non-clinical testing.

3.12 Consideration should be given to understandingdrug effects under conditions of ‘physiologicalchallenge’, for example, to improve our abilityto detect drugs that may cause posturalhypotension.

3.13 Changes in cardiovascular function can lead topathological consequences in toxicology testing.It is therefore important, and a currentchallenge, to integrate findings in cardiovascularsafety pharmacology studies with thoseobserved in regulatory toxicology studies.

Proposals and recommendations

3.14 Considerable emphasis has been placed onstudying drug effects on the QT interval.However, the real risk to human safety is therisk of causing TdeP, for which QTprolongation is a biomarker. The relationshipbetween the magnitude of QT prolongationand TdeP is poorly defined and should be thesubject of further non-clinical and clinicalresearch. In vitro animal models are beinginvestigated such as the rabbit Langendorff andventricular wedge preparation to identifypotential non-proarrhythmic QT prolongingagents. Such studies may support the safedevelopment of novel therapies that mayotherwise be discarded during non-clinicaltesting.


12


Current situation

4.1 The ICH S7A guideline for safety pharmacologyclearly defines a need for evaluation of potentialeffects upon the respiratory system prior to entryinto man.

4.2 The respiratory system has approximately 60different cell types involved in its homeostasisand so it is particularly complex to study andinterpret drug-induced effects. It has beensimpler to define effects upon the respiratorysystem as mediated through the structuresinvolved with movement of air or through gasexchange. Broad classes of pharmacologicalresponsiveness can be discriminated throughthe functional changes induced. For instancesensory irritants generally decrease respiratoryrate whereas pulmonary irritants increase rate.

4.3 There are major challenges in extrapolatingbetween species. For example aspirin inducedasthma is common in humans with 21% ofadults and 5% of children affected yet there is nopredictive animal model. Evidence from crosssensitivity between other non-steroidal anti-inflammatory (93–100% of aspirin sensitivepatients responding to ibuprofen naproxen anddiclofenac) and paracetamol (2% crossreactivity) suggest clear mechanistic differencesand possibly a genetic involvement. Theseobservations taken together with the availabilityof over 25 years of published clinical findingswith potentially fatal aspirin inducedbronchoconstriction suggest effects upon therespiratory system are under reported.

4.4 The absence of frequent or profound adversereaction in the respiratory system during initialclinical evaluation suggest non-clinical testing, ascurrently performed, appears to prevent majorrespiratory reactions in early clinicaldevelopment. In a study reported by Sibile et al(1998) there were no life threatening events in1015 healthy volunteers during 54 Phase 1studies. However, the profound consequence ofan adverse event (even if rare) is illustrated by thedeath of a volunteer following administration of

hexamethonium to the respiratory tract withoutadequate non-clinical evaluation by this route ofadministration. Adverse events in the respiratorysystem that led to withdrawal from the marketare rare, accounting for only 2 of 121 ifanaphylaxis is excluded.

4.5 The incidence and extent of adverse reactionsin the respiratory system are probably poorlyunderstood, particularly the influence of disease,pharmacogenetics and drug-drug interactions.Complications from agents affecting respiratorypattern during sleep is similarly poorlyunderstood. The animal models available toassess respiratory function are complicated byclear species differences. For instance alveolarsurface area shows marked difference betweenhumans and rats leading to very differentconcentrations present in lung fluid followingan equivalent inhaled dose. Experience withinsoluble particulate compounds has shown therodent is good at hazard identification but poorin defining risk because of the alveolarmacrophage ‘overload’ phenomenon when thiscell type is key for clearance.

4.6 Technology to measure lung mechanics,particularly in unrestrained conscious animals,is under-developed. The tests identified in ICHS7A regarding respiratory function monitoringdefine (Murphy, 2002) a basic minimumrequirement but in some instances suggestinappropriate insensitive measures such asblood gas analysis.


4.7 The ability of non-clinical studies to detectchronic pulmonary effects is poorly understood,together with our ability to interpret the clinicalconcordance of the data. The ability to detectpotential risks to human safety would beenhanced following future enhancements inrespiratory technologies. Many features of themulti component respiratory system are beststudied in the integrated whole animal. In vitrostudies while useful to explain specific cellulartarget toxicities are unlikely to be helpful inassessing functional respiratory function. The

Chapter four - Respiratory Safety Pharmacology


13


focus should therefore be on development andrefinement of animal models of humanrespiratory function. Murphy (2002) hasclassified the respiratory system into twocompartments, the ventilatory pump and thegas exchanger. Respiratory depression isaccompanied by changes in ventilatoryparameters (rate, tidal volume, air flow andairway resistance together with lung complianceas a measure of ‘elasticity’) which althoughtechnically demanding can be measured.Greater adoption of these methods in safetypharmacology will facilitate identification ofdrugs that adversely affect the respiratorysystem. It is clear that improved methods fordetecting airway obstruction are needed.

4.8 The development of micro – sensors andtransponders has led to opportunities indeveloping telemetric methods for use inunrestrained conscious animals. This will openopportunities to broaden the species used inrespiratory function testing. Experience to datehas focused upon the rodent with expertise inother species restricted to a few specialistcentres. The validity of these new methodswould benefit from a wider evaluation of drugsknown to affect the respiratory system in manthrough comparison of the changes in FEV1(routinely carried out in clinical evaluation ofdrugs known or suspected of adverse respiratory

effects) in man with changes in non-clinicalparameters.


4.9 The low incidence of known severe respiratoryADR’s suggests the current general approachshould not be minimized.

4.10 The present regulatory focus within ICH S7A ofusing blood gas analysis techniques (or pulseoximetry) solely as a means of non-clinicalrespiratory safety assessment is inappropriatedue to the insensitivity of these approaches.Hence, the content of ICH S7A is suboptimaland needs further modification to provide thebest non-clinical datasets for regulatorysubmissions.

4.11 An increased willingness to share data for drugsnot taken into human clinical development andthose with clinical experience to fullyunderstand the frequency and significance ofchanges in non-clinical models would greatlyimprove the understanding of their predictivevalue. This could be facilitated by creation of acentre dedicated to safety evaluation ofmedicines. The routine inclusion of simplerespiratory function testing during early clinicalevaluation should be considered.


14


Current situation

5.1 Effects upon the CNS are a significant cause ofadverse drug reactions ranging from non-compliance with drug therapy as a consequenceof, for example, weight gain, nausea or anxietyto severe reactions such as convulsions. Lessreadily recognizable effects are also important asa consequence of an aging population resultingin a higher prevalence of neurodegenerativediseases. Effects mediated through an adversereaction in the CNS have been consideredunder the following headings:

Abuse Potential and Addiction Liability

5.2 The potential of centrally active drugs to beassociated with abuse potential/addictionliability is an area of increasing regulatoryinterest. Although studies to investigate theseproperties are not undertaken on a routinebasis in safety pharmacology studies theyshould be considered as part of the investigationof the broader secondary pharmacologyproperties of a NCE. Although the non-humanprimate is studied extensively there isconsiderable evidence to support the use of therat in self-administration studies to predictabuse potential in man.

Drug induced Dizziness

5.3 The causality of dizziness is often ascribed todrug treatment, despite the subjectivity of thediagnosis and the fact that it rarely causes drugwithdrawal. There does not appear to be a goodnon-clinical model to predict this effect inhumans. It is unlikely, given the difficulty indesigning objective human measures, that arealistic animal model will be valuable andaccepted since the mechanisms associated withdizziness are diverse.

Nausea and Vomiting

5.4 Nausea and vomiting are common adverseeffects of drugs often mediated through the CNS

that affect drugs in all stages of development. Itis estimated that nausea and vomiting occur inapproximately 30% of initial drug trials and inapproximately 10% of phase 1 studies this effectdose limiting. The non-clinical safetypharmacology studies did not reliably predictthe effect seen in man suggesting better modelsare required.

Food intake, weight gain and weight loss.

5.5 Several classes of drugs have been associatedwith disturbance of food intake and/orbodyweight. The effects are generally of greaterconsequence in the very young, elderly orseverely ill patients. These parameters areroutinely measured in conventional toxicologystudies. Furthermore, specific models to assessappetite are also available and appear to havegood predictive utility. There does not appear tobe a need for development of further moresophisticated tests other than to follow upspecific drug related findings.

Neuropsychiatric disorders

5.6 Non-clinical evaluation of drugs that do notpenetrate the blood brain barrier does notroutinely include any sophisticated assessmentfor centrally mediated CNS effects such asanxiety, depression, mood disturbance, learningand memory or dependence. The currentapproach to assess specifically drug-inducedchanges is prompted by evidence of brainpenetration or signals for the initial broad-spectrum assessments – Irwin or FunctionalObservational battery. In addition, animalmodels are widely used to detect primarypharmacological effects that may have utility inthe treatment of anxiety and depression.However, the potential of these models to detectuntoward drug effects (anxiogenic anddepression) is poorly defined. Nevertheless,given the prevalence of neuropsychiatricdisorders in both clinical development and inclinical use post marketing this approach shouldbe re-considered.

Chapter five - Central nervous system (CNS) safety pharmacology


15



5.7 Given the high incidence of nausea and vomitingobserved in clinical trials either greater emphasison observations in non-clinical safety studies isrequired or improved models to predict theseeffects in man are required. The ferret, shrew andmarmoset show promise as useful species. This isan area where better prediction aiding design ofnew drugs may yield considerable human benefit.

5.8 The potential of animal models (currently usedto detect agents that may be of benefit in thetreatment of anxiety and depression) to identifypotential anxiogenic and prodepressive effectsis unknown and should be studied further,although it must be recognized that animalmodels used to detect, for example,antidepressant drugs are not animal models ofdepression.

5.9 The value of the rat as a model to detectpotential abuse potential of CNS active drugs,in comparison to the non-human primate,should be thoroughly investigated.

5.10 The potential of new techniques to study CNSeffects of NCEs should be considered. Forexample, the measurement of high-frequencyultrasonic vocalisations (USVs) in the rat.Ethological studies have revealed that rats make

USVs in many different social and emotionalsituations. Measurements of USV’s in e.g. theopen field assessment as part of the functionalobservational battery (FOB) may lead to theidentification of untoward drug effects.


5.11 The potential for centrally active drugs to haveabuse potential is an area of increasingregulatory interest. Although non-humanprimates are frequently used for these studies,literature data suggest that the rat has similarpredictive value to man. A systematiccomparison between the rat and non-humanprimate is required to identify the preferredspecies to predict this liability in man. Inaddition to these effects there is increasingconcern related to the potential for CNS activedrugs to cause adverse affective disorders(anxiogenic effects and suicidal tendencies), aswell as abuse potential and drug withdrawalreactions. Animal models are used to detectdrugs with the potential to have axiolytic andantidepressant activity, although the value ofthese animals to detect untoward affectiveeffects is unknown. This should be an area offurther research to develop animal models thatcan be used to test for anxiogenic/depressantactivity.


16


Current situation

6.1 Mechanisms of drug induced liver dysfunctionare complex and remain relatively poorly definedand require better mechanistic understanding.Liver toxicity and functional disturbance isprobably the most common serious adverse eventseen in drug development and has account for 9%of drug withdrawals from the US market.

6.2 Drug induced liver dysfunction is detectedfrequently during both non-clinical and clinicalevaluation and is the cause of significantattrition and project delay.

6.3 Despite non-clinical models readily identifyinghepatoxicity there are many examples wherenon-clinical models have failed to adequatelyidentify or characterise the risk to man. Many ofthe drugs currently on the market are known tohave the potential to cause adverse effects onliver function in certain susceptible individuals(Pritchard et al 2003, Olson et al 2000;Kaplowitz 2001; Boelsterli 2003; Lee 2003.). Insome cases the incidence and severity warrantsdrug withdrawal (e.g. Troglitazone) in othersdysfunction is mild and remains asymptomatic(eg Gentamycin).

6.4 Relatively few case examples of liver toxicityhave been studied in depth and even where thishas been undertaken our understandingremains incomplete e.g. paracetamol.

6.5 Liver toxicity that occurs with a very lowincidence (idiosyncratic) is poorly understoodand in particular the individual susceptibilityfactors are sparsely known. There is a lack ofappropriate experimental models.

6.6 Greater understanding of the role of drugtransporters in the function of the liver isemerging and will potentially give betterinsight to mechanism and predictability ofhepatotoxicity. In particular the role oftransporters in drug induced cholestatic injuryand the abnormal expression of bile salttransporters in disease and drug-induced

injury is emerging. Hereditary defects inindividual bile salt transporters have beenassociated with familial intrahepatic cholestasistype 1 (FIC1 gene), type 2 (BSEP gene), type 3(MDR3 gene) and Dubin Johnson Syndrome(MRP2 gene). It is notable that several drugsthat cause liver dysfunction have been shownto impair the transport activity of BSEP(Stieger et al 2000).

6.7 Diagnosis of liver toxicity is problematic. This islinked, at least in part, to our incompleteunderstanding of underlying mechanisms.Some compounds cause pathological alterationsthat do not result in overt toxicity as assessed byevaluation of plasma levels of “liver enzymes”(e.g. biliary hyperplasia), while sporadic mildelevations of liver enzyme levels that are notaccompanied by significant liver dysfunctionoccur frequently in the human population. The“gold standard” criteria for assessing toxicity tothe liver in animals plus man are liver histologyplus elevated plasma liver enzyme levels.Evaluation of liver histology has great value butcan be difficult to assess in man because ofethical and safety issues. Also, evaluating smallbiopsy samples can be misleading – sodiagnostic liver biopsies in man are lessfavoured now for diagnostic purposes than theywere 20 years ago. ALT assesses hepatocellulardamage but is not truly liver specific. Significantelevations in ALT exceeding small multiples ofnormal control range (with changes less than1.5-fold generally considered not significant) areconsidered suggestive of liver damage. Acombination of a 3-fold elevation in ALT plus 2-fold elevation of bilirubin is generally taken toindicate liver toxicity in man. Other plasmamarkers of liver toxicity that are widely usedinclude �-GT, AST, ALP (indicative ofcholestatic damage) and bilirubin.

6.8 Distinction between toxicity and adaptiveresponse by measurement of traditional clinicalchemistry parameters is currently notachievable. Identification of better biomarkers,chemical or imaging for liver injury wouldgreatly improve diagnostic capabilities.

Chapter six - Hepatic Function


17



6.9 There is a clear need for improved biomarkersfor liver dysfunction. The current methods arewidely used but old and in some circumstancesunspecific. Opportunities to use improvedimaging together with proteomic andmetabonomic methods should yeild newmarkers for evaluation.

6.10 Development of improved in silico and in vitroapproaches will aid design of drugs withreduced hepatotoxicity potential. Structure-toxicity databases need to be defined – butbefore this is feasible we need an improvedunderstanding of the role played by chemicalstructure as a determinant of drug-induced liverdysfunction. This should follow from a betterunderstanding of mechanisms. The mostpromising avenues currently are prediction offormation of chemically reactive metabolitesand transporter interactions. More complex invitro 3-D models (e.g. liver spheroids) and liverslices, which have the advantage of morerelevant cell architecture, are an attractiveoption but require further development.

6.11 Searching for genetic and environmentalsusceptibility factors that can pre-dispose certainpatients to drug induced liver injury. Althoughsome progress is now being made in identifyinggenetic factors, progress to date has been slow.

6.12 Improved in vivo models for assessingcompound-induced liver dysfunction and forcircumventing species variability. These shouldbe linked to mechanistic considerations.Transgenic technology could give us“humanised” experimental animals, whichimprove the human relevance of non-clinicalsafety testing in animals. Perhaps generatingmice that express humanised bile canaliculartransporters (BSEP, MDR1, MRP2 etc.) couldprovide a relevant model system for improvedassessment of hepatobiliary toxicity. In thisregard, an improved understanding of ADMEproperties in animals vs. man should also be ofgreat value.

6.13 Improved ability to define mechanisms of livertoxicity, once this is observed in man.

Appropriate use of the new approachesalongside conventional toxicity investigations,to define key molecular events andsusceptibility factors offer the best opportunitiesto understand mechanism. The role of biliaryefflux transporters in hepatobiliary toxicity is aparticularly promising avenue to explore, as isthe role played by reactive metabolites.


6.14 More consistent and accurate detection/diagnosis of drug induced liver dysfunctionduring clinical trials (rather than post-marketing) using existing technologies withconventional liver function tests supported bymore sophisticated causality assessments can beexpected to lead to better and earlieridentification of liver toxicity. This should becomplemented by research focused on moresophisticated application of existing approaches(most notably causality assessments), and ondetection of novel biomarkers of liverdysfunction.

6.15 An improved ability to predict and avoid drug-induced liver dysfunction prior to introductionof drugs into man. Significant progress in thisarea is likely to require a more sophisticatedunderstanding of the key molecular and cellularmechanisms that result in drug-induced liverdysfunction and that influence speciesdifferences in toxic responses. Particularattention should focus on the roles played bydrug disposition, chemically reactivemetabolites, and transporters interactions. Theoutcome from this work needs to be linked todevelopment and validation of structure-toxicity databases, and of in vitro experimentalapproaches, that can be used by medicinalchemists and bioscientists to design and selectsafer compounds during drug discovery.

6.16 Identification of susceptibility factors thatdetermine why some individuals are markedlymore susceptible to drug-induced liverdysfunction than the general population. Thesesusceptibility factors are likely to bemultifactorial, and could well involve a complexinterplay between genetics and environment.


18


Current situation

7.1 Adverse reactions in the gastrointestinal tractappear to occur at all stages of drug developmentand in subsequent clinical use. Drug-induceddisturbances of GI function negatively impactpatient safety, compliance, quality of life andclinical benefit. The impact of GI ADRs in termsof drug withdrawals from the market has beenminimal, with only one example, pirprofen, inthe period from 1960 to 1999, however, drugswith labeling restrictions are commonplace, forexample Lotronex (Fung et al., 2001).

7.2 GI ADRs are some of the most frequentlyreported in all phases of clinical drugdevelopment and for marketed products, asillustrated by the 700 drugs that are implicatedin causing diarrhoea. Furthermore, GI ADRsaccount for approximately 18% of all reportedclinical adverse drug reactions and 20 to 40% ofthose in hospitalised patient. Given, however,that symptoms of disturbed GI functionare encountered in everyday life, the actualincidence of drug-related effects are mostlikely extensively under-reported. The majorityof reported ADRs are functional in nature(nausea, vomiting, dyspepsia, abdominalcramps and diarrhoea or constipation) with alesser number related to lesions (e.g., ulceration)or enhanced susceptibility to infection (e.g.,pseudomembranous colitis). Of these, it isestimated that approximately 80% are Type Apredictable pharmacological reactions. Some ofthe most numerous and serious GI ADRs areattributed to chronic use of nonsteroidal anti-inflammatory drugs (NSAIDs). WidespreadNSAID use and associated reports of gastriccomplications, has given rise to the acceptancethat NSAID use carries a significant risk ofdeveloping peptic ulceration. In a recentprospective analysis of 18,820 patients in a UKhospital, 29.6% of all ADR cases were relatedto NSAID. Furthermore, NSAID use in theUS alone is estimated to be responsible forover 100,000 hospitalisations and 1,700 deathsper year.

7.3 The challenge of defining the non-clinical GIliability of a potential new medicine is not onlyhampered by biological complexity of thesystem under scrutiny, but also by fundamentaldifferences in terms of GI function between therat, the most commonly used species inresearch, and man. For example the rat does notposses a gall bladder and does not vomit. Thisraises important questions about the relevantspecies for conducting GI assessment. Based onGI functional homology for man, especiallymotility, gastric emptying and pH, particularlyin the fasted state which is analogous to theconditions prevailing in many Phase I trials, thedog, is perhaps a more relevant species.Moreover, the dog was a better predictor ofclinical GI ADRs than the monkey for 25anticancer drugs. Although physiologicalsimilarity is an important requirement it is onlyone of many factors that must be consideredduring species selection.

7.4 GI ADRs are some of those most frequentlyencountered in healthy volunteers and patients,yet assessment of potential new medicines on theGI system is not a regulatory requirement beforeconducting Phase I trials, nor even for marketingapproval. It is, however, recommended in theICH S7A Guideline for Safety PharmacologyStudies (CPMP/ICH/539/00) that the GIsystem should be studied on a case-by-case basis.The ICH guideline does not classify GI system asa “core” battery organ system, unlike therespiratory, central nervous system andcardiovascular systems, of which assessment ismandated prior to Phase I trials. Moreover, theGI system is relegated to a group of supplementalorgan systems to be studied only where there is acause for concern.

7.5 Further to the list of parameters in ICH S7A, arecent publication has reviewed GI models andtechniques applicable for use in safetypharmacology. One important element omittedfrom both ICHS7A and the review by Harrisonet al., (2004) of particular relevance toanticancer drugs is nausea and vomiting. There is no doubt this omission reflects the

Chapter seven - Gastrointestinal (GI) safety pharmacology


19


complexity of the symptom and that there areno non-clinical models able to relay nausea.There is, however, a growing understanding ofthe causes nausea and vomiting and non-clinicaldog and ferret models amenable for safetypharmacology testing.

7.6 The influence of GI intolerance on patientcompliance to medication is probably underreported.


7.7 A comprehensive review of non-clinical GIendpoints to determine the concordancebetween non-clinical GI safety models andman, to identify and promote appropriatemodels/species and relegate and excludeirrelevant misleading models/species is needed.Current methods to assess GI function despitebeing well established have poorly understoodpredictive value. A clear need exists to developa better understanding of the models availableand their applicability to drug evaluation.

7.8 Given its complexity the GI tract is a systemprobably best studied in vivo. Despite thiscomplexity there are clear advantages inunderstanding molecular targets associated withGI dysfunction and this should be an area forfuture research. There are opportunities tocombine key GI parameters (eg secretion andmotility) into single in vivo studies. To facilitateunderstanding of GI risk in humans a betterunderstanding of the importance of local

concentration rather than systemicconcentration/response is needed.

7.9 Two common major dose limiting effects,nausea and emesis, are poorly understood andpredicted. Better models are required.


7.10 Placing greater importance on GI-relatedobservations (e.g., emesis) in acute and repeateddose toxicology studies, including seeminglyisolated incidences may help trigger formalstudies of GI function if not conductedroutinely. The opportunities to use non-invasiveimaging techniques may enhance the predictivecapability of these tests and should beencouraged.

7.11 There is a clear need to develop and validatenon-clinical methods for detecting nausea andemesis potential. Reducing these important sideeffects would be expected to facilitate drugdevelopment and also improve patientcompliance for marketed drugs.

7.12 Encourage and nurture links between GIspecialists in academia and industry. Inparticular there is a need to understand betterthe predictive value of the non-clinical models.Establishment of a Safety Sciences centre withinthe UK that can act as a repository of datawould greatly help understanding of theexisting data available within individualcompanies.


20


Current situation

8.1 Adverse drug effects in man on the mainsensory organs, sight, hearing, taste and smellare relatively poorly understood and occur witha relatively low frequency when compared withthe incidence of headache, nausea & vomiting,diarrhoea and dizziness. However, untowardeffects on the sensory organs, especially hearingand sight, could have a significant impact onvolunteers/patients well being.

8.2 Behavioral pharmacology studies may detectuntoward effects on balance and sight, althoughit is probably that significant deleterious effectson these systems would be required beforechanges in behavior would be detected. Intoxicology studies a detailed investigation ofdrug effects on the eye are conducted as part ofthe clinical examination of animals is performed(corneal reflex (dog), anterior chamber and lensexamination using a slit lamp and examinationof the vitreous body and the retinal fundus usingan indirect ophthalmoscope and a magnifyinglens). In addition, histological examination ofpotential effects on the structure of the retinaand optic tract are conducted.

8.3 Sophisticated non-clinical tests are available toinvestigate drug effects on the retina (theelectroretinogram, ERG) and visual pathway(visual evoked potential, VEP). Pharmacologicaleffects on the ERG have been reported with arange compounds with varying mechanisms ofaction. Such studies, when used in conjunctionwith histopathology in toxicology, can be usedto establish a reversible pharmacological actionon the retina that would be less of a concernwith respect to human safety, or an irreversibletoxic effect on the retina that would be of greatconcern for human safety. In the event of visualadverse events being reported in man, the ERGin both man and animals is a valuableinvestigative tool to establish the site of action ofa compound and to provide a link between non-clinical and clinical data to build confidence inthe long-term safety of a drug. The challengesfor the future would be to further characterise

the relationship between drug effects on visualfunction in man and in animals to establishedthe predictive value of existing test systems andto develop novel assays.

8.4 Dizziness is a frequently reported adverse eventin man. This may arise through changes inblood pressure (postural hypotension), an effecton the central nervous system (CNS) or aneffect in the inner ear. Non-clinical models areavailable to investigate drug effects on balanceand motor coordination, such as the rota-rodand activity box, although the sensitivity ofthese models to detect adverse drug effects onbalance via an effect on the ear has not beensystematically studied.

8.5 Ototoxicity is thought to be rare, but possiblyunder reported (essentially confined to loopdiuretics, aminoglycosides, anti-neoplasticcytotoxics, aspirin and quinidine) although theimpact on the individual if affected is significant.There is no safety pharmacology test used todetermine potential effects of NCEs on hearingand there is no regulatory requirement toevaluate ototoxicity unless there is a clear causefor concern. However, early signs of ototoxicity(tinnitus) can be detected in volunteers andpatients that would be a signal to stop therapyprior to irreversible damage. This is an area thatrequires basic research to systematicallyinvestigate the predictive value of non-clinicalassays using a diverse range of pharmacologicalagents. Until such tests are in place, non-clinicaltesting for effects on hearing cannot berecommended.

8.6 Electrocochleography measures signals at thebeginning of the vestibulocochlear nerve. Animmediate reduction in action potential has beendetected within minutes of an intravenous doseof aminoglycoside. Brain Stem Auditory EvokedResponse (BAER) is a technique for measuringhearing loss. These techniques can be used inboth humans and animals. However, this is anarea that requires basic research to establish ifpredictive non-clinical assays can be established.Until such tests are in place, non-clinical testingfor effects on hearing cannot be recommended.

Chapter eight - Sensory Safety Pharmacology


21



8.7 As discussed above pharmacological effects onthe ERG have been reported with a rangecompounds with varying mechanisms ofaction. Results from non-clinical ERG studiesare best interpreted in conjunction withhistopathology. The challenges for the future isto further characterise the relationship betweendrug effects on visual function in man and inanimals to established the predictive value ofexisting test systems and to develop novelassays. Furthermore, a more comprehensiveunderstanding of the relationship betweenhistological changes to the visual pathway andchanges to the ERG and VEP is required.

8.8 The ototoxicity potential of drugs is notroutinely evaluated non-clinically which reflectsthe low incidence of ototoxicity in man and thefact that clinical signs of tinnitus can be detectedin clinical trials. Furthermore, the predictivevalue using a diverse range of pharmacologicalagents of the existing animal models to man isunknown. The challenge, therefore, is to furtherour understanding of the concordance betweeneffects of ototoxic drugs in animals and in manand to identify appropriate non-clinical testingstrategies appropriate for the risk.


8.9 Although safety pharmacology methods (ERGand VEP) are available to detect visual effects of

drugs, routine testing of NCEs prior to FIH is notrecommended in the absence of effects intoxicology. However, the challenge for the futureis to further characterise the relationship betweendrug effects on visual function in man and inanimals to establish the predictive value ofexisting test systems and to develop novel assays.

8.10 Ototoxicity is a potentially important safety issuefor drugs, but is hard to detect. Although somedrugs are classically associated with ototoxicity,the extent of the issue is probablyunderestimated since the symptoms may appeargradually and are often indistinguishable fromunderlying disease processes. Any impact onhearing is temporary and reversible uponremoval of the drug. Animal pharmacologymodels are available, but their utility in screeningis yet to be determined as few mechanisms ofototoxicity have been determined and theforward prediction of the available tests is notestablished and can be insensitive.

8.11 Overall, the current model, where evidence ofhearing impairment in toxicology studies or inclinical trials, is considered more appropriate.Nevertheless more attention should be paid tothis under-reported safety concern. In toxicologyand pharmacology laboratories, animal handlersand toxicologists should pay attention tobehavioral changes in animals that could suggestan auditory dysfunction. Should these data, oremerging data in humans suggest furtherinvestigation of auditory function is warranted,BAER or histopathology of cochlea could beused for investigational studies in animals.


22


Current situation

9.1 The available methods to assess the interactionof drugs with the immune system do not meet allthe needs for safety evaluation. Approximately6% of drug withdrawals are ascribed to adverseimmune responses, with a further 7% associatedwith bone marrow toxicity for which animmune response may be contributory. Thedifficulty with prediction of these responses is aconsequence of their low clinical incidence,poorly understood mechanism of action, andthe inability of non-clinical models to reliablyidentify hypersensitivity reactions in man.

9.2 Current methods for the non-clinical assessmentof immunosuppression appear to identify theclinical risk adequately. For this purpose one ormore of the following approaches are used: (a)examination of the histopathological appearanceand/or weight of key lymphoid organs (thymus,spleen and lymph nodes), together withconsiderations of haematological parametersand of bone marrow cellularity, (b) testsdesigned to measure the functional integrity ofthe immune system in exposed animals. Of themethods available, tests for assessment of theintegrity of antibody production (to either sheepred blood cells [SRBC] or keyhole limpethemocyanin [KLH]), are commonly used, butother approaches (such as for instanceimmunophenotypic analyses) are also employed.

9.3 Methods, albeit poorly characterized, to identifyhost resistance are available and have beenemployed to assess defense against infectiousmicro-organisms or transplantable tumour cells,although the ability of these models to assesssubtle effects is less confidently assured. Theadvantage of host resistance assays is that, intheory at least, they provide an holistic view ofchanges in susceptibility that may reflectcompromised immune function. However, itmust be appreciated that other forms of toxicity(such as, for instance, liver damage), andgeneral ill health, may be reflected by changesin host resistance in the absence of a primarylesion in the immune system.

9.4 Functional assays; SRBC/KLH assays orimmunophenotype analyses, particularly aimedat defining the quality of the immune responseand the selectivity of changes in type 1/type 2 Tlymphocytes, have been increasingly used tocharacterise adaptive immune function.

9.5 Methods for the identification andcharacterisation of chemical allergens havebeen developed and are widely used in safetyassessments. Validated methods for theevaluation of contact (skin) sensitising potentialare available. In addition, as yet unvalidatedapproaches have been developed forassessment of the potential of chemicals andproteins to cause allergic sensitisation of therespiratory tract and (in the case of proteins)gastrointestinal tract.

9.6 Approaches to identify compounds that induceautoimmunity, or to provoke idiosyncraticreactions are not available. These types ofresponse will continue to be the most difficult toassess and predict. One general method that isavailable (in several guises) is the popliteal lymphnode assay (PLNA). This has been, and remainsstill, the subject of investigations but has not yetbeen established as a reliable indicator of hazard.An understanding of genetic and environmentalsusceptibility factors will probably be the mostfruitful area of research.


9.7 Although there are available methods that canbe used for investigating the ability ofxenobiotics to interact with and/or perturb theimmune system, it is clear that they do not (atleast as currently deployed) meet all needs ofdrug safety assessment. Thus, a common causeof attrition following launch of a new drug is theappearance of idiosyncratic drug reactions,often manifest at only low incidence, that arethought to have an immune/allergicpathogenesis. The ability of candidate drugs toprovoke such reactions has proven extremelydifficult to predict. There are a number of

Chapter nine - Immuno-safety pharmacology


23


factors that contribute to this difficulty, including:(a) the fact that such reactions normally occuronly at low incidence, (b) there is frequently noattempt to define the mechanisms throughwhich such reactions are provoked, and as aresult there is uncertainty regarding the relevantimmunobiological processes, and (c) it is likelythat, as currently constituted, non-clinicaltoxicology and safety pharmacology studies arenot appropriate for identification of chemicalsthat have the potential to elicit hypersensitivityreactions.

9.8 With respect to the last of these points, twoseparate but related potential deficiencies canbe identified. The first of these is thatappropriate assessments are not conducted. Thesecond is that at least some of the assessmentsthat are conducted currently may yieldinformation of importance, but that the relevantquestions are not addressed, nor the correctdeductions made.


9.9 There is confidence that current non-clinicalmodels are able to detect immunosuppressionand impaired host resistance despite thevalidation of these methods being incomplete.

9.10 Methods to identify allergenicity have beenwidely used and for contact sensitization severalapproaches are validated. There remains a needto gain full acceptance of the methods availablefor detection of respiratory and gastrointestinalsensitisers. The local lymph node assay (LLNA)is a fully validated, OECD guideline method forthe identification of chemicals that have thepotential to cause skin sensitisation. The test ispredicated on the fact that skin sensitisingchemicals will provoke the stimulation ofspecific T lymphocyte responses in regionallymph nodes draining the site of exposure. Thismethod therefore has the potential to identify

chemicals that have an intrinsic potential tointeract with the adaptive immune system, orwhich can be metabolically converted to such.

9.11 Other approaches developed initially for theassessment of industrial chemicals includemethods designed to identify materials with thepotential to cause allergic sensitisation of therespiratory tract: the mouse IgE test and cytokinefingerprinting. These tests are designed to definethe quality of immune response that is provokedby exposure to a chemical allergen, andspecifically whether selective type 1 or type 2 Tlymphocyte responses are elicited. Thisapproach allows determination of whether achemical is likely to stimulate an IgE antibodyresponse – the major effect molecule in manyforms of allergic disease.

9.12 One potential innovation might be to considerthe coordinated and structured application ofthe above methods, with or withoutincorporation of one or other version of thePLNA, to facilitate an holistic assessment oflikely stimulation of an immune response. Usedin concert these methods could provide: (a) anassessment of the inherent potential to provokean immune response, and of the level ofimmunogenicity, and (b) an indication of thetype (quality) of immune responses that will beelicited preferentially.

9.13 Approaches for the identification ofautoimmune responses or idiosyncraticresponse have proven to be elusive and demandmore intensive study and should includeconsideration of genetic susceptibility factors,polymorphisms and environmental factors(diet, overall health status, life style etc).Furthermore, it is pertinent to ensure thatinformation relevant for a consideration ofimmune function from conventional toxicitystudies, and an appreciation of cytokineexpression patterns, are included in an overallassessment of impacts on immune status.


24


Current situation

10.1 In most regions of the world, the majority ofdrugs used in children have never been testedin this target population. Moreover, they wereapproved only on the basis of data obtained inadults, either animals or man (although off-label use is high). Nevertheless, in both theUnited States and Europe, health authoritiesare encouraging paediatric studies by means oflegislation such as the Best Pharmaceuticals forChildren Act (2002) and the Pediatric ResearchEquity Act (2003) or the European draftlegislation “Better Medicines for Children” (tobe finalized, 2005). The aim of all of theseconsiderations is to make valuable medicinesavailable to children as soon as possible,addressing what is seen widely as an importantmedical need.

10.2 The need to extend the access of children intonew therapies, and to develop paediatricformulations of existing drugs reinforces theurgency to implement strategies for earlyprediction of specific safety aspects in thispopulation. In this context, the use of juvenileanimals may be helpful and is usuallyconsidered on a case-by-case basis.


10.3 Traditionally, drug development in children hasbeen performed once sufficient numbers ofadults have been studied to define risks.Therefore, drug approval has not required thesame level of evidence for paediatrics as it hasfor adults, given the approvals already securedon the basis of adult safety data. Evidence frompaediatric development programmes recently,however, has suggested that children aredynamic and variable and reliance on adult datamay underestimate the risks of the drug tochildren. Moreover, in the future, especially inthe area of paediatric oncology, there will likelybe an emphasis on drugs designed specifically

for children and adult human safety data willnot be available as background to define risks.

10.4 Potential differences between adults and childreninclude pharmacokinetic (e.g. clearance pathway)differences, and developmental differences intissue structure and function, for example invision and other senses, thyroid function, CNSplasticity and the haematopoietic system (with aspleen and thymus focus, rather than bonemarrow in adults).

10.5 Clearly, non-clinical studies used to support aninvestigational new drug, or even a marketeddrug for which paediatric use is desirable, maynot directly address the concerns or thepotential interactions of age, gender and therelative development of the various organsystems. Age- or gender-related kinetic andmetabolic differences in response are notgenerally addressed. When performed, moststudies in juvenile animals use neonatal rats,dogs or swine, assuming equivalent postnataldevelopment to newborn infants.

10.6 In Europe, a guidance document addressing theuse of juvenile animals on safety assessment ofpaediatric drugs is being prepared by the SafetyWorking Party under request of the CHMP. Aconcept paper including the several items underdiscussion has been published in the EMEAwebsite. FDA/CDER is in the process ofdeveloping Guidance to Industry on Non-clinicalSafety Evaluation of Pediatric Drug Products.This document provides guidance on the roleand timing of animal studies in the safetyevaluation of therapeutics intended for paediatricuse and when such studies might be needed. It isintended to serve as a general resource in testingand provide specific recommendations based onthe available science and pragmaticconsiderations. There are no specific guidelinesto address the use of juvenile animals for safetypharmacology studies of paediatric drugs,although there is a general statement in ICH S7ASafety Pharmacology guideline thatconsideration should be given to selection ofrelevant animal models, including age of animals.

Chapter ten - Paediatric indications


25



10.7 Given the issues described above, the utility ofsafety assessment in juvenile animals is a subjectthat continues to be debated. Due to interspeciesdifferences regarding the development of severalorgans and systems, the value of studies injuvenile animals to predict the drugs safety inchildren needs to be considered on a case-by-case basis, taking into consideration theinformation from studies in adult animalsincluding reproductive toxicology studies.

10.8 This group proposes that the adequacy andappropriateness of all supporting non-clinicalsafety data (including safety pharmacology)should be considered as part of a paediatricinvestigation plan. Given the limited modelsavailable, the extent of non-clinical safetypharmacology testing to support paediatricclinical development will inevitably be decidedon a case-by-case basis. Information from adultanimal studies including reproductivetoxicology should also be taken into account.


26


Current situation

11.1 Considerable experience in the developmentof biotechnology-derived compounds has beengained over the past 20 years. Clinicalexperience has revealed toxicity frequently notpredicted from the initial non-clinical studies.The traditional safety pharmacology studiesneed to be used with care with biotechnology-derived compounds with the priority beingassessment of functional indices of toxicity. Thefollowing areas should be considered:

11.2 In vitro testing for tissue cross reactivity andtissue distribution is essential in speciesselection and interpretation of pre clinical data.

11.3 Many humanized biotechnology derivedmolecules are immunogenic in animals.Animal species may produce neutralizingantibodies, animal anti-human response(characterized by increased clearance and forloss efficacy), toxic antibodies which maymanifest responses disconnected frompharmacokinetics and anti-idiotype antibodiesthat result in high unexplained exposure notcorrelated withpharmacodynamic/toxicodynamic actions.


11.4 Since prediction of immunogenicity ofbiotechnology-derived products in man is verypoor, improvement of immunogenicityreduction strategies is necessary. Continuingprogress in prediction, detection andprevention of harmful immune responsesbrings the promise of safer and moreefficacious compounds in the future.

11.5 Considering that factors other than proteinsequence are equally important forimmunogenicity, and that immune responsesare genetically determined and, therefore,highly individual, sequence analysis may notbe sufficient to predict and avoid antibodyformation although a number of companies

claim success in this area. Ex vivo T-cellactivation assays and specialized animalsmodels including genetically engineered miceand MHC defined primates are promisingfuture directions.

11.6 Given the poor predictivity of the immuneresponse the default has often been to use aprimate model. It seems reasonable to challengethis premise and studies to establish its validityand explore whether non-primate species canbe more widely used should be encouraged.


11.7 Increase human sequence content: Severalapproaches are pursued to minimize murinesequence content: chimeric antibodiescomprising mouse variable regions and humanconstant regions; humanized antibodies inwhich murine CDRs are grafted onto a humanframework and fully human antibodiesproduced by phage display or in transgenicanimals. Nonetheless, antibody formation canoccur with all these modifications. Abbot’sHumira®, the first FDA-approved fully humanantibody, elicits an immune response in 12 % ofpatients when administered withoutimmunosuppression. Fully human proteins canbe immunogenic, as immune tolerance can bebroken under certain circumstances.

11.8 Improving solution properties: Proteinaggregates are typically more immunogenicthan dissolved proteins. Poor protein solubilitycan be overcome by optimizing expression,purification, formulation and solutionconditions. An alternative method is to userational solubility engineering.

11.9 Removing antibody epitopes: Antibodyepitopes are often located at a small number ofdiscrete sites on the surface of a protein.Modification of residues can reduceimmunogenicity and binding of existingantibodies. Modified variants of Factor VIII andstaphylokinase are examples.

Chapter eleven - Biotechnology compounds


27


11.10 PEGylation: The steric block of antibodybinding by derivatizing the protein withpolyethylene glycol (PEGylation) can decreaseimmunogenicity. PEGylation also increasessolubility and often changes pharmacokineticparameters which may permit less frequentdosing. PEGylation has been used successfullyto minimize immunogenicity of enzymes.Protein design approaches may help to furtheroptimize the balance between reducedimmunogenicity, improved pharmacokinetics,and activity maintenance.

11.11 Identifying and removing class II MHCagretopes: The production of IgG antibodiesresponsible for clinically relevantimmunogenicity can be minimized byidentifying and removing class II MHCagretopes from the therapeutic molecule. Thebinding site of antigen-derived peptides at theclass II MHC molecule and their bindingspecificities are well described. Followingcomputational or experimental identificationof MHC agretopes, a mutagenesis approachcan be used to produce variant sequences thatdo not interact with MHC.


28


Current situation

12.1 The development of sophisticated technologiesto screen large numbers of chemicals has led toan increased use of human tissue derivedtargets. Consequently there is an increasedlikelihood that the drug candidates identifiedwill be highly selective for human tissue/disease.While this may lead to more efficacioustreatments it does give difficulty in speciesselection for non-clinical safety evaluation.

12.2 There is increasing evidence that around 50% ofthe toxicities seen with candidate drugs indevelopment can be ascribed to thepharmacology (either primary or secondary)associated with the target receptor or protein. Ittherefore follows that an increasing number ofprojects will have compounds that do not crossreact with the species commonly used for safetyassessment. While studies in conventionalspecies will provide information on intrinsictoxicity related to the chemistry of thecompound a gap exists in understanding theevaluation of human selective or specific targets.

12.3 The issue of non-clinical safety evaluation ofhuman specific compounds has beenpreviously considered with biotechnologicallyderived pharmaceuticals. These have usuallybeen large molecules that have beenmanufactured to replicate an endogenoushuman protein or are an antibody to a humanspecific target. Many of these biotechnologycompounds were directed towardsreplacement therapy. The toxicity produced bythese molecules has been mediated through thespecific target receptors and selection ofappropriate species has been imperative tounderstanding human hazard. Toxicity with thebiotechnology compounds has so far beenunderpinned by an extensive literaturesupporting the understanding of the function ofthe specific target – examples being G-CSF, IL-1, IL-2, Erythropoietin and Insulin. For futuretargets that are exploited using small moleculechemistry it is likely that this prior knowledgewill not be abundant. Furthermore the biotech

compound developed to date are frequentlyaimed at life threatening diseases whereashuman specific or highly selective targets arenow being identified for disease that are not lifethreatening.

12.4 In assessing the selectivity of a compound for atarget two scenarios exist. The compound canhave no cross reactivity with the animal targetor cross reactivity can exist but the potency ofthe compound at the animal target issubstantially less than for human.Pragmatically if the drop off is more than 20fold it is unlikely that the dose levels can beachieved in an animal species that will enablesufficient assessment of pharmacologicallyassociated toxicity.


12.5 Conventional evaluations will describe thesafety pharmacology/toxicity associated withthe chemistry of the molecule to be assessedbut the following may be considered to assesspharmacologically related adverse effects:

12.6 Profiling in established laboratory species toconfirm selectivity and selection of the mostappropriate species.

12.7 Demonstrate that human specific moleculespossess high enough potency for animal target.

12.8 Target distribution in normal and diseasedtissue. Where targets are expressed in normaltissue it is worth considering its function ifknown to assess the potential deleterious effectsof pharmacological modulation.

12.9 Sensitivity – in some instances the laboratoryspecies may be less sensitive to compounds withhigh potency to human targets whilst having asimilar physiological role. In such cases use ofconventional high dose toxicology may permitestablishment of satisfactory safety margins.

12.10 Transgenics – increasing access to transgenicmodels will in the future give many more

Chapter twelve - Compounds selective for human targets


29


options to evaluate the human specific targets.Approaches that may have utility will includeknock-out/knock in models where theendogenous animal gene is replaced by acloned human gene. Information from straightknock out models may also be useful if thefunction of the gene in the animal model isassured to be similar in man. Where known,information from human populations witheither malfunctioning or absent genes isparticularly useful. An important caveat whenconsidering knock out models is the potential tooverstate the toxicity of inhibiting the target orvice a versa. In these circumstances it may beappropriate to consider use of a conditionalknock out. It should also be established that thetransgenic model functions in a similar waywith the correct response elements operationalwhen drawing conclusions on the utility of atransgenic model.

12.11 Homologous systems i.e. animal specificmolecule to assess the pharmacology in aresponsive animal species. Understanding thefunction of the target in both the animal speciesin comparison to man may also be important.With small molecules the option to usemonoclonal antibodies for cell surface targets isalso available and increasingly used. Thisapproach should also consider the effect onregulation of the target. If the target is nota receptor approaches which target mRNAmay yield useful information. Synthesis of ananti sense oligonucleuotide may provide suchtools but consideration of the chemistryassociated with the oligonucleotide is alsoworthwhile.

12.12 Disease models – Toxicity and safetypharmacology are traditionally assessed innormal animals. It is conceivable that

understanding physiological response would beassisted by use of disease models. Experiencewith recombinant erythropoietin wherehypertension was seen in patients and uremicrats despite no change observed in normal ratssupport this view. Furthermore, in situationswhere the target is expressed only in the diseasesituation (frequently seen with targets thought tobe important in inflammation) the use of diseasemodels may be helpful in establishing truetherapeutic indices.

12.13 Statistically or mechanistically it may not bepossible to reproduce all human idiosyncraticreactions in non-clinical studies. It is importantto understand the strengths and limitations ofsuch studies.


12.14 Evaluation of human specific targets can beexpected to increase in complexity over thenext 5 to 10 years. The understanding ofsimilarities and differences between the humandisease, animal model and responsiveness ofthe normal laboratory animal will increasesubstantially. It is therefore unwise to regulatethis area until a much better understanding hasbeen achieved. Precedence from thedevelopment of biotech large molecules shouldbe used to guide each development programmeindividually with the selection of species andmodels justified rationally.

12.15 Where small molecules are directed towardshuman specific targets it will be necessary tofully evaluate the intrinsic toxicity associatedwith the chemistry and assessment of thepharmacologically related toxicity will need tobe supplementary and rationally developed.


30


Current situation

13.1 Practitioners in the field of safety pharmacologyderive from a variety of core disciplines. Themajority are trained as pharmacologists,primarily non-clinical but some clinical whilsttoxicologists, biochemists and biologicalscientists have all achieved success in thediscipline. There is a strong integrative role of thesafety pharmacologist in assessment of all theavailable safety data as well as their own specialtystudies and also a need to relate to the expectedclinical utility of molecules under investigation. Itfollows then that core skills need to besupplemented with continuing development andthis, in the main, is provided in house by unitsproviding safety pharmacology services.


13.2 The current trend to reduce practical training atundergraduate level is the initial fail point forscience-oriented students who may wish todevelop a career in pharmacology. Lack ofopportunity to explore the range of scienceavailable in the field limits take up of furtherstudies at Masters or Doctorate levels.Furthermore, facilities for Masters courses inthis and cognate areas are minimal, largelydriven by problems associated with animalexperimentation.

13.3 Post graduate and postdoctoral training is also aproblem, which various Professional Societies(notably the British Pharmacological and BritishToxicological Societies) are trying to address inconcert with industry. Whilst industry, notablythe pharmaceutical industry, can and doesmaintain state of the art facilities and providesongoing training for its staff, the academic baseof the discipline is essential for both the widergood of the community but also to ‘educate theeducators’ of the future.

13.4 Training has two elements: technical andinterpretative and in the field of safetypharmacology lies across two domains. Data on

new medicines will be generated at the non-clinical level to permit the initial human dose.However, the clinical pharmacologist receivingthe data must be able to relate to the science andthe scientist to discuss the interpretativeelements of the data set. Training, therefore,must be available to meet both types of need.The availability of such training is reducing inthe UK and this was highlighted in the PICTIFreport, in has been confirmed in varioussubmissions from ABPI and ProfessionalSocieties to government and by thepractitioners themselves.

Recommendations

13.5 The training provision for both non-clinical andclinical pharmacologists needs core supportfrom government to help address the currentlimitations. Numbers of adequately trainedclinical and non-clinical pharmacologists arefalling which is a serious issue impacting on theperformance of a major industry contributing toUK plc and to the underpinning academic basewhich supports it. New integrative ContinuingProfessional Development (CPD) courses areneeded to use both non-clinical and clinicalpharmacologists more effectively in thedevelopment of new medicines.

Value added and measures of success

13.6 Compound attrition related to adverse effectsdetectable in safety pharmacology studies willbe reduced e.g. nausea and vomiting, CNSeffects and latter stage attrition related tohepatotoxicity. In addition, increasedtolerability may improve compliance increasingdrug efficacy.

13.7 Although Phase I studies are very safe, in thesestudy doses are escalated to identify themaximum tolerated dose. It is possible that insome circumstances tolerability issues preventthe achievement of exposures to predictedtherapeutic levels that may be related to effects

Chapter thirteen - Education and training


31


detectable in safety pharmacology studies e.g.dizziness, nausea and vomiting etc. Therefore,improving safety pharmacology testing mayreduce the Phase I attrition due to poortoleration in volunteers.

13.8 Investments in understanding further therelationship between prolongation of the QTinterval and proarrhythmia may support thedevelopment of NCEs that prolong the QTinterval that are not associated with causingTdeP. This in turn will increase the ‘chemicalspace’ available for Discovery chemistry toexploit against novel targets thus negatingthe negative impact of hERG affinity.Furthermore, the incidence of late stage drugwithdrawals related to arrhythmias will bereduced/eliminated.

13.9 Drugs selective for human targets areassociated with specific challenges for thepharmaceutical industry. Improved testingstrategies are vital in preventing thediscontinuation of compounds and potentiallyloss of interest in targets based upon incorrectand misleading adverse safety pharmacologyor toxicity derived from inappropriate models.The key value to gain from supporting

evaluation of human specific targets is anincreased linkage between the target and thedisease leading to more efficacious compoundsin humans.

13.10 Further refinement of safety pharmacologystudies (e.g. simultaneous telemeteredrecoding of cardiovascular and pulmonaryparameters and the application of PK/PDmodeling) will reduce the numbers of animalsused in non-clinical safety testing. This willcontribute to the 3Rs.

13.11 Although compound attrition due to effects onthe sensory organs appears to be low, improvingnon-clinical testing strategies may reduce furtherthe incidence of adverse events related to thevisual and hearing system. Furthermore suchstudies would raise confidence on the lack ofdrug effects on the sensory organs and anypotential long-term consequences.

13.12 Through data sharing and increasing ourknowledge of the concordance between non-clinical studies and human outcome wouldincrease our confidence to make decisions ondrug development.


32


ICH S7ASafety Pharmacology Studies for Human Pharmaceuticals(2000) Note for Guidance on Safety Pharmacology Studiesfor Human Pharmaceuticals(CPMP/ICH/539/00).

ICH S7BThe Non-clinical Evaluation of the Potential for DelayedVentricular Repolarization (QT Interval Prolongation) byHuman Pharmaceuticals (2004) Note for Guidance ICHStep 2 Revisionwww.fda.gov/cber/ichqt.htm

Boelsteri, U. A. (2003)Animal models of human disease in drug safety assessmentJournal of Toxicological Sciences, 28, 109–121.

Fung, M., Thornton, A., Mybeck, K., Hsiao-Hui, J.,Hornbuckle, K. and Muniz, E. (2001)Evaluation of the Characteristics of Safety Withdrawal ofPrescription Drugs from Worldwide PharmaceuticalMarkets – 1960 to 1999Drug Information Journal, 35, 293–317

Greaves, P., Williams, A. and Eve, M. (2004)First dose of potential new medicines to humans: howanimals helpNature Reviews: Drug Discovery, 3, 226–236

Jenkins, C., Costello, J. and Hodge, L. (2004)Systematic review of prevalence of aspirin induced asthmaand its implications in clinical practiceBritish Medical Journal, 328, 434–440

Kapolowitz, N. (2001)Drug-induced liver disorders: implications for drugdevelopment and regulationDrug Safety, 24, 483–490

Lazarou, J., Pomeranz, B. H. and Corey, P. N.(1998)Incidence of adverse drug reactions in hospitalised patients –a meta-analysis of prospective studiesJournal of the American Medical Association, 279,1200–1205

Lee, W. M. (2003)Drug-induced hepatotoxicityNew Engand. Journal of Medicine, 349, 474–485

Marshall, E. (2001)Human subjects: Volunteer’s death prompts review Science,292, 2226–2227

Murphy, D. L. (2002)Assessment of Respiratory Function in Safety PharmacologyFundamental & Clinical Pharmacology, 16, 183–196

Ogilvie, R. I. (2001)The death of a volunteer research subject: lessons to belearnedCanadian Medical Association Journal, 165,1335–1336

Olson, H., Betton, G., Robinson, D., Thomas, K.Monro, A. et al. (2000)Concordance of the toxicity of pharmaceuticals in humansand in animalsRegulatory Toxicology and Pharmacology, 32, 56–67

Pritchard, J. F., Jurima-Romet, M., Reimer, M. L.,Mortimer, E., Rolfe, B., and Cayen, M. N. (2003)Making better drugs: decision gates in non-clinicaldevelopmentNature Reviews: Drug Discovery, 2, 542–553

Schellekens, H. (2002)Immunogenicity of therapeutic proteins: clinicalimplications and future prospectsClinical Therapeutics, 24, 1720–1740

Sibille, M., Deigat, N., Jankin, A., Kirkesseli, S.and Vital Durand, D. (1998)Adverse Events in Phase-1 Studies: A Report in 1015Healthy VolunteersEuropean Journal of Clinical Pharmacology, 54, 13–20

stieger, B., Fattinger, K., Madon, J., Kullak-Ublick, G.A., Meier, P.J. (2000)Drug– and oestrogen-induced cholestasis through inhibitionof the hepatocellular bile export pump (Bsep) of rat liverGastrenterology, 118, 422–430

Stephens, M.B. (2004)Stephen’s Detection of Adverse Drug Reaction: FifthEdition. Appendix I: Drug products withdrawn from themarket for safety reasons pages 667–702Edited by john Talbot and Waller john Wiley andSons Ltd.

References


safer medicines report · such studies may support the safe development of novel therapies that may...

Documents