Real time release testing

Presentation prepared by Drug Regulations – a not for
profit organization. Visit www.drugregulations.org for the
latest in Pharmaceuticals.

www.drugragulations.org 1

 This presentation will cover
◦ What is Real Time Release Testing
◦ Batch Release & RTRT
◦ Organizational approach
◦ Examples
◦ End product testing Vs RTRT
◦ Process control : paradigm shift
◦ Benefits & challenges
◦ Relationship between QbD, PAT, Control Strategy &
RTRT
◦ Control Strategy – Conventional Vs RTRT
◦ ICH and other published examples of RTRT


 Medicinal products must comply with their
approved specifications before they are released
into the market.
 Compliance with release specifications can be
demonstrated by performing a complete set of
tests on the active substance and/or finished
product, according to the approved
specifications.
 Under certain conditions, an alternative strategy
to systematic end product testing is possible.


 So far this concept has been mainly applied
to sterility testing of terminally sterilized
products and has become associated with
parametric release applications.
 Recent guidelines adopted in the ICH context
(ICH Q8, Q9 and Q10) have made it possible
to apply a similar release decision process to
tests other than sterility, this approach has
been called Real Time Release Testing (RTRT).


 Real Time Release Testing (RTRT) is the ability to
evaluate and ensure the quality of in-process and/or
final product based on process data. ICH Q8(R2)
 Typically include a valid combination of measured
◦ Material attributes and
◦ Process controls


 The exact approach to RTRT will vary depending
on the process requirements.
 The RTRT strategy may be based on control of
process parameters, monitoring of product
attributes or on a combination of both at
appropriate steps throughout the process.
 Critically, the RTRT strategy should be based on a
firm understanding of the process and of the
relationship between process parameters, in-
process material attributes and product
attributes.


 Process monitoring may be applied to various
manufacturing steps or unit operations, such as
tabletting, on the basis of appropriate testing at
various stages in the process.
 Some parameters/attributes are usually checked
routinely at defined intervals regardless of the
design of the manufacturing process of a tablet.
 Uniformity of mass, crushing strength and
disintegration are such examples.


 The results of a comprehensive set of in-process
tests and controls in these cases may constitute
sufficient grounds for replacing the
corresponding end product testing.
 This may also offer greater assurance of the
finished tablet meeting certain criteria in the
specification, without the tests being repeated on
a sample of the finished product, as the amount
of data will in general be substantially larger.


 If testing of units is part of the RTRT a
sampling strategy should be defined that
provides the number of locations sampled
throughout the batch as well as the number
of dosage units tested at each location.


 RTRT will, in general, comprise a combination
of process controls which may utilise process
analytical technology (PAT) tools e.g.
◦ Near infrared spectroscopy (NIR) and
◦ Raman spectroscopy (usually in combination with
multivariate analysis),
◦ Together with the control of relevant material
attributes.


 Spectral data monitored on-line
◦ Controlling content of active substance,
◦ Polymorphism,
◦ water content,
◦ Blending homogeneity,
◦ Particle/powder properties or
◦ Film thickness
 could thereby replace end-product testing e.g.
◦ Uniformity of content,
◦ Tablet strength and
◦ Drug dissolution.


 In active substance manufacturing, RTRT can
apply to
◦ Continuous manufacturing processes, and
◦ Also to discrete unit operations such as
 Distillations,
 Hydrogenations,
 Crystallisations and
 All sorts of other chemical reactions or separations
(e.g. diastereoisomers).


Real time release testing is “moving the QC
lab into the process” and
“measure the CQAs where they are
generated”


Parametric Release: One type of RTRT.
Parametric release is based
on process data (e.g. temperature, pressure,
time for terminal
sterilization) rather than the testing of a
sample for a specific attribute
(ICH Q8 Q&A).


Real time release testing can replace end
product testing, but does not replace the
review and quality control steps called for
under GMP to release the batch.


 Batch release: Approved RTRT may form a
basis but
 More aspects needs to be taken into account
in the decision of a Qualified Person to
release a batch.
 These aspects could include batch results of
testing for an attribute not subject to RTR as
well as specific GMP requirements.


Formulation
Operations Quality
Development

Analytical
Regulatory RTRT Decision
Development

Technology Development Chemometrics

Multi-disciplinary / cross-functional teams are key to RTRt
New skill sets may be needed


 On-line or in-line measurements and/or controls,
◦ Tablet weight after compression
◦ Particle size measurement after granulation or milling
◦ Moisture measurement during drying
◦ Blend uniformity
 Fast at-line measurements,
◦ NIR for tablet assay
◦ Disintegration in lieu of dissolution
 Models as surrogate for traditional release tests,
◦ Multivariate model as a surrogate for dissolution
 Process signatures
◦ An evolving approach


Fixed Output
Input
Process

Disturbance:
Variation
due to
materials or
process Several days latter QC
End Product Testing


Process analyzers used to
NIR measure process
Interface parameters and adjust the
process

Adjustable Output
Input
Process

Disturbance:
Variation
due to
Immediate Feed back/
materials or
forward loop
process


 Reaction developed and
understood during development –
typical tools are IR, NIR and
Raman.
 At commercial scale NIR is used to
control the reaction.
 Stop the reaction at Maximal API
Concentration
 Stopping time differs from Batch to
Batch
 Real time release measurement of
the API assay and bi-product
(impurity)
 No sampling for in-process control
or end-product testing for this
CQA


Holistic Control Strategy e.g.:
Content Uniformity = Blend uniformity + Drug
concentration + Weight control

RTRT
1 = Blend Uniformity
2 = Granule particle size
3 = Weight, Hardness, Potency, Drug
concentration, Identity, Rate–controlling
polymer concentration


Process C ontrol Philosophy - Paradigm Shift
Conventional approach - lab based
End of phase testing of quality, to reduce the risk in
m oving to the next stage

O btain raw Mix active and Press tablets Package
m aterials excipeints

P.A.T approach - process based, at-line or on-line

O btain raw Mix active and Press tablets Package
m aterials excipeints

Continuously or m ore frequently test quality during each
phase, to rem ove the risk in m oving to the next stage


Granulation Fluidized Bed
Dispensation Dryer

Scale
Water Content – NIR
Identity-NIR Extent of Wet Air Particle size – FBRM
Massing - Power
Consumption
Raw Materials
Blending
Blend Homogeneity -
NIR

Multivariate Model (predicts
Disintegration)
Tableting
Content Uniformity NIR
Unit Operations
Attributes Packaging
Controls


 The outcome of a high level of process understanding
1. Controlling the process
2. Adjust for variability in raw and in-process materials
3. Increase yield, reduce waste, scrap
4. Reduce the risk of losing a batch
5. Reduced QC test
6. Increased control activity on the manufacturing shop floor
7. Reduced cycle time
8. Real time monitoring of CPPs and CQAs for free (must also be
included in continuous process verification and Annual Product
Review)
9. Quality of the finished product can be measured during
manufacturing – no surprises!
10. Regulators might be more interested in the beginning but this will
fade as process understanding has been demonstrated – reduced
inspection frequency


 New – not familiar to many
 PAT tools in place (in-line analysers, PAT data
management, multivariate data analysis, process control)
 Require new skills and reorganisation of work
 Risk associated with implementing PAT
 Installation of probes, representative sampling, failure of
instrument, failure of multivariate models, failure in feed
forward & backward controls, etc
 Backup strategy must be in place
 Models needs frequent update
 If RTRT fails it cannot be replaced by end-product testing
 Regulators might be very interested in the beginning...


QbD
Control
Design Strategy
Space

RTRT

CMA CPP CQA

RTRT

PAT


QbD is
really
about QbD and PAT links
the the patient,
patient product and process

Patient
1. Understanding what the patient
needs
2. Designing and developing a product
meeting these needs
Process
3. Designing and developing a Understanding
manufacturing process capable of
delivering the product that meets
these needs
Product Process


PAT
RTRT

A systematic approach to development that begins with
predefined objectives
and emphasizes product and process understanding
and process control,
based on sound science and quality
risk management


CQA’s
&
CPP’s
In Line
On Line
Process Analytical Technology is:
Analyzers
A system for designing, analyzing, and
controlling manufacturing through
timely measurements (i.e., during
processing) of critical quality and
performance
attributes of raw and in-process
Predictive materials and processes with the goal of
Models
ensuring final product quality

Real Time Real
Testing


Quality What is
Product Profile Target Identify critical to
Product CQA the
Profile
CQA’s Patient

QRM
PAT
Risk Assessments
Identify
CMA &
Design Space CPP

Design space Control Strategy
Control Strategy

Continual
Improvement

PAT , PAT RTRT
SOP PAT RTRT 32

 RTRT, when used, is part of the Control
Strategy
◦ Can include some or all of the final product CQAs
 QbD is not directly correlated to RTRT
◦ You can have QbD approaches without RTRT
◦ However, it would be difficult to justify RTRT
without a science and risk based approach


 Not all Process Analytical Technology (PAT)
leads to RTRT
◦ PAT systems can be designed to control CQAs of
raw materials or in-process materials and not
contribute to RTRT
 A design space is not required for RTRT
◦ Having a design space can increase operational
flexibility, without additional regulatory approval


 Control Strategy
◦ Planned set of controls
◦ Derived from current product and process understanding that
assures process performance and product quality
◦ The controls can include parameters and attributes related to
 Drug substance ,
 Drug product materials and components,
 Facility and equipment operating conditions,
 In-process controls,
 Finished product specifications, and
 The associated methods and
 Frequency of monitoring and control.’ (ICH Q10)


Manual

Automated & Advanced
Simple


Every product MUST have a control strategy

Minimal Enhanced approach

Drug product quality • Drug product quality
controlled primarily ensured by risk-based
by intermediates (in control strategy for
process materials) well understood
and end product and process
product testing • Quality controls
shifted upstream, with
the possibility of real-
time release testing or
reduced end-product
testing


 Identify CQAs
 Identify related CPPs and Material Attributes
(MAs)
 Develop the design space for the CPPs and MAs
 Develop the control strategy ensuring the CPPs
and MAS are always within the design space
 Based on risk-assessment plan how the control
strategy can be implemented
◦ This process starts in development
◦ It is a lifecycle activity and
◦ The Control Strategy can be updated as new knowledge
has been gained

NIR, at-line (id
raw materials)
IR, on-line
(purity, assay ) NIR, on-line
(Moisture, purity Assay (HPLC)
Purity, related

Conventional Testing
impurities, ((HPLC)
Residual solvent (GC)
Moisture content (KF)
Heavy Metals
Etc…

ID, Assay, CU (HPLC)
Purity, ((HPLC)
NIR, at-line (id raw Dissolution,
NIR, on-line materials)
Appearance
(reaction id) Moisture content (KF)
Etc
FBRM, on-
line (PSD)

NIR, on-line, blend
homogeneity

NIR, on-line, blend NIR, on-line
homogeneity (assay, CU, ID)

39

 NIR can be used for RTRT of water
determination
 Conventional lab-based NIR system
◦ Validated over range 1 – 6%
 Tablets dried and “spiked” to encompass
historical range and regulatory specification
 Prepare calibration curve
 In line NIR for water content determination


 CQA: CU, dissolution,
 Crystal size during formation - PSD
1. Focuses beam
reflectance
measurements can be
used to measure PSD
2. Measure crystal
diameter
3. Probe inserted into
reactor


 FBRM used to define the best cooling ramp
 FBRM used to measure PSD inline
 RTRT of PSD
 No sampling and
 QC test


Mock P 2 example
Design Space


In line Monitoring of drying Process


Sample
& Sample Sample
Test & &
Test Test

API
Pass
Excipient Blend Screen Blend Tablet or
Fail
Excipient

Fixed processes

Quality Criteria met if:
• Meets specification(s) (off-line QC tests)
• GMP Procedures followed

John Berridge, Pfizer www.drugragulations.org
46

Characterise Adaptive
processes

API
100%
Excipient Blend Screen Blend Tablet
Pass
Excipient
Real
PAT PAT time
release

Standards and acceptance criteria for a PAT/QbD approach are not
the same as a “Test to Document Quality” approach

www.drugragulations.org
47

 Liquid product, used to determine mix time
 CQA related to mix uniformity
 CPP’s (Critical Process Parameters) included agitator speed,
time after addition of one ingredient until the addition of
another, solution temperature, and recirculation flow rate.
 Process analyzer used was a refractometer
 Resulted in cost savings and quality enhancement

SCADA,
User
Mix
Interface
RI Tank
Sensor

Control Data
System Historian
Pump


Example from ICH case study
Blending Process Control Options
Decision on conventional vs. RTR testing

Key message: Both approaches to assure blend uniformity are valid in combination
with other GMP requirements


Example from ICH case study


RTRT of Assay and Content Uniformity
• Finished Product Specification – use for stability,
regulatory testing, site change, whenever RTR testing
is not possible
- Assay acceptance criteria: 95-105% of nominal amount
(30mg)
- Uniformity of Dosage Unit acceptance criteria
- Test method: HPLC
• Real Time Release Testing Controls
- Blend uniformity assured in blending step (online NIR
spectrometer for blending end-point)
- API assay is analysed in blend by HPLC
- Tablet weight control in compression step


RTRT of Assay and Content Uniformity

• No end product testing for Assay and Content Uniformity
(CU)
- Would pass finished product specification for Assay and Uniformity
of Dosage Units if tested because assay assured by combination of
blend uniformity assurance, API assay in blend and tablet weight
control (if blend is homogeneous then tablet weight will determine
content of API)


Investigation of the effect of API particle
size on Bioavailability and Dissolution
Drug Substance with particle size D90 of
100 microns has slower dissolution and
lower Cmax and AUC
In Vivo In Vitro correlation (IVIVC)
established at 20 minute timepoint

Early time points in the dissolution
profile are not as critical due to PK
results


Multifactorial DOE study of Exp No
1
Run Order
1
API
0.5
MgSt
3000
LubT
1
Hard
60
Diss
101.24

variables affecting dissolution 2
3
14
22
1.5
0.5
3000
12000
1
1
60
60
87.99
99.13

Factors:
4 8 1.5 3000 10 60 86.03
 5 18 0.5 12000 10 60 94.73
◦ API particle size [API] 6
7
9
15
1.5
0.5
12000
3000
10
1
60
110
83.04
98.07
unit: log D90, microns 8 2 0.5 12000 1 110 97.68

◦ Mg-Stearate Specific Surface Area
9 6 1.5 12000 1 110 85.47
10 16 0.5 3000 10 110 95.81
[MgSt] 11 20 1.5 3000 10 110 84.38

unit: cm2/g
12 3 1.5 12000 10 110 81
13 10 0.5 7500 5.5 85 96.85
◦ Lubrication time [LubT] unit: min 14
15
17
19
1.5
1
7500
3000
5.5
5.5
85
85
85.13
91.87
◦ Tablet hardness [Hard] unit: N 16 21 1 12000 5.5 85 90.72
17 7 1 7500 1 85 91.95
 Response: 18 4 1 7500 10 85 88.9

◦ % API dissolved at 20 min [Diss]
19 5 1 7500 5.5 60 92.37
20 11 1 7500 5.5 110 90.95

DOE design:
21 12 1 7500 5.5 85 91.95
 22 13 1 7500 5.5 85 90.86
◦ RSM design 23 23 1 7500 5.5 85 89
Note: A screening DoE may be used first to
◦ Reduced CCF (quadratic model)
identify which of the many variables have the
◦ 20+3 center point runs
greatest effect


Scaled & Centered Coefficients for Diss at 60min

• Key factors
0

influencing in-vitro
-1

dissolution: -2

- API particle size is -3

the dominating % -4

factor (= CQA of
API)
-5

-6

- Lubrication time has API Mg Lubricatio Tablet Mg St*LubT

MgSt*LubT
Hard
API

MgSt

LubT
n
Particle Stearate Hardness
a small influence Size N=23 SSA
R2=0.986
Blending
R2 Adj.=0.982

(= low risk DF=17 Q2=0.981 time
RSD=0.725 Conf. lev.=0.95

parameter)
MODDE 8 - 2008-01-23 10:58:52

Acknowledgement: adapted from Paul Stott (AZ) – ISPE PQLI Team


 Prediction algorithm
◦ A mathematical representation of the design space
for dissolution
◦ Factors include: API PSD D90, magnesium stearate
specific surface area, lubrication time and tablet
hardness (linked to compression pressure)

Prediction algorithm:
Diss = 108.9 – 11.96 × API – 7.556×10-5 × MgSt – 0.1849 × LubT –
3.783×10-2 × Hard – 2.557×10-5 × MgSt × LubT


 Account for uncertainty
◦ Sources of variability (predictability, measurements)
 Confirmation of model
◦ compare model results vs. actual dissolution results for
batches
◦ continue model verification with dissolution testing of
production material, as needed

Batch 1 Batch 2 Batch 3

Model prediction 89.8 87.3 88.5

Dissolution testing 92.8 90.3 91.5
result (88.4–94.2) (89.0-102.5) (90.5-93.5)


 Response surface plot for effect of API
particle size and magnesium stearate specific
surface area (SSA) on dissolution
Diss (% at 20 min)

Area of potential
Design risk for dissolution
Space failure

Graph shows interaction
between two of the variables:
API particle size and magnesium
stearate specific surface area
API particle size (Log D90)
Acknowledgement: adapted from Paul Stott
(AZ)

 Controls of input material CQAs
◦ API particle size distribution
 Control of crystallisation step
◦ Magnesium stearate specific surface area
 Specification for incoming material

 Controls of process parameter CPPs
◦ Lubrication step blending time
◦ Compression pressure (set for target tablet hardness)
 Tablet press force-feedback control system

 Prediction mathematical model
◦ Use in place of dissolution testing of finished drug product
◦ Potentially allows process to be adjusted for variation in API particle
size, for example, and assure dissolution performance


Impact on Assay and Content Uniformity CQAs
 QRA shows API particle size, moisture control, blending and
lubrication steps have potential to affect Assay and Content
Uniformity CQAs
◦ Moisture is controlled during manufacturing by facility HVAC control
of humidity (GMP control)
Drug Moisture
substance content in Blending Lubrication Compression Coating Packaging
particle size manufacture
in vivo performance
Dissolution
Assay
Degradation
Content uniformity
Appearance
Friability
Stability-chemical
Stability-physical

- Low risk
- Medium risk
- High risk


Decision on conventional vs. RTR testing


DOE for the Blending Process Parameter Assessment to
develop a Design Space
◦ Factors Investigated:
Blender type, Rotation speed, Blending time, API Particle size

Blending time Rotation speed Particle size D90
Experiment Run Condition Blender
(minutes) (rpm) ( m)
No.
1 2 varied 2 10 V type 5
DOE design

5 6 varied 2 10 Drum type 40
9 3 standard 9 20 V type 20
10 12 standard 9 20 Drum type 20
11 9 standard 9 20 V type 20
12 4 standard 9 20 Drum type 20


Blend uniformity monitored using a process
analyzer
 Control Strategy to assure homogeneity of the blend
◦ Control of blending
end-point by NIR
and feedback control
of blender
◦ API particle size
In this case study, the
company chooses to use
online NIR to monitor blend
uniformity to provide
efficiency and more
flexibility


 On-line NIR spectrometer 0.045

used to confirm scale up of

mean spectral standard deviation
0.04
blending 0.035
 Blending operation complete 0.03
Pilot Scale

when mean spectral std. dev.
Full Scale
0.025
reaches plateau region 0.02
◦ Plateau may be detected
0.015
using statistical test or rules Plateau region
0.01
 Feedback control to turn off
blender
0.005

Company verifies blend does
0
 0 32 64 96 128
not segregate downstream Revolution Revolutions of Blender
Number of (block number)

◦ Assays tablets to confirm
uniformity
◦ Conducts studies to try to Data analysis model will be provided
segregate API Plan for updating of model available
Acknowledgement: adapted from ISPE PQLI Team


Conventional automated control of Tablet Weight using
feedback loop:
Sample weights fed into weight control equipment which sends signal to
filling mechanism on tablet machine to adjust fill volume and therefore tablet
weight.


NIR Spectroscopy
NIR Monitoring Laser Diffraction (At-Line)
Blend Uniformity Particle Size • Identity
• Assay
Raw materials & • API to Excipient
API dispensing ratio
• Specifications
based on product

Roller Tablet Pan
Dispensing Blending Sifting
compaction Compression Coating


 Real Time Release Testing Controls
◦ Blend uniformity assured in blending step (on-line NIR spectrometer
for blending end-point)
◦ API assay is analysed in blend by HPLC
 API content could be determined by on-line NIR, if stated in filing
◦ Tablet weight control with feedback loop in compression step
 No end product testing for Assay and Content
Uniformity (CU)
◦ Would pass finished product specification for Assay and Uniformity
of Dosage Units if tested because assay assured by combination of
blend uniformity assurance, API assay in blend and tablet weight
control (if blend is homogeneous then tablet weight will determine
content of API)


 Before a medicinal product is released for sale,
the Qualified Person responsible for its release
should take into account, among other aspects,
the conformity of the product to its specification.
 In the case of approved RTRT, this conformity
would not routinely be supported by results of
end product testing.
 Nevertheless a specification has to be established
and each batch of a product should comply with
it if tested.


 The application for RTRT should be supported by
adequate validation of the RTR test method.
 The relationship between the RTR test, including
acceptance criteria, and the end product test and
associated specification should be well
understood and, where applicable, supported by
substantial comparative data at commercial scale
(parallel testing).


 When RTRT has been approved this should be
routinely used for batch release.
 In the event that the test results of RTRT fail or
are trending toward failure, RTRT may not be
substituted by end-product testing.
 Any failure should be investigated and trending
should be followed up appropriately.
 Batch release decisions will need to be made
based on the results of these investigations, and
must comply with the content of the marketing
authorization and current GMP requirements.


 Attributes (e.g. uniformity of content) that is
indirectly controlled by approved RTRT
should still appear in the Certificate of
Analysis for batches.
 The approved method for end-product
testing should be mentioned and the results
given as ”Complies if tested” with a footnote:
”Controlled by approved Real Time Release
testing”.


 In case of equipment failure the control strategy
provided in the application should include a
contingency plan specifying the use of alternative
testing or monitoring approaches on a temporary
basis.
 In this situation, the alternative approach could
involve use of end-product testing or other options,
while maintaining an acceptable level of quality.
 Testing or monitoring equipment breakdown needs
to be managed in the context of a deviation under
the Quality Management System and can be covered
by GMP.


In principle, end product testing should not be substituted
for failure of an RTRT release method. The failure should
be investigated and followed up appropriately.


 When RTRT is applied, the attribute that is indirectly
controlled (e.g. sterility, uniformity of content)
together with a reference to the associated test
procedure, should still be included in the
specification as “Conforms if tested”.
 The relationship between end-product testing,
material attributes, process monitoring and
acceptance criteria, should be fully explained and
justified.
 In addition, the use of any prediction models should
be fully explained, justified and verified at the
commercial site.


 Batch release is the final decision to release
the product to the market regardless of
whether RTR testing or end-product testing is
employed.
 End-product testing involves performance of
specific analytical procedures on a defined
sample size of the final product after
completion of all processing for a given batch
of that product.


 Results of real-time release testing are handled
in the same manner as end-product testing
results in the batch release decision.
 Batch release involves an independent review of
batch conformance to predefined criteria through
review of testing results and manufacturing
records together with appropriate good
manufacturing practice (GMP) compliance and
quality system, regardless of which approach is
used.


 Real-time release testing does not necessarily
eliminate all end-product testing.
 For example, an applicant can propose RTR
testing for some attributes only or not all.
 If all critical quality attributes (CQAs) (relevant for
real-time release testing) are assured by in-
process monitoring of parameters and/or testing
of materials, then end-product testing might not
be needed for batch release.
 Some product testing will be expected for certain
regulatory processes such as stability studies or
regional requirements.


 Product specifications (see ICH Q6A and Q6B)
still need to be established and met, when
tested.


 Even where RTR testing is applied, a stability
monitoring protocol that uses stability
indicating methods is required for all
products regardless of the means of release
testing (see ICH Q1A and ICH Q5C).


 RTR testing, if utilized, is an element of the
control strategy in which tests and/or
monitoring can be performed as in-process
testing (in-line, on-line, at-line) rather than
tested on the end product.


 Traditional sampling plans for in-process and
end-product testing involve a discrete sample
size that represents the minimal sampling
expectations.
 Generally, the use of RTR testing will include
more extensive on-line/in-line measurement.
 A scientifically sound sampling approach
should be developed, justified, and
implemented.


 In principle the RTR testing results should be
routinely used for the batch release decisions
and not be substituted by end-product testing.
 Any failure should be investigated and trending
should be followed up appropriately.
 However, batch release decisions should be made
based on the results of the investigations.
 The batch release decision should comply with
the content of the marketing authorization and
GMP compliance.


 In-process testing includes any testing that
occurs during the manufacturing process of
drug substance and/or finished product.
 Real-time release testing includes those in-
process tests that have a direct impact on the
decision for batch release through evaluation
of critical quality attributes.


 RTR testing can be based on measurement of
a surrogate (e.g., process parameter, material
attribute) that has been demonstrated to
correlate with an in-process or end-product
specification (see ICH Q8(R2); Annex, section
II.E (2.5)).


 Is there a potential for the measured CQA to
change downstream from the measurement point?
For example,
◦ Blend desegregation
◦ Loss of weight (e.g., chipping) after weighing step
◦ Hydrolytic degradation during aqueous film coating
 Is identity determined at a point that is visually
unique?
◦ Mitigation of potential human and/or system error
◦ Unique identifiers on the intermediate when measured
(e.g., embossing, size, shape)
 Risk assessment is valuable to exploring potential
failure modes


 Probe/sample location representative of entire
vessel
 Sample frequency representative of entire batch
 Sample acquisition time
◦ Suitable for system dynamics/mixing
 Sample volume/mass
◦ Determine amount of sample measured
◦ Representative of unit dose
 Sample interface
◦ Remains constant over the process (e.g., no fouling)
◦ Environmental factors (e.g., temperature, humidity)


 Specification still required in an RTRT approach
◦ (CFR §314.50(d) and CFR § 211.165(a))
 Should be representative of actual measurement
◦ Can include in-process measurements (e.g., NIR
measurements for assay of uncoated tablets)
◦ Can include surrogate measurements (e.g., models for
dissolution)
◦ Methods should be appropriately validated (including
models used as surrogate measurements)
 Alternatives can be included for stability
monitoring
 Appropriate statistical criteria for large sample
sizes


 Calibration models for spectroscopic analysis
◦ NIR, Raman, FTIR
◦ Typically use chemometric models
 Surrogate models for time consuming measurements
◦ Dissolution models relating process parameters and/or
material attributes to dissolution
 Design space models
◦ Surface response plots
◦ Mechanistic models
 Process control models
◦ Tunable controllers for individual unit operations
◦ Statistical process control & multivariate statistical process
control
 Other models


 Calibration data
◦ Include potential sources of variance (e.g., operating conditions,
raw materials, scale)
◦ Cover intended areas of operation/design space
◦ Appropriate distribution of spectra over the analysis range
 Model development
◦ Appropriate data pre-treatment
◦ Appropriate spectral ranges
◦ Number of model factors justified (avoid overfitting)
 Model validation
◦ Internal validation using subsets of calibration data
◦ External validation using an independent data set
 Robust and representative reference method


 NIR model results may change with time as new
sources of variability are introduced.
◦ Changes in raw material suppliers, process or analyzer
changes
 Evaluation of outliers as part of maintenance.
◦ Can detect bad spectra or interface problems
◦ Usually implemented through examination of residuals
 Procedures in place to monitor and update the model
◦ Done under the manufacturer’s quality system
◦ Include frequency and methods of periodical model
evaluation
 Depth of validation done on updated model,
depending on level of change


 Robust calibration model
◦ Use an appropriate reference method
◦ Include variations in raw materials
◦ Cover the entire design space
 Include an independent dataset for validation
 Demonstrate model performance at
commercial scale
◦ Understand and work within the model limitations
and model assumptions
◦ Compare model results to a reference method for a
statistically acceptable number of batches


 Develop and document procedures on how to
evaluate and update the calibration model
◦ How to deal with OOS results
◦ Develop criteria for model re-calibration
 Verify or recalibrate the model for process changes:
◦ Revising the operating ranges
◦ Change in raw materials
◦ Change in manufacturing equipment or measuring
instrument
 Include plans for model maintenance/update in the
firm’s Quality System
◦ Tracking/trending (for process monitoring) included within
the Quality System


 Level of detail in submission should depend on the
importance of the model to the overall control strategy
 Low Impact Model (e.g., Models for development)
◦ General discussion of how model was used to make decisions
during process development
 Medium Impact Model (e.g., Design space models)
◦ More detailed information about model building, summary of
results and statistical analysis
◦ Discussion of how the model fits into the control strategy
 High Impact Model (e.g., RTRT models)
◦ Full description of data collection, pretreatment and analysis
◦ Justification of model building approach
◦ Statistical summary of results
◦ Verification using data external to calibration set
◦ Discussion of approaches for model maintenance and update


Quality What is
Product Profile Target Identify critical to
Product CQA the
Profile
CQA’s Patient

QRM
PAT
Risk Assessments
Identify
CMA &
Design Space CPP

Design space Control Strategy
Control Strategy

Continual
Improvement

PAT , PAT RTRT
SOP PAT RTRT 97

Real time release testing

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Viewers also liked

Viewers also liked (20)

Similar to Real time release testing

Similar to Real time release testing (20)

More from GMP EDUCATION : Not for Profit Organization

More from GMP EDUCATION : Not for Profit Organization (20)

Recently uploaded

Recently uploaded (20)

Real time release testing