Safer Medicines

A report from theAcademy of Medical Sciences FORUM

November 2005

The Academy of Medical Sciences

10 Carlton House Terrace

London SW1Y 5AH

Tel: 020 7969 5288

Fax: 020 7969 5298

Email: apollo@acmedsci.ac.uk

Web: www.acmedsci.ac.uk

Date of publication: November 2005

Jenny-Cover.qxd 11/9/05 11:29 PM Page 2

The independent Academy of Medical Sciencespromotes advances in medical science and campaignsto ensure these are translated as quickly as possibleinto benefits for patients. The Academy’s Fellows arethe United Kingdom’s leading medical scientists fromacademia, hospitals, industry and the public service.

The aims of the Academy are to:

� Give national and international leadership in themedical sciences

� Promote the application of research to the practiceof medicine and to the advancement of humanhealth and welfare

� Promote the aims and ethos of medical sciences withparticular emphasis on excellence in research andtraining

� Enhance public understanding of the medicalsciences and their impact on society

� Assess and advise on issues of medical science ofpublic concern

The Academy of Medical Sciences was established in1998 following the recommendations of a workinggroup chaired by Sir Michael Atiyah, OM, FRS,HonFMedSci, Past President of the Royal Society. TheAcademy currently has a Fellowship of over 800.

There is an elected Council of 23 Fellows that includesthe five Honorary Officers of the Academy:

President Sir Keith Peters, FRS, PMedSci

Vice-President Sir John Skehel, FRS, FMedSci

Vice-President Sir Michael Rutter, CBE, FRS, FBA,FMedSci

Treasurer Sir Colin Dollery, FMedSci

Registrar Professor Patrick Vallance, FMedSci

The Academy’s Executive Director is Mrs MaryManning.

The FORUM is an active network of scientists fromindustry, the healthcare sector and research funders. Itcurrently focuses on pharmaceutical, biotechnology,medical engineering and health product sectors, withrepresentation from relevant trade organisations.FORUM members have a demonstrable commitmentto the pursuit of research excellence and qualityhealthcare. The FORUM is Chaired by ProfessorBarry Furr, OBE, FMedSci.

For more information about the work of theAcademy or the FORUM, see please www.acmedsci.ac.uk

The Academy of Medical Sciences is a companylimited by guarantee.

Registered Charity No. 1070618

Registered Company No. 3520281

Registered in England

ISBN No. 1–903401–10–0

Printed by Latimer Trend Ltd, Plymouth, UK

Safer Medicines

A report from the Academy’s FORUM with industry

November 2005

Acknowledgements

The Academy of Medical Sciences is most grateful toDr Geoffrey Schild, FMedSci, Dr Gwyn Morgan, SirColin Dollery, FMedSci, and Sir John Skehel, FRS,FMedSci, and to the members of the Working Groupsfor undertaking this report. It would also like to thankthe Review Board, Chaired by Professor PatrickVallance FMedSci, for their instructive comments andsupport.

4

Safer Medicines

Disclaimer

This report is published by the Academy of MedicalSciences and has been endorsed by its Officers andCouncil. Contributions by the Working Groups andrespondents to the consultations are made purely in anadvisory capacity.

The members of the Working Groups participated inthis report in an individual capacity and not asrepresentatives of, or on behalf of, their individualaffiliated hospitals, universities, organisations orassociations (where indicated in the appendices). Theirparticipation should not be taken as an endorsementby these bodies.

© Academy of Medical Sciences 2005

All rights reserved. Apart from fair dealing for thepurpose of private study, research, criticism or review,no part of this publication may be produced, stored ina retrieval system or transmitted in any form, or byany means, without prior permission of the Academyof Medical Sciences.

Cover image: Selection of medicinal pills in a hand.Cristina Pedrazzini/Science Photo Library

5

Safer Medicines

Contents

Page

Summary and recommendations 7

Chapter one: Background 11

Chapter two: Pre-clinical evaluation of safety 15

Chapter three: Safety assessment during pre-marketing clinical trials 19

Chapter four: Safety assessment after marketing 23

Chapter five: Biologicals: vaccines and gene therapies 27

Chapter six: Assessment, communication and management of risk 31

Chapter seven: Building capacity for better safety assessment 37

Appendix 1: Remit 39

Appendix 2: Working Group membership 41

Appendix 3: Glossary and abbreviations 44

Appendix 4: References 45

Appendix 5: Working Group reports on CD-ROM Inside back cover

6

Safer Medicines

7

Safer Medicines

The development of a drug that is both safe andefficacious is a complex and challenging process.While the prime concern of the general public is thata drug is safe and works well, it is the unavoidable casethat drugs bringing genuine benefit will always carrysome degree of risk. This report seeks to explain therisks involved and to describe approaches that mightbe taken to minimise them. Greater transparency andcommunication about the risks and benefits associatedwith drugs would allow the identification of real riskswhere they exist and promote sustained publicconfidence in the safety of medicines.

Achievements in basic science and medical research,along with an ever-faster pace of advance in drugdiscovery and development, have made considerablecontributions to the improvement of human healthover the past 50 years. This is most strikinglydemonstrated in the increasing longevity ofpopulations in the developed world and in theimproved quality of life for many people worldwide.But the traditional tools used to assess drug safety havechanged relatively little over the past 30 years. This isperhaps unsurprising given that the validation of newtechniques is lengthy and complex and requires long-term investment. That said, technologies derived fromthe ‘omic sciences (genomics, proteomics andmetabolomics), as well as the validation of novelbiomarkers and advances in medical imaging arepotentially useful additions to the safety assessmenttoolkit of the future.

This project, initiated by the Academy of MedicalSciences and its industry FORUM, is distinctive inthat it has brought together experts from across therelevant constituencies to take an integrated view ofthe evidence. The recommendations in this report willhave implications for many stakeholders, among themindustry, academia and the regulatory agencies, aswell as patients themselves. The report’srecommendations focus on the following key areas:

� The application of new technologies, specifically topre-clinical and clinical safety assessment

� The need to provide for a rapid and thoroughinvestigation of emergent clinical safety issues

� The role of information, in particular large patientdatabases, in the detection and reduction of adversedrug reactions

� The opportunity to improve the safety of medicinesthrough public engagement on associated risks andbenefits

� The importance of building research capacity insafety assessment

Recommendation 1: Expedite theapplication of new technologies to pre-clinical and clinical safety assessmentthrough collaboration between industry,academia and regulatory agencies.

The ability to analyse, very rapidly, large numbers ofgene transcripts, proteins and metabolic products inblood, urine and tissues offers great promise.Advances in these areas have the potential to aid thedetection of safety signals or characteristic fingerprintsthat define potential risk by improving the speed,sensitivity and specificity of analysis. Such advancesmay make it possible to achieve some reduction in thenumber and duration of exposure of animals used insafety tests, in addition to their ability to contributedirectly to patient safety. These advances may alsohelp to diminish the present high rate of attrition in thedevelopment of new medicines.

The application of new technologies to ensure thatbiological products, including vaccines, are free frompotentially harmful contamination has already hadsuccess. Molecular genetic studies of vaccineorganisms will also provide essential information inpre-clinical studies and improve the detection ofpotential safety problems.

Extensive collaboration will be required betweenresearch-based pharmaceutical companies, academicexperts and regulatory bodies in order to carry out thethorough validation of the new technologies. Aprerequisite is the willingness of industry to sharesafety data. There is an urgent need to facilitate suchpre-competitive joint research, both nationally andinternationally.

Summary and recommendations

Recommendation 2: Rapid and thoroughinvestigation of emergent clinical safetyissues should be facilitated by internationalnetworks between countries with advancedhealthcare systems.

Emergent, severe, safety issues, such as liver necrosisand severe cardiac arrhythmia, are a medicalemergency both for the patients suffering the adverseeffect and for all other patients who are, or might be,treated with the drug. Vaccination and gene therapycan also have potentially serious adverse effects.Currently, relevant information on these safety issues isoften incomplete, collected too late (or not at all) andaccess may be barred by data confidentiality and legalconsiderations.

Countries with advanced healthcare systems have aresponsibility to develop international networks toinvestigate emerging safety issues (liver injury would bean appropriate pilot project). Such networks wouldenable doctors caring for patients with possible newdrug-induced toxicity to respond quickly by contactinga coordinating centre. The centre would then use acommon protocol and could suggest additionalinvestigations including DNA sampling. Such anapproach might allow a useful drug (or class of drug) toremain on the market by identification of geneticpredisposition, drug or disease interactions, andconsequent screening of potential recipients of thedrug. Where an adverse reaction has occurred with abiological product, this should also be vigorouslyinvestigated, including identification of product andhost-related factors in the reaction.

Recommendation 3: Build and use largedatabases of patients to speed the detectionof adverse reactions that increase theincidence of common diseases.

Recent highly publicised examples of adverse drugreactions, such as the increased risk of suicidal thinkingin adolescents taking serotonin re-uptake inhibitors(SSRIs) and the increase in heart attacks in patientstaking COX-2 inhibitors, illustrate the difficulty ofrecognising adverse effects that change the incidence ofa common disease. This problem could be addressedthrough the effective use of information stored on largepatient databases. These databases could be scrutinised

8

Safer Medicines

by sophisticated computer software, searching fordisproportionate changes in disease incidence. Suchdatabases should be used in combination with large-scale, long-term clinical trials and patient follow-up.

The UK National Health Service (NHS) offerscapabilities for combining clinical and prescribinginformation. In principle the whole NHS could beavailable to seek and test safety hypotheses usinganonymised data. The UK clinical research networksfor clinical trials, currently in development, haveconsiderable potential for collecting long-term data onsafety as well as efficacy, because of their ability tomaintain long-term follow-up. Further investment inNHS capacity to evaluate drug safety should be apriority for Government and industry and might wellattract international investment.

Recommendation 4: Reduce the risk ofadverse drug reactions through greaterpublic engagement.

Although adverse reactions to newly marketed drugsreceive much publicity, some of the most frequent andpotentially dangerous reactions arise with long-established drugs. Examples include the anticoagulantwarfarin, the heart stimulant digoxin and non-steroidalanti-inflammatory drugs such as aspirin. Theincidence and/or severity of many of these adversereactions could be reduced by more intelligent use ofmedicines by medical practitioners and betterunderstanding of their risks and benefits by patients.Access to the internet and the new NHS network offeropportunities to explore new methods ofcommunicating intelligible, up-to-date information,and of checking on patient progress and safety.

There is considerable individual variation in theunderstanding of benefits and risks associated withtreatment thus research will be needed to findoptimum methods of communication. Developingbetter means of communicating absolute and relativerisk to patients should be a priority. This is theresponsibility of all stakeholders; industry andregulators must improve communication withphysicians, and physicians must improve incommunication with their patients. We recommendthe development of an agreed, standardised system ofpresenting the risks and benefits of medicines.

9

Safer Medicines

Recommendation 5: Steps should betaken to address the decline in capacity insafety assessment.

Historically, pathological examination of animalsexposed to high doses of a particular drug has been animportant component of pre-clinical safety evaluation.More recently, the development of the safetypharmacology discipline and the use of drugmetabolism to measure exposure levels in blood(rather than dose) have played an increasing role. Inpractice, the safety assessment of medicines requiresinput from many disciplines.

The application of new technologies requires skills inpharmacology, genomics, proteomics, biochemistryand bioinformatics, in addition to the established skillsin pathology, drug metabolism and toxicokinetics. For

clinical safety assessment, skills in pharmacokinetics,epidemiology, genomics, statistics and risk/benefitanalysis are required. With such a variety of disciplinesinvolved in safety assessment, the skill of integratingmany different sources of information to make abalanced judgement is becoming increasinglyimportant.

So far, only limited effort has been made to train non-clinical and clinical scientists to adopt an integratedapproach. Training in safety assessment withinacademia has not kept pace with developments inindustry and it has become an orphan discipline in UKhigher education. The education and training ofspecialists in safety assessment will require exposure toboth active academic research centres and industrialexpertise with encouragement from a nationalsponsoring body. The Academy proposes to examinehow best this can be achieved.

10

Safer Medicines

11

Safer Medicines

1.1 The pharmaceutical revolution, which is stillonly about 70 years old, has transformedmedical care and made a major contribution tothe increased longevity and quality of life in thedeveloped countries. For example, survivalrates in childhood have been improveddramatically over the past 20 years; the deathrate following heart attack has nearly halvedsince the introduction of effective treatments;and patients with human immunodeficiencyvirus (HIV) infection have a substantiallyimproved life expectancy. In addition, theprevention of many serious and life-threateninginfections of children and adults by specificvaccination has led to profound improvementsin child survival and public health.

1.2 Despite these advances, there still remains alarge unmet need for effective therapeuticagents in diseases such as most cancers,degenerative diseases of the nervous system andosteoarthritis. New developments in therapeuticapproaches, such as those involving genetherapy, offer promise in the treatment orprevention of diseases for which currentapproaches are ineffective. Vaccines are likelyto have an increasingly important role in theprevention and therapy of some cancers.

1.3 Spectacular developments in pharmacologyand therapeutics have been accompanied byprogressive improvements in safety assessmentby pre-clinical testing in animals and carefulmonitoring of patients. Most pharmaceuticaltherapy is now very safe if the drugs are usedappropriately. The exceptions are mainly in thefield of cancer chemotherapy where the marginbetween killing malignant cells and damagingnormal tissues may be small or non-existent.

1.4 Patients generally expect that the medicinesthey are prescribed will be highly effectiveand very safe. The reality is that many usefulmodern medicines given to large numbers ofpatients can cause serious harm in a smallminority. For the best results improved methodsof patient selection are required, and safetyissues also need to be identified early in the

course of the introduction of new therapies.Medical care with drugs is a continuum frompharmaceutical research to patient. Whilemanufacturers and regulatory authorities bear aheavy responsibility for ensuring that medicinesare effective and safe, major responsibilities alsorest on doctors and patients. Many instances ofdrug toxicity are potentially avoidable bygreater skill and knowledge in the prescribingdoctors and greater care and understanding bypatients.

1.5 The tools used to assess safety of medicines(animal toxicology and clinical trials prior tomarketing) have changed relatively little overthe past 30 years. The basic processes of safetyassessment involve exposure of animals tograded doses of the drug for up to a year,including levels higher than those that would beused in humans. The animals are monitoredclinically, tested for changes in blood and urinecomposition and, after they are killed, detailedexamination of 40 or more different tissues iscarried out. In the early clinical phase patientsare carefully monitored and, again, regularmeasurements of blood and urine compositionare taken. In addition, if any specific concernsare raised either by the mechanism of action ofthe drug or potential problems identified inanimals, appropriate parameters will bemonitored wherever feasible. These methodsshould not be undervalued because severetoxicity in human has become rare, but there isevidently room for improvement.

1.6 Concerns about safety remain the biggest causeof failure of potential drug molecules to progressas medicines. One measure of the success ofexisting methods for the assessment of safetyonce a drug is entering the development phaseis the proportion of new chemical entities(NCEs) that progress from ‘first-in-human’studies to registration with regulators. Thenumbers vary according to therapeutic targets,but the average success rate of all therapeuticareas is estimated at 11%. In the year 2000,toxicology and clinical safety issues were

Chapter one - Background

1 Kola and Landis, 2004.

estimated to account for approximately 30% ofall instances where a potential new drugfailed to reach the registration phase.1 The totalcost of developing a new drug has beenestimated to be in excess of 800 million dollars.2

If a drug is withdrawn from use before thesedevelopment costs are recouped, there maybe serious financial implications for themanufacturer.

1.7 While current methods of safety assessmentappear to have been quite successful inscreening out compounds that might causetoxicity in a substantial proportion of patients,they have sometimes been less successful atpredicting serious adverse effects that occuronly in a relatively small minority of patients(less than 1 in 1000) and those where the adverseeffect is to increase the incidence of a commondisease. Examples of the former includetroglitazone-induced liver injury and skeletalmuscle toxicity (rhabdomyolysis) fromcerivastatin. Examples of the latter includeincreases in suicidal thinking in depressedadolescents treated with SSRIs and the increasein heart attacks in patients taking certain COX-2 inhibitors used to treat arthritis. These eventshave shaken public confidence in the presentmethods of assessing safety.

1.8 It should be emphasised that, over time,differences of emphasis have emerged in thepre-clinical evaluation of drugs in comparisonwith biologicals. For drugs, the generalapproach indicated above provides commonground for evaluation of different types of drug.In the case of vaccines, classical toxicology is oflimited relevance in many instances, and animaland other pre-clinical tests vary according to thespecific nature of the agent involved, includingits pathological characteristics. Particularly inthe case of live vaccines (containing liveattenuated infectious agents), animal models areof value to investigate that the infectious agent isstably attenuated and lacks pathogenicity orreversion to virulence. They also contribute toinformation that the product is free fromcontamination with infectious agents derived

12

Safer Medicines

from the source biological materials and doesnot possess potentially harmful immunologicalproperties.

1.9 Advances in science and technology over thepast decade have opened new possibilities fordetecting and, in some instances, predictingsusceptibility to drug toxicity but most of thesemethods have yet to be tested in large-scaletrials or accepted by the regulatory bodies thatdecide whether a new drug can be marketed.There is a disparity between what is recognisedas a ‘safe’ drug by regulators and the perceptionof the general public. The launch of a new drugis often accompanied by a large amount offavourable publicity, which may encourageunrealistic expectations from prescribers, themedia and the general public about its safetyand efficacy. Drug withdrawals after licensingare then seen as failures of the system, ratherthan an inevitable consequence of a process thatcan never guarantee complete safety.

1.10 The pharmaceutical industry has a clearobligation to society to ensure that drugs are as‘safe’ as they can possibly be, as it is society thatpays the price when serious adverse eventsoccur. In the most serious cases, the after-effects of drug safety failures can be felt formany years after the event and may delay thedevelopment of important new therapeuticagents. Studies suggest that adverse drugreactions are responsible for over 6% of acutehospital admissions, leading to a projectedannual cost to the NHS of almost £0.5 billion.3

However, even highly toxic drugs do have arole when used under the right conditions andknowledge – for example, thalidomide isregarded as a ‘bad’ drug, but is useful intreating leprosy.

1.11 Regulatory agencies are subject to muchscrutiny following announcements of drugwithdrawals, whether they relate to safety,efficacy or both. The regulatory environment isincreasingly challenging, with rapid growth inknowledge resulting in increasing demand forspecialisation. Significant cross-disciplinaryskills and knowledge are needed to integrate

2 Tufts Center for the Study of Drug Development, Impact report 5: http://csdd.tufts.edu (accessed 5 October 2005).

3 Pirmohammed et al., 2004.

13

Safer Medicines

diverse sets of data from often unrelated fields.These skills are in short supply both inregulatory agencies and industry and arepoorly developed in universities. Regulatoryauthorities are under such pressure to ensuresafety that they may be viewed as increasinglyconservative; this has the potential to increasedrug development time and withholdpotentially useful new medicines in areas ofunmet medical need.

1.12 In the light of all these circumstances, theAcademy of Medical Sciences, through itsFORUM with Industry, resolved to re-examinepresent methods of assessing drug safety and tomake recommendations that might secureimprovements. This project is distinctive onseveral counts: it has brought togetherinterested parties from industry, academia andthe regulatory agencies; addressed issuesrelevant to both pre-clinical and clinicaldisciplines; and covered biological as well aschemically synthesised products.

1.13 The scope of the project ranges widely acrossthe life sciences, from cell biology,pharmacology, pathology and drug metabolismto pharmacogenetics, pharmacogenomics,bioinformatics and molecular and clinicalmedicine. The project remit can be found inAppendix 1. A total of seven Working Groups

were set up to look at various areas in the safetyassessment of drugs, namely:

1. Pre-clinical toxicology

2. Safety pharmacology

3. Pre-marketing (clinical phase I, II and III)assessment

4. Vaccines

5. Gene therapy

6. Risk-benefit assessment

7. Post-marketing surveillance

1.14 Full versions (in electronic form) of eachWorking Group report can be found inAppendix 5 on the attached CD-ROM.

1.15 This report represents a distillation of theimportant conclusions from these workingparties and is one step in a longer-term strategy,the object of which is to advance the progress ofsafe and efficacious new therapeutic modalitiesfor the benefit of all patients. The Academy ofMedical Sciences will continue to build links withother interested parties in taking forward thepresent recommendations and to support, whereappropriate, the initiatives of other bodies. TheAcademy greatly welcomes feedback on any ofthe points raised in this report.

14

Safer Medicines

15

Safer Medicines

2.1 The objective of this chapter is to review, briefly,the present methodology available to assessdrug safety before administration to humansand consider opportunities to apply newtechnologies.

2.2 Pre-clinical safety assessment of drugs reliesheavily on macroscopic and microscopic exam-ination of organs and tissues from at leasttwo species of animals, using a range of drugdoses for periods of up to six months or more.Application of these methods requires highlyskilled scientists who are familiar with bothanimal and human pathology. The mainimprovement in methodology in recent yearshas been to measure the concentrations andkinetics of the drug in the blood and tissues of theanimals used in the safety tests. This is necessarybecause small animals often metabolise drugsmuch more rapidly than humans. Equivalentdoses on a bodyweight basis might thereforeinvolve much lower exposure to the drugs inanimals than would be achieved in humansgiven the same dose per kilogram. Much higheroral doses may be needed to match humanexposure. It is now standard practice to calculatethe maximum exposure in animals that causesno adverse effect. This is known as the NOAEL,meaning no observable adverse effect level.Human exposure is normally kept below theNOAEL, and five- to tenfold below if the initialtoxicity is of a serious kind, such as injury to theheart or liver.

2.3 For drugs and some biological products thesestandard methods are supplemented by a widerange of more specialised safety tests in animalsor isolated cells. These include testing forteratogenic effects upon the embryos of pregnantanimals, genotoxicity assessed by damage toDNA in vitro and in vivo, and very long termexposure (years) of animals to asses the risk ofcarcinogenicity. Assessment of the risksassociated with reproductive toxicity andcarcinogenicity is made almost entirely using pre-clinical data, as it is not possible to obtaincorresponding data in clinical trials or in a timelyway through patient monitoring.

2.4 Although traditionally the application ofconventional toxicological tests has not beenregarded as generally relevant to vaccines, thereare instances where specialised safety tests are ofvalue, particularly for novel biologicals such asthose being developed for gene therapy or thosewhich contain agents that are potentiallyteratogenic.

2.5 Advances in chemical analytical methodology,particularly mass spectrometry, mean that it isusually possible to have a sensitive and specificanalytical method available for measurement ofplasma concentrations in animals and humans.It is normal practice to synthesise a species of thedrug molecule labelled with carbon-14, a long-lived radioactive isotope of carbon. This can beused both to study the tissue distribution of themolecule in rodents and to obtain more detailedinformation about routes of metabolism.

2.6 Studies of the rates and routes of metabolism ofthe drug molecule will also have beenundertaken, using freshly isolated liver cells(hepatocytes) or hepatic microsomal fractionfrom humans and animals and in vivo studies inthe animals. These data are of particular value inmaking predictions about initial doses inhumans but may also indicate potentialproblems from formation of reactive entities.Such in vitro studies are subsequently followedby studies of metabolism in animal species usedin toxicity testing, and in humans, to ensure thatexposure to metabolites has been achieved inanimal species. Early findings in humans mayrequire further testing in animals, for example ifhumans produce abundant quantities of a drugmetabolite that was seen only in much lowerconcentrations in animals.

2.7 An increasingly important area of safetyevaluation before administration to humans istermed safety pharmacology. The aim of thisstep is to make sure that there are no importantfunctional effects upon vital systems, such as theconducting system of the heart, blood pressure,breathing, consciousness, liability to fits, etc. atexposures above those likely to be achieved inhumans. Safety pharmacology studies can also

Chapter two - Pre-clinical evaluation of safety

be conducted to provide a mechanisticunderstanding of adverse effects observed inhumans that were not predicted from earliertesting.

2.8 Everyone concerned with assessing drug safetywishes to minimise the number of animals usedbut it is important for all concerned tounderstand that there are, in most cases, noalternatives. Isolated cells, such as hepatocytes,are very useful for limited purposes such asassessing metabolic routes, but they quickly losemany of their specialised properties whengrown in culture. Furthermore, the mechanismleading to toxicity may require the interaction ofseveral different cell types; consequently muchdrug toxicity depends on long-term complexeffects upon integrated functions that can onlybe assessed in a living animal.

2.9 For several vaccines and other biologicalmedicines there are examples of the use ofsensitive and specific molecular assays, such asthe gene amplification (polymerase chainreaction, PCR) test for the detection ofextraneous contaminating viruses, reducing theneed for biological and animal tests. They arealready also routinely in use for the detection ofblood-borne viruses in blood donations used inthe manufacture of blood products where theyprovide a highly sensitive and reliablemethod for detecting hepatitis B and C virusesand HIV. There are examples of vaccineswhere characterisation of the product byphysicochemical methods is of value for qualitycontrol and related safety; one of these is thestructural analysis by nuclear magneticresonance (NMR) of the polysaccharide-basedmeningitis vaccines now in common use.

Problems and gaps in the presentapproaches

2.10 Inferences drawn from very high exposure. Thehighest doses used in animal safety tests areintended to cause some toxic effect to elucidatewhat type of adverse effects may occur inhumans. Some of these may not precludeadministration to humans but will requirespecial additional monitoring, whereas othersmay be so serious that it would be unacceptable

16

Safer Medicines

to risk their happening in humans, for exampleirreversible damage to cells in the retina. Therehas been a long debate about the significance oftoxic effects seen only at very high exposures.At very high exposures, a compound maygenerate a substantial amount of a chemicallyreactive metabolite that swamps the capacity ofthe normal mechanisms that exist in the body’scells, particularly the liver, to inactivate them.At lower doses the same compound producingthe same metabolite, but in much smalleramounts, may be non-toxic because the reactivespecies is removed efficiently, for exampletrapping by glutathione. It is thereforeimportant to understand that the fulldose–response curve for each compoundshould be characterised.

2.11 Interspecies comparisons. Predictions of safetybetween species (for example rat, dog andhuman) are good but not perfect. A longer-termhope is that knowledge of the genome of themain species of laboratory animal, and ofhumans, will increase the reliability ofpredictions. However, the mechanismsinvolved in toxicity are a complex mix of injury,defence and repair so progress in understandingmay be slow. A special case is when a drug is soselective for the human target that it has littleeffect in the animal species used for safetyassessment, as is the case for most monoclonalantibodies and some organic chemicals. Sometypes of pathology are more common inanimals used for safety assessment than inhumans, and a few pathologies seen in animalsdo not have any known human counterpart.Changes in the bile ducts are relatively commonin rats probably because they are more efficientat excreting drugs in bile than are humans.These situations put a premium on the skill ofexperienced pathologists who have knowledgeof the typical pathology in different species, andon the ability to integrate data on exposure withthe microscopic findings.

2.12 A suitable animal model may not exist forstudies of the safety and efficacy of a newvaccine under development. However, modernmouse genetics provide possibilities ofdeveloping transgenic mice strains with therequired degree of susceptibility to the organismagainst which the vaccine is being developed. A

17

Safer Medicines

successful example for this approach is the useof transgenic mice susceptible to polio (normalmice are not susceptible) which are now widelyused for safety tests of poliovaccine.

2.13 Mechanism of toxicity. Toxic events can bedivided broadly into those caused by the mainpharmacological action of the drug and thosethat are unrelated to it (off target). Thepharmaceutical industry is exploring manynovel targets derived from the human genome,and background knowledge about thebiological role of the target receptor or enzymeis often fairly limited. This limits to some extentthe ability to predict the possible unwantedeffects of the pharmacological action, althoughthe use of modern techniques to define thedistribution of the target in the body provides aninsight into the likely organs that may beaffected by a new medicinal product. Greateropenness with the academic community wouldbe mutually beneficial. Non-target-relatedtoxicity poses even greater problems: a detailedunderstanding of the mechanism of toxicity at amolecular level is rarely available althoughconsiderable progress is being made concerningthe role of chemically reactive metabolites.Even so, the fact that a reactive metabolite isgenerated is far from proof that there will besafety issues in humans. There is a need forgreater scientific investment to expandknowledge in this area.

Opportunities with new technology

2.14 Application of the ’omics. This term is often usedto describe a range of new technologies. Themain ones are: (a) transcriptomics, a method forexamining the expression of thousands of genesusing a DNA chip, often from the Affymetrixcompany; (b) proteomics, the identification ofchanges in the pattern of the thousands ofpeptides and proteins present in plasma or otherbiological samples; and (c) metabolomics, theinvestigation of changes in the pattern ofendogenous chemical products of metabolismin plasma and urine.

2.15 Several pharmaceutical companies are known tohave undertaken studies using transcriptomics

to examine the effects of known hepatotoxiccompounds after short-term administration torodents. These methods appear promising ontwo grounds: a signal can be detected before themicroscopic injury is visible, and the changes inthe pattern of gene expression give valuableclues about the mechanism involved that maynot be readily apparent under the microscope.To move this area forward inter-companycollaboration and collaboration with regulatorybodies should be a priority.

2.16 Proteomics and metabolomics have made lessprogress, partly because the analyticalthroughput of present methods is much lowerthan transcriptomics. However, because of thedifficulty of obtaining human tissue fortranscript analysis, these methods may comeinto their own when translating animal findingsto humans. Eventually such approaches maylead to a database of characteristic ‘omicfingerprints that warn of likely hazard.

2.17 Imaging. The great advantage of imagingtechniques is that they are non-destructive andso the same animal can be examined at severaltime points to assess the evolution andreversibility of a drug-induced lesion. Thistechnology is also often directly transferable tohumans.

2.18 Advances in the application of ’omics andimaging methods require a broadercommitment to the principles and practices ofknowledge management. Full exploitation willrequire the development of computationalresources and people capable of curating andanalysing the resulting data. These databasesshould enable efficient interrogation of all theinformation generated and permit therelationships between animal toxicology dataand clinical drug data to be ascertained. Suchresources will also develop; as toxic signalsemerge datasets can be re-probed to re-assesssafety of molecules in the pipeline or already inthe marketplace.

2.19 The more established technologies are alsoadvancing. Wider use is being made in safetyassessment of more sophisticated microscopictechniques including electron microscopy and

immunocytochemistry. Rapid advances in theknowledge of the role of transporter moleculesin cells are being used to better understandwhat determines the concentration of a drugmolecule within cells - where the toxic effect isusually manifest. For example, many drugs areorganic acids (e.g. the statins) and these canachieve concentrations within liver cells of tento a hundred times higher than in plasma by theaction of organic anion transporters in liver cellmembranes.

2.20 Sophisticated mathematical modelling (SystemsBiology) has an important part to play in

18

Safer Medicines

integrating information gathered about the drugfrom many sources and making predictionsthat may be difficult and/or expensive to testexperimentally.4

2.21 The proliferation of promising new approacheshas not thus far been matched by acorresponding increase in academic research ornational research investment outside industry.Several of the Academy Working Groupsargued that a national centre for drug toxicologywas needed to give impetus to research on thenew approaches and maintain internationalcompetitiveness.

4 The Academy of Medical Sciences is currently undertaking a project with the Royal Academy of Engineering on Systems Biology and these opportunities and challenges willbe addressed further in that project.

19

Safer Medicines

Current process and procedures

3.1 This chapter deals largely with safety issues inclinical trials but it must never be forgotten thatit is the balance between efficacy and safety thatis most important. Patients will accept quitesevere drug toxicity for potential benefit indeadly diseases such as cancer and HIV;conversely drugs intended for less seriousconditions such as the treatment of obesity ormild allergy must be very safe and welltolerated. The question of risk assessment andthe communication of risks and benefits is thesubject of a later chapter.

3.2 Phase I. Before a new drug is given to humansfor the first time, usually in healthy volunteers,there is an extensive review of the safety data onNOAEL in animals (paragraph 2.2) to identifyany concerns that might require exceptionallyintensive or specialised monitoring. The predic-tions made about a likely effective dose will alsobe reviewed, and the starting dose is normallyset at a fraction of the dose that is anticipated tohave any observable effect.

3.3 It is a testimony to the effectiveness of themethods used in pre-clinical safety assessmentthat serious, unexpected, toxicity in humansduring early phase studies is very rare indeed.The effects most likely to be missed are lessserious ones such as headache, sedation,inattention, nausea and fainting that are difficultto detect in animals.

3.4 The first phase consists of a placebo-controlledsingle dose-rising study in groups of 12–16volunteers that will stop when either a pre-setexposure level has been reached (say one fifthof the NOAEL), there are dose-limitingsymptoms, such as nausea and vomiting, orthere are biochemical/haematological safetytests that can raise an alert. Blood samples aretaken at frequent intervals for full blood countsand biochemical tests that may give anindication of damage to the liver, thekidney, skeletal muscle, etc. Urine and

electrocardiogram (ECG) examinations are alsomade and any reported symptoms recorded.Pre-set upper limits for these tests at whichdosing must be stopped will have been defined.An important part of this early study is to obtainreliable data on the plasma concentrations andduration of exposure.

3.5 The next stage is to give a series of doses, usuallyfor 14 days, at levels based on the data from thesingle dose study. Similar safety measurementsare made. It is well recognised that smallincreases (up to two to thee times higher thanthe upper limit of normal) of the ‘liver’ markeralanine aminotransferase (ALT) activity may beobserved on both active drug and on placebo.In many cases these appear to be more relatedto the circumstances of housing in aninvestigational unit and dietary changes than toa drug effect. This illustrates the difficulty indistinguishing between adaptation and toxicity.

3.6 Phase II. The course of phase II varies dependingupon whether the compound has a precedentedaction or whether its action is so novel thatthere is no precedent. For a precedented action,studies will be undertaken in patients with thedisease in which that type of activity is known tobe effective. For an unprecedented action, moreexploratory studies in experimental medicinewill be undertaken using intensive measuringtechniques, sometimes in patients with severaldifferent possible target diseases. As the objectiveis to demonstrate an action, if there is one, thegroups of patients are carefully selected as beinghealthy in other respects and, increasingly often,are chosen as having features that increase thelikelihood that they will respond. The number ofpatients involved is usually small in phase II,ranging from approximately 30 to 200, and theduration of exposure is usually short, rarely morethan 6 weeks.

3.7 One of the main challenges at this stage is tofind the right dose or range of doses that will beeffective but not run a needless risk of toxicityfrom overdose. For some diseases this is

Chapter three - Safety assessment during pre-marketingclinical trials

relatively easy (e.g. asthma or hypertension),whereas in others it can be very difficult(e.g. schizophrenia, Alzheimer’s disease).Where the response is difficult to measure,advanced technologies such as positronemission tomography (PET; to measurereceptor occupancy) may be employed. It is atruism that the most common property of a newdrug that the industrial sponsor gets wrong is thedose. As the choice of dose is often ultimatelybased on the most favourable balance of efficacyto safety, obtaining good dose-response data forboth is of critical importance.

3.8 Several special safety issues have to beaddressed in phase II. The two most commoninvestigations are evaluating the effect of thenew drug on the ECG, and studying its potentialto cause drug interactions by speeding up orslowing down the rate of metabolism of othercritically regulated drugs such as theanticoagulant warfarin.

3.9 Several commonly used drugs with unrelatedactions (e.g. the antihistamine terfenadine) havebeen found to prolong the re-polarisation phaseof the ECG (long QT) and this predisposes to aserious irregularity of the heart beat called‘torsades de pointes’. It is now a requirementthat compounds should undergo an exhaustivetesting in healthy subjects with multiple ECGsto characterise and exclude those compoundscausing anything more than minimal changes incardiac re-polarisation.

3.10 The magnitude of effect and duration of actionof most drugs is determined by their speed ofelimination from the body. In many cases thisinvolves metabolism by enzymes in the liver. Ifa new drug speeds up or slows down the rate ofmetabolism by one of these enzymes it willdecrease or increase the concentration of allother drugs that are metabolised by the sameenzyme, which can have serious consequences.If pre-clinical studies with human drug-metabolising enzymes suggest that such effectsmay occur in the range of expected therapeuticconcentrations, a panel of test substancesmetabolised by different members of theenzyme family chiefly responsible (usuallycytochrome P450s) will be administered withthe test compound to normal volunteers.

20

Safer Medicines

3.11 The first part of phase II (IIA) ends when it isshown that the drug has an action withpotential utility in a human disease, termedproof of concept or PoC, or that it has not. If thePoC is positive it is followed by phase IIB,which is used to finalise the dose and make abetter assessment of therapeutic activity. PhaseIII involves very large confirmatory trials toachieve the necessary data for registration of theproduct. Phase IIB/III involves much largernumbers of patients, usually several thousands.

3.12 Phase IIB/III. These large trials determinewhether a drug will be accepted by regulatorybodies for general use. Their primary aim is todemonstrate efficacy but they add valuableinformation about safety because of the largernumber of patients.

3.13 Patients in these late-phase trials are monitoredat intervals in a similar fashion to those in theearlier studies, although not necessarily sofrequently. Each trial usually has its own safetydata monitoring committee comprising severalexperts who are independent from the mainstudy but have access to, and conduct a regularreview of, data from the study. If serious safetyissues arise, a trial may be suspended while theissue is clarified. A recent example is thesuspension of trials with drugs that affectadhesion molecules, which are necessary forwhite cells to migrate from the blood intotissues. One compound in this class, amonoclonal antibody (Tysabri), was associatedwith several cases of a rare and serious viraldisease of the brain. Studies with compoundswith a similar mechanism of action were placedon clinical hold.

3.14 A limited amount of work is done on the effectsof intercurrent disease of vital organs; forexample it is usual to study changes in thekinetics of the new drug in patients withmoderately severe kidney or liver disease.

Problems and gaps in the present approaches

3.15 Exposure. Although phase IIB/III trials ofteninvolve 5000–10,000 patients, the duration ofexposure is often only a few weeks or months.

21

Safer Medicines

The total number of ‘patient years’5 of exposuremay only be in the mid-hundreds and only asmall minority will have been on the drug for aslong as a year. If toxicity is cumulative overtime, pre-marketing trials are not sensitiveenough to detect it. When toxicity arisesbecause of a low-frequency human geneticpolymorphism, the number of individualsinvolved in the trials and the limited duration ofexposure make it unlikely that it will bedetected. Yet once a compound is marketed,human exposure may quite quickly increase byhundreds or even thousands of fold.

3.16 A recent example relating to the size of thephase III trials concerned a rotavirus vaccineunder development. Phase III trials involvingseveral thousand subjects failed to detect aserious neurological adverse reaction, Guillain-Barré Syndrome. This only became apparent bypost-marketing surveillance during the widerroutine use of the vaccine, which wassubsequently withdrawn. There is an increasingawareness that recombinant therapeuticbiologicals, although of human sequence, mayafter repeated application induce specificantibodies in recipients. This is the case withsome interferons and growth factors.

3.17 Generalisability. The careful selection of patientsfor clinical trials means that the trials may notdetect safety issues arising from seriousintercurrent disease, environmental factors,such as high social drug intake, oradministration of multiple other medicinessimultaneously.

3.18 Biomarkers. It is usually impractical to obtainsamples of human tissue to assess safety duringclinical trials, so reliance has to be placed onsubstances that can be measured in accessiblefluids, mainly blood and urine. Thesesubstances are often referred to collectively as‘biomarkers’. When safety issues arise, such as asignals of liver toxicity, they are almost alwaysdetected by established biomarkers in clinicaltrials before significant tissue injury has takenplace, because of the intensity of monitoring.However, there are some types of pathology

that are more difficult to detect at this earlystage. A good example is a condition termedphospholipidosis.

3.19 Phospholipidosis is a relatively common findingin animal safety tests and involves infiltration ofmany tissues with phospholipids, probablybecause their normal breakdown bymetabolism has been impaired. Despite itsfrequency in animal safety testing it is rarelyfound in humans, although it is often sought.The only reliable way of detecting the conditionin humans, short of finding tissue infiltrationwith phospholipids, is to make electronmicroscopic examination of circulating whitecells in blood. Phospholipidosis is used hereonly as an example; there are others.

3.20 Even when there are existing safety biomarkersthere is often room for improvement. Themarkers used for detecting liver injury, such asALT, are not liver specific, and liver biomarkersthat are both more specific to that organ andthat give a better indication of the type of liverinjury would be of value.

Opportunities with new technology

3.21 Development of new biomarkers for bothefficacy and safety (if feasible) could be of greatvalue in improving the accuracy of conclusionsreached in the clinical phase, particularly in itsearly stages. Acceptance of such biomarkers foranything more than internal managementdecisions will require extensive collaborationbetween the pharmaceutical industry andregulatory authorities. That progress is possibleis beyond reasonable doubt but its extent andtimescale is rigorously debated.

3.22 Imaging methods continue to advance. Althoughimaging is mainly used for efficacymeasurements, this does affect safety. Forexample it is now technically possible tomeasure the effect of treatment on the size of amalignant tumour, the penetration of aradiolabelled drug into it, changes in the tumourblood flow and the viability of the cells within it.

5 Number of patients � fraction of year exposed to drug.

In time these methods will make it possible todetect tumour response earlier, optimise dosesbetter and achieve a better balance betweenefficacy and safety. Imaging is expensive and itis difficult for any but very well fundedacademic centres to compete, but it is of greatimportance for the future.

3.23 Developability. Many of the problems associatedwith predicting drug interactions, variableabsorption, reactive metabolites and the like arebeing addressed efficiently by pharmaceuticalcompanies who screen their compound librariesfor these properties. But the ability to makeaccurate mathematical models of interactionswith drug-metabolising enzymes or the cardiacpotassium channel (which is largely responsiblefor cardiac re-polarisation) may ultimately meanthat such problems can be avoided whenmolecules are designed by medicinal chemists.

3.24 Mathematical modelling of clinical trials. Asnoted earlier, sophisticated modelling has greatpotential and it is possible to envisage a timewhen models could be used to test a greaterrange of possible situations than it is practical toaddress in affordable clinical trials. Usingdemographic, physiological, genetic and in vitroenzyme/transporter kinetic data, the knowledgebase on a compound can be extrapolated andscaled up through biomathematical modellingto predict population pharmacokinetic–pharma-codynamic response. Such an approach permitsthe evaluation of heterogeneity and the activeexploration of those who may be at risk. Theprerequisite science base to conduct thesestudies is established and has FDA support butit is not widely used. A significant hurdle is thelack of sufficient scientists with appropriateexpertise.

22

Safer Medicines

3.25 Genomics. The potential applications ofgenomics to safety are vast. The study of DNAsequence variation in relation to differentialdrug response is the basis for the application ofpharmacogenetics in the development ofpersonalised patient safety.6 Pharmacogeneticshas the potential to transform pharmaceuticalR&D processes and create new industrystandards in efficacy and safety, in terms offocusing effort in phase II–III trials andproviding the tools for ADR profiling inpharmacovigilance. A gene chip (Amplichip;Roche) has recently been made availablecommercially that makes it possible to screenindividuals for common polymorphisms of twoof the important drug-metabolising enzymes(CYP2D6 and CYP2C19).

3.26 No doubt chips to detect enzymepolymorphisms that lead to low frequencysevere toxicity with important drugs such asazathioprine and 5-fluorouracil will follow. Theimportance of single nucleotide polymorphism(SNP) profiling in safety assessment has beendemonstrated by the association of thehypersensitivity reaction to abacavir (a drugused in the management of HIV) with twogenes, TNFalpha-238 and HLA-B57. However,it would be wrong to assume that this approachwill abolish unwanted effects. Indeed it isunlikely to provide a ‘safe’ or ‘not safe’ signal,but rather to give an indication of the alteredprobability of experiencing an unwanted effect.Of particular importance is the possibility toreduce, or largely avoid, toxicities that have animmunological component, as animal safetytests are of little value in predicting them.

6 For more information please see the summary of Allen Roses’ presentation to the Academy’s inaugural FORUM meeting at www.acmedsci.ac.uk/p50evid7.html (accessed 5 October 2005).

23

Safer Medicines

4.1 Once a drug is approved and licensed formarketing, the most challenging test of drugsafety begins. Many of the patients to whom thedrug is prescribed may have other diseaseconditions and be taking other medicines. It isalso likely that these patients will be less carefullymonitored than those involved in clinical trials.Almost all recent safety concerns have arisenafter the compound had been on the market foran appreciable period of time and been taken bylarge numbers of patients. There are severalprocesses in existence for detecting toxicity aftera compound is marketed:

4.2 Spontaneous reporting by healthcare professionalsor patients. An example is the ‘Yellow card’system in the UK and ‘Medwatch’ in the USA.These are maintained by the national drugregulatory agencies, in these cases the MHRAand the FDA, respectively. Manufacturers alsoreceive reports directly and from their fieldrepresentatives and are required to makeregular reports to regulatory agencies.Sophisticated computer programs have beendevised to scrutinise these databases for adisproportionate excess of clinical eventsrelated to a particular form of treatment. Theproblem is that only a very small proportion ofserious adverse effects are reported, even inSweden where reporting is mandatory.

4.3 Formal post-marketing surveillance is anincreasingly common requirement by regulatorybodies, in the form of post-marketing clinicaltrials or observational studies to increase thesafety and efficacy databases. Such trials involvecarefully supervised patients and the incidenceof adverse events is usually low, therefore thestudies need to be very large.

4.4 Monitoring large patient databases. It isincreasingly recognised that large-scaleobservational datasets may be of particular valuein assessing safety signals as they represent what

happens under normal conditions of clinicalpractice. Some have been establishedspecifically for the purpose of monitoring safetysignals, but the greatest opportunity lies in thosethat simply collect routine clinical data.

� Prescription event monitoring (PEM),operated by the Drug Safety Research Unit,uses NHS prescription data to collectinformation from GPs about clinical eventsexperienced by patients taking a selection ofnewly marketed drugs.7

� Databases such as the General PracticeResearch Database (GPRD), which is apopulation-based primary care researchdatabase of computerised, anonymised patientrecords from the major GP software systems.8

The GPRD includes about 300 UK familypractices. Other databases of this type includeTHIN,9 Mediplus and DIN-link (both ofwhich are commercially run)10 and Q research(run by the University of Nottingham).11

� MEMO, a record-linkage database run by theTayside Medicines Monitoring Unit,incorporates data on dispensed medicationfrom general practice, inpatient data andinformation from other national datasets forpatients in the Tayside region.12

� Disease registries: there are over 100 databaseslisted in the directory of clinical diseases.13

4.5 The information in these databases is of variablequality. They do, however, reflect routine usagein the general population and providedenominator data. Several UK databases,particularly the GPRD, are used internationally,which indicates the strength of the informationthey hold.

4.6 In North America, the databases used forpharmacoepidemiological research are generallyhealth maintenance organisations (HMO) claimsdatabases and as such are more designed foradministrative purposes and contain limited

Chapter four - Safety assessment after marketing

7 www.dsru.org/main.html (accessed 5 October 2005).

8 www.gprd.com (accessed 5 October 2005).

9 www.thin-uk.com (accessed 5 October 2005).

10 www.imshealth.com (accessed 5 October 2005).

11 www.nottingham.ac.uk/~mczqres/ (accessed 5 October 2005).

12 www.dundee.ac.uk/memo/ (accessed 5 October 2005).

13 www.docdat.org (accessed 5 October 2005).

clinical information. As a result of this, thespecific information needed to assess a safetyissue may not be available.14

4.7 Important initiatives in Europe include Eudra-Vigilance, a European pharmacovigilancedatabase set up as a result of the EU Clinical TrialsDirective (in addition to the EudraCT trialregistration database). This will include data froman annual safety update, but is not currentlyintended to be accessible to other researchers.This seems like a missed opportunity, and webelieve this database should be made morewidely accessible.

4.8 NHS connecting for Health, which isimplementing the NHS National Programmefor Information Technology (NPfIT), plans toimplement a single system linking NHS patientrecords from any source in an electronic HealthRecord. This presents a unique opportunity tolink drug prescribing to adverse events at alllevels of care, and to allow an analysis of drugsafety at a national level. The NPfIT is limited toEngland, while other systems are beingdeveloped in parallel in the devolved countries.It is important that the different systems arecompatible to allow a UK-wide approach andmaximise the opportunities for major advancesin pharmacovigilance.

4.9 Existing databases also provide opportunitieswhich have yet to be exploited. PEM is mainlyused for signal detection, with MEMO and theGPRD being used to strengthen and refutesignals, to quantify absolute and relative risks,and to identify subpopulations at risk. Someimportant issues, such as the impact of maternaldrug exposure on pregnancy outcome, neonataland early childhood health, and the use ofmedicines in children, could be addressed usingthe systems (with the appropriate ethical reviewand approval).

Problems and gaps in the present approaches

4.10 The main problem with all the spontaneousreporting systems is the very small percentage

24

Safer Medicines

of serious adverse events that are reporteddespite energetic attempts to increase them.There are several explanations. One of themost important is that if an event is wellunderstood, for example bleeding on ananticoagulant or a low white cell count duringcancer chemotherapy, physicians see no needto report it. Cancer specialists might need tofile a report on every second patient onintensive chemotherapy. The second problemis that a small change in the incidence of acommon disease caused by a drug is unlikely tobe recognised unless a sequence of such eventsin a short time forces it upon a doctor’sattention. There is a better chance that an eventwill be reported if it is unusual or known to bea sign of serious toxicity, such as jaundice or avery low white cell count for no apparentreason.

4.11 Formal post-marketing studies are likely to runinto the same problem as pre-marketing trials.They are performed in selected patients who areclosely supervised and may be less likely tosuffer adverse events; if they do the drug willprobably be stopped promptly before theybecome very serious. The most severe adverseevents are likely to occur when a patient persistswith a drug despite the onset of symptoms orwhen the drug is reintroduced after a shortperiod without it. This was the case with someinstances of severe liver toxicity with the anti-diabetic drug troglitazone and serious skinreactions with the anti-epileptic druglamotrigine.

4.12 Large-scale controlled clinical trials. Thecardiovascular safety issues with rofecoxib(Vioxx) and celecoxib (Celebrex) were detectedbecause large-scale and long-durationcontrolled clinical trials were in progress forefficacy endpoints such as Alzheimer’s diseaseor prevention of colon carcinoma. Theseprovide the most secure scientific evidencebecause of the randomly allocated comparisongroup. However, to launch large-scale trialswhose sole endpoint is safety presents ethicaldifficulties and it might be hard to persuadepatients to take part.

14 These databases include Puget Sound, Kaiser Permanente, United Health, MediCaid and Saskatchewan.

25

Safer Medicines

4.13 Large databases. The use of large clinicaldatabases for safety evaluation is increasingrapidly and the more comprehensive theclinical information the more useful they are.Their only real drawback is that patients are notallocated to treatment randomly and an excessor deficit of events might reflect the prescriber’sjudgement about the disease features in thatpatient. To some extent this problem can bemanaged by making statistical adjustments fordisease severity. Unless the databases are verylarge they are of limited value for less widelyused drugs.

Opportunities with new technology

4.14 The expansion and enrichment of searchableclinical safety databases is of the utmostimportance in improving the safety monitoringof marketed medicines. An important factor inthe use of these datasets is improving the linksbetween them. The implementation of NPfIT ispotentially of great importance, but it will alsobe necessary to maintain and strengthen the useof current databases, such as the GPRD, untilNPfIT can at least replace their researchcapacity.

4.15 There is significant frustration within theresearch community about the constraints onsharing and research use of patient data inpursuit of public health objectives. It is importantthat clear guidelines for the interpretation of theData Protection Act are developed to allow thesharing of valuable drug safety data whilepreserving patient confidentiality.

4.16 One approach to the issue of consent,particularly in circumstances where it isimpossible or very difficult to obtain consentfrom all individuals but it is important that all areincluded (as in the case of pharmacovigilance), isto ensure that all data are completelyanonymised. From a legal standpoint, there isprovision within legislation to obtain patient datawithout consent where there is ‘overridingpublic interest’. It may be argued that disclosure

of information relating to the safety of drugs isin the public interest, provided that suchdisclosure is to a specific authority and patientconfidentiality is respected.15

4.17 The process of ethical review may benefit frombeing more streamlined. At present, a researchermay have to go through several approvalprocesses including the appropriate fundingbody, an NHS ethics committee, an independentscientific panel (for example when usingdatabases such as the GPRD), data protectionofficers and sometimes the R&D departmentswithin each hospital trust. There have also beensignificant costs associated with using some ofthese datasets derived from NHS patients. Forexample, for several years access to GPRD datawas so expensive that the bulk of the work usingit could only take place in groups in the USA. It isessential that NHS datasets that could be used topick up safety signals, verify safety profiles or testhypotheses related to safety can be accessed byresearchers.

4.18 The limitations of these databases are largelyrelated to the breadth and quality of the datathey contain. Both the GPRD and PEMschemes contain information from GPs only,and therefore are unable to provideinformation on drugs which are mainly used inhospitals, for example cancer chemotherapy.PEM is a voluntary scheme which does notprovide any financial incentives, and as a resultat least 30% of GPs choose not to complete theappropriate paperwork. This creates a potentialbias in the data, the effects of which areunknown.

4.19 Another limiting factor is that for some of thedatabases GPs do not record social data such asoccupation, marital status and employment statusin a routine and standardised format. As a resultthe accuracy of these data in terms ofcategorisation is variable.16 Investment inexpertise to adjust data for social class anddevelop techniques to allow stratification for othersocial factors could allow better use of theresource. It may also form the foundation for

15 The Academy of Medical Sciences is undertaking a project ‘Personal Data for Public Good: using health information in medical research’ to develop proposals on key issues ofconsent, security of data, confidentiality and public engagement; therefore these issues will not be covered in detail in this report.

16 Wong, 1999.

beginning to use NHS datasets to pick up safetysignals as well as testing specific hypotheses.Analysis of large routine clinical datasets in theUSA demonstrated that the Vioxx risk wasdetectable once the risk had been identified inclinical trials, but development of statistical and

26

Safer Medicines

bioinformatics tools will be needed in order to usesuch resources to identify warning signals. Werecommend investment in such expertise, someof which is already available within the defenceindustry in the scanning of electronic traffic forunexpected signals.

27

Safer Medicines

5.1 Biologicals make increasingly importantcontributions to the prevention and therapy ofdiseases. The rate of development andintroduction of new biologicals has acceleratedover the past two decades. New products includevaccines and recombinant-derived therapeuticsincluding interferons, hormones, immunoglob-ulins and blood factors. Monoclonal antibodiesfor therapy, targeting of drugs and diagnosis areother examples of the expansion of biologicals.In some cases recombinant-derived productshave replaced substances extracted from humanor animal tissues, for example human growthhormone, with clear advantages for productsafety and consistency.

5.2 There are contrasts in the approaches taken tosecure the safety of biologicals, compared withchemical drugs, arising out of their nature andproduction. In general, the toxicologicalapproaches applied to chemical drugs are ofonly limited relevance to biologicals. Biologicalsare molecularly complex and oftenheterogeneous in composition. They cannot,with few exceptions, be characterised in precisechemical terms and require biological assays fortheir characterisation. For drugs, detailedchemical analysis is the key approach to qualityand product consistency. Biologicals have agreater potential for batch-to-batch variation inproduction. Arising from the biological natureof the source materials, there are special safetyissues concerning potential contamination withextraneous agents and special attention needs tobe given to ensuring that harmful agents are notpresent.

5.3 Despite the historical differences in approach tosecuring the safety of biologicals and chemicaldrugs, recently there has been a trend towards acommon approach, so that similar principles areapplied to both groups of medicines by themajor international and national regulatoryauthorities. This is a positive developmentwhich will favour the adoption of the mostvaluable aspects of the approaches taken forbiologicals and chemical drugs.

5.4 It is not the intention to address the whole rangeof biological medicines here, but to use vaccinesas examples of products widely used and genetherapy as an example of new medicaldevelopments.

Vaccines

5.5 As vaccines are used to prevent infection anddisease in healthy individuals, primarilychildren, there is a major emphasis on safetyby manufacturers, vaccinators, regulatoryorganisations and the public. In particular,standardisation is recognised as a key element inthe assurance of quality and safety, reflectingimportant lessons learned from past events.International guidelines relating to themanufacture and testing of well-establishedvaccines are provided by the World HealthOrganization (WHO)17 and, for example, theEuropean Pharmacopoeia. The production ofvaccines and securing their efficacy and safetyhas much benefited from the availability ofinternational biological reference preparationsdistributed on behalf of the WHO. Theseprovide ‘yardsticks’ for internationally acceptedunits of measurement and are used as keyreagents in the standardisation of laboratorytests. Several hundred biological standards,relevant to a wide range of products, aredistributed by the UK National Institute forBiological Standards and Control on behalf ofthe WHO.

5.6 Another special feature of the regulation ofvaccines is the requirement in the UK and otherEU countries for batches of each production lotof vaccine to be independently tested by anational or international authority forcompliance with licence specifications. Thisincludes biological assays relevant to safety.

5.7 To complement and in some instances toreplace biological tests, progress is being madein the application of advanced physicochemicalmethods to characterise the structure of vaccine

Chapter five - Biologicals: vaccines and gene therapies

17 See WHO Guidelines on Principles for Non-clinical Evaluation of Vaccines.

materials. NMR spectroscopy has, for example,been found to be of value in the characterisationand quality control of meningitis vaccinesinvolving bacterial polysaccharide–proteinconjugates. It is recommended that efforts bemade to expand such structural studies, whichare likely to make a contribution to safety as wellas efficacy.

5.8 Testing in animals is required for some vaccines.However, there are important opportunities forreplacing some animal models by using, forexample, molecular markers for virulence, as isthe case for poliovaccine. In addition, in vitroassays in cell culture are being developed toassess the presence of toxins for some vaccines,thus replacing animals.

5.9 Research work to develop and evaluate non-animal alternatives should be given highpriority by industry and funding bodies. Thereis currently a strong contention that thevalidation of surrogate tests is crucial butunderfunded.

5.10 The quality and safety of a vaccine depends to alarge extent on the source materials andmanufacturing conditions under which it ismade. Methods for the control of startingmaterials for production, as for the vaccinesthemselves, and the detection of potentialcontamination exist in many cases, but thereremains an important need to develop andintroduce additional and more sensitive tests,for example for the detection of prion materials.

5.11 Many vaccines are based on attenuated, liveviruses or bacteria. The securing of safetydepends on an understanding of the mechanismof attenuation and its genetic stability. Inaddition, better understanding of thepathogenesis of the natural infection, includingsites of replication and mechanism of cell andtissue damage, is important to safety andresearch in this key area and should beincreased.

5.12 Regarding clinical trials of vaccines, moreinformation could be extracted from these bystudies of the response to vaccination using, for

28

Safer Medicines

example, the ’omics technologies describedearlier in this document. The use of these newerapproaches may be expected to improvepredictions of the likelihood of adverse events.Long-term follow-up of vaccine recipients withactive surveillance for potential adversereactions is important and will benefit fromlinkage to vaccine databases. The availability ofsuch long-term information would have been aconsiderable advantage in dealing with therecent issue of the safety of the measles, mumpsand rubella (MMR) vaccine, which was basedon speculation derived from a handful ofclinical observations.

5.13 Valuable input into the content and conclusionsof this report was provided by a symposiumProgress Towards Assuring the Safety of Vaccinesorganised by the Academy of Medical Sciencesin collaboration with the UK Health ProtectionAgency in April 2004.18

Gene therapy

5.14 Gene therapy is perhaps the most recent noveland highly innovative approach to developingtreatment for a variety of diseases includingsome that have proven difficult to deal with byexisting approaches. The strategy is to transferfunctional genetic material into human somatictarget cells, in order to supplant the defectivegene function of the patient. Genes relevant totherapeutic or preventative functions of the hostare delivered in specifically designed vectors,generally of viral nature.

5.15 No gene therapy product is yet licensedanywhere in the world, but a large number ofsmall clinical trials are in progress in the USA,Europe and Japan. There is much researchand developmental work in progress inacademia, biotechnology companies and thepharmaceutical industry. Current clinical trialscover therapy of monogenic inherited diseases,primarily haematological disorders, cancers,cardiovascular disease and intractable infectionssuch as HIV. Gene therapy has the potential tobe the source of a valuable class of medicinal

18 See www.acmedsci.ac.uk/p50evid4.html for further details (accessed 5 October 2005).

29

Safer Medicines

products of the future should the clinical studiesshow successful results.

5.16 The main general safety concerns for genetherapy area as follows:

� The possibility of insertional mutagenesis in aregion of the recipient genome thatpredisposes to malignancy.

� Adverse effects induced by the proteinsexpressed by the introduced gene. There is apotential for autoimmune responses andother harmful immune response as well asother manifestations of the expressedproteins, which may be produced inphysiologically unusually large quantities andin abnormal sites.

� Exposure of germ cells to the introducedgene, which is specifically prohibited bycurrent UK and EU rules.

� Contamination of the gene therapy productsby potentially harmful viruses used in theproduction process.

5.17 Clinical trials of gene therapy approaches needto pay particular attention to measures to avoidadventitious exposure of the treated person withthe genetically modified organisms used as genevectors.

5.18 In relation to safety it is clearly important todevelop and apply appropriate test methodsfor individual vectors, which will include studiesin cell cultures and, where appropriate, inanimal models. For example, recent clinicalstudies in very young children with severeimmunodeficiency in relation to SCID-XI

treatment, which were associated with the onsetof leukaemia, provided the stimulus for researchin mouse models relating to integration andleukaemia development. The ability topredetermine favoured sites of integration in thegenome so as to avoid harmful mutagenesis is amajor challenge for long-term research.

5.19 Gene therapy has considerable potential to leadto effective treatments for conditions wherenone exist at present. It is important that theproducts themselves should be wellcharacterised by laboratory tests for purity,specificity and consistency. Novel biologicalassays will be required for use in pre-clinicalstudies.

5.20 There is much to be done in studies assessingthe risks of randomly inserted genes and incomparing vectors in relation to various cellpopulations. It is important that regulatoryauthorities address the complex issues of safetyfor gene therapy, that clinical studies are welldesigned to yield the maximum possibleinformation and that great attention be given tothe health and welfare of individuals involved inthe studies. As yet it is too early to predict in abroad sense the adverse effects that may beencountered when more and larger clinicaltrials are undertaken.

5.21 It is a positive consideration that several of theregulatory and testing approaches which haveserved us so well in securing safe and effectivevaccines, and some of the lessons learned fromthese, are entirely relevant to safety in the fieldof gene therapy.

30

Safer Medicines

31

Safer Medicines

The present situation

6.1 Although great efforts are made to ensure thatmedicines are used safely, risk is inherent in anytherapeutic intervention. There is no such thingas an absolutely safe medicine. Regulatoryauthorities pay close attention to the efficacyclaims made, and precautions needed, forproper use of new medicines, and theserequirements are reflected in the labelling of themedicine and in promotional material. But,under time pressure in medical practice andactive promotion by company representatives,the detailed indications and warnings may beoverlooked or forgotten. Considering thepotency of many medicines, and their potentialfor harm if not used correctly, relatively littleeffort is expended on balanced, impartialeducation of medical students and practisingdoctors in therapeutics.

6.2 Publications of the results of clinical trials areprone to describe the efficacy of the treatmentregimes in detail but often give only a relativelybrief account of safety issues. This problem isnot confined to industry-sponsored trials andmay reflect the views of authors and editors thatreaders are less interested in such information.The lack of detail on safety results makes itdifficult or impossible to carry out the kind ofmeta-analysis of safety data across many trialsthat are performed routinely for efficacy results,for example in Cochrane systematic reviews.Posting full safety data on company websiteswould be very helpful and academicallysponsored trials should aim to do no less.A standardised format for the presentation ofbenefit and risk would help.

6.3 Medicines are often launched onto the marketamid a flood of generally favourable publicity,including lectures by leading medical expertswho have worked on the drug (known inindustry as Key Opinion Leaders or KOLs). No

doubt most are motivated by genuineenthusiasm but they rarely convey a balancedview. The volume of publicity may encourageunrealistic expectations in prescribers and thegeneral public about the safety and efficacy ofthe new drug.

6.4 As the time taken to develop a drug increases,there is a consequent reduction in the periodbefore patent expiry for marketed drugs. Thishas led companies to attempt to grow sales asfast as possible to maximise returns - assisted inthe USA by direct to consumer advertising.However, unless companies are allowed topromote new products responsibly patients mayfail to benefit from them and the UK is oneof the slowest to adopt new products in thedeveloped world. The appropriate period ofcommercial exclusivity for a new drug needsinternational consideration in conjunction withproposals for more controlled entry into themarket for a new drug.

6.5 In the UK, advertising from pharmaceuticalcompanies focuses instead on prescribers, ashighlighted in a recent House of Commonsreport.19 Drugs that are available over thecounter (OTC) may be advertised to patients,but those that are available by prescription onlymay not.

6.6 While considerable efforts are made byregulatory authorities and responsiblepharmaceutical companies to monitor safetyissues and alert prescribers about them, untilrecently this has been regarded as an activitysomewhat separate from the critical need todemonstrate efficacy.

6.7 In the rush to achieve maximum sales it isimportant that safety concerns are notminimised. The perception that importantsafety data on some COX-2 inhibitors were notrevealed in a timely manner has the potential to

Chapter six - Assessment, communication and management of risk

19 The influence of the pharmaceutical industry, 2005.

cause problems for the entire class of drug andmore than outweigh the marketing advantageinitially created by overemphasising thepotential benefits.

6.8 Carefully planned risk management can be veryeffective, and several companies have been ableto keep on the market important medicines witha safety profile which might otherwise havebeen considered unacceptable. An example isthe use of intravenous aminoglycosideantibiotics such as gentamycin in severeinfections. These drugs can severely damagehearing and balance if the concentration is toohigh, as can easily happen in a very ill patientwhose kidney function is impaired. Physicians’knowledge of this problem and routinemonitoring of the plasma concentrations of theantibiotic have largely avoided severe toxicity.Another, more recent, example of effective riskmanagement is the use of the HIV drug,abacavir (see section 3.26).

6.9 One of the most difficult areas is the provision ofintelligible information for patients. A minorityof patients read the package inserts providedwith medicines and a very few read them intheir entirety.20 These leaflets present a long listof side effects with no attempt to categorise byincidence or severity let alone likelihood for anindividual. The short average contact timebetween family doctors and their patientsrestricts the amount of person-to-personinformation imparted, although this type ofcontact probably has the highest impact.

6.10 Specialist patient support organisations such asCancerBACUP provide some of the mostdependable information. More patients areturning to the Internet for information abouttheir diagnosis and their treatment, but themany sources available on the web vary greatlyin reliability. A basic difficulty is that mostpatients have a very limited knowledge ofbiology and medicine: words that have a veryprecise meaning to a doctor may conveyrelatively little to a patient.

32

Safer Medicines

Opportunities for improvement

Informing prescribing doctors and involving patients

6.11 The front line of drug safety is in the doctor’sclinic when he or she writes a prescription andhands it to the patient. Clear warnings andcarefully worded advice at this stage can have areal impact. Many serious adverse reactionsoccur not with new products (although theseachieve most of the publicity) but with longer-established ones that are supplied largely bygeneric manufacturers. High on the list areanticoagulants such as warfarin, non-steroidalanti-inflammatory drugs and drugs used to treatheart disease such as digoxin. Given the verywide variety of human disease and the greatlyexpanded range of treatment options, somemistakes are inevitable but great efforts must bemade to keep these to a minimum. The USAInstitute of Medicine report To Err is Human:Building a Safer Health System reviews many ofthe issues arising from medical errors.21

6.12 One option that already exists but needs furtherdevelopment is to use intelligent software withinthe prescriber’s IT system to remind her or himwhen a problem seems likely. This is not as easyas it sounds because the software has to beintelligent enough not to keep issuing warningswhen the problem is non-existent or relativelyminor, for if it does it will eventually be turnedoff or ignored. This is an area that needs furtherdevelopment. Given the rapid expansion ofknowledge about mechanisms of toxicity it isreasonable to assume that most practitioners arenot fully up-to-date on safety issues andeducational efforts are needed. In thisconnection the Academy notes with regret theapparent decline in teaching of clinicalpharmacology and therapeutics in manymedical schools.

6.13 The patient is the person who cares most aboutthe outcome of treatment. Development ofimproved methods of communicating the risksand benefits of drugs, in order to assist the

20 Raynor and Knapp, 2000.

21 Kohn et al., 1999.

33

Safer Medicines

doctor’s prescribing and the patient’sunderstanding, is important. For example,relating the risks and benefits to commonsituations and giving absolute as well as relativerisks may aid understanding. Industry,regulators and the National Institute for Healthand Clinical Excellence (NICE) ought to worktogether to develop standardised approaches topresentation of risk and benefit. A recent articlein the New England Journal of Medicine useda matrix of open circles, some of which werefilled in black to illustrate degrees of risk.22 Agenerally agreed method of portraying seriousrisks and major benefits, extensively tested withpatients and doctors, would be very valuable.

6.14 Most households now have Internet access.Many patients now use the web as an importantsource of information. Its capability is only justbeginning to be exploited as a means of helpingpatients on long-term drug therapy and there isa need to help patients identify reliable sources.It is readily possible to foresee the day when aswell as a ‘Dear Doctor’ letter from apharmaceutical company advising physicians ofa drug safety problem there will be aconfidential ‘Dear Patient’ letter sent almostsimultaneously via the Internet by doctors to alltheir patients (by name) who are recorded ashaving an active prescription for the drug inquestion. The use of the web as a means ofperiodic bidirectional communication betweendoctors and their patients, to check on progressand to remind and advise, could also haveimportant applications in achieving safer use ofdrugs. Communications between doctors andpatients in the UK NHS are very slow andprimitive by current standards.

Regulatory and industrial oversight

6.15 Considerations of efficacy and safety should beinseparable for all medicines. For that reason theAcademy does not consider that a separateagency dealing with safety is desirable, but itwould like to see greater emphasis on safetywithin existing structures. In the USA, thecreation of a Drug Safety Oversight Board withinthe FDA’s Centre for Drug Evaluation and

Research has recently been announced. The UKand the EU should consider whether such asystem would be helpful and how we could workclosely with the USA. Within pharmaceuticalcompanies there is a need for close integration ofthose staff dealing with product efficacy andthose with safety.

6.16 This new American Board will be responsiblefor providing enhanced oversight andtransparency on drug safety issues: managinginformation on risks; resolving scientificdisagreements on safety data; developing drugsafety policies; and in increasing the flow of databetween doctors and patients. The newcommunication channels are: the Drug Watchweb page, to include emerging safetyinformation on newly approved and olderdrugs; information sheets for healthcareprofessionals; and patient information sheets tocontain new safety information as well as basicinformation about how to use the drug. Placingdata in the public domain is not enough: thereneeds to be a credible independent body toanalyse the data, identify genuine safetyconcerns and dismiss false safety concerns. Tobe credible, such a board would need to be seento be independent.

6.17 In the USA, collaboration on a new electronicmedicine safety monitoring system ‘MedNet’across government agencies and the privatesector will create an active surveillance andevaluation programme for monitoring marketeddrugs. It is of interest that the FDA has a contractwith the UK GPRD to obtain safety information.The NHS provides an unrivalled opportunity tocollect and analyse computerised data on clinicalcare that could be analysed to detect safetysignals for drugs.

6.18 The new functions of the Drug Safety OversightBoard will be integrated with the currentresponsibilities of the FDA rather than creatingan entirely new regulatory agency to monitordrugs after approval. Clearly, there are alsoimplications for ensuring that pharmacovigilanceand communication are optimised in Europe.Since drugs are used globally, it seems sensible

22 Elmore and Gigerenzer, 2005.

that safety systems are also globalised, and closerworking between European and USA agenciesshould be a priority.

Conditional approval?

6.19 Other radical options are also being entertainedas part of the continuing dialogue betweencompanies and regulators. One of these is aprovisional licensing system, with drugapproval confirmed subject to evidence ofefficacy and safety from widespread use.Defining the nature of the probationaryrequirements may not be easy, but lessons canbe learned from the pilot schemes alreadyagreed for risk sharing after approval. Forexample, studies on disease-modifying multiplesclerosis drugs23 and a statin24 representpartnership between the manufacturer andhealth services to deliver agreed performance interms of cost effectiveness. However, ifconditional approval becomes another burdenof cost and delay in marketing new drugs, it willbe counterproductive.

6.20 Conditional approval with monitoring ofsubstantial numbers of patients might allowdrugs to be approved with less onerous pre-marketing trials. It will also requireconsideration of the patent life for a new drug toensure companies have an opportunity toobtain appropriate benefit from R&Dinvestment. Currently many new drugs are usedwith little attempt to collect data. A conditionalapproval system could allow early entry of newdrugs onto the NHS with the real prospect ofgenerating important safety data.

6.21 While the focus of existing studies of this typehas been on efficacy, the determination of safetyis probably best suited to this approach whereasefficacy is better assessed in randomised trials.There should be further discussion of the costsand benefits of conditional approval. There willbe difficult points to resolve in ensuring thatpost-marketing studies cover all of the relevantpatient population, for example to reflect the

34

Safer Medicines

natural incidence of co-morbidities, rather thanimposing exclusion criteria (excluding patientsat higher risk) so as to accumulate a ‘clean’ dataset.

Risk management

6.22 A vital part of product stewardship is riskmanagement. The risk management strategyought to go far beyond collection and analysisof safety reports. An essential part of riskmanagement is to attempt to foresee the possibleproblems a compound may encounter when itcomes into general use, including secondarypharmacological effects, formation of reactivemetabolites, failure to observe contraindications,mistakes in dose, serious concurrent diseases andtheir treatment and, increasingly importantly,genetic polymorphisms. It is likely that new dataon the impact of genetic polymorphisms anddebates about their application will come todominate future discussion of low frequency butserious adverse effects. Ought a pharmaceuticalcompany to consider the impact of apolymorphism, or more difficult still, acombination of polymorphisms affecting thesafety of their product present in 10%, 1%, 0.1%or 0.01% of a population? In the past productshave been withdrawn (e.g. the antibioticchloramphenicol that rarely causedagranulocytosis) when the life-threateningadverse effect occurred in only about 1 in100,000 patients treated.

6.23 These are perplexing issues that deserveinformed public debate. Companies are nowrequired to have a risk-management strategy foreach marketed product and one consequence islikely to be closer integration of the safety andefficacy aspects of product development.Regulatory agencies should make public theagreed risk-management strategy for newproducts. An effective, well implemented, risk-management strategy should allow manyproducts with safety issues to remain on themarket because the means to avoid the risk invulnerable patients has been developed.Pharmaceutical companies, as well as patients,

23 Miller, 2003.

24 Chapman et al., 2003.

35

Safer Medicines

have much to gain from effective riskmanagement and clear presentation of theapproach.

Crisis management

6.24 During the seven decades or so of moderntherapeutics there have been several seriouscrises. For example, deformed babies withthalidomide, serious eye, skin and peritonealchanges with practolol, liver toxicity withtroglitazone, increased incidence of heartattacks with rofecoxib and other COX-2inhibitors, etc. Several major crises in relation tovaccination have occurred. One in the USA inthe late 1950s was the ‘Cutter incident’ thatinvolved several hundred cases of polio inrecipients of Salk poliovaccine. This occurreddue to incomplete inactivation of the virusduring vaccine production, and wassubsequently remedied by improvements inproduction techniques and testing. The eventprecipitated a major improvement in emphasison quality control and safety testing. Themisconception in recent years that measles,mumps and rubella (MMR) live vaccine wasassociated with chronic inflammatory boweldisease and autism in children provoked seriouspublic concern and had a lasting negativeimpact on vaccine uptake and public confidenceand vaccination safety.

6.25 Handling these crises has proved difficult. Theinitial reaction of pharmaceutical company staff,some of whose lives’ work has been thedevelopment of the compound, may be one of

disbelief or denial. Pharmaceutical companiesand regulatory authorities appear to react slowlybecause the information reaching them at theoutset is often limited and poorly authenticated.News media create a storm of concern that maylead patients to discontinue needed treatment.Lawyers begin trawling for patients who mayhave been harmed to launch class actions fordamages and bar access to their clients whenadditional clinical details are sought. Thereought to be a better way of handling crises.There is a need to be prepared, and an essentialelement of a speedy and sensible reaction is togain access to reliable, detailed, clinical data onindex cases as soon as possible.

6.26 Large clinical databases are a promising sourceof information for crisis management but theexisting ones all have some limitations. HMOdatabases in the USA have limited clinicalinformation, and the UK GPRD does notinclude hospital information. As most severereactions end up in hospital, and it is to someextent predictable what vital organs are mostlikely to be affected, for example liver, blood,kidney or heart, clinical networks organised bymedical specialist societies (national andinternational) in collaboration with regulatoryagencies could play a critical role, withappropriate IT support. In some places thesearrangements already exist but they need to beput on a firmer basis. A truly independent bodycharged with responsibility for safety might bea mechanism to ensure that true safety concernslead to prompt action and that false ones arerapidly dismissed.

36

Safer Medicines

37

Safer Medicines

Infrastructure, training, manpower and research requirements

7.1 Safety assessment for pharmaceutical productsis a rapidly developing and evolving, multi-disciplinary activity. As populations age it islikely that in future approaching 50% of thepopulation will be taking a therapeutic drug andmost older patients will be taking severaldifferent medicines at any one time. Given thishigh exposure of the population it is surprisingthat drug safety does not figure higher on thenational medical research agenda.

7.2 It is a testimony to the considerable efforts madeby the pharmaceutical industry and theregulatory bodies that most medicines arerelatively safe when used as recommended. But,as the use of medicines extends to less severemedical conditions and enhancing quality oflife, the safety requirements become ever morestringent. The boundary between social drugs,such as caffeine and alcohol, and therapeuticagents designed to enhance intellectual orphysical performance in older people maybecome ever more blurred.

7.3 The number of disciplines that contribute tosafety assessment has continued to increase andit is readily foreseeable that human genetics andsocial and environmental sciences, particularlythe former, will play an increasing role. Whilethe largest pharmaceutical companies haveexpanded their capacity in safety assessmentacross a wider range of disciplines, theregulatory bodies and the major academiccentres have lagged behind. The skill ofintegrating many different sources ofinformation to make a balanced judgement isrelatively rare and only limited effort has beenmade to train non-clinical and clinical scientiststo adopt an integrated approach.

7.4 At present most matters in relation to drugefficacy and safety are regarded by governmentand research funding agencies as, primarily, afinancial responsibility of the research-based

pharmaceutical industry. As a greater proportionof the medicines in use become generic, withactive government encouragement, this positionbecomes less tenable.

7.5 It is doubtful that a major initiative on thevalidation of new methods of evaluating safetycould be carried out in a single Europeancountry. Studies are likely to involveinterdisciplinary and cross-sector collaboration.One example of this type of collaboration is thatof a large consortium of pharmaceuticalcompanies which has bid for funding from theEuropean Framework Programme. The bidincludes a proposal (called PredTox) to establishprocesses to improve the predictability oftoxicology experiments by integrating the new‘omics technologies.

7.6 PredTox involves the construction of anintegrated database with information on animalexperiments of compounds with known toxicityprofiles to compare traditional endpoints with‘omics data. One aim of the project is tofacilitate access to new technologies for EUregulatory authorities, and to help defineguidelines for use and interpretation of the data.This should be made available in the publicdomain.

7.7 There are a great many opportunities toprogress better sharing of company safety dataacross similar classes of compounds and toembark on collaborative research. Acommitment to partnership across thestakeholder constituencies can be expected toyield considerable progress in animal andmodel development and accept-ance, clarification of contribution by newtechnologies and identification of ADRmechanisms. The assessment of safety needs tobe seen as ‘pre-competitive’ and is tooimportant to be left as a competitive activitywith secrecy preventing progress. Thiscollaboration could usefully be extended totraining the next generation of safety scientistsand physicians as a joint venture of companies,

Chapter seven - Building capacity for better safety assessment

regulatory agencies and academia. But to makethis possible there would have to be acontribution of government funds. Werecommend that systems to explore sharing ofpre-competitive safety data are explored as amatter of urgency.

7.8 Academia needs to develop expertise in thebiology and medicine of safety assessment. Thiswill require establishing a limited numbers ofcentres of excellence with requisite critical massin the ‘omics, bioinformatics, systems biologyand imaging technologies, as well as expertise inin vivo studies in animals and humans such asexperimental medicine to conduct mechanisticresearch on potential or real adverse effects inhumans.

7.9 Several of the Academy working parties arguedthat there was a need for a UK National Centrefor Drug Safety to span the wide range ofdisciplines involved and sponsor research, forexample on new biomarkers from animals tohumans and on genetic factors in susceptibilityto toxicity. Such a centre could also draw uptraining programmes in consultation withinterest parties in industry. They argued that itsexistence would enhance the competitiveness ofthe UK as a centre for pharmaceuticalresearch.25

38

Safer Medicines

7.10 The UK already has a substantial number ofgovernmental (the Biotechnology and BiologicalSciences Research Council (BBSRC), MedicalResearch Council (MRC), NHS R&D) andcharitable bodies (the Wellcome Trust, CancerResearch UK (CRUK), the British HeartFoundation (BHF), the Association of MedicalResearch councils (AMRC), etc.) supportingbiomedical research and the Academy is hesitantabout recommending formation of anotherwithout very careful consideration. Safetyassessment transcends the remit of the maingovernmental research funding agencies as wellas the regulatory agency, the MHRA. TheAcademy’s FORUM with Industry could form amodel of how to develop a dialogue about howbest to address a growing national need for abetter coordinated approach to education andresearch in safety assessment and to convincegovernment that investment would add value tothe country as a research base for industry. TheAcademy proposes to invite representativesfrom the interested bodies for discussion abouthow best to progress this national need. TheAcademy does not, at this stage, rule outrecommending formation of a national centrebut will re-evaluate the question afterconsultations.

25 The Academy notes with interest the proposals from the European Plaform on Innovative Medicines to establish a European Centre for Drug Safety as part of the 7th Framework Programme.

39

Safer Medicines

Introduction

Animal research plays an essential part in theunderstanding of normal and disordered biologicalprocesses and in the development and evaluation ofnovel medicinal products. It will continue tocontribute significantly in the foreseeable future, butthere is wide acceptance of the principles of reduction,refinement and replacement, where appropriate, ofthe use of animals in scientific procedures. Advancesin human genomics research, together with thesequencing of the main laboratory species, provide animportant opportunity to understand the comparativedeterminants of toxicity and adverse reactions. It is theaim of this project to survey the current ‘state of the art’in the assessment of the safety of medicinal productsand to identify scientific opportunities, across a broadfront, to improve safety assessment.

The safety of medicines remains a major publicconcern, and the goal of the Project is to improve theprediction of hazard while enhancing developmentopportunities and improving risk assessment byidentifying newer scientific approaches.

The Project draws on the resources of theAcademy’s FORUM to bring together interestedparties from academia, industry and the regulatoryagencies, covering all relevant scientific disciplinesand addressing issues for animal and clinical studies.

Project remit

1. The Project will examine and evaluate the methodsavailable to assess and predict potential risk ofadverse events associated with medicines;addressing both technologies and the process ofassessment.

2. Case Study Analysis will be based on chemicaldrugs, conventional and recombinant biologicalproducts, vaccines and gene therapy products, inorder to identify the principles needed forconstruction of the evidence base and for thedevelopment of risk assessment procedures.

3. A survey of current, relevant activities anddevelopments in academia and medical research,industry and other bodies will be conducted inorder to identify the research strengths,

weaknesses, opportunities and threats and to serveas the basis for developing recommendations.

4. The Project Steering Committee will take accountof relevant activities by other bodies and willidentify the particular value to be added by theAcademy-based activity, for example in sharingbest practice, identifying options and disseminatingawareness more broadly.

5. Recommendations will cover R&D requirementsand priorities and the issues for translationalresearch, including the development of knowledge,skill sets and competencies and the value of public-private collaborations, in order to characterise thekey elements of a national/European/internationalstrategy to strengthen capability in this area. TheProject will also advise on an appropriate approachto monitor subsequent outcomes.

6. The Steering Committee will consider the potentialto initiate proof-of-concept studies in riskassessment in agreed target areas.

Target audiences and specific outputs

The primary deliverable is an authoritative Reportand designated outputs are intended to be relevant forkey stakeholder groups (in the UK and internationally)across industry, academia, government, regulatoryauthorities and policy-makers:

1. Improving safety of medicines by identifying andunderstanding the defects in current riskassessment approaches capitalising on newopportunities in science and technology.

2. Explaining role and responsibility of industryleadership in developing improved models inpredictive toxicology – including alternativesto animal research – and importance of standard-isation of tools, metrology and quality assuranceprocedures.

3. Informing and assisting Regulatory Authorities inthe adaptation of new technologies; and identifyingresources to provide expertise to advice regulators.

4. Educating public and media on the importance ofscience and technology and new approaches insafety evaluation.

Appendix 1 - Remit

5. Identifying skills and training needs in regulatoryscience; providing strategies for the training andadvice for the next generation of toxicologists,pharmacologists and other relevant sciences andfor the continuing professional development of thecurrent generation.

Fields of science involved and style of working

The Project is expected to range widely in the lifesciences, including cell biology, pharmacology,pathology, drug metabolism, genetics and genomicsand informatics-based technologies, and molecularmedicine, and to adopt interdisciplinary perspectivesin developing the integrated mechanism-basedapproach to risk assessment.

The cross-sectoral Steering Committee will beconstituted to function with transparency and willinclude experts from academia, industry and theRegulatory Authorities; predominantly drawn fromthe UK but with broad awareness of globaldevelopments and, in particular, European policymatters.

40

Safer Medicines

The Steering Committee will have the responsibilityfor collecting evidence across a wide front, analysingissues and generating options for recommendations.The study of several specific areas of the Project willbe delegated to Working Groups and may beaddressed in various ways (e.g. by organising a seriesof Working Group meetings, holding workshops orother consultative actions). Topics for Working Groupsinclude: (1) Pre-clinical toxicology; (2) Safetypharmacology; (3) Pre-marketing (clinical phase I, IIand III) assessment; (4) Vaccines; (5) Gene therapy;(6) Risk–benefit assessment; and (7) Post-marketingsurveillance. Working Group Chairs will be membersof the Steering Committee, to ensure coordination andleadership, and cross-sectoral membership of theWorking Groups will be agreed by the SteeringCommittee. Further specification of working style,associated seminar events and timetable will be apriority next step.

As an activity initiated under the auspices of theAcademy’s FORUM, it will be important for theSteering Committee to draw upon the FORUM, bothas an initial resource and as a stakeholder group withwhom to test emerging recommendations.

41

Safer Medicines

Editorial team

Dr Geoffrey Schild, FMedSci (Chair)

Dr Gwyn Morgan (Scientific Advisor)

Sir Colin Dollery, FMedSciSenior ConsultantGlaxoSmithKline

Sir John Skehel, FRS, FMedSciDirectorMRC National Institute for Medical Research

Review Board

Professor Patrick Vallance, FMedSci (Chair)Head of Division of MedicineUniversity College London

Sir Alasdair Breckenridge, FRSE, FMedSciChairmanMedicines and Healthcare products RegulatoryAgency

Dr Joseph DeGeorgeVice President Safety AssessmentMerck & Co., Inc. (USA)

Professor Roderick Flower, FRS, FMedSciProfessor of Biochemical PharmacologyWilliam Harvey Research Institute

Dr Robert TempleAssociate Director for Medical PolicyFood and Drug Administration

Professor Jack UetrechtCanada Research Chair in ImmunotoxicityUniversity of Toronto

Pre-clinical toxicology

Dr Sandra Kennedy (Co-chair)Vice President, Safety Assessment UKGlaxoSmithKline

Dr Lewis Smith (Co-chair)Global Head of Product DevelopmentSyngenta

Dr Graham BettonSenior Principal Scientist, Experimental PathobiologyAstraZeneca

Dr John HaseldenHead, Metabolic ProfilingInvestigative Preclinical ToxicologyGlaxoSmithKline

Dr E. Clive JosephSenior Director, Safety EvaluationPfizer

Dr George OrphanidesHead, Investigative Toxicology and Stage 1 Toxicol-ogySyngenta

Dr Christopher PowellDirector, Safety Assessment EuropeGlaxoSmithKline

Mr Henry StemplewskiPharmacotoxicologistMedicines and Healthcare Products RegulatoryAgencyDr Jonathan TugwoodPrincipal Scientist, Molecular Toxicology GroupAstraZeneca

Safety pharmacology

Dr Rob Wallis (Co-chair)Executive Director, Discovery BiologyPfizer Global Research and Development, SandwichLaboratories

Dr Tim Hammond (Co-chair)Vice President Safety Assessment UKAstraZeneca

Dr Peggy GuzzieSenior Director, Safety SciencesPfizer Global Research and DevelopmentConnecticut

Appendix 2 - Working Group members

Dr Peter HoffmannHead Non-Clinical Drug SafetyPharmaceuticals DivisionF. Hoffmann – La Roche

Professor Ian KimberHead of ResearchSyngenta Central Toxicology Laboratory

Dr Ian MacKenzieHead of PharmacologyCovance Laboratories

Professor Beatriz Silva LimaAssociate Professor, Department of PharmacologyUniversity of Lisbon

Dr Andrew SullivanDirector of Safety Pharmacology, Safety AssessmentGlaxoSmithKline

Dr Konrad TomaszewskiSenior Director, Safety and Risk-managementPfizer Global Research and Development

Professor William DawsonManaging DirectorBionet

Pre-marketing (clinical phase I, II and III)assessment

Professor David Webb, FRSE, FMedSci (Co-chair)Christison Professor of Therapeutics and ClinicalPharmacologyUniversity of Edinburgh

Professor Martin Wilkins (Co-chair)Professor of Clinical PharmacologyImperial College London

Professor Adam CohenDirectorCentre for Human Drug ResearchUniversity of Leiden Medical CentreThe Netherlands

Professor Colin GarnerChief Executive OfficerXceleron

42

Safer Medicines

Professor Paul Grasby, FMedSciChairman, Brain Disorders SectionMRC Clinical Sciences CentreHammersmith Hospital

Professor Mervyn Maze, FMedSciChair, Department of Anaesthetics and IntensiveCareImperial College London

Professor David NewellProfessor of Cancer TherapeuticsNorthern Institute for Cancer ResearchUniversity of Newcastle-upon-Tyne

Professor Walter NimmoVice ChairmanCharles River Laboratories

Professor Paul RolanMedical DirectorMedeval

Professor Geoff TuckerProfessor of Clinical PharmacologyHead, Academic Unit of Clinical PharmacologyUniversity of Sheffield

Vaccines

Dr Philip Minor (Chair)Head of VirologyNational Institute for Biological Standards and Control

Dr Mike CorbelHead, Division of BacteriologyNational Institute for Biological Standards and Control

Dr Elwyn GriffithsAssociate Director GeneralBiologics and Genetic Therapies DirectorateHealth Canada

Professor Adrian Hill, FMedSciProfessor of Human GeneticsWellcome Trust Principal Research FellowUniversity of Oxford

43

Safer Medicines

Dr Christopher JonesHead, Laboratory for Molecular StructureNational Institute for Biological Standards and Control

Professor Elizabeth MillerHead, Immunisation DepartmentCentre for InfectionsHealth Protection Agency

Professor Kingston MillsProfessor of Experimental ImmunologyDepartment of BiochemistryTrinity College, Dublin

Professor Peter Openshaw, FMedSciProfessor of Experimental MedicineNational Heart and Lung Institute

Dr Clive SweetSenior Lecturer, School of BiosciencesUniversity of Birmingham

Gene therapy

Professor Klaus Cichutek (Chair)Vice President, Paul Ehrlich InstituteLangenGermanyProfessor George DicksonProfessor of Biomedical and Molecular Cell BiologyRoyal Holloway, University of London

Professor Bernd GänsbacherDepartment of SurgeryUniversity of Munich

Dr Anthony MeagerNational Institute for Biological Standards and Control

Dr James RobertsonNational Institute for Biological Standards and Control

Professor Adrian ThrasherProfessor of Paediatric ImmunologyInstitute of Child HealthUniversity College London

Risk–benefit assessment

Dr Gwyn Morgan (Co-chair)

Professor Alan Boobis, OBE (Co-chair)DirectorDepartment of Health Toxicology UnitImperial College London

Professor John AshbySenior Syngenta FellowSyngenta Central Toxicology Laboratory

Dr John EvansSenior Director, PathologySafety AssessmentAstraZeneca

Professor David HarrisonHead, Division of PathologyUniversity of Edinburgh

Professor Brian LakeHead of Molecular Sciences DepartmentBIBRA International Limited

Professor Peter MoldéusVice President, Global Safety AssessmentAstraZenecaSweden

Professor B. Kevin Park, FMedSciHead, Department of PharmacologyUniversity of Liverpool

Dr Jennifer SimsGlobal Head of Safety Profiling and Assessment Project SupportNovartis Pharma

Dr Per SjöbergPartnerEureda KBSweden

Post-marketing surveillance

Professor Munir Pirmohamed (Co-chair)Professor of Clinical Pharmacology and HonoraryConsultant PhysicianUniversity of Liverpool

Professor Janet Darbyshire, OBE, FMedSci (Co-chair)DirectorMRC Clinical Trials Unit

Dr Richard AshcroftHead of Medical Ethics Unit and Reader in Biomedical EthicsImperial College London

Dr Keith BeardConsultant Physician and Hospital Prescribing AdviserNHS Greater Glasgow

Professor Alison BlenkinsoppProfessor of the Practice of Pharmacy, Department of Medicines ManagementKeele University

Dr Kevin CheesemanGlobal Director of Development PharmacogeneticsAstraZeneca (UK)

Professor Stephen EvansProfessor of PharmacoepidemiologyLondon School of Hygiene and Tropical Medicine

Dr Stephen InglisDirector, National Institute for Biological Standardsand Control (NIBSC)

44

Safer Medicines

Dr Alun McCarthyHead of Clinical Pharmacogenetics and Psychiatry,Medical GeneticsGlaxoSmithKline (UK)

Dr Frances RotblatSenior PhysicianMHRA

Professor Saad ShakirDirectorDrug Safety Research Unit Southampton

Dr Lincoln TsangAttorney, Arnold and Porter LLPNon-executive member of the National BiologicalStandards Board, NIBSC

Professor Tom WhalleyDirector of HTA ProgrammeDepartment of Pharmacology and TherapeuticsUniversity of Liverpool

Dr Louise WoodDirector of GPRD DivisionMHRA

45

Safer Medicines

ADR Adverse Drug ReactionALT Alanine AminotransferaseAMRC Association of Medical

Research CharitiesBBSRC Biology and Biological

Sciences Research CouncilBHF British Heart FoundationCelecoxib A COX-2 inhibitor used to(Celebrex) reduce pain and inflammation in

osteoarthritis and rheumatoid arthritis

COX-2 inhibitor A class of non-steroidal anti-inflammatory drugs (NSAIDs),thought to have fewer side-effectsthan traditional NSAIDs

CRUK Cancer Research UKDNA Deoxyribonucleic acidECG ElectrocardiogramEudraCT A database of all clinical trials

commencing in the EuropeanUnion from May 2004 onwards

EudraVigilance A data processing network andmanagement system for pharmacovigilance in the EuropeanEconomic Area

FDA Food and Drug AdministrationGenomics The branch of genetics that studies

organisms in terms of theirgenomes (their full DNA se-quences)

GP General PractitionerGPRD General Practice Research Data-

baseHIV Human Immunodeficiency VirusHMO Health Maintenance

OrganisationIT Information TechnologyKOL Key Opinion LeaderMEMO The Medicines Monitoring UnitMetabolomics The study of changes in metabolite

profiles as a result of a biologicaldisruption (such as disease or physi-ological stress).

MHRA Medicine and Healthcare Products Regulatory Agency

MMR Measles, Mumps and RubellaMORI Market & Opinion Research

Institute

MRC Medical Research CouncilNCE New Chemical EntityNHS National Health ServiceNICE National Institute for Health

and Clinical ExcellenceNMR Nuclear Magnetic ResonanceNOAEL No Observable Adverse Effect

LevelNPfIT The National Programme

for Information Technology coordinated by NHS Connecting for Health

OTC Over the Counter - a medication that is available from a pharmacy without a prescription

PCR Polymerase Chain ReactionPEM Prescription Event MonitoringPET Positron Emission TomographyPharmacogenetics The study of genetic factors that

influence an organism’s reactionto a drug

PoC Proof of ConceptPolymorphism A variation in the DNA that is

too common to be due merelyto new mutation. A polymor-phism must have a frequency ofat least 1% in the population

Proteomics A branch of biotechnologyconcerned with analysing thestructure, function and interactions of the proteins produced by the genes of a particular cell, tissue or organism, organising the information in databases, andapplications of the data

R&D Research and DevelopmentRofecoxib (Vioxx) A type of COX-2 inhibitor used

in the management of acutepain and osteoarthritis

SCID-XI X-linked severe combined immunodeficiency

SNP Single Nucleotide Polymor-phism

SSRI Selective Serotonin Re-uptakeInhibitor

WHO World Health Organization

Appendix 3 - Glossary and abbreviations

46

Safer Medicines

Chapman, S., Reeve, E., Rajaratnam, G. and Neary, R. (2003)Setting up an outcomes guarantee for pharmaceuticals: newapproach to risk sharing in primary care.British Medical Journal, 326, 707–709

Elmore, J. G. and Gigerenzer, G. (2005)Benign breast disease – the risks of communicating riskNew England Journal of Medicine, 353, 297–299

House of Commons Health Committee (2005)The influence of the pharmaceutical industryLondon: The Stationery Office Limited

Kohn, L., Corrigan, J. and Donaldson, M. (eds)(1999)To err is human: building a safer health system.Institute of Medicine. Washington DC, National Academy Press

Kola, I. and Landis, J. (2004)Can the pharmaceutical industry reduce attrition rates?Nature Reviews: Drug Discovery, 3, 711–715

Miller, M. A. (2002)Chemical database techniques in drug discovery.Nature Reviews: Drug Discovery, 1, 220–227

Pirmohamed, M., James, S., Meakin, C., Green, C.,Scott, A. K., Walley, T. J., Farrar, K., Park, B. K. andBreckenridge, A. M. (2004)Adverse drug reactions as a cause of admission to hospital:Prospective analysis of 18 820 patientsBritish Medical Journal, 329, 15–19

Raynor D. K. and Knapp, P. (2000)Do patients see, read and retain the new mandatory medicines information leaflets?Pharmaceutical Journal, 264, 268–270

Wong, I. C. K. (1999)Pharmacovigilance resources in the United Kingdom.Pharmaceutical Journal, 263, 258–288

Appendix 4 - References

Safer Medicines

A report from theAcademy of Medical Sciences FORUM

November 2005

The Academy of Medical Sciences

10 Carlton House Terrace

London SW1Y 5AH

Tel: 020 7969 5288

Fax: 020 7969 5298

Email: apollo@acmedsci.ac.uk

Web: www.acmedsci.ac.uk

Date of publication: November 2005

Jenny-Cover.qxd 11/9/05 11:29 PM Page 2