Questions and Answers about Cleaning Validation - Part 2


GMP/GDP – On Demand Online Training

You can book the desired online training from our extensive database at any time. Click below for more information.


Stay informed with the GMP Newsletters from ECA

The ECA offers various free of charge GMP newsletters  for which you can subscribe to according to your needs.

In March 2022, ECA Academy conducted an online training on cleaning validation - during which participants had many questions. Due to the large number of questions answered, we had split this post. Below you can find the second part of the questions answered by the speaker Robert G. Schwarz from FH Campus in Vienna.

All answers reflect the opinion of the speaker and are based on his experience.

41. Would a bracketing approach be appropriate for a biotech process?

Generally spoken - yes, it could be an option.

42. How to handle with cleaning validation for shared facilities (with pharmaceutical products and biological active additives)?

First evaluation if shared facility is possible. Shared facility is not necessarily shared equipment, especially when we have closed processes and/or CIP cleaning it is less critical. Shared equipment is sometimes better than dedicated equipment cleaned in shared COP-washer - segregation issue in the washer itself - changeover cleaning with empty chamber required. For Multi-purpose equipment the cleaning process development and the lab scale studies is more difficult than the cleaning validation itself. A "hard" bracketing could be performed including product and equipment bracketing with a profound and traceable product-equipment-matrix. There the cleaning validation effort could be shrunk down significantly - one worst-case product with one worst case equipment covers "everything else". Disadvantage is the lack of an optimized cleaning process for the different product equipment combinations. If this is the desired, way, less bracketing is possible because bracketing over different cleaning processes is only accepted in very rare cases by authorities. Additionally, if all products are macromolecules/ proteins I'd highly recommend a degradation-strategy over using HBELs.

43. Do we need to have full CFU count with identified microbes for non-steriles or is a general screen acceptable?

The cfu count is needed, usually identification is not necessary because objectionable microorganism are only relevant for product bioburden testing. In addition, I would recommend also performing endotoxin testing.

44. How do you set bioburden limits - what rationale is used?

Usually, the basis is the allowed bioburden in the product. It is calculated back to the surface from the absolute amount of microorganism allowed in the smallest batch. For rinse sample the volume of the finale rinse step and the sample volume needs to be taken into consideration. Additionally, if an open storage of equipment is possible a cross-check of the bioburden limit limits for the Cleanroom Class (CRC) of the storage area for surfaces are needed.

45. For excipient manufacturers it is difficult to judge what means "doesn't harm the patient" as the application is mostly not exactly known. Do you have any recommendation?

Limits are the requirements of customers, also internal, that use the excipients for final formulation of medicinal products. Those are usually defined in a quality agreement resulting in incoming goods specifications.

46. EMA HBEL guideline instructs to calculate PDE with NOAEL and NOAEL is chosen based on available human and animal data. Why prefer NOEL over NOAEL?

I suggest using NOEL, because a desired effect of product 1 with patient A could be an adverse effect for patient B using product 2.

47. Regarding automatic cleaning: We want to implement a washbox with our cleaning validation - when we adapt the programs later, do we have to repeat the CV with all 3 full runs?

Depending on the extent of the adaption it can be risk assessment which identifies the adaption with minor impact and maybe only one verification run (to support this hypothesis with data) is needed or it could be a major impact where full revalidation is needed-

48. Where is reliable toxicological study data available? Question also applicable for pharmacological data and study data?

If you are a CMO you get the data from the license holder of the product you produce. It is part of clinical development to generate those data in the clinical phases for manufacturers of pharmaceutical products. If you produce generics, you may get data from the patent holder.

To my knowledge no freely available literature for HBEL are available. Selection of suppliers for "common" small molecule APIs:

49. Typically, who should determine NOAEL, F1-F5 and calculate PDE?

According to EMA "shared facilities Guideline" it is the task of toxicologist with requirements also mentioned in the Q&As to this Guideline also published by EMA:
"Health-Based Exposure Limits should be determined by a person who has adequate expertise and experience in toxicology/pharmacology, familiarity with pharmaceuticals as well as experience in the determination of health-based exposure limits such as Occupational Exposure Levels (OEL) or Permitted Daily Exposure (PDE).
Where experts are contracted to provide the HBEL, contractual agreements in compliance with Chapter 7 requirements should be in place prior to work being conducted. It is not considered acceptable for manufacturers to 'purchase' HBEL assessments without recording an assessment of the suitability of the provider (including the specific technical expert) as a qualified contractor."

50. Is there any possibility to calculate MACO only based on NOEL without therapeutic dosis?

The only option I see would be that you assume that the therapeutic dose is the volume of the whole batch which usually leads to a very low MACO. Sometimes even under the LOQ of the analytical method for some acceptance limits.

51. Volume rinsing solution - is this equal to sample volume?

No, the rinsing solution volume is the whole volume used for the rinsing step where the rinse sample is drawn. For final rinse it is determined by the cleaning process for solvent rinse it should be a minimum volume to cover the whole product contact surface.

52. Is concurrent validation acceptable?

Yes, based on a profound risk assessment evaluating the specific circumstances it is acceptable.

53. DEHT must be validated during initial validation but can be eliminated during cleaning monitoring? Is this true? Is it required to perform CEHT studies periodically?

Usually, DEHT it is not part of the cleaning monitoring every time. For CEHT I would recommend performing a sampling after a maintenance shutdown which should be part of the preventive maintenance program. 

54. Does equipment sterilization fall under the cleaning validation scope if it is a last step of "cleaning"?

No, it only impacts the definition of CEHT.

55. If after a sanitization process (CIP Chromatographic column) one/several parameters are OOL, is it mandatory to register it as deviation or could we execute a re-cleaning process until an acceptable value?

No, this would violate the validation of the cleaning process and it could be assumed that the process is not robust. Cleaning until the acceptance limits are met is comparable to testing into compliance.

56. Which studies of NOAEL determination are preferred to use (human/dogs/ rats) while calculating PDE? If they all have different duration and values of NOAEL?

Although I'm not a Toxicologist either the value for the species the medicinal product is applied is used or the "nearest" species, i.e., species with the most comparable physiology or, what I usually recommend, if I get the chance to discuss the tox assessment with its author, to calculate all the different PDEs and use the strictest value. Sometimes even it is necessary to re-evaluate HBELs based on data collected during clinical trial phase IV (pharmacovigilance) when the product is already on the market. This reflects the life cycle approach of every validation strategy.

57. Cleaning monitoring when you're not manufacturing using worst-case product. Do you need a risk assessment for that?

Usually, it is not necessary to use the worst-case product for cleaning monitoring. If there is no cleaning process being performed in the period where a cleaning monitoring needs to be performed, I'd recommend describing this scenario in the MVP or the cleaning monitoring SOP and perform the cleaning monitoring the first time the cleaning process is performed. Some colleagues even do this three times if the overdue of the monitoring frequency is significantly.

58. Macromolecules/peptides: but could the remnants may serve as 'food' source for microorganisms?

Yes, but this is covered via CHT.

59. DEHT: is this some sort of plastic type? (e.g., dioctyl terephthalate - non-phthalate plasticizer)?

No, it is the abbreviation for dirty equipment hold time, which is often used synonymously for DHT (dirty hold time)

60. When no DHT is validated, is it ok to use 24h as a standard?

No, DHT needs to be addressed in the cleaning validation according to Annex 15 of EU-GMP Guideline. You can't justify 24h with data if you'd do so.

61. Do you recommend cleaning the equipment before the process even if it is within the holding time of the previous process cleaning?

Usually this is not necessary but depends on the storage conditions. I'd rather optimize those than perform a re-cleaning

62. When equipment is dedicated, is it ok to stop testing after 3 conform CV runs (results and cleaning cycle ok), and the report is not finalized (e.g., very tight production schedule)?

If the three runs are defined in the cleaning validation risk assessment/ protocol as sufficient and only the report is missing from a formal point of view every produced batch needs to quarantined until the final report is approved, but the routine production can start.

63. Is it possible to perform a cleaning validation and then take process control samples instead of validating long campaigns?

You can always perform cleaning verification instead of cleaning validation, but if a sample is OOL you need to open a deviation and, in worst case, end the campaign immodestly. Additionally, you then need to sample to the full extent after every cleaning process.

64. Do you have to perform inactivation studies with each manufactured product?

For macromolecules it should be performed for each substance where you don't follow the HBEL/PDE-strategy.

65. The rinse sample matrix in biotechnological processes should be taken in WFI? can it be taken in equilibration buffer? Can other buffers such as NaCl 0.9% be used?

This is depending on the parameter to be measured and the analytical method if the matrix components interfere with the analysis. Generally speaking, and from a regulatory perspective this is possible.

66. How do you define the TOC limits if during cleaning your resin releases the organic material coming from its own ligands?

You need to assume that the organic carbon's source is the most toxic substance in this production step and use the adequate limit. If this is not practically then another parameter, usually a component specific analytical method, needs to be established.

67. What can you recommend establishing an acceptance limit for rinse samples from Chromatography columns if that equipment isn't in contact with PW? Is it possible to prove cleaning just from API/excipients not from solvents/eluent?

This is no issue, because you can perform the rinsing sample with the eluation buffer if it is covered in the analytical method validation.

68. What can be simplified for CV (dedicated equipment)?

No bracketing approach for products, DHT and CHT needed to be justified, no lab scale studies for cleanability. Also, maybe unspecific methods are needed. Limit is usually higher. Starting with a cleaning verification at the beginning to get data and stop after enough samples to show the process is validated is easier.

69. Drug product manufacturing: Any experiences with HA (especially LATAM (Brazil, Peru, ...) countries) if we have several bio products on the same equipment? And some products are monoclonal antibodies, other protein molecules, mRNA, cleaning validation is in place?

Have a good risk assessment for DHT and CHT (incl. different sampling points for CHT), favor degradation-strategy HBEL-concept. Evaluate, if degradation products are potentially sensitizing and have a rational for that. Proper scientific risk assessment. According to my experience ANVISA (HA of Brazil) is at a lot of topics comparable stringent to US-FDA, especially when it comes to "numbers" (e.g., ANVISA is the only HA in the world that specifies minimum recovery of contact plates with 50%). Justify why your recovery rates are acceptable for your purpose.

70. For multi-purpose large molecules equipment: Is a worstcase concept still sufficient or should we have TOC method (with recovery for each product)?

I'd consider a well-documented worst-case strategy with a riskbased approach is still state of the art. You have the TOC-limit specific for every product anyway because it is based on the TOC-content of your API-molecule.

71. Is just the worst-case enough, or should we have to sample at least 1x after each substance?

Depending on the reliability of the small-scale lab study for cleanability it could be necessary to prove that lab scale is representative for production scale that a verification run with one or more other products than the worst-case is reasonable. It is at least recommended if small-scale lab study doesn't identify one clear worst-case product (e.g., it is 50% harder to clean than the second product).

72. When introducing a new product, is sampling at least 1x (or 3x) after manufacturing the new product sufficient and if everything is OK, can it go to the matrix and do only the worst-case?

First it needs to be evaluated if the product could be produced on a multi-purpose equipment train based on its toxicity and cleanability. For cleanability lab scale studies are recommended. Additionally, it needs to be considered that every other product produced in between needs to be quarantined until the final CV report is approved. Only after that it could go to the matrix being included in the OCV (cleaning monitoring).

73. What tests are recommended to be performed in the manufacture of biological products?

Usually if you clean with inorganic cleaning agents like alkalis and acids unspecific methods like TOC are sufficient for product residues especially when you could demonstrate full inactivation of active components. Cleaning residues are detected via conductivity or pH. Bioburden and Endotoxins especially important in biotech production.

74. Regarding sampling: Any guideline on how to consolidate different plastic surface materials?

No Guideline available for such a bracketing approach, Strategy should be the minimization of variety of plastic materials that are in direct product contact. Generally, a worst-case assumption is performed via different surface roughness, if no specific extractables need to be considered. According to my experience chloro- or Bromo butyl-rubber materials without any coating are hardest to clean because of their porous surface structure. Besides that favor only PTFE (Teflon), Viton and silicone as materials for hoses or gaskets that are not single use.

75. If equipment is really hard to reach visual cleanliness but all other acceptance criteria are ok, what could we do?

This is usually the case with CIP-systems and especially for piping. This is covered in a risk assessment. Some companies perform endoscopic checks during preventive maintenance during production shutdowns.

76. There are also hold times in between cleaning steps. Sometimes they are very long. How do you validate that? (Sometimes it just hold in alkaline solution or something similar)

If the equipment is covered in a cleaning solution, even if it is just PW or WFI this is to be considered as a cleaning step but with defined duration. It also needs to be addressed in the CV risk assessment if a worst-case cleaning cycle is, if this process step time varies, with the shortest (i.e., worst cleaning capability) or longest (potential impact on equipment and microbial proliferation) process step time.

77. Is it possible to validate the CEHT only 2 times for 3 CV runs, because parts of the equipment train are needed for another production (tight production schedule)?

Not based on the argumentation "tight production schedule" and I wouldn't recommend arguing this in a risk assessment.

78. Regarding campaigns: Do we need 3 CV runs with the full campaign length or is one full campaign length run and 2 shorter CV runs sufficient?

No, if the parameters vary they are no three consecutive runs anymore. Could be argued in a risk-assessment. Based on my experience I wouldn't recommend that.

79. When you decide not to manufacture the product anymore. do you have to perform cleaning monitoring with last manufactured product batch?

I would recommend that. It goes align with concepts of calibrated instruments and should be part of a change control for GMP-compliant retirement.


About the Author
Robert G. Schwarz is a lecturer at the FH Campus in Vienna. He has 20 years of practical experience in aseptic processing, contamination control and cleanroom technology.

Go back