The words of George Santayana — “Those who do not remember the past are condemned to repeat it” — ring especially true for companies regulated under good manufacturing practices (GMPs). Learning from and reacting to lessons from past inspections (both your own and those of other companies) is one of the best ways to prepare for future inspections. Regular review and close study of 483 notices issued during inspections can be an efficient and accessible means of identifying and absorbing those lessons.   483 “101”: An IntroductIon   A 483 is officially an FDA Notice of Deficiency. It is written by a member of an inspection team when he or she observes a situation, practice, or activity that in his or her opinion does not comply with current good manufacturing practice (CGMP) or other regulations in Title 21 of the US Code of Federal Regulations. The nickname “483” derives from the US government form (Form FDA 483) on which investigators or their team members record such observations. Numerically, Form 483 comes after Form 482, which is a Notice of Inspection presented by the lead investigator when a team arrives, usually unannounced for US inspections, to begin their work. A 483 is not a notice of violation; it is an observation (an opinion). A 483 observation is a notification that your company’s practices do not meet an inspection team’s expectations for CGMP (or other regulatory) compliance. Such a notice should be taken seriously. Several types of situations can result in an observation of deficiency, which is important to keep in mind when you are reviewing 483s. First and foremost, a 483 is written in response to an apparent CGMP compliance deficiency. But biopharmaceutical and biologics manufacturers must comply with many regulations besides CGMPs, and a 483 can be written for apparent noncompliance with those other regulations as well. Finally, there is also what is often called a commitment deficiency. If a company is not complying with one of its own written policies or procedures, that can also result in a 483 observation. While reviewing 483s, if you find yourself thinking, “Where in the CGMPs does it say you have to do that?” then you’re probably looking at a commitment violation.   Reviewing 483s   Options for obtaining 483s to review vary widely in cost, complexity, and time spent. The easiest is to simply “watch this space”: Several times a year, I plan to provide topical lists of 483s in a series of articles that will follow this introduction. The simplest way to keep up regularly is to subscribe to a paid-subscription newsletter. One example is GMP Trends, which covers a wide range of 483s. Another example is my own newsletter, BioQuality, which focuses exclusively on those issued during biopharmaceutical and biologics company inspections.

An easy, free, but limited option is to read 483s as they are posted in the FDA Office of Regulatory Affairs (ORA) Freedom of Information Act (FOIA) Reading Room on the agency’s website: www.fda.gov/AboutFDA/CentersOffices/ORA/ORAElectronicReadingRoom/default.htm. The main problem with this strategy is the small number of 483s that are posted, typically only those requested with exceptional frequency or considered most important by the FDA. You can request copies of past 483s directly from the FDA under the Freedom of Information Act. But the process can be time consuming, and you may be charged a fee. If you decide to go this route, here is a place to start: www.fda.gov/RegulatoryInformation/FOI/HowtoMakeaFOIARequest/default.htm. Finally, you can purchase 483s from a company that specializes in providing Freedom of Information searches and documents. This option is especially valuable if you are seeking information about inspections of specific companies— for example, your competitors. My favorite search firm is FOI Services (www.foiservices.com). In my opinion, the top-level lesson to be learned from 483 review is an awareness of where the FDA is currently focusing its attention. Even though 483s are not typically available for several months after an inspection, they are still a very valuable resource in this regard. Warning letters, all of which are posted on the FDA website, can also be reviewed for this purpose. Unfortunately, they are typically available much later than are 483s. Few inspections have associated warning letters, whereas 483s are issued after nearly all inspections. The only exception would be for a company that appears to be in complete compliance — a rare breed, to be sure. Many specific lessons can be learned from reviewing 483s. The following discussion comes from my study of a decade of biopharmaceutical and biologics deficiency notices. The main generator of deficiency observations over the past decade has been inadequate investigation, which accounted for 20% of total 483 observations in that period. Next, unsurprisingly, come validation issues (15%), record-keeping deficiencies (11%), and equipment–facilities– operations (EFO) issues (11%). Following is a review of some typical 483 observations from these four generators of regulatory problems, quoted directly from the observation forms themselves.   Four Lessons on Investigations   Inadequate investigations have traditionally been a major cause of regulatory problems with the Center for Biologics Evaluation and research (CBER). Here are some lessons to be learned and example 483s that typify the problems companies continue to have with these ineluctable and inexorable add-ons to their already full plates. Investigations Lesson 1: Always ensure that your investigations are complete, that they uncover root causes, and that effective preventive and corrective actions are implemented and tracked for effectiveness. “Sterility failure investigations are incomplete. Extent of contamination of media lot was not determined. Retain sample of media lot was not tested for sterility. Corrective actions were not implemented until two years after occurrence.” “Company failed to adequately study causes for the acceptable quality limit (AQL) failures that occur across product lines.” “Several issues not adequately investigated to determine root cause. Instead, problems were blamed on system upgrades, although there was inadequate evidence to support this.” “Written records of investigations are deficient in that they do not include conclusions and follow-up.” “Corrective action following growth failure during media testing was limited to retraining of operators, and there was no long term corrective action or CAPA initiated.” Investigations Lesson 2: Be sure to include trending as part of your investigations program. “Initial OOS results are not always evaluated for trend s such as analyst problems and/or test-method problems.” “A formal system has not been implemented for tracking, trending, and initiating corrective and preventative actions for manufacturing incidents.” Investigations Lesson 3: Never retest without an investigation. “The company’s out-of-specification procedures are deficient. Procedures indicate that release can be performed on retesting without an investigation. If the result is OOS, and a lab investigation does not show an assignable cause, then a retest is performed. If the retest fails, then the test fails and the original results are reported. But if the retest passes, then the original result is deemed invalid and a new test is performed. Additionally, no manufacturing investigation is performed.” “The out-of-specification test result investigation procedure is deficient in that it allows following laboratory investigation without assignable cause. Two retests are executed. If they pass, the average of the original OOS result and two retests is reported. Following a laboratory investigation without an assignable cause, if the original analyst obtains two failing results and two additional analysts obtain passing results, then the average of the two passing results is reported. There is no scientific justification for invalidating the original analyst’s result. It is assumed, without evidence, that the original analyst is improperly trained.” “Noncompliant OOS procedure: SOP instructs to disregard the initial OOS after three passing results. SOP does not assure that reported test results are representative of batch. SOP lacks detail to assure full investigation of initial OOS result, including requirement for analyst interview. SOP does not include requirement to analyze all test results to assure statistically valid result.” “No documentation that laboratory investigations were conducted to determine reason for out-of-specification result before retests were conducted.” Investigations Lesson 4: Always extend your investigations to other batches and (if appropriate) other products. “Investigations did not extend to other batches manufactured during same time period as batch failures.” “Investigations did not extend to other complaints received during same time period.” “Company failed to extend investigations to include product currently in distribution.” “Investigations into unexplained discrepancies did not always extend to other lots/products that may have been associated with the discrepancy. Specifically, the company failed to quarantine/assess all product or process intermediates affected by atypical events pending completion of investigations as required by company’s policies and procedures.” “Discrepancy and failure investigations do not extend to other drug products that may have been associated with the specific failure or discrepancy.”   Five Lessons About Validation   Inadequate validation is an expressway to regulatory problems for many companies. Validation Lesson 1: Remember that when a process or procedure changes, validation or revalidation is often expected. “Changes were made to equipment cleaning procedures subsequent to validation, but equipment cleaning procedures were not revalidated to establish the impact of the changes on the processes.” “The cleaning validation for product-contact equipment for drug products and their respective drug substances is inadequate. Process changes were made to the routine cleaning process for the reactor used in the manufacture of the drug substance. These changes have not been validated: The volume was decreased approximately fourfold. Previously the pressure was not controlled, now it is controlled. A filter was added to the outlet valve on the bottom of the reactor used for circulation. The time of water-for-injection rinses was increased, and the volume of water for injection decreased for rinses.” Validation Lesson 2: Be certain that you document all changes and modifications that occur during validation. “Process validation study was flawed. No documentation of the times samples was collected to ensure uniformity across the batch. Process parameters were changed during validation, but master batch record was not changed to reflect these modifications, and the precise nature and times of these changes were not included in the validation report.” “Inadequate validation report: Deviations were not adequately documented.” “Validation report includes acceptance criteria that were not included in the validation protocol.” Validation Lesson 3: When planning validations, always take the worst case into account. “The company is unable to demonstrate that worst-case cleaning validation was executed.” Validation Lesson 4: Make sure your validations are comprehensive and conducted under dynamic conditions.

“Validation/qualification of classified aseptic areas is inadequate. Studies failed to assess/evaluate nonviable particulates and air quality conditions simultaneously at different locations within the aseptic area being qualified. Studies were conducted using only one nonviable particulate probe, which collected and tested samples at different locations at [redacted] intervals. In addition, reports indicate studies were conducted under dynamic conditions but do not specify the conditions at which the studies were done (e.g., line speed, number of operators working in the area at the same time). Additionally, smoke studies were found inadequate in that not all employee interventions are simulated and/or videotaped during the studies. No evidence that smoke study was conducted with the maximum number of filling-line operators allowed to be working at the same time, which is indicated by a sign in the filling room. Behavior of air flow is not simulated/evaluated under other dynamic conditions, such as loading.” “Smoke study was incomplete: did not include dynamic simulation of operator postfilling activities.” Validation Lesson 5: Don’t forget to have your protocols and reports reviewed by your company’s Quality unit. “QA failed to adequately review validation protocols and reports, resulting in batches released that were manufactured with a process that showed significant variability and was not adequately validated.”   Three Lessons on Documentation   Documentation and recordkeeping deficiencies always loom large in causing compliance problems. Here are some typical ways that companies come to grief. Documentation Lesson 1: When things change, make sure your documents change with them. “The master batch record was not revised following process revision and revalidation to incorporate the modified steps.” “Master and batch production records were not revised after an incorrect process pa rameter was found.” Documentation Lesson 2: Complete records are a must. “Not all critical process parameters are recorded in the master and batch production records.” “Laboratory records are deficient in that they do not include a complete record of all data obtained during testing.” “Batch production and control records do not include the identification of the persons performing each significant step in the operation for each batch of drug product produced.” Documentation Lesson 3: Make sure your documentation systems are accurate and controlled. “Paper-based sample tracking system is not maintained accurately (e.g., testing not done was logged in as completed).” “Assay worksheets are not controlled.” “Company failed to verify and ensure accuracy of information sent to FDA in license application. Discrepancies were noted during review of QC raw data, test records, and stored test results; and batch records from process validation lots. Following are examples of discrepancies and problems encountered: Tests were performed but submitted as not performed. Raw data, test records, and stored test results were missing. Calculation errors, transcription errors, and rounding errors were found. Failures to perform required repeat testing after test control failures occurred. Batch size and yield discrepancies were noted. Information was not available to support all batch record data submitted.”   Four EFO Lessons   Equipment, facilities, and operations leave plenty of room for compliance problems and here are a few for you to ponder. EFO Lesson 1: Be sure to have a written and followed calibration and maintenance program in place. “Routine calibration of automatic, mechanical, and electronic equipment is not performed according to a written program designed to assure proper performance.” “Procedure used to calibrate analytical and microbalances is inadequate: does not assess performance under actual conditions of use; fails to show measurement uncertainty. Repeatability is determined using only one standard at the top of the range.” “Routine calibration and checking are not performed for automatic and electronic equipment.” “Calibration of instruments and apparatus is not done at suitable intervals with an established written program and provisions for remedial action in the event that accuracy and/or precision limits are not met.” EFO Lesson 2: Always have adequate facilities. “Building lacks adequate space for the orderly placement of equipment and materials to prevent mix-ups between different components, in-process materials, and drug products — and to prevent contamination.” EFO Lesson 3: Keep your vendor qualification program in order. “Company continued to use packaging from a vendor that had been disqualified during a previous audit.” EFO Lesson 4: Have an adequate program to deal with inevitable deviations. “The procedure documentation of process deviations does not contain a procedure for change control or its assessment. The following examples demonstrate a lack of an adequate change control system: Production made a permanent change to a new reactor mixing speed for drug substance, but the change control does not document the evaluation for whether this change would affect product quality. Engineering modified wiring connections on the same cooling relay that services the cleanroom complex. This change caused the relative humidity in the cleanroom complex to exceed the established specification of 80% relative humidity. Therefore, Production and Quality Control approved a relative humidity specification change. A change control was not executed to scientifically evaluate how this specification change would affect operations and quality of the cleanroom complex used to fill product. Two planned deviations allowed reduction in batch size, and a change control was not initiated to determine the potential impact of this change and whether validation was necessary.”   More to Come   I hope that you find this 483 primer to be useful and informative. Please contact me with your concerns and questions, and I will cover them in a future installment of this series.

About the Author

Author Details Tom Pritchett, PhD, is cofounder and publisher of the BioQuality industry newsletter, PO Box 7087, Citrus Heights, CA 95621; 1-916-729-0109, fax 1-916-729-2602; [email protected], www.bioquality.biz. He is an active industry consultant and trainer focusing on CGMP and technical issues and a member of BPI’s Editorial Advisory Board.