N E W S & A N A LY S I S

FROM THE ANALYSTS COUCH

Factors influencing non-approval

of new drugs in Europe
Michelle Putzeist, Aukje K. Mantel-Teeuwisse, Bo Aronsson,
Malcolm Rowland, Christine C. Gispen-de Wied, Spiros Vamvakas,
Arno W. Hoes, Hubert G. M. Leufkens and Hans-Georg Eichler
The high proportion of non-approved
marketing authorization applications for new
medicinal products in the European Union is
of serious concern and has raised questions
about the efficiency of drug development
and the regulatory system1. To investigate
the potential underlying factors, we analysed
all marketing applications (and assessment
reports) for new active substances considered
for approval at the European Medicines
Agency (EMA) from 2009 to 2010 (n=68)
and evaluated the extent to which deficits in
the drug development plan, disappointing
clinical outcomes (both efficacy and safety)
or doubts about the clinical relevance of
the data in the application were associated
with failure to be approved. In these three
assessment categories, we identified 15 key
variables; definitions of these variables and
scoring methods are described in detail
in Supplementary information S1 (table)
and Supplementary information S2 (box).
Analysis
Of all 68 applications evaluated by the
EMA Committee for Medicinal Products
for Human use (CHMP), 45 (66%) were
approved, whereas 23 (34%) were not;
6 received a negative opinion and 17 were
voluntarily withdrawn by the applicant before
a final opinion was made by the CHMP.
The summary scores in the three categories
development plan, clinical outcome and
clinical relevance are shown in TABLE1,
stratified for approved and non-approved
applications. All eight drugs with a positive
rating for the development plan, a positive
clinical outcome and convincing clinical
relevance were approved, whereas 12 out
of 14 applications for which all three scores
were negative were not approved.
Univariate and multivariate logistic
regression analyses were conducted that
addressed the associations between the scores
in the three assessment categories and the
way the CHMP expressed its views on the
dossier (TABLE2). A disappointing assessment
of the clinical outcome seems to be a major

driver for non-approval (odds ratio: 21.7;

95% confidence interval: 5.094.0). Negative
scores on clinical relevance contribute less
to the likelihood of non-approval (odds
ratio: 4.6; 95% confidence interval: 1.120.0)
than negative scores on clinical outcome.
Conversely, positive scores on clinical
relevance can help overcome deficiencies
with regard to clinical outcome, particularly
when good alternative treatment options are
lacking for the disease, such as in the case
of pirfenidone (Esbriet) for the treatment of
idiopathic pulmonary fibrosis.
Moreover, the data in TABLE2 indicate
how important the underlying development
plan is for increasing the likelihood that a
medicinal product is approved (odds ratio:
6.1; 95% confidence interval: 0.942.7).
We distinguished between learning-phase
studies (early-stage trials, for which we
assessed mode of action, proof of concept,
pharmacokinetics, dose finding and safety
pharmacology) and confirmatory studies
(late-stage trials, for which we assessed study
design, choice of primary end point, target
population, trial duration and statistical
analysis) in the development plan. We
explored four scenarios defined by the
presence of major objections at day 120 of
the EMA centralized procedure for assessing
marketing authorization applications, at

Ernest pouffe by Donna Wilson, courtesy of www.scp.co.uk

which point the CHMP adopts its list

of questions for the applicant (TABLE3).
A negative learning phase combined with a
negative confirmatory phase was significantly
associated with non-approval (relative risk:
5.3; 95% confidence interval: 1.223.6).
In applications that were approved
despite major objections at day 120 regarding
confirmatory studies, the objections were
addressed at a later stage in the procedure.
For example, study design issues were
addressed through convincing justification
by the applicant of the chosen design or by
the submission of results of a just on time
completed trial in the final stage of the
centralized procedure. Moreover, restriction of
the indication, or the acceptance by the CHMP
of lower efficacy owing to improved tolerability
by patients, was often sufficient to overcome
initial concerns about PhaseIII efficacy results.
Examples include: gefitinib (Iressa), which
was approved for a restricted indication the
treatment of adults with locally advanced or
metastatic non-small-cell lung cancer with
tumours harbouring activating mutations in
the epidermal growth factor receptor; and
dronedarone (Multaq) for rhythm control
in stable patients with atrial fibrillation, for
which concerns about efficacy in comparison
with amiodarone were outbalanced by an
advantageous tolerability profile.

Table 1 | Summary scorecard of EMA assessment of 68 MAAs*

Development
plan

Clinical
outcome

Clinical
relevance

Non-approved
(n=23)

Approved
(n=45)

18

12

EMA, European Medicines Agency; MAA, marketing authorization application. *Definitions of positive (+)
and negative () scores are given in Supplementary information S2 (box).

NATURE REVIEWS | DRUG DISCOVERY

VOLUME 11 | DECEMBER 2012 | 903

2012 Macmillan Publishers Limited. All rights reserved

N E W S & A N A LY S I S
FROM THE ANALYSTS COUCH

Table 2 | Univariate and multivariate analysis of EMA assessments of 68 MAAs*

Category

Nonapproved
(n=23)

Approved
(n=45)

Univariate OR
for non-approval
(95% CI)

Multivariate OR
for non-approval
(95% CI)

2 (11%)

16 (89%)

Reference

Reference

21 (42%)

29 (58%)

5.8 (1.227.9)

6.1 (0.942.7)

4 (9%)

38 (91%)

Reference

Reference

19 (73%)

7 (7%)

25.8 (6.799.1)

21.7 (5.094.0)

7 (18%)

31 (82%)

Reference

Reference

16 (53%)

14 (47%)

5.1 (1.715.1)

4.6 (1.120.0)

Development plan

Clinical outcome

Clinical relevance

CI, confidence interval; EMA, European Medicines Agency; MAA, marketing authorization application; OR,
odds ratio. *Definitions of positive (+) and negative () scores are given in Supplementary information S2 (box).

Positive scores are taken as reference categories. The odds ratios indicate the association between a
negative score and non-approval.

Deficits in the learning phase were more

strongly associated with non-approval than
deficits in any of the confirmatory studies, in
line with recent publications that emphasize
the importance of appropriate preclinical and
early PhaseI and II studies to reduce attrition
rates (for example, see REF.2). More in-depth
analysis of the learning-phase studies
identified that insufficient evidence on the
mode of action, proof of concept and dose
finding were also significantly associated with
licensing failure (Supplementary information
S1 (table)). For most non-approved
applications with major objections on
proof of concept, a statistically significant
effect on the primary efficacy end point in
PhaseIII trials was also lacking. Conversely,
some products with deficits identified in
the learning-phase studies were ultimately
approved because immature or lack of
proof-of-concept data were overruled
by convincing PhaseIII results. However,

a robust learning phase may be of major

importance for novel products, for which the
benefitrisk profile is not clearly understood.
An example is plerixafor (Mozobil) for the
mobilization of haematopoietic stem cells,
for which data on the mode of action had
an important role in finding the right target
population in terms of the mobilizer status of
the patient, resulting in a positive benefitrisk
profile in a restricted population.
Summary
We dichotomized the assessment of recent
European registration dossiers into positive
or negative scores, thereby reducing the huge
amount of data and the subtle regulatory
weighting of all the information possibly
relevant for patients into simple binary terms.
This is also what happens when regulators
determine the benefitrisk profile of a dossier
for a new medicinal product: the system
requires a yes or a no. Overall, negative

Table 3 | Summary table of EMA assessment* of development plan

Development plan
Learning phase Confirmatory phase

Non-approved Approved Relative risk of non(n=23)

(n=45)
approval (95% CI)

2 (11%)

16 (89%)

Reference

5 (25%)

15 (75%)

2.3 (0.411.6)

3 (37%)

5 (63%)

3.4 (0.620.2)

13 (59%)

9 (41%)

5.3 (1.223.6)

CI, confidence interval; EMA, European Medicines Agency. *Definitions of positive (+) and negative ()
scores are given in Supplementary information S2 (box). Categorized for learning and confirmatory phase.
A positive score in both phases was taken as the reference.

904 | DECEMBER 2012 | VOLUME 11

clinical outcome results seem to contribute

most significantly to current non-approval
rates. Our study also indicates that relevant
learning-phase studies are valuable in
reducing the number of failed dossiers and
speeding up pharmaceutical innovation.
Drug developers are encouraged to increase
investments in such studies before moving to
large and more costly PhaseIII trials.
Michelle Putzeist, Aukje K. Mantel-Teeuwisse and
Hubert G. M. Leufkens are at the Utrecht Institute
for Pharmaceutical Sciences, Division of
Pharmacoepidemiology and Clinical Pharmacology,
Utrecht University, Universiteitsweg 99,
3584 CG Utrecht, The Netherlands.
Michelle Putzeist, Christine C. Gispen de Wied
and Hubert G. M. Leufkens are at the Medicines
Evaluation Board, Graadt van Roggenweg 500,
3531 AH Utrecht, The Netherlands.
Bo Aronsson, Spiros Vamvakas and Hans-Georg Eichler
are at the European Medicines Agency,
7 Westferry Circus, London E14 4HB, UK.
Malcolm Rowland is at the School of Pharmacy and
Pharmaceutical Sciences, University of Manchester,
Oxford Road, Manchester M13 9PT, UK.
Arno W. Hoes is at the Julius Center for Health
Sciences and Primary Care, University Medical Center
Utrecht, Universiteitsweg 100, 3584 CG Utrecht,
The Netherlands.
doi:10.1038/nrd3894
1. Eichler, H.G. etal. New drug approval success rate in
Europe in 2009. Nature Rev. Drug Discov. 9, 355356
(2010).
2. Paul, S.M. etal. How to improve R&D productivity:
the pharmaceutical industrys grand challenge.
Nature Rev. Drug Discov. 9, 203214 (2010).

Competing interests statement

The authors declare competing financial interests: see Web

version for details.

Disclaimer

The views expressed in this article are the personal views of the
authors and must not be understood or quoted as being made
on behalf of or reflecting the position of the European Medicines
Agency or one of its committees or working parties.

FURTHER INFORMATION
European Public Assessment Report for Esbriet:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/
medicines/human/medicines/002154/human_med_001417.
jsp&mid=WC0b01ac058001d125&murl=menus/medicines/
medicines.jsp
European Public Assessment Report for Iressa:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/
medicines/human/medicines/001016/human_med_000857.
jsp&mid=WC0b01ac058001d124
European Public Assessment Report for Mozobil:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/
medicines/human/medicines/001030/human_med_000910.
jsp&murl=menus/medicines/medicines.jsp&jsenabled=true
European Public Assessment Report for Multaq:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/
medicines/human/medicines/001043/human_med_001207.
jsp&mid=WC0b01ac058001d124

SUPPLEMENTARY INFORMATION
See online article: S1 (table) | S2 (box)
ALL LINKS ARE ACTIVE IN THE ONLINE PDF

www.nature.com/reviews/drugdisc
2012 Macmillan Publishers Limited. All rights reserved