Upload
shein-chung
View
214
Download
0
Embed Size (px)
Citation preview
This article was downloaded by: [The University of Manchester Library]On: 24 November 2014, At: 07:43Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
Journal of Biopharmaceutical StatisticsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/lbps20
BRIDGING STUDIES IN CLINICAL DEVELOPMENTJen-pei Liu a & Shein-Chung Chow ba Department of Statistics , National Cheng-Kung University , Tainan, Taiwanb StatPlus, Inc. , Heston Hall, Suite 206, 1790 Yardley-Langhorne Road, Yardley, PA, 19067,U.S.A.Published online: 05 Oct 2011.
To cite this article: Jen-pei Liu & Shein-Chung Chow (2002) BRIDGING STUDIES IN CLINICAL DEVELOPMENT, Journal ofBiopharmaceutical Statistics, 12:3, 359-367, DOI: 10.1081/BIP-120014564
To link to this article: http://dx.doi.org/10.1081/BIP-120014564
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
BRIDGING STUDIES IN CLINICALDEVELOPMENT
Jen-pei Liu1,2,* and Shein-Chung Chow3
1Department of Statistics, National Cheng-Kung University,
Tainan, Taiwan2Division of Biostatistics and Bioinformatics, National Health Research
Institutes, Taipei, Taiwan3StatPlus, Inc., Heston Hall, Suite 206, 1790 Yardley-Langhorne Road,
Yardley, PA 19067
ABSTRACT
Global development of pharmaceutical products has become the key to the
success of any pharmaceutical sponsors. It is therefore crucial to address the
efficacy and safety variations of a new test pharmaceutical product among
different geographic regions due to ethnic factors. Recently, geotherapeutics
has attracted much attention from sponsors as well as regulatory authorities
from different geographic regions. To address this issue, the International
Conference on Harmonization (ICH) has published a guideline entitled
“Ethnic Factors in the Acceptability of Foreign Clinical Data,” which is
known as ICH E5 guideline. The ICH E5 guideline provides a general
framework for evaluation of the impact of ethnic factors on the efficacy,
safety, dosage, and dose regimen. We provide an overview of ICH E5
guideline including ethnic sensitivity, necessity of bridging studies, types of
bridging studies, and assessment of similarity between regions based on
bridging evidence. In addition, challenges on the establishment of regulatory
requirements, the assessment of bridging evidence, and design and analysis of
bridging studies are addressed.
359
DOI: 10.1081/BIP-120014564 1054-3406 (Print); 1520-5711 (Online)Copyright q 2002 by Marcel Dekker, Inc. www.dekker.com
*Corresponding author. E-mail: [email protected]
JOURNAL OF BIOPHARMACEUTICAL STATISTICS
Vol. 12, No. 3, pp. 359–367, 2002
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
Key Words: Ethnic factors; Bridging studies; Extrapolation; Similarity
INTRODUCTION
For marketing approval of a medicine, sponsors are required to provide
substantial evidence of effectiveness and safety from adequate and well-
controlled clinical trials. The U.S. Food and Drug Administration (FDA)
recommends that at least two clinical studies (the so-called pivotal trials) in the
same targeted patient population be performed to confirm the reproducibility of
the evidence on efficacy, safety, and dose response of the study medicine.
However, after a study medicine is approved by the regulatory agency in the
original region (e.g., the United States of America), sponsors might seek
registration of the approved medicine in a new region (e.g., European Community
or Asian Pacific countries). The possible differences in ethnicity, culture, and
clinical practice between the regions and their impacts on the safety, efficacy,
dose, and dosing regimen have limited the willingness of the regulatory authority
in the new region to accept the clinical data generated in the original region.
Consequently, the regulatory authority in the new region often requests the
sponsors to repeat similar studies for obtaining all or much of the clinical data in
the new region. This extensive duplication of clinical evaluation in the new region
not only demands valuable development resources but also delays the availability
of new medicine to needy patients in the new region. To resolve this dilemma, the
International Conference on Harmonization (ICH) has recently published a
tripartite guideline entitled “Ethnic Factors in the Acceptability of Foreign
Clinical Data,” which is usually referred to as the ICH E5 guideline, to address the
above issues.[1]
The objective of the ICH E5 guideline is to provide a framework for
evaluation of the impact of ethnic factors on the efficacy and safety of a study
medicine at a particular dosage or dose regimen. In addition, it describes
regulatory strategies of minimizing duplication for clinical data and requirement
of bridging evidence for extrapolation of foreign clinical data to a new region. The
purpose of this article is to provide an overview of the ICH E5 guideline including
ethnic sensitivity, necessity of bridging studies, types of bridging studies, and the
assessment of similarity between regions based on bridging evidence.
In the next section, ethnic sensitivity, necessity of bridging studies, types of
bridging studies are discussed. The third section provides an overview of the
definition of similarity and statistical methods for assessment of similarity
between regions based on bridging evidence. A brief concluding remark regarding
challenges on the establishment of regulatory requirements, the assessment of
bridging evidence, and design and analysis of bridging studies is given in the last
section.
LIU AND CHOW360
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
ETHNIC SENSITIVITY, NECESSITY OF BRIDGING STUDIES,AND TYPES OF BRIDGING STUDIES
Ethnic Sensitivity
The ICH E5 guideline lists critical properties of a compound for assessment
of sensitivity to ethnic factors. These critical properties include linear phar-
macokinetics (PK), flat pharmacodynamic (PD), therapeutic range, degree of
metabolism, extent of bioavailability, potential for protein binding, potential for
interactions, genetic polymorphism, inter-subject variability, systemic mode of
action, and potential for inappropriate use. However, the ICH E5 guideline also
points out that no one property of the medicine is predictive of the compound’s
relative sensitivity to ethnic factors. Because of the complexity due to possible
interaction among drug’s pharmacological class, indication, and demography of
patient population, the ICH E5 does not provide a precise and definitive criterion
for evaluation of ethnic sensitivity. As a result, no probability statements can be
made for the errors resulting from the decision-making on sensitivity to ethnic
factors. Therefore, both regulatory authority in the new region and the sponsor do
not have a criteria and a method for an objective and impartial evaluation of ethnic
sensitivity and necessity of a bridging study.
Necessity of Bridging Studies
Since no well-defined and scientifically justifiable criteria for assessment of
ethnic sensitivity are indicated in the ICH E5 guideline, any proposed approach for
the assessment of the necessity of bridging studies could be subjective and
controversial and may not be accepted by the regulatory authority and sponsors in
the new region.
However, several approaches have been proposed in the literature based on
some statistical criteria. For example, Shao and Chow[2] proposed the following
reproducibility probability as a statistical guide for quantifying the likelihood that
under the same experimental conditions, a second trial conducted in the new
region can reproduce the same results of the first trial in the original region:
PðTðxÞÞ ¼ 1 2 tn22ðtn22jTðxÞÞ þ tn22ð2tn22jTðxÞÞ;
where x represents the observed data from the clinical trial conducted in the
original region, T(x ) the value of the unpaired t-statistic based on x, n the total
number of subjects, tn22 the 97.5% percentile of a central t-distribution with n 2 2
degrees of freedom, and tn22( j( ) denotes a noncentral t-distribution with n 2 2
degrees of freedom and noncentrality parameter u. The information of repro-
ducibility probability provides a guide regarding statistical inference on the
chance that clinical results observed in the original region are reproducible at the
new region.
BRIDGING STUDIES IN CLINICAL DEVELOPMENT 361
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
To address the possible impact of ethnic factors, Chow et al.[3] suggested
performing a sensitivity analysis on the reproducibility probability with respect to
a sensitivity index D, which is defined as
D ¼ ð1 þ 1=mdÞ=C;
where md is the mean difference between the treatment group observed in the
original region, 1 the change of md in the new region and C represents the change
in variability in the new region. Chow et al.[3] refer to the reproducibility proba-
bility with respect to the sensitivity index, i.e., p(DT(x )) as the generalizability
probability. The concept of generalizability probability may be useful in providing
regulatory authorities with an opportunity to choose a statistical criterion for
determining whether a clinical bridging study is necessary.
On the other hand, as indicated in the U.S. Drug Master File, thousands of
medicines have been approved for various indications for different patient
populations by various regulatory authorities in different geographic regions.
Sufficient pre-approval and post-marketing experience on the critical properties
regarding ethnic sensitivities and the impact of intrinsic and extrinsic factors on
efficacy, safety, dosage, and dose regimen have been accumulated for these
medicines. Based on these data, an instrument consisting of three domains can be
developed to determine the degree of the impact of difference in ethnic factors on
the efficacy, safety, dose, and dose regimen of these medicines and consequently
the necessity of bridging studies. The first domain may include the critical
properties of the compound mentioned earlier. The second domain may consist of
intrinsic factors as discussed in Appendix A of the ICH E5 guideline. The third
domain may comprise the extrinsic ethnic factors as described in Appendix A of
the ICH E5 guideline. Within each domain, a scoring scheme for each property or
factors may be designed to characterize the degree of the impact on efficacy,
safety, dose, and dose regimen. For example, a possible scoring scheme could be a
5-point system such as 1 (no), 2 (mild), 3 (moderate), 4 (strong), and 5 (complete).
An algorithm can then be developed to provide a summary index for an overall
assessment of the impact on the efficacy, safety, dosage, and dose regimen of the
study medicine. In practice, the database of these compounds can be divided into
two data sets, namely, a training set and a validation set. Based on the summary
indices computed from the medicine in the training set, a threshold can be
determined to classify these medicines into two groups. One group consists of
medicines that are insensitive to ethnic factors and hence do not require bridging
studies. The other group contains medicines that are ethnic sensitive and hence
require bridging studies. The probability of classification error can then be
evaluated based on the validation set. Within the group of medicines with the
necessity for bridging studies, further cutoff points can be estimated for different
types of bridging studies. When a new medicine is applied for registration in the
new region, regulatory authority in the new region and the sponsor can calculate
the summary index of the study medicine to determine whether a bridging study is
necessary and what type of the bridging study is warranted. The above approach
LIU AND CHOW362
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
seems logical and acceptable to the pharmaceutical industry and regulatory
agencies. However, it requires a joint effort and close collaboration among
researchers/scientists from multiple disciplines in pharmaceutical research and
development.
Types of Bridging Studies
In general, the types of bridging studies required depend upon the ethnic
sensitivity of the study medicine, clinical experience of the drug class, extrinsic
ethnic factors, and ethnic differences between the new and original regions. As
stated in the ICH E5 guideline, a bridging study could be a PK/PD study or a
controlled clinical trial (CCT). Table 1 provides a summary of types of bridging
studies with respect to the factors of region, medical practice, drug class, and
clinical experience.
As it can be seen from Table 1, there are at least two fundamental issues in
the ICH E5 guideline: (i) sensitivity of study medicines to ethnic factors, necessity
of a bridging study, and the nature and type of bridging studies and (ii) assessment
of similarity based on bridging evidence.
ASSESSMENT OF SIMILARITY BASED ON BRIDGING EVIDENCE
According to the ICH E5 guideline, a bridging data package consists of (i)
selected information from the so-called complete clinical data package (CCDP)
that is relevant to the population of the new region and (ii) if needed, a bridging
study to extrapolate the foreign efficacy and/or safety data to the new region. In
other words, bridging evidence is actually provided either in the CCDP generated
during clinical drug development program for submission to the original region or
in a bridging study conducted in the new region after the pharmaceutical product is
approved in the original region. When the bridging evidence provided in the
CCDP could not allow extrapolation of foreign clinical data to a new region, then a
bridging study should be conducted in the new region to generate a limited amount
Table 1. Types of Bridging Studiesa
Medicine Region
Medical
Practice
Drug
Class
Clinical
Experience
Bridging
Studies
Insensitive — Similar — — No
Sensitive Similar — Sufficient No
Sensitive Dissimilar Similar Familiar — PD
Choice of dose — Different Unfamiliar Insufficient CCT
aPD, pharmacodynamics; CCT, controlled clinical trials.
BRIDGING STUDIES IN CLINICAL DEVELOPMENT 363
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
of clinical data to bridge the clinical data between the two regions. The ICH E5
guideline clearly states that assessment of the ability of extrapolation of the
foreign data relies on the similarity of dose response, efficacy, and safety between
the new and original regions, either with or without dose adjustment. However, the
ICH E5 guideline does not provide a precise definition or criteria for evaluation of
similarity. For assessment of similarity, a number of different statistical proce-
dures have been proposed based on different definitions or concepts of similarity.
For example, batch similarity in stability analysis for shelf-life estimation, simila-
rity in drug release for comparison of dissolution profiles between drug products,
similarity in drug absorption for assessment of bioequivalence between drug
products, and the concept of consistency between clinical results. Among these
definitions and concepts of similarity, bioequivalence in drug absorption and
consistency between clinical results are most relevant.
Population Similarity
Based on of the concept of similarity in drug absorption for assessment of
bioequivalence between drug products, Chow et al.[3] proposed to adopt the
concept of population bioequivalence for assessment of similarity of clinical
results between the original region and the new region. The idea is to establish
similarity according to the following aggregated equivalence (similarity) criteria,
which is similar to the one proposed in the recent FDA guidance for establishment
of population and individual bioequivalence:[4]
u ¼ðm0 2 m1Þ
2 þ s 2T1 2 s 2
T0
s 2T0
;
where s 2Tk ¼ s 2
Bk þ s 2Wk is the total variance in region k and s 2
Bk and s 2Wk are the
between-center variance and the within-center variance in region k, respectively.
We claim that clinical results in the new region are similar to those observed in the
original region if the 95% upper confidence bound of the following linearized
criterion is less than 0:
6 ¼ ðm0 2 m1Þ2 þ s 2
T1 2 ð1 þ uUÞs2T0;
where uU is the similarity margin set by the regulatory authority in the new region.
Consistency Among Studies
For the concept of consistency between clinical studies, Shih[5] suggested the
use of a predictive probability function for measuring the consistency between
clinical results from different studies. Shih’s approach is briefly outlined below.
Suppose that there are a total of H reference studies and each compares the same
two treatment groups. Let W denote the vector of standardized between-group
LIU AND CHOW364
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
differences of these H reference studies, W ¼ ðw1; . . . ;w11Þ and v is the
standardized between-group difference for the bridging study. The result of v from
the bridging study is consistent with the previous results W if and only if
pðvjWÞ $ min{pðwhjWÞ h ¼ 1; . . . ;H};
where p(vjW) is the predictive probability function that provides a measure of
plausibility of v given previous result W.
Hierarchical Model
Under a hierarchical model, on the other hand, Liu et al.[6] proposed the use
of the concept of equivalence or noninferiority for evaluation of similarity
between the new and original regions. Their approach is summarized below.
Step 1: From the CCDP, under a hierarchical model, use the clinical data
from the original region to obtain the estimate of relative efficacy
and its estimated standard error.
Step 2: From the data of the bridging study, obtain the estimate of relative
efficacy and its estimated standard error in the new region.
Step 3: Based on the estimated relative efficacy and its standard error from
both new and original regions and equivalence limit, perform the
usual two one-sided tests procedure or one-sided noninferiority
procedure (or confidence interval).
However, one can argue that the results of clinical data in the new region are
similar to those in the original region if the new region also demonstrates a
positive treatment effect.
Bayesian Approach
Liu et al.[7] suggested an empirical Bayesian approach to provide the evi-
dence of a positive treatment effect. The results on dose response, efficacy, or
safety of the original region can be incorporated prior to evaluating a positive
treatment effect by the bridging data in the new region. In other words, for a pre-
specified significance level a, 0 , a , 1; one can conclude similarity based on
the concept of a positive treatment effect if the posterior probability is
P{a positive treatment effectjdata and prior} . 1 2 a:
Despite the above different definitions for similarity, a direct interpretation
of the ICH E5 guideline on similarity requires performing a between-region
(study) analysis to evaluate the treatment-by-region interaction. It is then very
clear that the sample size required for the test based on the treatment-by-region
interaction will be much larger than that for detection of the treatment effect
BRIDGING STUDIES IN CLINICAL DEVELOPMENT 365
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
alone.[6] This statement is true for all types of studies and for all types of
endpoints. On the other hand, one only wants to verify whether the evidence of
efficacy or safety or PK/PD properties observed in the original region can be
reproduced in the new region. In this context, for example, a statistical significance
based on a particular endpoint can also be obtained from the bridging study
conducted in new region if it had been observed in the original region. However,
an equal or even larger sample size is required to reproduce a similar statistical
significance for detection of treatment effect in the new region (see, e.g.,
Refs. [2,3,8,9]). Therefore, these arguments indicate a fundamental conflict
between the evaluation of similarity and the objective of minimizing duplication
of clinical data in the ICH E5 guideline.
Consequently, Bayesian methods have been suggested to synthesize the data
from both the bridging study and the original region to resolve this conflict.[5 – 7]
However, some difficulties also arise using the Bayesian method. First, a medicine
was approved in the original region due to its substantial evidence of efficacy and
safety based on a sufficiently large sample size. The result of the bridging studies
using empirical Bayesian approach will be overwhelmingly dominated by the
results of the original region due to an imbalance of sample sizes between the
regions. In other words, it is very difficult, if not impossible, to reverse the results
observed in the original region and even the result of the bridging study is comple-
tely opposite. In addition, the Bayesian method for evaluation of probability for
error of decision-making on similarity is still needed to work out. This error
probability is extremely crucial for the regulatory authority in the new region to
approve a medicine in their jurisdiction.
CONCLUDING REMARK
The ICH E5 guideline provides a rationale for assessment of ethnic factors in
the acceptability of foreign data for regulatory strategies of minimizing dupli-
cation of clinical data and it also describes the requirement of bridging evidence
for extrapolation of foreign clinical data to a new region. It is, however, too prema-
ture to develop statistical methods for regulatory implementation unless well-
known and scientifically justifiable criteria for (i) the evaluation of sensitivity of
medicines to ethnic factors, (ii) the assessment of necessity of a bridging study,
(iii) the determination of the nature and type of bridging studies, and (iv) the
assessment of similarity based on bridging evidence are addressed in the future
revision of the ICH E5 guideline.
ACKNOWLEDGMENTS
The first author wishes to thank the support and encouragement of the National
Health Research Institutes, Taiwan during the preparation of this manuscript. The views
expressed in this article are personal opinions of the authors and may not necessarily
LIU AND CHOW366
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14
represent the position of the National Cheng-Kung University and National Health
Research Institutes, Taiwan and StatPlus, Inc., Yardley, Pennsylvania.
REFERENCES
1. ICH. International Conference on Harmonization Tripartite Guidance E5. Ethnic
Factors in the Acceptability of Foreign Data, The U.S. Federal Register, 1998;
Vol. 83, 31790–31796.
2. Shao, J.; Chow, S.C. Reproducibility Probability in Clinical Trial. Stat. Med. 2002,
21, 1727–1742.
3. Chow, S.C.; Shao, J.; Hu, O.Y.P. Assessing Sensitivity and Similarity in Bridging
Studies. J. Biopharm. Stat. 2002, 12 (3), 385–400.
4. FDA. Guidance for Industry on Statistical Approaches to Establishing Bioequiva-
lence; Center for Drug Evaluation and Research, Food and Drug Administration:
Rockville, MD, 2001.
5. Shih, W.J. Clinical Trials for Drug Registrations in Asian–Pacific Countries: Proposal
for a New Paradigm from a Statistical Perspective. Control. Clin. Trials 2001, 22,
357–366.
6. Liu, J.P.; Hsueh, H.-M.; Chen, J.J. Sample Size Requirement for Evaluation of
Bridging Evidence. Biometrical J. 2002, accepted.
7. Liu, J.P.; Hsueh, H.-M.; Hsiao, C.F. Bayesian Approach to Evaluation of the Bridging
Studies. J. Biopharm. Stat. 2002, 12 (3), 401–408.
8. Hung, H.M.J.; O’Neill, R.T.; Bauer, P.; Kohne, K. The Behavior of the P-Value When
Alternative Hypothesis Is True. Biometrics 1997, 53, 11–22.
9. Chow, S.C.; Shao, J. Statistics in Drug Research—Methodologies and Recent
Development; Marcel Dekker, Inc.: New York, 2002.
BRIDGING STUDIES IN CLINICAL DEVELOPMENT 367
©2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 0
7:43
24
Nov
embe
r 20
14