Process Characterization – It’s Not A Science Project March 1, 2017 BioProcess Technology Consultants www.bptc.com CONFIDENTIAL
Why Do Process Characterization? What’s the Goal? Standard answers from clients • Increase process knowledge ― A correct, if esoteric answer. It begs the question, “Then what?” • Establish Proven Acceptable Ranges (PARs) ― Better, but for what immediate and ultimate purpose? • Design of Experiments • Creating mathematical models relating process variables to Critical Quality Attributes It’s Not A Science Project! 2 CONFIDENTIAL
Why Do Process Characterization? What’s the Goal? From PDA Technical Report #60: Process Validation: A Life Cycle Approach, 2013 has the answer • From Section 3.7 ― ( What PC is ): “Process characterization is a set of documented studies in which operational parameters are purposefully varied to determine their effect on product quality and process performance” ― (Almost there!) : “The approach uses the knowledge and information from risk assessments to determine a set of process characterization studies to examine proposed ranges and acceptance criteria” ― (Here’s the Goal or Application of PC): “ The resulting information is used to define the PPQ ranges and acceptance criteria ” 3 CONFIDENTIAL
From BPTC Standard Templates for PPQ Protocols PC studies needed to ensure that operation at the NORs specified in PPQ protocol (and at values somewhat outside of the NOR to account for measurement error and possible deviations) will not impact CQAs and process performance. 4 CONFIDENTIAL
From BPTC Standard Templates for PPQ Protocols In addition to historical process data from full scale GMP manufacturing lots, PC studies are required to specify acceptable CQA limits for PPQ campaign. 5 CONFIDENTIAL
From BPTC Standard Templates for PPQ Protocols In addition to historical process data from full scale GMP manufacturing lots, PC studies are required to specify acceptable performance for PPQ campaign. 6 CONFIDENTIAL
With Goal Now Defined, How Do We Design PC Program? What variables do we study? • Those that are at the greatest risk of screwing up your PC campaign What variables would those be? • Those that were ranked the highest during the risk assessment (e.g., the RPN score from an FMEA) OK, we have the variables we want to study, what ranges do we study for each variable • Study values a little bit outside of the Normal Operating Ranges (NORs) to account for measurement error and deviations How do we determine what the NORs are? • ……. 7 CONFIDENTIAL
What is an NOR and How Do We Establish Them? One definition of NOR • From TR60: “a defined range, within (or equal to) the Proven Acceptable Range, specified in the manufacturing instructions as the target and range at which a process parameter is controlled, while producing unit operation material or final product material meeting release criteria and CQAs” ― NOR < PAR (That’s the only guidance TR60 gives) Not a helpful definition. For example ….. • Suppose we have data to know that a temperature between 34 and 39 o C has no impact on product quality or process performance (i.e., the PAR is 34 to 39). How does this help us define NOR? ― Is it 34 – 39? 35 – 38? 36.5 – 37.5? ??????? ― Key term in NOR is “normal”. Where does the process normally run? What is a reasonable range within which to control a process parameter regardless of PAR? 8 CONFIDENTIAL
Establishing Normal Operating Ranges Another, more useful, definition of NOR • “A defined range within (or equal to) the Proven Acceptable Range, specified in the manufacturing instructions as the target and range at which a process parameter is controlled, and that can be achieved with a high degree of assurance and reproducibility ”. • One way to establish NOR is from historical data ― Use process data from historical batches to calculate 95%/99% tolerance intervals 7.35 99%/99% BioRx pH or Elution pH or ??? 7.25 95%/99% 7.15 Example: 30 batches are shown and plotted. 95%/99% 7.05 tolerance interval is 6.77 to 6.95 7.22. Thus, NOR is 6.77 – 6.85 7.22 (as long as PAR is wider). 6.75 6.65 0 5 10 15 20 25 30 Run # or Sample # 9 CONFIDENTIAL
Establishing Normal Operating Ranges Best way to establish NORs is from historical at-scale data • Use process data from historical batches to calculate 95%/99% tolerance intervals But what if only a few historical lots have been manufactured, how does one establish NOR? • Same idea as previous slide – propose ranges where you suspect process can operate 90% or 95% of the time (regardless of impact on process or product as that will be discovered during PC) • Idea is to minimize the need to execute process deviations and investigations during commercial manufacturing • Example: for process buffer pH, given the equipment, procedures and instrument tolerances, propose a pH range for the process buffer that can be met 90% or 95% of the time (regardless of impact on process or product as that will be discovered during PC) 10 CONFIDENTIAL
We’ve Established NORs: What’s Next for PC Program? We know what variables we want to study We know the NORs for the variables What ranges do we study for each variable? • Study values a little bit outside of the Normal Operating Ranges (NORs) to account for measurement error and deviations How far outside the NORs should be studied in our DOEs? • One answer: Potentially very far outside because we want to see an impact on CQA and process attributes so that the DOE results generate a mathematical model relating process variables to ……. It’s Not A Science Project! 11 CONFIDENTIAL
Mathematical Models “Don’t Do No Harm” ……… The need for mathematical models reminds me of a line from the movie Western “Unforgiven” • Gene Hackman as Little Bill Daggett ― “ Look son, being a good shot, being quick with a pistol, that don't do no harm, but it don't mean much next to being cool-headed. A man who will keep his head and not get rattled under fire, like as not, he'll kill ya ” • Having mathematical models relating process variables to CQAs and performance attributes “don’t do no harm”, but it is NOT the point of the PC program. • The point of the PC program is to: ― Provide experimental support for the NORs specified in the PPQ protocols; ― Provide some wiggle room to operate in the event the NORs are exceeded (i.e., deviations) & to account for measurement variability 12 CONFIDENTIAL
We’ve Established NORs: What’s Next for PC Program? So, how far outside the NORs should be studied in our DOEs • One answer: Potentially very far outside because we want to see an impact on CQA and process attributes so that the DOE results generate a mathematical model relating process variables to ……. • Far enough to account for measurement variability & to provide wiggle room to operate in the event the NORs are exceeded so as to provide valuable information for investigations to deviations • Example: If NOR for pH of column load is 6.7 to 7.3 ― Explore 6.5 to 7.5 during PC program NOR pH 6.7 7.3 Range studied for PC 6.5 7.5 to establish PAR 13 CONFIDENTIAL
What is an PAR and How Do We Establish Them? One definition of PAR • From TR60: “A characterized range of a process parameter for which the operation within this range, while keeping other parameters constant, will result in producing a material meeting relevant quality criteria.” • “…. producing Range of acceptable performance material meeting relevant quality criteria”. What is relevant quality criteria? • Again, can rely NOR NOR on historical performance of clinical batches • Graph shows historical clinical batches 14 CONFIDENTIAL
Establishing Proven Acceptable Ranges Example • Characterization data shows all experiments studied met all criteria of acceptable performance defined as that seen in clinical batches • Thus, PAR is the entire range studied • Note: No mathematical models required! ― No need to look at wider ranges to see an impact on performance NOR NOR PAR PAR Process Characterization Results Manufacturing Scale Results 15 CONFIDENTIAL
Establishing Proven Acceptable Ranges Another Example • Characterization data shows some experiments failed at values less than the lowest boundary of the NOR • Thus, PAR is range for which parameter gave acceptable results ― Data shown could be actual data or “data” obtained from a model fit • Lower end of NOR in danger of providing unacceptable quality ― Need more data at lower end NOR PAR Process Characterization Results Manufacturing Scale Results 16 CONFIDENTIAL
Establishing Proven Acceptable Ranges Another Example • Characterization data shows some experiments failed at values less than the lowest boundary of the NOR • Thus, PAR is range for which parameter gave acceptable results ― Data shown could be actual data or “data” obtained from a model fit • Lower end of NOR in danger of providing unacceptable quality ― Need more data at lower end; seems like critical process parameter NOR PAR Process Characterization Results Manufacturing Scale Results 17 CONFIDENTIAL
Recommend
More recommend
