It is now possible, as demonstrated by Janssen (1), for drug manufacturers to convert their processes to adopt continuous production of oral solid-dosage forms produced in compact, closed units, with a higher level of automation and minimal manual interventions. In continuous manufacturing, the sequential production steps that are part of a classic batch process are integrated into a continuous process. To ensure GMP compliance, cleaning is required on some frequency between batches of the same product and when switching between different products.

	It is important to define continuous manufacturing and to understand the characteristics that differentiate it from a typical batch process. The classical definition involves the following:

		All crucial steps involved to produce something from its starting materials to the end-product must be integrated into a steady uninterrupted process.

		The systems should operate continuously (non-stop) targeting at least 50 weeks per year.

		There should be no significant downtime in product or process changeover.

	Some variations of this processing mode may be foreseen over the years. Processes that do not fit the above definition, such as infrequent manufacture (e.g., operates only once or twice per year) or a multi-product process train (i.e., same equipment used for many different products), are generally not considered continuous manufacturing (2).

	Although most companies do not have immediate plans for implementation of continuous manufacturing, many GMP professionals have questions concerning the impact of this type of process on cleaning and cleaning validation. People involved in quality, engineering, operations, and validations are asking about applying cleaning validation to pharmaceuticals produced in a continuous manufacturing process. The purpose of this article is to provide a review of the regulatory expectations on cleaning and cleaning validation and to help drive efficiency by rethinking the design of the cleaning process in continuous manufacturing.

	Professionals interested in understanding cleaning and continuous manufacturing must become familiar with the following aspects:

		The regulatory perspective on cleaning validation and applicability to continuous manufacturing

		The importance of the lifecycle model and risk assessment tools to successfully design and implement a cleaning procedure for continuous manufacturing

		Testing, sampling, process analytical technology (PAT), and equipment design concepts that should be reviewed and considered to optimize cleaning in continuous manufacturing

		How to establish cleaning residue limits explicitly for continuous manufacturing.

	In the United States, the regulatory expectation is that equipment be cleaned prior to manufacturing to prevent contamination or adulteration of products (3). Similarly, in Europe, Canada, Asia, Latin America, and other territories, equipment cleaning is a regulatory requirement cited in their applicable GMPs. According to 21 Code of Federal Regulations (CFR) Part 211.67 Equipment cleaning and maintenance, “Equipment and utensils shall be cleaned, maintained, and sanitized at appropriate intervals to prevent malfunctions or contamination that would alter the safety, identity, strength, quality, or purity of the drug product beyond the official or other established requirements” (3). Also, in 1993, FDA issued the Guide to Inspections of Validation of Cleaning Processes to assist the industry in compliance with GMP requirements (4). The European Medicine Agency (EMA), Health Canada, and others subsequently issued similar guides to assist the industry in complying with the relevant regulations (5–6).

	Based on these guidance documents, cleaning validation is clearly a necessity when the equipment is used to manufacture more than one drug product due to cross-contamination concerns. A question remains as to whether cleaning validation is required for dedicated equipment, which may be the case for most continuous manufacturing processes. The need to validate the cleaning procedure when the equipment is dedicated to one product is generally left to the company’s discretion, and it must be supported with proper justification. Nevertheless, cleaning validation of dedicated equipment has been discussed in other publications, and the rationale behind this can also be applied to continuous manufacturing (7). In the case of continuous manufacturing, cleaning validation of dedicated equipment should be done to demonstrate that the cleaning procedure can effectively remove residue build-up and undesirable residues (including microbial) produced during a specific length of manufacturing (i.e., campaign) that may compromise product quality and patient safety.

	As of the time of publication, no specific regulation or guidance document for continuous manufacturing has been released. FDA authorities have presented a general perspective on continuous manufacturing, which addresses the concerns around “lot” and “batch” definitions (8). According to the CFR, “batch” and “lot” refer to a quantity of material with uniform characteristics and do not specify mode of manufacture; no regulations or guidance documents forbid the adoption of continuous manufacturing. In fact, continuous manufacturing seems to be consistent with philosophies, such as quality by design (QbD) and the lifecycle approach to process validation, found in current guidance documents (9–12).

	In 2011, FDA issued a process validation guide (12) focusing on three main elements:

		Quality needs to be built into the process

		Quality is not guaranteed by in-process or final process testing

		A manufacturing process needs to be defined and continuously monitored to ensure consistent quality.

	The elements of the lifecycle model are the building blocks to a harmonized approach to process validation and subsequently, cleaning validation. The lifecycle elements of design, validation, and monitoring are also the building blocks to make continuous processing an efficacious mode of manufacturing. The process lifecycle approach as discussed in the guidance document focuses on understanding the processes and ensuring that they are meeting the requirements set forth in the design stage. The elements of the lifecycle approach should not be limited to manufacturing since they may also be applied to other processes including cleaning. With this in mind, the lifecycle model can be used to improve or optimize cleaning procedures by having a better understanding of the input variables and the output attributes.

	Process efficiency is important in continuous manufacturing and cleaning is not an exception. For example, at some pre-established schedule, the continuous manufacturing facility must be shutdown to perform equipment cleaning and maintenance. Cleaning procedures are expected to be done quickly and correctly the first time in order to meet the optimum changeover time as established in the production schedule. Also cleaning procedures should be systematically designed to reduce waste within the cleaning process.

The design stage of the lifecycle model (Stage 1) is of particular importance because the cleaning process is defined based on knowledge gained through development and scale-up activities. This stage ensures that the variables are identified and their criticality to the cleaning process is assessed. For example, the design inputs or critical cleaning process parameters for the wash step include the cleaning agent, concentration, temperature, time, cleaning method, water quality, and environmental factors (13–14). Laboratory, pilot, or field studies should be used to help define the cleaning process and identify conditions that would lead to the desired fast, effective, and lean cleaning process (see 

). These studies may involve cleaning evaluations with wet, baked-on residue found in the drier as well as dry, compacted residue for tablet pressing equipment (see 

	A formal risk assessment for the cleaning process is also recommended using a system for identifying and managing risk such as fault tree analysis (FTA), hazard analysis and critical control point (HACCP), or failure modes and effect analysis (FMEA) (15–16). Risk assessment should be based on the knowledge gained through the design stage and focus should be placed on the issues that have potential impact on product quality and patient safety. 

 includes items to consider for building a cleaning process knowledge base. A design of experiments could be used to identify those parameters that have a significant impact to cleaning within a specified range (17).

	Cleaning processes are often viewed as time- and resource-consuming activities that only add to the operational costs of product manufacturing. Delays in equipment readiness due to cleaning failures, lengthy manual cleaning procedures, or off-line sampling wait times can challenge continuous manufacturing schedules and result in costly production delays.

	Manufacturers using batch processing, including cleaning processes, perform laboratory testing conducted on pulled samples to evaluate quality attributes. PAT, however, can be used in continuous manufacturing to provide real-time continuous analysis and release of the cleaned equipment. PAT is a system for designing, analyzing, and controlling manufacturing through timely measurements, process understanding, and process control (18). In cleaning applications, PAT may be applied to complement the cleaning performance qualification and later, to support continued verification. For example, concentration-versus-conductivity plots are pre-established to monitor the final water rinse for residual cleaning agent (19). Total organic carbon (TOC) analysis can also be correlated to process residues. In a published case study, a cleaning process was evaluated using an on-line TOC analyzer integrated into the return line of a clean-in-place (CIP) system (20).

	There are sampling options (e.g., rinse and swab) universally acceptable for monitoring the cleaning performance of the targeted residues; each option has advantages and disadvantages to consider. For most cleaning applications, rinse sampling is typically expected to be faster and easier compared to swab sampling, which may require access to equipment locations through disassembly or confined-space entry and consequently compromises personnel safety.

	Analytical methods can be specific or non-specific to determine the amount of residue on the surface through direct and indirect sampling methods. Non-specific detection methods, such as pH, conductivity, or TOC, can be used to measure multiple residue types. These types of methods are preferred due to the quick turnaround time, minimum waste generation, and simplicity of the assay.  Ultra high-pressure liquid chromatography (UHPLC) instruments measure residue based on detection of a specific analyte and is a faster alternative over traditional HPLC (21).

 lists examples of at-line and in-line methods that could be used for detection of residual cleaning agent. For continuous manufacturing, it is recommended to review testing technologies and select one that makes the most sense given the analytical resources, type of residue and carry-over risk, speed of analysis, and/or adaptability to PAT.

	Batch pharmaceutical processes employ a variety of methods to clean process equipment. Manual cleaning methods using wipes and brushes are common in legacy batch-mode processes while CIP systems are popular at newer facilities. Even though both manual and automatic cleaning methods are accepted by drug regulatory agencies around the globe, companies considering continuous manufacturing processes should opt for CIP systems because they are effective, consistent, and reliable. Consistent cleaning results are achieved because there is minimal operator intervention, reduced likelihood of human error, and consistent control of critical parameters when combined with PAT technologies. The principal objective of a CIP system is to achieve the desired cleanliness level without disassembling the process equipment. Generally, CIP cleaning is done by circulating cleaning solutions through pipes, pumps, valves, and spray devices that distribute the cleaning agent over the surface areas of the equipment. PAT technology can be used to monitor cleaning steps, such as the preparation of cleaning solution to a pre-established concentration and the rinsing of the equipment down to pre-established residue limits.

	With adequate sanitary design, such as coverage, diaphragm valves, pitch, and dead-leg orientation, CIP systems can deliver faster and lean cleaning processes suitable for continuous manufacturing. All sanitary design concepts must be thoroughly reviewed to ensure equipment cleanability and minimize water consumption. Multiple sources provide details on sanitary options (22–24). In summary, a laboratory evaluation to determine critical cleaning parameters combined with a review of equipment design concepts should help improve speed of the cleaning procedure, reduce waste generated during cleaning execution, minimize cleaning agent and water consumption, and reduce human interventions.

	In setting acceptable residue limits within continuous manufacturing, two situations should be considered (25). In the first situation, the manufacturing process for Product A is interrupted (a scheduled or unscheduled stoppage) due to a minor maintenance event or light cleaning such as vacuuming or dry wiping the work surfaces; when complete, the manufacturing line is back up and running. In this situation, there is no concern of residual carry-over of the active ingredient because only one product is being manufactured; therefore, cleaning validation is not applicable. If a product impurity, cleaning agent, equipment lubricant, or similar processing aid was introduced, however, a cleaning procedure should be performed to remove those residues to safe levels prior to resuming production. This type of interruption will fall into the second situation.

	In the second situation, Product A production has stopped and change-out is occurring to begin continuous manufacturing of Product B. In this scenario, there is a need to perform an in-depth or heavy cleaning to make sure that active components from Product A do not affect the quality of Product B, as well as ensuring that the non-active components (any cleaning agents used, and microbial residues remaining on the surface) are within acceptable levels. Similar to traditional batch processing, possible pharmacological and toxicity effects of Product A, as well as residue impacting the stability of Product B, need to be considered.

	In establishing residue limits, safe concentration of the target residue in the subsequent product (i.e.,  Product B), referred to as Limit 1 (L1), needs to be understood. The safe dose (L0) or a scientifically derived health-based limit (i.e., an acceptable daily exposure [ADE] or permitted daily exposure [PDE]) of the residue (i.e., of Product A) is calculated using 

, and L1 is calculated using Equation 2 (26–27).

 L0 = Toxicity Assessment x Body Weight x Minimum Daily Dose of Product A

 L1 = L0 / Maximum Daily Dose of Next Product

, multiplied by the batch size of the subsequent product (which, as previously discussed, must be defined in continuous manufacturing by the manufacturer), provides a maximum allowable carry-over value (MAC or MACO value, also known as L2).

	From the L2 or MAC value, the limit per surface area can be calculated by dividing by the shared surface area of the equipment train.

	This calculation assumes a uniform distribution of the residue and is considered a conservative approach to setting limits per surface area. Most cleaning validation swab sampling plans would include sampling the hardest-to-clean locations based on a risk assessment.

	Stratified and non-uniform residue limits

	The concept of uniform distribution is an ideal scenario and is not always true (28). This assumption, therefore, may lead to failing results on select equipment or sampling locations.

	Stratified residue distribution on equipment splits the calculated L2 or MAC residue among the different pieces of equipment within the manufacturing process based on a risk assessment. It is important to document the risk assessment used to determine the stratification because it will be reviewed by auditors. The stratification scheme can also vary based on the target residue. For example, microbial limits may be set lower for the wet granulation and coating steps because there is a greater risk for microbial proliferation during these steps.

	Non-uniform residue contamination in the product means that the residue on the surface of the equipment concentrates into the first units of production and is then reduced in subsequent units. In a tablet press, for example, the residue from the previous product will be transferred to the first tablets (or round of tablets) pressed at a higher concentration than the second and third tablets pressed. In a filling line, as another example, the residue from the previous product will be transferred to the first vials (or series of vials) filled at a higher concentration than the second and third vials filled.

 shows a calculation for setting limits in a non-uniform contamination example. The limit (L0) of Residue A in Product B has been calculated (or defaulted) to be 10 ppm or approximately 10 μg/mL. The total surface area of the filling equipment and piping is 10,000 cm

, and the limit per surface area of Residue A has been predetermined to be 1 μg/cm

. Product B is filled in 10 mL vials. If the contamination of 1 μg/cm

 of the total shared surface area of 10,000 cm

 were concentrated into the first vial filled of 10 mL, then the residue level would be 1000 μg/mL or 1000 ppm, which is well above the 10 ppm limit. The first 10 or 100 vials should be discarded because the residue in the vials would be less than 100 or 10 ppm, respectfully.

	The understanding of non-uniform and stratified sampling is important for setting residue limits within a continuous manufacturing process because a small quantity of product may migrate through a manufacturing process, concentrating residue from one piece of equipment to another and defaulting to a traditional uniform residue limit, which can adversely impact the quality of the product and potentially impact the health of the patient.

	In the lifecycle approach to cleaning validation, Stage 2, the performance qualification stage, is a readiness check to ensure the cleaning process is able to be validated. This stage will involve checking that suppliers have been approved, analytical methods have been validated, personnel performing the cleaning have been trained, standard operating procedures and validation documents are ready to be performed or executed, and process equipment and utilities have been successfully qualified and ready for use (29).

	If the critical process parameters and critical quality attributes were well characterized during the cleaning design stage, then the performance qualification (Stage 2) should be performed using normal operating parameters.

	The third stage of the lifecycle model is continued process verification. Implementation of process controls such as change control, preventative maintenance, and corrective and preventative action systems ensure a validated state. The cleaning process continuously operates in a state of control. A periodic review of the cleaning program is also crucial to ensure that the cleaning process remains in control and is flexible to change when required. This periodic review of the cleaning validation program should be reviewed in a similar manner as the yearly product quality review. A review of product cleaning is generally part of the yearly product quality review, but it is generally not thorough enough and often doesn’t review similarities and difference between multiple products manufactured in the same equipment.

	Items that should be reviewed include change control data, monitoring data, deviations, corrective and preventive actions, maintenance, quality records, and retraining events.

	If the review shows control and consistency, then summarize the findings and conclude that the cleaning program is operating in a state of control (i.e., a validated state). The review can also identify high-risk or high-waste areas that need to be improved. A corrective action plan or lean manufacturing event can be used to develop a plan to correct this deficiency.

	Depending on the type of change being proposed, it may be necessary to go back to the design stage to determine the impact of the change to the cleaning process. For this reason, validated cleaning procedures must be included in the change control management system. This ensures that any proposed changes are evaluated fully for their impact on the validated state of the procedure.

 is not an all-inclusive list of possible changes but helps provide an idea of the type of changes and their potential impact (29). The impact of the change to the cleaning process may have already been assessed during the design stage; otherwise additional testing within the design stage may be warranted to mitigate risk prior to implementation of the change. This action is depicted in 

 as arrows from Stage 3 to Stage 1 or 2 as a result of the change (29). 

	Managing change is an important process because the goal of the design stage and continuous monitoring stage of the cleaning validation lifecycle approach for continuous manufacturing production facilities is to operate within a state of control and to drive out waste to improve efficiency and maintain flexibility of the cleaning process.

	Continuous manufacturing focuses on streamlining production while minimizing the process footprint and waste from non-value activities. To ensure cGMP compliance, cleaning is required on some frequency between batches of the same product and when switching between different products. Developing a cleaning validation program using the process lifecycle approach provides a firm understanding of the critical cleaning process parameters as well as the critical quality attributes that need to be monitored.

	The equipment used for continuous manufacturing may vary based on the type of product manufactured as well as the method of cleaning. It is important to evaluate cleaning during the drafting of the user requirement specifications and functional requirement specifications of the production equipment. The development and proof of concept of a laboratory-scale cleaning model during the design phase of the cleaning process and equipment fabrication will aid in the validation of cleaning processes. This model will also help in the ability to conduct additional investigations to support process and cleaning changes to reduce waste. The use of health-based limits, such as an ADE or PDE value, for any route of administration, along with a rationalized use of non-uniform and stratified residue limits, allows for setting practical, achievable, and justified acceptance limits.

 Title 21, Part 211.67 "Equipment Cleaning and Maintenance" (Washington, DC).	

Guide to Inspections Validation of Cleaning Processes

		Pharmaceutical Inspection Convention/Pharmaceutical Inspection Co-Operation Scheme (PIC/S). 

Validation Master Plan Installation And Operational Qualification Non-Sterile Process Validation Cleaning Validation

Cleaning Validation Guidelines (Guide-0028)

		S. Chatterjee, “FDA Perspective on Continuous Manufacturing,” Presentation at IFPAC Annual Meeting (Baltimore, MD, January 2012).	

Guidance for Industry, Q8 Pharmaceutical Development

Guidance for Industry,Q9 Quality Risk Management

Guidance for Industry,Q10 Quality Systems Approach to Pharmaceutical cGMP Regulations 

Articles

Troubleshooting

A lifecycle approach can be used to develop GMP-compliant cleaning procedures for continuous manufacturing of solid-dosage pharmaceuticals.

Cleaning Validation in Continuous Manufacturing

	The industrial Internet of things (IIoT)-the network of equipment connected by sensors communicating data through the Internet-promises benefits such as the ability to run and maintain equipment more efficiently. IIoT platforms make a manufacturing plant part of a broad “digital ecosystem,” which can include the infrastructure of devices, hardware, and software, as well as the expertise and support from the infrastructure provider, equipment manufacturers, and third-party application or analytical tool suppliers.

	“The key to an effective IIoT ecosystem is to have three things,” said Andrew Hird, vice-president and general manager of Honeywell Process Solutions’ Digital Transformation business, in a press release (1). “First, you need to have secure access to the data being collected; secondly, the capability to analyze that data; and finally, you need domain knowledge to understand how to deploy information to benefit the operation.”


	The Honeywell INspire program harnesses this domain knowledge by bringing together a community of participants-including customers, equipment vendors, process licensors, and Honeywell experts-to jointly develop solutions for leveraging the IIoT to help customers minimize unplanned shutdowns, maximize output, minimize safety risk, and optimize supply chain strategies, says the company (1). On October 26, 2016, Honeywell announced that Flowserve, which provides domain expertise in flow control solutions (e.g., pumps, valves, seals) will be part of this collaboration to provide IIoT solutions. Flowserve instrumentation and services are already co-located with Honeywell automation and controls technologies at many manufacturing locations.

	“Companies have been trying to use big data for improving equipment reliability, but previous tools were limited. This new platform takes data collection and collaboration to the next level,” says Ben Blanchette, with Honeywell Process Solutions’ Digital Transformation business. “All the equipment reliability data are in one place for everyone-at the site and across the enterprise-to view and use to make decisions together. For example, if the operations group wants to increase throughput, what impact will this have on maintenance?” 

	In pharma manufacturing, there is a need to distinguish between the data used for analytics (e.g., to optimize a process or track equipment health) and the data being used to run the process (which is collected through the distributed control system [DCS] or manufacturing execution system and becomes part of the electronic batch record), explains Torsten Winkler, leader at the Honeywell Process Solutions’ Center of Excellence Life Sciences EMEA. Big data used for process analytics and improving equipment reliability does not need to be part of the batch record, so the IIoT infrastructure (instead of the DCS) can be used to securely collect and aggregate these data for analysis.

Emerson recently expanded its Plantweb digital ecosystem, which it describes as “a scalable portfolio of standards-based hardware, software, intelligent devices, and services for securely implementing the Industrial Internet of Things” (2). Plantweb includes process control, safety, and asset management systems. Emerson added sensing technologies for corrosion detection and hazardous gas leak detection, among others, in a portfolio called Pervasive Sensing. Another new feature is “Secure First Mile,” which refers to system architecture that connects secure data from the plant floor to operational performance applications in the cloud. Emerson says that this ecosystem, combined with the company’s Operational Certainty program, which provides a business case for IIoT investment, can help companies begin the move to using new digital technology.

	“There is a lot of talk about the promise of Industrial IoT,” said Peter Zornio, chief strategic officer, Emerson Automation Solutions, in a press release. “Emerson is putting the ROI [return on investment] in IoT.”

	Another part of Plantweb is the new AMS ARES Platform, which collects data from multiple sensing technologies to show the real-time “health status” of plant assets. The company says it is “designed to prioritize and deliver asset and device health data, enabling maintenance decisions that increase operational and asset availability” (3).

	The increase in available data resulting from interconnected systems could make finding specific data more difficult, but “dashboards” are often used to display just the data that are pertinent to the person viewing it. The AMS ARES Platform delivers information, based on responsibility, using streamlined dashboards. Embedded tools prioritize actions based on criticality, which allows “users to determine quickly the urgency of an issue, facilitating faster workflows to increase plant performance” (3).  

	The platform uses OPC UA open architecture, which will allow it to more easily evolve as the industry changes, according to Emerson. The platform can connect with human-machine interfaces (HMI) and with the plant historian. Connecting the platform to business intelligence and reporting tools makes asset health part of the business dashboard.

		Honeywell, “Honeywell and Flowserve Collaborate on Industrial Internet of Things (IIoT) Solutions,” Press Release, Oct. 26, 2016.	

		Emerson, “Emerson launches new Plantweb digital ecosystem,” Press Release, Oct. 24, 2016.  

https://www.emerson.com/en-us/news/automation/1610-plantweb

		Emerson, “Emerson Launches Holistic Asset Monitoring Platform to Improve Reliability And Plant Availability,” Press Release, Oct. 25, 2016. 

http://www.emerson.com/resource/blob/176678/a320c8df312ac5ee1dae1c58469dad67/ams-ares-platform-data.pdf

Production Trends

New software platforms aim to help manufacturers implement connected technology and use data efficiently.

Equipment and Processing Report-11-16-2016

Cleaning Validation in Continuous Manufacturing

Using the Internet of Things to Manage Manufacturing Equipment