The CM system and its control strategy were designed to control parameters that impact the 602
manufacture and quality of the drug substance, including impurity profile and physicochemical 603
properties. The overall control strategy was developed in accordance with the main guideline and 604
ICH Q7–Q11.
605
2.1. Equipment Design and Integration 606
Reaction 1: Starting materials 1 and 2 are coupled in a PFR to produce Intermediate 1.
610
Diversion Point D1 is located after the PFR to permit material diversion when PFR 611
conditions are outside predefined acceptance criteria. The reaction is quenched as an 612
integrated operation after the PFR, and unwanted by-products are removed by liquid-liquid 613
Reaction 2: Intermediate 1 and Intermediate 2 (prepared upstream through separate batch 617
unit operations) are coupled in a second PFR to form the crude drug substance. The online 618
PAT near the reactor exit (T1) monitors conversion of Intermediate 1 to the crude drug 619
substance. Diversion Point D2 located after PAT is used to divert non-conforming material.
620 621
Drug Substance Isolation: The crude drug substance is purified by carbon filtration and 622
continuous two-stage crystallisation. The crystal slurry is filtered by using two identical 623
filtration units running in an alternating fashion. This setup enables continuous processing 624
of the drug substance after crystallisation by allowing the collection of crystallised products 625
on one filter unit at the same time product isolated on the second filter is discharged.
626
Diversion Points D3 and D4 allow for material diversion at the crystalliser and just before 627
batch operations, respectively. A batch dry milling operation is used to achieve the desired 628
particle size distribution of the crystallised drug substance.
629 630
Three surge points (each containing multiple surge tanks) are used: one before Reaction 2, another 631
before the two-stage continuous crystallisation, and one just before final batch operations. These 632
are important components of the system design and control strategy, as they improve process 633
robustness and mitigate temporary differences in mass flow rates by decoupling upstream and 634
downstream operations.
635 636
The design of the overall system and each unit operation, along with the control strategy, optimise 637
material quality. For example, PFR design elements (i.e., dimension and configuration) allow 638
precise control of temperature, mixing and reactant flows. These parameters were shown during 639
development to be important to the drug substance impurity profile.
640
2.2. Process Control and Monitoring 641
Holistic controls used across Reactions 1 and 2 ensure consistent operations and quality of the 642
resulting crude drug substance. The stoichiometry of Reaction 1 is controlled precisely via control 643
of concentrations and flow rates of the feeds. Conversion of starting materials to Intermediate 1 644
18
with minimal impurity formation is ensured through control of the reaction temperature. Reaction 645
2 is controlled through feedback control of the addition rate of Intermediate 2 based on the PAT 646
measurement of Intermediate 1 levels. This ensures correct stoichiometry for that reaction and 647
minimises the impact of variability of the Intermediate 1 feed solution on drug substance purity.
648
The PAT also measures levels of crude drug substance and process impurities, which confirm 649
successful operation of all preceding steps and consistent product quality.
650 651
RTD was used to develop a suitable strategy for disturbance detection, corrective actions, and 652
material diversion. RTD characterisation was based on mathematical modeling of all unit 653
operations and surge points across the entire CM process over planned mass flow rates. The RTD 654
was then confirmed through experimental tracer studies for appropriate segments of the 655
commercial equipment. Decisions for triggering material diversion are based on comparing 656
process parameters and PAT measurements to predefined acceptance criteria with timing and 657
Understanding of process dynamics and its impact on quality attributes of material produced 661
throughout the entire process was also used to guide start-up and shutdown strategies. For example, 662
during start-up of Reactions 1 and 2, a small amount of Intermediate 1 or crude drug substance is 663
diverted at Diversion Points 1 or 2, respectively, to allow those materials to reach the target 664
concentrations before processing into subsequent operations. The criteria for diversion were 665
established based on time considering the RTD. This approach was supported by development 666
studies and confirmed in commercial process equipment. PAT monitoring after Reaction 2 667
provides additional verification that appropriate criteria have been met during start-up. Collection 668
of material proceeds to the end of the process as subsequently described.
669 670
Sampling and process measurement needs were evaluated, considering relevant factors such as 671
residence times (RTs)/RTD, surge points, process dynamics, and the type and purpose of the 672
measurement. The measurement frequency of the PAT at Reaction 2 is sufficient to detect 673
disturbances, inform process adjustments, and ensure timely diversion of material based on 674
predefined criteria. The criteria for material diversion are based on the magnitude and duration of 675
the disturbance, an understanding of process dynamics and RTD for downstream unit operations 676
and surge points, and the impurity purging capability of the crystallisation operation. As a result 677
of this control strategy, all crude drug substance solution that enters continuous crystallisation 678
meets acceptable quality criteria and can be forward processed through the crystalliser.
679 680
Appropriate controls and monitoring requirements for the continuous crystallisation were 681
extensively investigated during development in similar, but smaller scale equipment and verified 682
using commercial equipment. Process development included spiking studies using impurity-683
enriched feed solutions and intentional perturbations in process parameters (i.e., feed flow rates, 684
their ratios, and temperatures). An evaluation of the encrusted solids in the crystalliser over 685
extended run times demonstrated the solids were the same form and purity as the free-flowing drug 686
substance slurry. The set of process parameters and ranges identified by these studies were 687
appropriately scaled up. Implementation of these controls along with post-crystallisation material 688
tests (e.g., crystal form, purity) ensure consistent quality of the resulting drug substance throughout 689
continuous crystallisation and filtration.
690
19 691
The resulting material is collected at Surge Point 3 and is dried and milled using batch operations 692
to provide a drug substance of the appropriate particle size for use in drug product 693
manufacturing. Procedures were developed to allow diversion of material at Diversion Points D3 694
or D4 in the event desired process conditions or material attributes are not met. However, diversion 695
of the drug substance from the crystalliser was found to be unnecessary either during start-up or 696
shutdown.
697
2.3. Consideration of Other Controls 698
Process robustness and performance over time are important considerations. A risk assessment 699
was performed to ensure that adequate controls are in place to support the proposed run time 700
(which can be up to several months). It identified a number of considerations and corresponding 701
controls/measures. Examples are summarised in Table 2.
702 703
Table 2: Examples of other controls for consideration 704
Consideration Controls/Measures
Cleaning and fouling potential
Establishment of a risk-based cleaning strategy, including
understanding of the impact of build-up on drug substance quality
Additional monitoring to assess fouling and cleanliness (e.g., pressure sensors at the discharge of feed pumps, periodic visual checks for the continuous crystalliser)
Reduction of other risk factors (e.g., filtering feed streams to further reduce fouling risk)
Stability of in-process materials
Hold times at key points in the process (e.g., feed streams;
accumulated material at the surge points, reactors, and crystalliser) managed through batch record and process automation
Risk assessment of microbiological growth (i.e., negligible risk based on the nature of the process materials and conditions)
Calibration and potential for changes/drift in instrumentation
Periodic checks at selected points (e.g., process parameter
measurements for the PFR, system suitability for the PAT analyser)
Dual sensors at selected locations (e.g., temperature probes for the PFR) so that appropriate corrective actions can be taken
Equipment maintenance
Maintenance requirements for target run time
Use of redundant equipment (e.g., backup pumps) at key locations to enable continuous operation
705
Additionally, specifications for input materials were evaluated during process development. There 706
were no differences between batch and continuous processing for this example.
707 708
Collectively, the process understanding developed along with implementation of the various 709
controls described provide a robust and reliable control strategy. This ensures consistent quality of 710
the resulting drug substance including the impurity profile, physicochemical properties, and ability 711
of the system to identify and appropriately react to unexpected events.
712
2.4. Process Validation 713
20
The combination of process controls, online PAT measurements, comprehensive monitoring of 714
process parameters and material attributes, and end-product testing results in a data-rich 715
environment for this process. Together with system understanding generated during development, 716
this enabled the use of a traditional process validation for commercial product launch and 717
continuous process verification to validate process changes over the product lifecycle.
718 719
A range of batch sizes was initially established based on material demands and the quantities of 720
material necessary to match input needs of the final batch unit operations. The process was 721
validated using a fixed number of batches. A single planned start-up and shutdown of the 722
commercial CM system was used to manufacture the process validation batches. This approach 723
was supported by the totality of evidence demonstrating the start-up and shutdown capabilities of 724
the system. This included development work on similar equipment, commercial equipment and 725
system qualification data, results of a prevalidation demonstration run, and extensive process 726
monitoring of the CM system that can verify success of each start-up and shutdown in real time.
727 728
Subsequently, a continuous process verification approach was adopted after product approval to 729
support increases in batch size with extension of run time. This approach used a risk assessment 730
for the longer run time, which concluded that process performance and material quality would not 731
be impacted. Under the continuous process verification approach, data generated during the 732
manufacture of each batch was used to support successful validation of that batch with the 733
extended run time. This included information such as system performance monitoring and data 734
logs along with other controls that ensure material quality with appropriate detection and corrective 735
action. Additionally, appropriate regulatory actions were taken to communicate this manufacturing 736
change and use of the continuous process verification approach.
737
3. REGULATORY CONSIDERATIONS 738
Refer to Section 4 of the main guideline. In consideration of the specific CM process design, 739
additional elements may need to be included in a dossier. For instance, in this example, the 740
influence of surge points on the material diversion and collection strategy, including the fate of 741
materials, was described.
742
21
ANNEX II: CONTINUOUS MANUFACTURING FOR DRUG PRODUCTS