-
UNITED STATES
EDITOR’s NOTE: Welcome to IPQ’s “Monthly Update” on key CMC/GMP
developments in the US, Europe, and internationally. The IPQ family
of publications includes “The News in Depth” and”Updates in Brief”
on our website as they occur, “Weekly News Alerts” sent via e-mail,
and the “Monthly Update.” IPQ’s suite of offerings support our
mission of helping read-ers understand, engage in and respond to
the dialogue and developments around evolving and harmonizing the
regulation of drug and biologic quality and manufacturing.
Subscribers and license holders to IPQ have access to all of these
sources of cutting-edge news and in-depth analysis as well as to
the full IPQ archives. Visit IPQpubs.com for further
information.
Bill Paulson, Editor-in-Chief
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM
INTERNATIONAL PHARMACEUTICAL QUALITYInside The Global Regulatory
Dialogue
VOL. 3, NO. 5
CMC/REVIEW
• Human Factors Need to Be Studied for Pre-filled Syringes 2 •
Post-Change Comparability, PEGylation Among Biotech CMC Issues
Addressed at AAPS NBC........................................5 •
QbD at Vertex
Role in Formulation and Manufacturing Cited..................9
New Course Charted with
CMOs......................................19
Bio-Modeling, Continuous Manufacturing on Agenda 26 • IQ
Consortium Spins Off Lab Interlinking Effort.................31
INTERNATIONAL CMC/REVIEW
• WHO Imprimatur, Overhauled Regulatory Structure Could Push
China To Global Vaccine Prominence.............47
UPDATES IN BRIEF - p. 49
ISPE Mission Refinements • MHRA and EMA In-spection Results •
Release Testing in EU for Russia
mmmmm
Breaking news at your fingertips -
IPQ’S RSS FEED
IPQ is now on FACEBOOK and TWITTER. Come join the global
regulatory dialogue.
GMP/INSPECTION • Genentech Sheds Light on Bacterial
Contamination of Rituxin
Process..........................................................................39
http://www.ipqpubs.com/category/news/feed/http://www.facebook.com/?ref=home#!/pages/International-Pharmaceutical-Quality/140205272670466http://www.twitter.com/jerrychapmangxphttp://www.ipqpubs.com/http://www.ipqpubs.com/
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 �
MONTHLY UPDATE - JUNE 2012 UNITED STATES
International Pharmaceutical Quality TM (ISSN 1937-6898) is
dedicated to helping its readers understand, engage in and respond
to the dialogue and developments around evolving and harmo-nizing
the regulation of pharmaceutical and biologic quality and
manufacturing.
Subscriptions and licenses include full access to IPQ’s news
updates and in-depth analysis plus the full archives. See
IPQpubs.com for site/company/organization license rates.
© 2012. All rights reserved. IPQ Publications LLC. Content
cannot be photocopied, stored or transmitted by magnetic or
electronic means. Authorization to photocopy items for internal or
personal use is ganted by IPQ when the fee of $2.00 (U.S.) per copy
of each page is paid directly to Copyright Clearance Center, 222
Rosewood Dr., Danvers, MA 01923, USA (+1 978-750-8400).
INTERNATIONAL PHARMACEUTICAL
QUALITY™
Editor-in-ChiefBill [email protected]
Senior EditorJerry [email protected] Chief
Financial OfficerRob [email protected]
ext.105
Sales/Marketing CoordinatorWayne
[email protected]
Editorial Staff Rebecca Stauffer
7920 Norfolk Ave., Suite 900, Bethesda, MD 20814
FDA Expectations for Studying Human Factors in Pre-filled
Syringes are Taking ShapeFDA is increasing its pressure on sponsors
to conduct pre-market human factors (HF) studies for combination
pro-ducts, including pre-filled syringes, and industry is getting
clearer on the agency’s expectations as more experience is
gained.
At the AAPS National Biotech Conference in late May in San
Diego, California, Douglas Mead, the Regulatory Affairs CMC
Director for Medical Device and Combination Products at Janssen
Biologics, confirmed that the Center for Drug Evaluation and
Research (CDER) is now “requiring formal summative human factors
studies for most delivery devices.” CDER, he added, is requesting
that human factor study protocols be “sent in for review prior to
implementation.”
Human factors/usability analysis has been gaining prominence in
the dialogue between combination product manufacturers and
regulators.
At issue is an effective pathway for companies to follow in
conducting and submitting the analysis and for regulators to follow
in reviewing it. (See IPQ “Monthly Update” April 2011, p. 12. The
story includes insights by Center for Devices and Radiological
Health (CDRH) Office of Device Evaluation (ODE) Ron Kaye, who leads
the office’s Human Factors and Device Use Safety Team.)
At the AAPS meeting, Mead discussed the experience Janssen and
other firms are gaining regarding: ● what information the agency is
looking for, when it is required and in what
form ● general study requirements ● expectations for Phase III
clinical trials, and ● challenges posed by biologics when
conducting human factor studies.
FDA Wants to Pre-clear Protocols
CDER wants HF protocols to include “the current instructions for
use and some summary information for the actual user risk analysis
that was used to determine which critical tasks in the use of your
product you are going to assess in these protocols,” Mead
explained.
“I know from my own experience and from my industry colleagues
that at least six requests from FDA have been made to submit human
factor study protocols in the last year” for prefilled syringes, he
commented.
The Janssen official characterized the agency’s effort as
applying the design validation aspects of medical device design
controls to a drug product.
The “essential aspects” that the agency is looking for include a
user task analysis based on the instructions for use (IFU) and
identification of the critical tasks that need to be conducted.
One or more “formative studies” – the “practice runs” for human
factor studies – should be conducted. During the conduct of the
studies, the sponsor “can interact with the subjects and understand
their thinking, what they are confused about and what design
characteristics of your device are problematic.”
http://www.ipqpubs.com/issues/ipq-monthly-update-april-2011/http://www.ipqpubs.com/issues/ipq-monthly-update-april-2011/http://www.ipqpubs.com
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 �
INSIDE THE GLOBAL REGULATORY DIALOGUE™
Following the formative studies, a “summative study” is
conducted, which is similar to a clinical trial pivotal study,
“where you provide the instructions that you intend to go to market
with and see, with no intervention, whether the subjects can use
the product,” Mead explained.
The studies are generally requested by CDER during a “Type C
meeting” or an “end-of-Phase II meeting” in response to the sponsor
stating that a usability study is planned.
At that point, the agency has been requesting submission of HF
study protocols. If the request is made by CDER, it “will confer
with the device experts at CDRH, who will look at it from a human
factors standpoint, and they will get back to you with comments
within 30-60 days,” Mead pointed out.
He characterized submission of the protocols as “generally a
good idea, because you want some input at that point before you do
the study to see if the study methodology would be acceptable to
them ultimately when you are seeking approval.”
Mead commented that the reviews are generally thorough and
helpful.
The agency is looking to ensure that identification of critical
tasks and a user FMEA have been conducted, and examining how those
are reflected in the user instructions.
“In other words, what are the controls? What are you comparing?”
Mead asked. The sponsor may be looking only for confirmation of
usability, but what studies actually get performed and submitted
are negotiated with the agency.
In turn, FDA is still working on “the process they are going to
use to approve or provide informal feedback on these protocols and
what they will accept at the end.”
FDA Guidance Addresses Subject Involvement
The Janssen official pointed out that FDA has a “very good”
draft guidance that explains its expectations for human factors
studies. The draft, issued in mid-2011 by CHRH, addresses the
applicability of “human factors and usability engineering to
optimize medical device design.”
One of the issues on which clarification is provided is the
number of subjects and subject groupings that an HF analysis should
encompass. “A minimum” of 25 test participants is recommended.
“That does not mean just 15-25 subjects – it means per group,”
Mead emphasized. “So, for an autoinjector study, you may define a
group that has an impairment, a group that does not have any
impairment, a group that is comprised
of healthcare providers in a home setting, and a group of
patients in a home setting. These groups can be any way that you
decide to define them.”
This is “an important point,” he emphasized. “If you just limit
it to 25 subjects and they all pass, you would seem pretty
satisfied. However, you have to deal with actual field experiences.
Say you take 100 subjects and 99% of them can use an autoinjector
with no problem at all. When you look at your complaint rates in a
marketed product, multiply that 1% of user difficulty by a million
autoinjectors on the market and you will see that you have a very
high complaint rate. So this only gives you a feel for what your
success in the market will actually be.”
Mead pointed to the uncertainty that exists in the studies and
their inability to predict what may actually happen once the
product is marketed.
“All kinds of things can happen,” he stressed. “We need to
consider human behavior and startle reactions. It is not always
predictable. Human factors studies will not actually tell you what
is going to happen in the field perfectly.”
One tact Janssen has pursued to aid with user understanding was
“an autoinjector trainer to make sure that patients were very
familiar with autoinjector functionality.”
Phase III and Biologics Expectations Taking Shape
Two issues that have been drawing attention in the HF arena are
the drug-device presentations that should be used in Phase III and
bioequivalence considerations, particularly in the biologics
context.
“New on the table,” Mead pointed out, is a move by CDER to
“require the final to-be-marketed presentations to be used in some
way in the Phase III trial.” While doing so is not a substitute for
a human factors study, the final configuration is expected to “be
assessed in a clinical trial in some way.”
Mead noted that there is a heightened concern with
bioequivalence data in the US.
“In Europe, they have a bioequivalence guideline that says that
for a single mode of administration you really only need to do, for
example, subcutaneous injection – [that] if you have a prefilled
syringe or an autoinjector, it really doesn’t raise a
bioequivalence question.”
Complex biologic products, he noted, present unique challenges
in assessing the bioequivalence implications of usage
variations.
As the patient response to biologic products can be “highly
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 �
MONTHLY UPDATE - JUNE 2012variable,” it is challenging to
determine in a bioequivalence study “whether any differences you
saw were related to a statistical nuance of patient variability or
were actually related to differences in the devices.”
In these cases, a number of questions can confound the results –
for example, “is the depth of injection the same? Are you
compressing a fat layer differently?”
Comparing the US and European requirements, Mead noted that “on
a practical basis, those things are not considered in Europe, where
they believe that sub-C is sub-C.”
DOWNLOAD FROM THE STORY:
• FDA Draft Guidance on Human Factors and Usability
IPQ wishes to thank the following sponsors:
For subscription and sponsorship information visit IPQpubs.com
or contact Wayne Rhodes — [email protected], Tel:
�0�-8�1-9�70.
http://www.ipqpubs.com/wp-content/uploads/2012/06/FDA_draft_human_factors_and_usability.pdfhttp://www.biotechlogic.com/http://www.usp.org/http://www.rx-360.org/http://www.cmcbio.com/http://www.westpharma.com/http://www.parexel.com/http://www.dba-global.com/usahttp://www.edqm.eu/http://www.ipqpubs.com/mailto:[email protected]
-
The Advantage ofUnrivaled Quality...by Design
Components
Ensuring a drug product maintains its safety and ef� cacy from
concept to commercialization while reducing total cost of ownership
requires high-quality packaging components.
West designed NovaPure components within a Quality by Design
framework. Our data-driven process provides a deeper understanding
of our product development and manufacturing processes. We apply
this knowledgeto components that can help customers build quality,
safety and ef� cacy into their drugs.
Our holistic approach encompasses a thorough process review,
from raw materials to commercial manufacturing to transportation of
components to the customer’s facility.
NovaPure serum stoppers, lyophilization stoppers and syringe
plungers are globally available with a reduced lead time, and are
providedin optimized packaging that helps ease the transition
through manufacturing environments.
West NovaPure®
#6731
The New Definition of High-Quality Components
West and the diamond logo and NovaPure® are registered
trademarks of West Pharmaceutical Services, Inc., in the United
States and other jurisdictions.
© Copyright 2012 West Pharmaceutical Services, Inc.
Unrivaled Quality…by Design
To learn more about West’s NovaPure technology, call your West
account manager or a Technical Customer Support representative.
Contact West today.
North America +1 800-345-9800Europe +49 2403 7960Asia Paci� c
+65 6860 5879
www.westpharma.com
6731 NovaPure Advert.indd 1 3/29/2012 8:46:01 AM
www.facebook.com/westpharmawww.twitter.com/westpharmawww.westpharma.wordpress.comwww.gplus.to/westpharmawww.linkedin.com/company/west-pharmaceutical-serviceshttp://www.westpharma.com
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 �
INSIDE THE GLOBAL REGULATORY DIALOGUE™
Analysis of Post-Change Comparability and PEGylation Among
Biotech CMC Concerns Drawing FDA Comment at AAPS National Biotech
ConferenceThe expectations for pre- and post-change lot comparisons
and for PEGylation analysis were among FDA CMC application filing
concerns addressed by CDER Office of Biotechnology Products (OBP)
official Susan Kirschner at an “ask the regulator” session at AAPS’
National Biotech Conference in late May in San Diego,
California.
Intended to probe into some of the issues on which firms are
looking for more clarity in putting together CMC applications for
biotech and other biological products, the questions to which
Kirschner responded had been submitted to FDA in advance of the
meeting by the AAPS’ Protein Aggregation and Biological
Consequences Focus Group.
In addition to OBP’s expectations for comparability studies and
PEGylation, the focus group sought clarification regarding: ● the
promotion of new analytical technologies ● preference of
technologies for examining sub-visible particles ● surfactant
specifications ● stability testing of sterile filtered bulk drug
product ● the use of disposables in manufacturing ● assessment of
leachable accumulation, and ● expectations for characterization of
what happens to biomolecules after injection.
While some of the questions were not the type that lend
themselves to easy answers, Kirschner made an effort to give
succinct responses and to summarize current agency thinking on key
areas of concern. She serves as Associate Chief of OBP’s Immunology
and Therapeutic Proteins Laboratory. [Kirschner’s responses to all
the questions posed by the AAPS committee are provided below.]
Compare Pre/Post-Change Lots Side-by-Side
Kirschner stressed that data from stability and degradation
studies of various types and side-by-side comparisons between pre-
and post-change lots were key elements that the agency looks for in
comparability studies.
Depending on the change, extended characterization studies/data,
real-time stability data, accelerated temperature and/or forced
degradation studies may be required. “Forced degradation studies
are less likely,” she commented. Stressed and accelerated
degradation data “can be important to understanding the kinetics of
degradation…that may reflect some change in your molecule that
wasn’t picked up otherwise.”
Kirschner explained the agency’s expectation for side-by-side
analyses of pre- and post-change lots.
“If you are running an SDS-PAGE gel, it is really hard if you
have different gels or gels from different time periods to compare
the data. So it is really good to have those in side-by-side
analyses.”
Recognizing that there are generally fewer post-change lots
available than pre-change lots, the CDER official emphasized the
importance of putting post-change data in the context of the
pre-change data. “That usually occurs in the form of control charts
or some kind of trending data.”
CDER is “looking to see if you are within trend or if there is
some indication that there has been a shift in a particular product
quality attribute. That has in some cases required additional
post-change data to see if there has truly been a shift and what
that shift may mean. If you have gels and chromatograms, we would
like to see pictures of them for multiple pre-change lots and some
post-change lots to put everything into context.”
PEGylated Products Require Detailed Analysis
CMC applications for PEGylated proteins – those that contain
chemically-bonded polyethylene glycol (PEG) to enhance structure
and/or function – should include detailed characterization of both
the molecule and the PEG used to produce it.
“PEG is a raw material and should be characterized,” Kirschner
emphasized. She explained that it is important to examine the
purity of the PEG – in particular, the chemical residuals from the
manufacturing process that may need to be removed. “Some PEG
manufacturing processes involve the use of cyanide, so you want to
measure the cyanide levels. You should also look at the structure
of the PEG…and the level of activation.”
The agency also expects an analysis of how and where PEG is
incorporated into the active pharmaceutical ingredient (API).
The CDER official stressed the need to “look at the number of
sites that are occupied by the PEG, where those sites are, and what
the level of occupancy is at any given site during the process.”
She added that “we have seen that…even some subtle changes in
levels of site occupancy can impact PK. So these are not just ‘nice
to knows,’ they are really ‘need to knows’ for PK purposes.”
Also recommended is quantitation of the free PEG in the product
and evaluation of the stability of the PEGylation. “We have seen
some products where the bond between the PEG and the protein is not
stable, and over time they separate. So you have to look at that
too.”
DOWNLOAD FROM THE STORY: • CFR Amendment on stability test
requirements for biological products
https://www.federalregister.gov/articles/2012/05/03/2012-10649/amendments-to-sterility-test-requirements-for-biological-products
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 �
MONTHLY UPDATE - JUNE 2012CDER’S SUSAN KIRSCHNER ON BIOTECH CMC
QUESTIONS
Q: What is the approach taken by FDA to evaluate a comparability
study?
Of course, there is ICH Q5E [“Comparability of
Biotechnological/Biological Products Subject to Changes in Their
Manufacturing Process”] that we follow. Our main interest is
whether it impacts drug substance and/or drug product. Depending on
the change, this may require extended characterization studies or
data, real-time stability data, stressed accelerated temperature
and/or forced degradation studies. Forced degradation studies are
less likely. Stressed and accelerated degradation data can be
important to understanding the kinetics of degradation of a change
that may reflect some change in your molecule that wasn’t picked up
otherwise.
You really should have side-by-side analyses with pre- and
post-change lots. If you are running an SDS-PAGE gel, it is really
hard if you have different gels or gels from different time periods
to compare the data. So it is really good to have those in
side-by-side analyses.
Generally there are fewer post-change lots available than there
are pre-change lots, but we would like to see the post-change data
put in the context of the pre-change data. That usually occurs in
the form of control charts or some kind of trending data. There we
are looking to see if you are within trend or if there is some
indication that there has been a shift in a particular product
quality attribute. That has in some cases required additional
post-change data to see if there has truly been a shift and what
that shift may mean. If you have gels and chromatograms, we would
like to see pictures of them for multiple pre-change lots and some
post-change lots to put everything into context.
Q: How does FDA decide when to ask for data from new
technologies?
First and foremost, safety is a critical driver. For example,
analysis of sub-visible particles was driven by a concern over
immunogenicity. The impact of immunogenicity of sub-visible
particles was really what drove pushing those technologies.
The other reasons that we may push for implementation of new
technologies is better characterization for product knowledge so
that you can understand product quality attributes and how process
changes may impact those quality attributes. Again, that would be
at the characterization level, not for release.
We do sometimes push including new technologies for in-process
and release tests because they are more accurate or sensitive or
better resolve differences, like electrophoresis methods. We have
pushed including new technologies when we think that the technology
the company is using has become too outdated or is not informative
enough.
To push something to be an in-process or release test, the
technology has to be ready for implementation in a GMP environment.
So sometimes there is a lag when you put something in for
characterization studies and when you might start incorporating it
into in-process and release testing, because it was just not ready
for GMP requirements yet.
Q: Does FDA prefer a particular technology for looking at
sub-visible particles?
The answer is no, not right now. If it falls between the 0.2 to
10 micron range, the nature of your particles need to drive the
technologies most suitable for your situation. Most companies use
size exclusion HPLC below 0.2 microns and light obscuration above
10 microns. That is a choice, not a requirement, with the exception
of parenterals, for which there is a particulate matter requirement
above 10 microns of using light obscuration, which is in one of the
recommended USP methods.
At the AAPS National Biotech Conference in late May in San
Diego, California, CDER Office of Biotech Products Laboratory of
Immunology and Therapeutic Proteins Associate Chief Susan Kirschner
pro-vided answers to a set of CMC questions submitted in advance of
the meeting by the AAPS Protein Aggregation and Biological
Consequences Focus group. The group sought clarification from OBP
on: ● submission and evaluation of process changes and
comparability protocols ● promotion of new analytical technologies
● preference of technologies for examining sub-visible particles ●
surfac-tant specifications ● evaluation of PEGylated products ●
stability testing of sterile filtered bulk drug product ● the use
of disposables in manufacturing ● assessment of leachable
accumulation, and ● expectations for characterization of what
happens to biomolecules after injection.
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 7
INSIDE THE GLOBAL REGULATORY DIALOGUE™
That specific number [0.2 micron] comes from where the gap
really starts. In most size exclusion chromatography analyses there
is a 0.2 micron filter. Anything larger than that tends to get
filtered out in those analyses.
USP starts at 10 micron, so most people were collecting data at
10 microns and above. But nobody was really systematically
collecting between 0.2 and 10. That is where the range comes from –
that is the range that wasn’t being measured.
Q: At which clinical phase is surfactant specification required?
If we can demonstrate stability over a range of surfactant
concentrations, is it still required to have surfactant content in
the specification?
There are clearly different interpretations and approaches to
this question. Thinking about the surfactant being one of the
excipients – as an excipient you would have to know how much is in
the product, and you should test for it. It could potentially be in
the bulk drug substance or the final dose product, depending on
when you know the level of the surfactant.
You cannot just raise your surfactant level without knowledge,
or else you have the potential to impact PK/PD and clinical safety
and efficacy, not just product stability. You can’t just change
your surfactant because the product is stable. If the same
excipient is present in different amounts in different lots, then
those are considered different formulations.
Q: What are the regulatory recommendations for PEGylated
products?
Actually the original question said regulatory ‘hurdles,’ and we
changed it to recommendations, because I don’t think there are any
regulatory hurdles.
PEG [polyethylene glycol] is a raw material and should be
characterized. PEG has a range of size distributions, so you should
know that. You should look at the purity – in particular, the
chemical residuals from the manufacturing process that may need to
be removed. Some PEG manufacturing processes involve the use of
cyanide, so you want to measure the cyanide levels. You should also
look at the structure of the PEG.
There are other things you might want to do for qualification as
a raw material as well, depending on what your specific needs are.
Sometimes PEG gets activated, so you might want to look at the
level of activation.
For the bulk drug substance we also have requirements. You
should look at the number of sites that are occupied by the PEG,
where those sites are, and what the level of occupancy is at any
given site during the process. We have seen that…even some subtle
changes in levels of site occupancy can impact PK. So these are not
just ‘nice to knows,’ they are really ‘need to knows’ for PK
purposes.
You should look at the amount of free PEG in your product and
see whether or not you are going to remove it. And then the
stability of the PEGylation is important. We have seen some
products where the bond between the PEG and the protein is not
stable, and over time they separate. So you have to look at that
too.
Q: What is the need for stability testing of sterile filtered
bulk drug product if in-line filtration is used?
We don’t actually review this in our group, so we checked with
our counterparts in the biologics manufacturing and assessment
branch, which does the micro reviews for BLAs as well as facility
inspections.
Effective May 3, 2012, the CFR has been amended regarding the
use of specific sterility tests, sample size requirements and the
need for sterility testing of bulks. So a lot of those requirements
have been removed or amended. You should check the Federal Register
to see how that applies to your particular product.
On stability, we really prefer container closure integrity tests
rather than stability testing as it is a better measure of what
your container closure is capable of rather than whether any
particular vial was actually contaminated.
Q: Do disposables bring more safety or risks into bioproduction?
Is there a platform approach to leachables testing?
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 8
MONTHLY UPDATE - JUNE 2012 Disposables can increase safety. If
you have a multi-drug facility they could help eliminate cross
contamination. If you have particularly dangerous products like
toxins, it may be very useful to use disposables. However, there
are different risks so far as leachables and extractables are
concerned…. The leachables may increase or decrease any particular
risk, but they are different and you need to understand them.
The other question is whether you can use a platform approach.
We have seen cases where you can use model media, which may give
you some reasonable idea of what your leachables are. However, the
protein itself can impact the leaching process. Most importantly,
there is some concern about the toxicity in and of itself of a
leachable. Most leachables are not present in levels that are toxic
to humans but are present in levels that can degrade product or
impact product quality.
From that standpoint, if you are looking at platforming, you
have to understand how the leachables impact your product. And that
isn’t really easily platformable, unless it is something that is
really like the drug product…with maybe only a change of a few
amino acids.
Q: Is an assessment of leachables accumulation while using
multiple products required?
FDA generally considers that leachables assessment of bulk drug
substance and final product reflects accumulation during the
process that should be removed by diafiltration steps or other
means. In-process hold time studies may also incorporate assessment
of leachables accumulation depending on how they are designed.
We have not yet asked for specific studies to look at
accumulated leachables all along the way and look at them
specifically in that manner.
Q: How concerned is FDA currently as regards the use of silicone
to lubricate packaging components – do we now have some regulatory
relief?
Probably not. FDA is concerned with silicone because of its
ability to nucleate aggregation. FDA expects aggregates to be
assessed, and to the extent possible, characterized. So during that
process for which there is no regulatory relief at this point, you
will be looking at your silicone.
Siliconization is required for syringes to function. So you need
to assess your syringes for their consistency and levels of
coverage to ensure consistent syringe function and also for the
stability of the syringe function. So I don’t see how we could get
away with providing regulatory relief on those kinds of
assessments.
Q: What are the regulators’ expectations as regards
characterization of what happens to the biomolecule (aggregation,
binding to local/plasma proteins, degradation, glycosylation) after
injection or infusion? Is there a set of standard tests available
or planned to be made available? Are there specific in vitro models
or assessments FDA prefers?
There isn’t an actual regulatory requirement to look at that. We
do consider it useful information when you are thinking about
product stability and product quality attributes and setting
limits.
These studies are really difficult to do. It is something we are
interested in, particularly the subcutaneous space, and whether or
not your product generates aggregates in that space when you
inject, and whether that impacts immunogenicity. So while we would
love for people to do those studies and be happy to look at them
and review them, we haven’t been requiring them.
There were also some questions about antibody-drug conjugates.
But I couldn’t get anybody from the Division of Monoclonal
Antibodies here, so I can’t talk about it.
What I can say is that antibody conjugates are reviewed as a
collaborative review with the Office of New Drug Quality Assessment
looking at the drug part and the Division of Monoclonal Antibodies
looking at the monoclonal antibody part. They work very closely
together. They are complicated products and complicated
reviews.
-
http://www.cmcbio.comhttp://www.cmcbio.com
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 9
INSIDE THE GLOBAL REGULATORY DIALOGUE™
Vertex Harnesses QbD to Solve Difficult Formulation and
Manufacturing IssuesVertex’ experience with the development and
approval of its oral solid hepatitis-C drug Incivek (telaprevir)
testifies to the power of quality by design (QbD) to help solve the
problems posed by molecules that are difficult to develop,
formulate and manufacture.
At a symposium sponsored by the International Consortium for
Innovation and Quality in Pharmaceutical Development (“IQ
Consortium”) in December in Cambridge, Massachusetts, Vertex Senior
VP for Pharmaceutical Development Patricia Hurter commented on the
close collaboration and cross-discipline efforts that were needed
to address the QbD challenges and bring the development and
commercialization process for Incivek to completion.
“Pharmaceutical development, which I am the head of, regulatory,
technical operations and quality lived in each other’s pockets for
months,” working together to solve the technical and communication
problems, the Vertex official said. “I think being a small company
helped us to be able to do that.”
In her presentation, Hurter discussed: ● Incivek development and
characterization ● understanding chemical stability ● advancing the
science of spray drying ● scaling down for new projects, and ● the
firm’s QbD filing.
Hurter also spoke about Vertex’ focus on biopharmaceutics
modeling and continuous manufacturing as potentially powerful tools
for advancing its drug development and quality by design (QbD)
program (see the story on p. 26).
[Editor’s Note: A third story in IPQ’s series on Vertex focuses
on the communication pathways that had to be created with its
contract manufacturing partners to achieve its Qbd objectives—see
story on p. 26]
Incivek, approved in May 2011 under a six-month review clock, is
Vertex’ first drug on the market. Approval of a second QbD-based
application followed in January 2012 for Kalydeco, a drug for
treating cystic fibrosis. In a speech delivered in February, FDA
Commissioner Hamburg touted Kalydeco as the first drug to “treat
the underlying mechanism of the disease rather than the symptoms”
(see IPQ “Monthly Update” March 2012, p. 2).
Vertex currently has around 2,000 employees and uses contract
partners to do all of its clinical and commercial manufacturing,
from bulk production through final packaging. The primary drug
substance manufacturer with whom Vertex has been working is
Hovione, which has been
an important contributor to the overall QbD effort.
Solubility and Bioavailability Problems Addressed
Hurter explained that Incivek is large for a small molecule, at
680 Daltons. She characterized it as a “very complicated” and
“extremely challenging” molecule with low solubility and low
bioavailability.
Telaprevir is “actually less soluble than marble,” she noted,
and finding a model for how to formulate it to get acceptable
bioavailability was “challenging.”
From solution, the bioavailability was about 2%; from a
crystalline suspension, 1%; and from a nano-suspension, 1.7%. By
creating “tight dispersions with different types of polymers,” the
bioavailability was increased to around 20% or 40% in rats. “That
was hugely promising,” Hurter stressed. “So that was the path
forward that they took.”
The initial amorphous formulation had a “limited tendency to
crystallize a bit too rapidly,” so Hurter’s group developed a more
physically stable formulation “that improved exposures
dramatically” and became the formulation used in the marketed
product.
When telaprevir was first dosed in man in 2004, the toxicity was
“lower than expected,” but exposure data was highly variable.
An investigation into the reason for the variable exposure data
indicated that the original stability studies performed on the
suspension had likely overestimated the stability time, that dosing
was performed near the end of the stability time, and that the
amorphous drug product appeared to be crystallizing.
Vertex scientists performed X-ray powder diffraction, solid
state NMR, thermally-stimulated current, differential scanning
calorimetry and solubility tests, but were unable to discover why
the material was crystallizing.
Testing with isothermal titration calorimetry, a technique used
in research to determine drug-protein binding constants, revealed
that the telaprevir suspension was crashing – changing from
amorphous to crystalline form over time – with a transition time of
about four-and-a-half hours.
The “induction time” – how long after the suspension is prepared
before it begins to crash – was found to decrease
http://www.ipqpubs.com/news/fda-commissioner-hamburg-highlights-the-role-of-fda-and-regulatory-science-in-drug-innovation/
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 10
MONTHLY UPDATE - JUNE 2012as the storage temperature went
up.
As a short-term solution, the firm decided to refrigerate the
suspension until it was time to dose. That action was shown to
improve the physical stability.
For a longer term solution, different polymers were investigated
for use in the dispersions, eventually leading to the formulation
that was used in the final commercial product, which stayed 100%
amorphous for at least 72 hours at 37°C.
“Once we got the physical stability nailed, we thought we were
good,” Hurter reported. “Then we discovered that we had a really
bad chemical stability issue, which basically meant that we would
have to have a refrigerated product, which is really not very cool.
Nobody wanted to have a refrigerated product.”
Excipient Impurities Impact Chemical Stability
“Huge” lot-to-lot differences in the stability of telaprevir
were investigated using accelerated stability tests conducted at
40°C and 75% relative humidity in an open dish. The differences in
the rate of lot-to-lot degradation were asso-ciated with the use of
different lots of a particular excipient.
After much experimentation, the differences seen were traced to
the slurry pH of the excipient, and its variability to a residual
impurity from the excipient manufacturing process.
Vertex worked with the excipient manufacturer and found that the
impurity level had increased over time. “What had happened was that
they had allowed their process to kind of drift off in a region
where they weren’t washing the excipient at the end quite as well.
By instituting a better washing procedure they were able to bring
the impurity down again.”
Vertex “set specifications and had the supply chain folks
negotiate an ordering spec at 100 ppm, even though our proven
acceptable range was up to 300 ppm,” Hurter explained. “In my mind,
this all worked together to produce an excellent result.”
The Vertex VP emphasized that solving the stability issue
required collaboration between a large num-ber of individuals and
disciplines.
“We had good collaboration between formulation, analytical and
process chemistry, who helped figure out what the degradation
mechanisms were from studying the molecule. Materials
characterization experts helped with all the differ-ent things that
we did to characterize the excipient. Technical operations handled
the packaging and the controls in the
manufacturing. Together they determined the root cause of the
instability and developed models to predict stability.”
Advancing the Science of Spray Drying
Because telaprevir is an amorphous product with a high melting
point, spray drying is required to produce the final API.
During development of Incivek, Vertex developed a “huge”
spray-drying database, with 162 data points from two similar models
of commercial scale spray dryers – “the one that was originally
there and the one that was built,” the Vertex VP explained.
“We did a total of 13 DoEs over three years to generate the
database. It was an awful lot of money, time and effort. Some of
the stuff that was made was used as clinical material, so it wasn’t
all development. It was a huge investment, and we wanted to be sure
to make the best possible use of the investment.”
Hurter reported that Vertex is collaborating with Bend Research,
a firm that specializes in spray drying and has published papers
and given presentations regarding a concept called “heat and mass
transfer ratio,” or HMT. HMT relates the speed of spray drying to
the properties of the material produced.
“When I heard about this, I thought that it was fantastic,”
Hurter commented. She reasoned that Vertex could use the data from
its database, calculate HMT ratios, and run the calculated
conditions to produce product with better particle size
control.
However, what she observed was variability in the HMT ratio
correlation as well as some variability between spray dryers.
In discussing the results with Bend, Vertex discovered that
Bend’s research had all been done with constant droplet size,
whereas Vertex’ database contained data resulting from varying a
variety of parameters, including droplet size.
Although Hurter admitted that “there are still things that we
need to understand about how to control these spray dryers and how
to collect the data that we haven’t figured out yet,” in the case
of telaprevir the work has led to a process that is “under
excellent control with really consistent control of particle size
and bulk density.”
New Projects Require Smaller Scale
One of the “major reasons” Vertex is working with Bend is to try
to figure out how to scale down for new projects
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 11
INSIDE THE GLOBAL REGULATORY DIALOGUE™
and address the differences between its various spray dryer
equipment models.
Vertex deployed a small scale dryer in its initial development
work, and then graduated up to a larger capacity model. “As we
changed the scale of the spray dryer we had much more drying time
and nozzle changes,” as well as different relationships between
particle size and bulk density, the Vertex official explained.
Vertex worked with Bend using a custom spray dryer nozzle test
apparatus that allows it to take nozzles from other firms’ spray
dryers and measure the droplet size as a function of atomization
pressure.
Bend was able to perform the measurements on Vertex’ various
spray dryers and predict the atomization pressure that would
reproduce the desired droplet size between equipment.
Using the data produced, Vertex was “able to get right in the
region we wanted to be in” and reproduce the desired droplet size
on its various scale spray dryers.
Hurter praised the contribution of QbD in creating a “flat line”
process.
All the QbD work the company did with telaprevir has resulted in
the production of “tons of product with high quality and low
variability.”
QbD Filing Proves Challenging
Vertex’ new drug application (NDA) for Incivek was the first the
company had filed. In addition, it was the first full QbD filing
FDA had received for a product containing a complex molecule with a
complicated synthesis and extended supply chain.
The QbD filing was “a little bit ambitious,” Hurter admitted, as
the product is “complicated in every way—from an analytical
standpoint, from an API standpoint, and from a formulation
standpoint.”
The virtual manufacturing process involves multiple players in
diverse geographic locations, she pointed out.
“We make API in one place and then we ship it somewhere else, to
a different continent, to spray-dry it to make the amorphous stuff.
Then we ship it somewhere else to a different continent and we make
tablets and film coat
them. Then we ship it somewhere else to be put in blister
packaging cartons.”
She pointed out that with all activity outsourced, Vertex had to
oversee the QbD implemention at manufacturing plants “that we don’t
control” (see the story on p. 19).
Another challenge in the QbD arena was an odd-shaped design
space that proved “very difficult” to explain to FDA.
“Because we had a square-shaped design space with one corner cut
off, we had to represent the specification with an equation – very
challenging to explain to the agency,” Hurter noted.
Vertex Introduces Model Updating Mechanism
Adding another element to the QbD dialogue with FDA was Vertex’
introduction of an “allowable model bias” (AMB) concept that
reflects the uncertainty inherent in the initial model and allows
it to be refined as more experience is gained.
For example, the AMB concept would allow a particular spray
drying operating parameter to be redefined as more data is
generated.
“You don’t want to keep using the wrong model to control the
spray dryer,” Hurter emphasized. “You want to be able to refine
your models to reflect what you now believe is the truth.”
She pointed out that FDA is “fully in favor of what they call
‘model maintenance,’ but when you actually ask how that is supposed
to be done, nobody has a clue. So we came up with the AMB.”
Hurter recounted that during a pre-NDA meeting when Vertex was
explaining the AMB concept, FDA asked for a literature reference
that explained how it was arrived at and defined.
“We told them there wasn’t one – we made it up. They looked at
us like we were a little bit crazy. So we said that we need to
publish it so we can reference ourselves.”
Coming up with ways to do model maintenance that everyone can
buy into “is one of the things that I am hoping the IQ Consortium
can help with,” Hurter emphasized. [Editor’s note: An update on the
IQ Consortium’s progress to date is provided on pp. 38-45]
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 1�
MONTHLY UPDATE - JUNE 2012VERTEX’ PATRICIA HURTER ON INCIVEK
PROCESS AND PRODUCT DEVELOPMENT
CHALLENGES
We talked this morning about how innovation is all about
culture. I completely agree with that. You have to make sure it is
not a program. It should be a way of just being.
Vertex has a pretty unique culture…. The company’s whole purpose
is to ‘innovate to redefine health and transform life with new
medicines,’ which is a pretty lofty core purpose.
The first of three Vertex values that go along with that is
‘fearless pursuit of excellence.’ I really like that one. Basically
it says that you have the ability to take risks and not be afraid.
In my experience at Vertex we have done some pretty risky things.
And when the stuff has hit the fan, so to speak, people did not
come down on you like a ton of bricks – they said, ‘OK, what is
plan B? Let’s move on.’
The second one is ‘innovation is our lifeblood.’ I think that
kind of speaks for itself – we are very in favor of innovation. My
boss, the chief scientific officer, is probably the most
innovation-friendly person I have ever met in my entire life. He
encourages people to be innovative, which I think is great.
The third one is ‘we wins.’ Initially when I heard these three,
I liked the first two, and the third one sounded a little boring,
like playing nice in the sandbox, which didn’t seem that exciting
to me. But when it was explained more fully, ‘we wins’ means
innovation happens at the vertices. It means that success is not
singular. Success is when people get together. And the fact that
innovation happens at the vertices means that basically we have
people from different backgrounds – different academic training,
different countries, different genders, people of different,
diverse backgrounds. And when you put them together that is when
you get innovation.
I am going to try to give you some examples of collaboration
across traditional boundaries – between research and pharmaceutical
development, between the company and contract manufacturing
organizations, between us and the regulators – where we get
innovation.
I am going to go through three things: ● development of Incivek
and some areas where collaboration drove innovation ● progress we
are making on biopharmaceutical modeling, and ● where we are
heading – continuous processing. Don’t worry that the first one
takes a while because the second two are much shorter.
2011 was really exciting. We filed our first new drug
application in the US in November 2010, and got approval in May
2011. It was followed by approval in Canada, Europe and Japan
shortly thereafter.
Incivek is an amazing drug. It cures people of hepatitis C,
which is a serious worldwide health problem. It is the first drug
that Vertex has taken all the way to market by themselves, after
being in business for 21 years and having spent about four billion
dollars.
We also filed our first NDA in late 2011 for cystic fibrosis. To
have two NDAs back-to-back, the first ones the company has filed,
in less than a year, was obviously quite a challenge.
Incivek Characterization
Incivek – which is also known as telaprevir – is actually an
extremely challenging molecule. It is supposed to be a small
molecule, but it is not that small – it is 680 Daltons.
At a symposium sponsored by the International Consortium for
Innovation and Quality in Pharma-ceutical Development (IQ
Consortium) in December in Cambridge, Massachusetts, Vertex Senior
VP for Pharmaceutical Development Patricia Hurter detailed the
product and process development, for-mulation and filing challenges
her firm overcame with its hepatitis-C drug, Incivek (telaprevir).
In her presentation, entitled ”Innovation Happens at the Vertices,”
Hurter discussed: ● Incivek development and characterization ●
understanding chemical stability ●, advancing the science of spray
drying ● scaling down for new projects ● remaining opportunities,
and ● the firm’s QbD filing.
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 1�
INSIDE THE GLOBAL REGULATORY DIALOGUE™
It is very complicated. It has a high surface area and high logP
[octanol-water partition coefficient, a measure of hydrophobicity].
It is incredibly insoluble – 4.7 micrograms per milliliter. The
main reason is that it is highly crystalline with a high melting
point of 246°C.
Telaprevir is actually less soluble than marble…. Finding a
model on how to formulate it to get good bioavailability, needless
to say, was challenging. Before I joined Vertex, a very creative
and innovative formulation development group tried absolutely
everything. It was a small group with limited resources, but they
tried everything to get this molecule to be bioavailable.
From solution, the bioavailability is about 2%; from a
crystalline suspension, 1%; and from a nano-suspension, 1.7%. But
they found that when they made pretty tight dispersions with
different types of polymers, they could increase the availability
to around 20% or 40% in rats. That was hugely promising, so that
was the path forward that they took.
If you take the 4.7 micrograms and make it into an amorphous
dispersion…it goes up to 0.15 milligrams per milliliter, which is a
[huge] increase. Fortunately for us, even though it is amorphous,
its glass transition temperature is 105°C. The higher the glass
transition temperature, the more stable something is and the less
likely it is to crystallize at room temperature. That was in our
favor.
The initial amorphous formulation that was discovered had a
limited tendency to crystallize a bit too rapidly…. We worked on
the formulation and came up with a more physically stable
formulation that improved exposures dramatically. That basic
formulation was what went into the later stage toxicology studies
and then into clinical studies and is in the marketed product.
When I first arrived at Vertex, there was an article that was
just published in Nature Biotechnology that described some of the
CMC challenges. This data came from that article.
The day I joined Vertex in June of 2004 was the first day that
telaprevir was dosed in man, so it was really exciting…. One of the
first meetings I went to was a meeting to discuss the tox studies,
which had been done in July, that used variable exposure, and [the
toxicity] was lower than expected. Shortly after that we got data
back from one of the Phase I panels where the formulation exposure
was zero. They were going up in exposure and getting good
exposures, and then one panel was really bad.
It turned out that they had originally done the stability
studies on the suspension and showed that it was stable for 24
hours. They had done it one time at room temperature in February.
Room temperature in February in Boston, even with a good HVAC
system, is different than room temperature in Arkansas in July with
a good air conditioning system, and is also different than room
temperature at a clinical site in Germany in the middle of a heat
wave. Germany rarely has heat waves, so they don’t have great air
conditioning.
I think what basically happened that we discovered later was
that the basic temperature was too high, so the 24-hour stability
wasn’t really 24-hour stability. Of course they dosed at 23 hours
and 59 minutes…. We found out all this later. When I got there we
did not know what was going on.
The powder that was used to prepare the suspension was amorphous
and we characterized it by X-ray – which is how we had
characterized it before – and it still looked amorphous. We were
not sure what was going on. Were there crystalline seeds in the
powder? Was it aging? We were thinking about a lot of different
ways to analyze the powder and looking at suspensions and trying to
get a handle on why it was crystallizing. We thought about X-ray
powder diffraction, solid state NMR, thermally-stimulated current,
differential scanning calorimetry and solubility. We did all these
things, but we still could not get a handle on it.
In the meanwhile, there was a man named Pat Connelly, who is a
physical chemist, and was with Vertex and had left the company,
then rejoined Vertex a month after I joined. I met with John
Thomson, a very famous character who is in the book The Billion
Dollar Molecule, who suggested I meet with Pat Connelly….
I went to a research meeting in San Diego and ended up meeting
Pat. We wound up having breakfast together and found out we both
live in the town of Harvard, which only has about 6,000 people. We
ended up carpooling together
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 1�
MONTHLY UPDATE - JUNE 2012 quite a bit. There is pretty bad
traffic between Cambridge and Harvard so it gave us a lot of time
to have really intense discussions about innovation and physical
stability problems.
He came up with an idea to use isothermal titration calorimetry,
which is used in research to determine drug-protein binding
constants. It had never been used for something like this before,
but he said, ‘let’s give it a try.’ So we put a little bit of
suspension in a vial and plunked it in the machine overnight and
came back in the morning.
What we saw was a lovely trace where…if you analyze the stuff at
the beginning by a number of methods it was amorphous, and if you
analyze it at the end, it was crystalline. We knew exactly what
happened and exactly when it happened. This gave us a lot of
insight.
We learned that the transition time was about four-and-a-half
hours. We always talked about the suspension ‘crashing.’ My mind’s
picture was like a snow globe – you turn it upside down and it all
crashes. So we learned that it took four-and-a-half hours to crash.
It was not an instantaneous event.
We also found that the induction time – how long it takes before
it starts this transition – is actually quite variable. Even at a
very constant temperature it is quite variable. I think a lot of
people know this about crystallization events – that there is some
variability about when they happen.
We used this to investigate the problem of what was happening in
the clinics and in the tox studies. The induction time – how long
it takes for that transition to start – really goes down as you
increase the temperature very rapidly. If you are at 40°C it
basically happens instantaneously, whereas if you are at 25°C it
can take ten hours to start.
Similarly, the rate of transformation – how quickly does the
‘crashing’ happen, whether it is four-and-a-half hours or much
shorter – goes up tremendously as you increase the temperature. In
terms of a short-term solution on how to get to the next tox study
and to the next clinical study, basically we just refrigerated the
suspensions. After the suspensions were prepared, we would
refrigerate them until it was time to dose, and that improved the
physical stability problem.
In the longer term, we wanted to use discrete different
formulation prototypes. The preliminary Phase 1 early tox
formulation took 12 hours at 37°C to all go to crystalline. So we
used a different polymer. In the final commercial formulation, the
polymer detection was run out to 72 hours at 37°C and it was still
100% amorphous. This was a huge improvement that led to the PK
results I spoke about earlier. By stabilizing it in the body –
keeping it amorphous in the body at 37°C – we were able to…get much
more drug into the body….
So we had a guy doing research that didn’t have anything to do
with pharma development, and me, who didn’t know anything about
amorphous stuff up to that point. I vaguely remembered learning
something about it in chemistry in school. We got together, had
long intense discussions, used a new technique, and it really gave
us a huge insight and led us to a really great place.
Understanding Chemical Stability
Once we got the physical stability nailed, we thought we were
good. Then we discovered that we had a really bad chemical
stability issue, which basically meant that we would have to have a
refrigerated product, which is really not very cool. Nobody wanted
to have a refrigerated product.
We noticed lot-to-lot differences. So we put it on accelerated
stability at 40°C and 75% relative humidity in an open dish. There
were huge differences in the rate of degradation lot-to-lot
depending on one particular excipient that we were using. We said,
‘gee, what is causing this?’
We investigated every single thing that we could characterize
that we could possibly think of. At the end of the day, the answer
was fairly simple: the slurry pH of the excipient correlated with
stability, and was traced to a residual impurity from the excipient
manufacturing process.
We decided we would do studies to try to characterize stability
as a function of temperature, moisture and excipient impurity
level. We used what are called ‘TRH’ studies, which are useful for
all kinds of purposes. We equilibrated the
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 1�
INSIDE THE GLOBAL REGULATORY DIALOGUE™
samples at different moisture contents and then put them in
little foil bags, got all the air out, heat-sealed the bags, and
put them at different temperatures. This way we did not have to
have a number of different ovens with different temperatures and
humidities. We had three ovens at 25°C, 30°C and 40°C, put the
samples in, and measured the chemical degradation rates.
We took that data and put it into a huge regression model. The
graph showed us, with respect to temperature and time versus
different levels of the bad impurity, that with low levels of the
impurity the degradation rates were very low. For each one there
were three moisture levels – low, medium and high….
Using all that data we could plot contour plots, including the
predicted shelf life at each of the conditions and the predicted
percent of epimer under ICH conditions. Then we could define the
proven acceptable range, or PAR, based on how much epimer we will
allow over time.
We defined a normal operating range where we keep the tablet
moisture under normal circumstances at less than 2% and the
excipient impurity level at less than 100 parts per million, or
ppm. That is the normal operating range, which is well within the
design space for the proven acceptable range.
With this control strategy you can also predict the stability of
packaged tablets by predicting the moisture increase over time. The
model was taken from the other experiments that were completely
separate from this and used to plot data that was in traditional
ICH stability studies.
The end result was excellent stability. It was almost concerning
in a way because our registration stability lots had an impurity
level of 21 ppm, which is very low. That is good from a stability
standpoint, but it was almost bad from a ‘the stability looks too
good’ standpoint. But the FDA was very reasonable about it and
bought the whole QbD argument and didn’t make us have tighter specs
because of the fact that we couldn’t.
We had good collaboration between formulation, analytical and
process chemistry, who helped figure out what the degradation
mechanisms were from studying the molecule. Materials
characterization experts helped with all the different things that
we did to characterize the excipient. Technical operations handled
the packaging and the controls in the manufacturing. Together they
determined the root cause of the instability and developed models
to predict stability.
We also worked with the excipient manufacturer and found that if
the excipient impurity level was plotted over time that there was
an increase. What had happened was that they had allowed their
process to kind of drift off in a region where they weren’t washing
the excipient at the end quite as well. By instituting a better
washing procedure they were able to bring the impurity down again.
We set specifications and had the supply chain folks negotiate an
ordering spec that was at 100ppm, even though our proven acceptable
range was up to 300ppm. In my mind, this all worked together to
produce an excellent result.
Advancing the Science of Spray Drying
As I mentioned, this is an amorphous product, and given the high
melting point we needed to spray dry it. When we started we did not
know much about it, but we learned a lot over the three or four
years that we were developing these products. We were thinking
about how to use what we learned to do it better next time.
We scaled up the process to a bigger spray dryer. And for other
products we want to be able to make stuff early on a smaller scale
that will be similar in terms of properties to what we will be
making later on the larger scale spray dryer….
During development of Incivek, we developed a huge spray-drying
database, with 162 data points on two commercial scale spray dryers
– the one that was originally there and the one that was built.
They were slightly different models. We did a total of 13 DoEs over
three years to generate the database. It was an awful lot of money,
time and effort. Some of the stuff that was made was used as
clinical material, so it wasn’t all development. It was a huge
investment, and we wanted to be sure to make the best possible use
of the investment.
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 1�
MONTHLY UPDATE - JUNE 2012 The end result was that the process
is under excellent control with really consistent control of
particle size and bulk density. The question was – how do we
maximize it and how do we do it better next time?
We now have a collaboration with Bend Research – I am sure many
of you in the room do as well. They have published papers and given
talks about the concept of using this thing called the ‘HMT ratio’
– the heat and mass transfer ratio. Basically, with spray drying,
the faster you dry the droplets the more quickly the solvent comes
out, and you make a balloon that is fluffy and hollow, and very
compressible….
However, it you dry really slowly it produces balloons that
slowly crumple as they dry up, and you get raisin-like particles
that are much more dense and are much more compressible.
Bend has defined it in a chemical engineering way – the SDD
compressibility is proportional to the heat transfer rate over the
mass transfer rate. We can calculate the HMT ratio in terms of what
we do in spray drying by using the variables that we have
captured.
When I heard about this, I thought that it was fantastic. If we
do this and take the data out of each database and calculate the
HMT ratio, then maybe we can develop a much better experiment,
where by controlling the HMT ratio we can get more orthogonal
particles to come out of the spray dryer.
We went ahead and did it, and I agreed to give a talk at the
AAPS symposium using the data, expecting that it was going to turn
out to be fantastic. What we got wasn’t fantastic. It was a bit of
a bummer as I got this data a week before I was supposed to give
the talk.
What was supposed to happen, was that when we plotted bulk
density versus HMT, we would have a lovely correlation. But it was
scattered…. There was an incredible variability in the HMT ratio
correlation. There was also some variability between spray
dryers.
We tried to figure out why. In talking to Bend, they said that
they did it at a constant droplet size. The way that they do their
development is they determine the droplet size that they think they
want, then they use the HMT ratio to figure out the spray drying
conditions. It is much more targeted. What we had done on the spray
dryers were factorial DoEs, where we turned the knobs up and down
all over the place. So we had different droplet sizes and different
everything. Basically, our database was a lot more diverse.
We tried to look at what happened if we chose a subset of data
of approximately constant droplet size. It did improve the
correlation somewhat when we constrained the data to a narrow range
of nozzle orifices and then looked at different feed pressures. But
it was still not that great. My take-away from this is that there
are still things that we need to understand about how to control
these spray dryers and how to collect the data that we haven’t
figured out yet. We are working with Bend on how to do this
better….
One of the things that I find frustrating about doing a spray
drying experiment is that usually if you just do a normal DoE on
things like outlet temperature, condenser temperature, low, medium
and high, the particle size and the bulk density change together.
So if you really want to understand independently what SDD particle
size does and what SDD bulk density does, you don’t really get that
from this kind of experiment.
What I am hoping – but we haven’t done this yet, so I don’t know
if it is going to work – is that if we change droplet size to be
small and large, and we change dry rate to be fast and slow, and
concentrate on those variables instead of the spray drying
variables, that in that case we may be better able to separate out
particle size and bulk density from each other. The next time I get
a chance to work on it, that is what we are going to try and
do.
Scaling Down for New Projects
One of the major reasons we are doing this work with Bend is to
try to figure out how to scale down for new projects. Previously,
when we did the initial development, we started off on a PSD2 then
we went to a PSD1, then we went up to a PSD4. As we changed the
scale of the spray dryer we had much more drying time and nozzle
changes, and in general the particle size and bulk density increase
as you go up in scale.
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 17
INSIDE THE GLOBAL REGULATORY DIALOGUE™
Our initial work on a PSD1 showed low particle size and low bulk
density, nothing at all like what we got on the FSD and PSD
commercial scale spray dryers. We wanted to figure out how to
operate the PSD1 in a certain way in order to get the same
properties we got on the PSD4.
Bend has a nozzle test [apparatus] that allows them to use a
genuine nozzle from a spray dryer and spray it and measure the
droplet size pattern. They took some nozzles that we had used on
the PSD4 and measured the droplet size as a function of atomization
pressure. They figured out what the droplet size most likely was on
the spray dryer. Then they figured out what the atomization
pressure needed to be to match the droplet size between the PSD4
and the PSD1.
We wanted to keep a constant HMT ratio from what we were using.
But then once we have that relationship between the solution flow
rate, the feed pressure, and the inlet and outlet temperatures, we
have a whole bunch of constraints. So from that, we can figure out
where we need to be in terms of atomization pressure and solution
feed rate to operate at a particular droplet size and HMT
ratio.
We went ahead and did that and used the mobile apparatus that
had a six foot extension, which lets us have a longer drying time
because we have a longer length of the drying chamber for the
droplets. Using the mobile extension and using those scale
parameters, we were able to get right in the region we wanted to be
in, where we were able to get the kinds of particles on the PSD1
that we had previously gotten on the PSD4 and PSD5. That was pretty
exciting for me….
Remaining Opportunities
We are currently scaling up to the PSD5, and we are trying to
figure out how to write the regulatory filing to justify the use of
some of these parameters so that we don’t have to generate as much
data on the PSD5 as we did on the PSD4. We want to be able to
leverage a bunch of that information.
One of the big differences between small scale and big scale
spray dryers is the density of the plume. There is always one
nozzle in a spray dryer. You can imagine that if you have a small
scale-spray dryer with one nozzle, the amount of stuff coming out
is a lot less than what you have with a large scale spray dryer,
which has a much higher throughput resulting in a much denser
plume. The heat and mass transfer rates in that plume are going to
be different between the large scale and small scale dryers. There
are chemical engineering methods that can be used to model heat and
mass transfer rates as a function of solvent saturation in the
drying gas. We are looking forward to doing that.
When I gave the talk a month ago, I was sitting listening to
people who I worked with, yet I still learned things that I didn’t
know…. I wondered how it could be that they know things that I
don’t know and I know things that they don’t know. We really need
to improve the collaboration with people we already work with –
sharing knowledge currently residing with different people. We will
continue to collaborate internally and externally.
The QbD Filing
A flat line in an EKG is not good. But in a pharmaceutical
process that you are running, having a flat line process that is
really boring is a good thing. With all the QbD stuff we did with
telaprevir, that is basically what we have – a flat line process….
We have produced tons of product with high quality and low
variability….
The QbD filing was a little bit ambitious considering this was
the first NDA that Vertex had ever written. We did a full QbD NDA….
We had a very complex filing. It was quite difficult for anybody to
get their arms around. We started off with a very complicated
molecule, which requires a complicated synthesis. It is complicated
in every way, from an analytical standpoint, from an API
standpoint, and from a formulation standpoint.
We have a very complex manufacturing process. We make API in one
place and then we ship it somewhere else, to a different continent,
to spray-dry it to make the amorphous stuff. Then we ship it
somewhere else to a different continent and we make tablets and
film coat them. Then we ship it somewhere else to be put in blister
packaging cartons. This is just one supply chain. We have another
supply chain and are working on a third one at the moment.
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 18
MONTHLY UPDATE - JUNE 2012 Given that we are a virtual company,
all this activity is outsourced. So we are dealing with managing
CMOs. We have a QbD filing with a very complex molecule, a very
complex process, with a bunch of companies involved, and we are
implementing QbD at a manufacturer that we don’t control. All of
that is very complicated. Working with FDA on this was quite
difficult.
We have a very high drug load – a 750mg dose three times a day.
Consequently, the SDD [spray dry dispersion] properties have a huge
effect on the tablet properties, because the tablet is basically
80% [spray dried material]. We do experiments on this using the
spray dryer, but we need to know what the interaction is between
those and what happens downstream. To do this, we had to do a DoE
of the tablet process to see what happens and how it all
interacts.
Trying to explain to FDA what we did with these huge databases
became quite difficult. We talked about our design space models. We
had models for everything. We had models for dissolution and spray
drying control that were multi-dimensional with multiple variables.
To make life even more fun, we did not have small little design
space squares – we insisted on claiming every corner that we
managed to carve out. Trying to explain this to FDA was extremely
difficult. For some reason the concept was very hard to get
across.
Because we had a square-shaped design space with one corner cut
off, we had to represent the specification with an equation – also
very challenging to explain to the agency….
To throw in a little more fun, we defined something called
‘allowable model bias [AMB],’ which reflects the uncertainty in the
model and allows us to refine it as more process experience is
gained.
So if you do these models and have a parameter of 23.15, it is
actually, for example, 23.15 plus or minus two. Especially for the
spray drying models, if over time you find out as you generate more
data that 23.15 wasn’t quite the right answer – it was actually
24.03 – you don’t want to keep using the wrong model to control the
spray dryer. You want to be able to refine your models to reflect
what you now believe is the truth. FDA is fully in favor of what
they call ‘model maintenance,’ but when you actually ask how that
is supposed to be done, nobody has a clue. So we came up with the
AMB.
We had a lot of trouble explaining this. There was one funny
pre-NDA meeting where we were explaining the AMB concept and FDA
asked what the reference was for how we defined it. We told them
there wasn’t one – we made it up. They looked at us like we were a
little bit crazy. So we said that we need to publish it so we can
reference ourselves. This is one of the things that I am hoping the
IQ consortium can help with – coming up with ways to do model
maintenance that everyone can buy into.
[Regarding] the issue with the funny particle size bulk density
spec that we had trouble explaining – late one night before an FDA
meeting the next day, my son showed me a little animation. I said,
‘let’s take five minutes and see if we can use this to explain it.’
So this is how we did it. If the bulk density is less than some
value, then the entire range of particle size is acceptable. But as
the bulk density changes, the particle size changes also.
In addition to the models I told you about, we also had a bunch
of non-traditional studies, including stability studies – a bunch
of work on physical stability predicting relaxation and
crystallization rates and other things. On the chemical stability,
we did things like an FEI analysis of blister thickness and
moisture vapor transmission rates. We used those to estimate shelf
life in different packaging.
We did Monte Carlo simulations to estimate combined
probabilities. For example, if you have things happening all the
way through the process like the chemical stability thing, you want
to be able to say that you put all the specs in place and got a
certain amount of variability and at the end of the day, you know
what to expect. Even though the specification for the epimer is 4.5
based on how we defined the model, there is only one-in-a-billion
chance that you will be at 4.5. What you would expect most of the
time is 1% epimer, and it has actually moved further to the left
since then.
The point of this is to show that it was an incredibly complex
NDA. In hindsight, it should have been simpler. We also had a
priority review, so it was a six-month review clock, which is a
challenge as well.
-
MONTHLY UPDATE - JUNE 2012 MONTHLY UPDATE - JUNE 2012 MONTHLY
UPDATE - JUNE 2012
JUNE 2012 19
INSIDE THE GLOBAL REGULATORY DIALOGUE™
Virtual Innovator Vertex Pioneers New QbD Course with CMO
Partners
A clear communication pathway with contract partners for
escalating issues during and after development has been key to
Vertex’ success in building state-of-the-art, quality-by-design
(QbD) applications as a virtual company.
Eda Montgomery, who as CMC/QbD quality director helped Vertex
develop and gain clearance for two seminal QbD applications over
the past year, emphasized the importance of these communication
pathways in presentations on the firm’s experience in CMO
management at IFPAC’s annual conference in Baltimore in January and
again at an ISPE supply chain conference held there in June.
Montgomery is now working with Shire to help advance its QbD
program.
The two prominent Vertex approvals were for Incivek (telaprevir)
in May 2011 – the first drug approved for curing hepatitis C – and
Kalydeco – another oral solid, approved in January for treating
cystic fibrosis. In a speech delivered in February, FDA
Commissioner Hamburg touted Kalydeco as the first drug to “treat
the underlying mechanism of the disease rather than the symptoms”
(see IPQ “Monthly Update” March 2012, p. 2).
[See pp. 16-26 and 33-37 for more on: ● Vertex’ use of QbD to
help address Incivek’s development, formulation, and manufacturing
process, and ● the firm’s focus on biopharm modeling and continuous
manufacturing to help advance its drug development and QbD efforts
in the future.]
The first step in developing a QbD-oriented contract
relationship is to put a joint sponsor/CMO project team in place,
Montgomery explained. Vertex’ experience indicates
that it is then “crucial to have a process for and an agreement
on an open line of communication regarding how to escalate issues,
how to manage the relationship, how to talk about what is working
well and what needs to be improved.”
The escalation pathway established during the QbD development
process, in turn, she pointed out, provides “a nice stepping stone
to activities around communication of out-of-specification [OOS]
and out-of-trend [OOT] results, conducting investigations,
discussions around oversight and how closely managed certain
activities are going to be, and finally evaluation metrics and
frequency.”
Vertex currently has around 2,000 employees and uses contract
partners to do all of its clinical and commercial manufacturing,
from bulk production through final packaging.
Montgomery stressed that, with three different CMOs involved in
making its marketed products, there are four companies, quality
systems and cultures “that have to work well together” for the
process to gel. The primary drug substance manufacturer with whom
Vertex has been working is Hovione, which has been an important
contributor to the overall QbD effort.
QbD Development to Manufacturing Roadmap Established
In her presentation, Montgomery provided a roadmap for the CMO
relationship and knowledge management processes needed for a
virtual company to take a product from QbD
They approved us on May 23, 2011, which was very nice. It really
was an intensive collaboration between Vertex and FDA. There are
certain champions of QbD at FDA like Christine Moore and Sharmista
Chatterjee who were very encouraging and very supportive. They had
many meetings with us and really talked with us a lot to try to
help us get through this.
Within Vertex itself, pharmaceutical development – which I am
the head of – regulatory, technical operations and quality lived in
each other’s pockets for months, answering all of these questions
and pulling together last-minute presentations. I think being a
small company helped us to be able to do that.
We definitely still need ongoing dialogue. There are many things
that we had to pull out of the NDA because we just couldn’t get
agreement on the timeline. We are hoping that we can work with
those off-line as post-approval changes, and we hope that the IQ
consortium can play a role there. So instead of just us giving FDA
feedback, we can give them more consolidated feedback from the
industry as a whole.
http://www.ipqpubs.com/issues/ipq-monthly-update-march-2012/http://www.ipqpubs.com/issues/ipq-monthly-update-march-2012/
-
MONTHLY UPDATE - JUNE 2012
WWW.IPQPUBS.COM JUNE 2012 �0
MONTHLY UPDATE - JUNE 2012development through commercial
manufacturing.
Significant landmarks of the journey include: ● defining the
contract governance process ● implementing and managing QbD at the
CMOs ● managing the knowledge ● putting the knowledge gained to use
● handling the key performance indicators, and ● the ongoing
challenges. [Montgomery’s complete analysis of these various
components in the Vertex/CMO relationship is provided on pp.
28-32]
The governance process, Montgomery explained, includes a
documentation and a quality system piece. Vertex’ approach involved
building in the QbD documents to ensure the smooth transition in
the transfer from development to commercial manufacturing.
A risk assessment document is another “critical piece to
managing the ongoing routine commercial manufacturing as well as
managing the changes,” she pointed out.
Overarching these is a control strategy document that explains
how the quality systems of the two companies work together on a
product-specific basis. And trend reports round out the
process.
While its CMO partners may have had limited exposure to
manufacturing under QbD, Vertex found them “amazingly receptive to
the concept” and they “provided a lot of good input and a lot of
good clarification on how we were to do this.”
Non-Conformance and Knowledge Management Need Definition
One of Vertex’ focal points was the non-conformance handling
process and how deviations would be defined and dealt with in a QbD
context.
To make it manageable for the operators to keep the process in
its sweet spot within the traditional manufacturing context, the
batch records included the design space for critical and key
process parameters with clear manufacturing instructions on how to
address excursions.
Another key component of the relationship was defining knowledge
management and establishing a coordinated approach to trending key
performance indicators and any confirmed OOS deviations,
observations and complaints.
Montgomery explained how a coordinated, periodic sharing of
results was built in with the suppliers, leading to the shared
continuous quality improvement objectives. She also
pointed to the role a “trending protocol” and trending reports
play in this communication channel. She went on to provide concrete
examples of the improvements in knowledge and process performance
that have been gained through this trending program.
In her analysis of the role that tracking of key performance
indicators has played in optimizing the CMO processes, Montgomery
explained how the QbD investment has paid off in understanding root
causes and facilitating working with the CMOS to improve batch
records, eliminate human errors through automation and improve
equipment performance.
A significant “unintended consequence” for Vertex was to help
drive improvements in the batch releas