Markus Schwind, PhD, is HSE manager at Sanofi-Aventis Deutschland GmbH, HSE Germany, Industriepark Hoechst, Bldg. H831, Room A.04.72.2, D-65926 Frankfurt am Main, Markus.Schwind@sanofi.com. 

Markus Schwind

Submitted: July 25, 2016.
	Accepted: August 19, 2016.

A commonly used method in cleaning validation (CV) is 0.001 of the minimal daily dose (MinDD). This method lacks a consensus definition, as well as a source document where MinDD can be obtained. In November 2014, the European Medicines Agency (EMA) published a guidance document requiring the calculation of permitted daily exposure (PDE) limits. In this study, the authors investigated the relationship between the 0.001 MinDD and the PDE values for 140 drug substances as an attempt to identify high-risk groups of products for patient safety. This comparison can serve as a method for prioritization of APIs for development of PDEs.

	Chapters 3 and 5 of the European Union’s 

EudraLex Guidelines for Medicinal Products for Human and Veterinary Use

 require a toxicological evaluation to assess and control cross-contamination risks presented by the drug products manufactured in shared production facilities (1, 2). The European Medicines Agency (EMA) regulatory guideline on the setting of permitted daily exposure (PDE) values came into effect for all new human pharmaceutical products on June 1, 2015, and for all existing human pharmaceutical products on Dec. 1, 2015 (3). Presently, pharmaceutical companies are investing significant effort into providing the requested PDE values based on toxicological and pharmacological data for all their products manufactured in multipurpose facilities.

	In the past, several other methods were used in the pharmaceutical industry for determining an acceptable level of carryover with uncertain levels of patient health protection (4). One of the most frequently used methods was 0.001 of the minimum daily dose (MinDD). The use of 0.001 MinDD was proposed as the first of a set of three criteria, each of which had to be met, for determining acceptance limits for cleaning validation (5):

		No more than 0.001 of the minDD of any product will appear in the maximum daily dose of another product.

		No more than 10 ppm of a product will appear in another product.

		No quantity of residue will be visible on the equipment after cleaning procedures are performed.

	The dose criterion makes no distinction between solid or liquid forms, or between the different routes of administration (topical, parenteral, inhalation). The core issue with the first criterion, however, is the absence of a clear and health protective definition of the term MinDD, in particular, one that resolves the operational issues in deriving the MinDD. The terminology and definitions were recently discussed (6). In the absence of an operational definition from a regulatory or consensus organization, the answer to what constitutes the MinDD has been subjectively established by each pharmaceutical company based on explicit or implicit decisions and judgments on these key issues:

		Is the minimum daily dose the smallest therapeutic daily dose administered to healthy adults, or should it encompass all potentially sensitive subpopulations and requisite dosing reductions (e.g., patients with underlying disease [such as renal or hepatic impairment], patients taking concomitant medications with potential drug interactions, pediatric patients, or pregnant or nursing women)?

		How do you deal with complex posology (dose, frequency, and method/route of administration)?					

				For drugs that are administered chronically on an intermittent basis (e.g., once a week or once a month), is it appropriate to divide the bolus dose by the number of days between doses (e.g., 7 or 30) in the calculation of the MinDD?

				For medications that are administered only once or only a small number of times in a lifetime (e.g., anesthetics or medications for reversal of neuromuscular blockade after surgery) but could potentially cross-contaminate drugs that are administered daily, can this single dose value be used, or would it be appropriate to include a factor for lifetime daily exposure in the calculation of the MinDD?

				For medications with differences in routes of administration between the initial and subsequent drug product, should differences be considered in relative bioavailability in the calculation of the MinDD, as for instance, the carryover of a parenteral drug with low oral bioavailability into a subsequent oral drug?

		How do you calculate the MinDD for a drug that is not approved or is contraindicated for a subpopulation that could potentially receive it through cross-contamination? As an example, drugs previously classified as FDA Category D or X should not be administered to pregnant women. Does it make sense to use the MinDD of nonpregnant females or healthy males (e.g., angiotensin-converting enzyme inhibitors) as the basis of MinDD in such carryover calculations?

		What confidence is there in 0.001 MinDD for a drug that may cause serious adverse effects from a single or small number of doses (e.g., anaphylaxis from a residual beta-lactam antibiotic, embryo-fetal toxicity from an antineoplastic such as thalidomide)?

	In practice, varying responses to these questions has resulted in different cleaning limits with the MinDD criterion. These questions also highlight the ambiguity of the former methodology, and how arbitrary decisions or judgments could result in profound differences in carryover limits.

	Others have recently proposed methods to establish expedited PDE values based on occupational exposure levels (OELs) and occupational exposure bands (OEBs) (7). The availability of OELs and OEBs may be helpful to prioritize these substances for PDE-setting, but are of limited validity as surrogates for PDEs given that OELs and OEBs are derived to protect healthy adult workers and focus principally on the inhalation route of exposure. Consequently, they are subject to some of the same issues as MinDD described (e.g., differences in therapeutic routes of administration and protection of sensitive sub-populations). In the present article, the authors investigated the mechanism of action (MOA) or therapeutic area of 140 drug substances (DS) and compared their PDE values with the 0.001 MinDD in effort to help identify the critical drug product types that are of highest priority with regard to expert toxicological evaluation and PDE determination. This prioritization tool can be helpful for those companies that have large portfolios of legacy APIs that need to be assessed by expert toxicologists as well as the auditors that are looking for a tool to identify patient safety critical substances.

	The study evaluated 140 drug substances manufactured by Novartis. PDE values were calculated following the EMA guideline and recently published criteria (8-11). PDEs were compared to the MinDD used in therapy divided by a factor of 1000 (0.001 MinDD). The quotient of these two values was termed R. The minimum therapeutic dose for these calculations was based on the lowest known therapeutic or the starting dose identified in the published literature. To avoid the confounding effects from adjustments necessary to account for different routes of administration (e.g., topical, parenteral, inhalation), the same route of administration was assumed for a subsequent drug in these comparisons (e.g., PDE oral in μg/day with 0.001 MinDD in μg/day used in oral treatment). For the purposes of simplification, drugs that do not have a daily dosing regimen were excluded from this evaluation.

	The comparison of the PDE value by route of administration with the respective 0.001 MinDD is shown in 

. The ratio R=PDE/0.001 MinDD for the 140 drug substances investigated was plotted in decreasing order. Red bars indicate therapeutic substances where R < 1.

	There were 9% (n=12) out of the 140 drugs evaluated in the group with R < 1, and 11% (n=16) of the group with 1 ≤ R <10; 26% (n=37) of the drugs evaluated had 10 ≤ R <50, and 53% (n=75) of substances had R≥50. Depending on a company’s drug portfolio, these percentages would be different. For example, a company that has a large oncology portfolio might have a higher percentage of substances with R <10. 

 presents all of the drug substances with R<10, which includes all of the antineoplastic medications. The drugs that really stand out in this comparison of PDEs and MinDD derived carryover limits are the antineoplastic agents and the organ transplantation anti-rejection drugs.

	This observation makes sense as most older antineoplastic agents (e.g., alkylating agents, antimetabolites, antibiotics, and topoisomerase inhibitors) interact directly with DNA or its precursors, causing indiscriminant damage to normal and malignant cells (12). These agents, with several other mode of action (MOA), are known to be able to affect fertility or embryo-fetal development (EFD). Additionally, many of them, in particular, the first- and second-generation antineoplastics, are genotoxic. Generally, antineoplastic drugs are administered to cancer patients at high doses, often at the maximum tolerated doses for patients, while effects on EFD may occur at much lower doses. Therefore, PDEs are lower than 0.001 MinDD for most anticancer substances, and use of 0.001 MinDD as a carryover limit could potentially pose a risk to patients.

Doxorubicin is an example of an antineoplastic that has a PDE 10-fold lower than 0.001 MinDD. Doxorubicin causes nucleotide base intercalation and has apparently prolonged tissue binding activities that affect its excretion. Independent of the primary mechanism of cell damage, the cellular response is frequently apoptosis. Doxorubicin was embryotoxic and teratogenic in preclinical studies. It is positive in several genotoxicity studies and is classified by the International Agency for Research on Cancer (IARC) as Group 2A (probably carcinogenic to humans) based on sufficient data from animals (13). Its lowest therapeutic dose is 20 mg/m

 weekly and was, therefore, excluded from the evaluation in 

. Doxorubicin is not suitable for oral administration as less than 5% of the drug is absorbed. Its terminal half-life is approximately 20-48 hours. Impairment of liver function results in slower excretion and, consequently, increased retention and accumulation in plasma and tissues (14). This example shows a number of factors that would lower the PDE value below the 0.001 MinDD, such as adverse effects in reproduction/development, genotoxicity, carcinogenicity, nondaily administration, potential for accumulation, and sensitive subpopulation of patients (e.g., those with liver damage or pregnant women). Clearly, there could be a significant risk to patients from carryover if the MinDD, which is based on once per week dosing, were used as the basis for the calculation.

	Another group of substances that have R<10 are sex hormone modulators indicated as antineoplastics. Examples in this group are estrogen receptor antagonists and estrogen receptor modulators, which can cause adverse effects on reproduction and EFD. In this group, estrogen receptor agonists, in addition to the EFD effects, were found to also have carcinogenic potential. Aromatase and androgen receptor inhibitors, as well as antiandrogens, are also consistently found in the group where R<10. For example, tamoxifen citrate is a selective estrogen receptor modulator (SERM). Tamoxifen and several of its metabolites are thought to act as estrogen antagonists by competitively binding to estrogen receptors on tumor and other tissue targets, producing a nuclear complex that decreases DNA synthesis. This mechanism appears to cause cells to accumulate in G0 and G1 phases and may induce apoptosis independent of estrogen receptor expression. It is also recognized that tamoxifen acts as an estrogen agonist on endometrium, bone, and lipids. Primary uses of tamoxifen are brain tumors, breast cancer, melanoma, and soft tissue sarcoma. Other uses are carcinoid tumor, endometrial cancer, and pancreatic cancer. Reproductive toxicity and non-genotoxic carcinogenicity have been reported for other SERMs such as raloxifene hydrochloride and lasofoxifene tartrate. Raloxifene and lasofoxifene have been shown to be teratogenic whereas tamoxifen may be potentially teratogenic (15).

	In contrast to the mentioned hormone or hormone modulating substances, there are other agents from the same group that pose a lower risk to patients. Progestogens, progestins, and corticosteroids generally have PDE values greater than 0.001 MinDD. Equally low priority drugs are thyroid hormones and their modulators or peptide hormones such as vasopressin, oxytocin, and prolactin even though their absolute PDE values may be low.

I, there are several immunosuppressive or immunomodulating substances indicated for use in organ transplantation, such as cyclosporine A, mycophenolate mofetil, and mycophenolic acid. These substances may cause adverse effects in pregnancy and are dosed in relatively high doses in patients undergoing organ transplantation. Cyclosporine A is a calcineurin inhibitor, and mycophenolate mofetil and mycophenolic acid are purine synthesis inhibitors. An additional adjustment factor is assigned to these medicines because of concern for their potential to cause adverse effects on reproduction or development. Conversely, mammalian target of rapamycin (mTOR) inhibitors such as tacrolimus and everolimus have R > 100 because adverse effects on reproduction-development are not driving the PDE values. If the adverse effect profile of these substances would be entirely due to the pharmacological action of the substance, the likelihood that the PDE would be lower than 0.001 MinDD is low. However, as these adverse effects are pharmacologically-mediated, their R-ratios are higher. The PDE is established to prevent any adverse effects, whether pharmacologically- or toxicologically-mediated.

	Antiviral substances may potentially have genotoxic, developmental, or reproductive effects. These effects lower the PDE values because an additional adjustment factor for severity of effect is applied. One example is ribavirin, which has a PDE value lower than 0.001 MinDD due to adverse effects on fertility and EFD and a long elimination half-life of more than 150 hours, and therefore, a potential for accumulation with the assumption of repeated daily exposure required for PDE calculations.

	The authors evaluated the 14 antibiotics in their company’s portfolio, none of which were beta-lactams. Some of these antibiotics have very high MinDD values, including neomycin, with a MinDD of 33 g/day. The R was typically around 10 because these antibiotics may be skin or respiratory sensitizers, which typically result in low PDE values.

	There were four antibiotics out of 14 that had R<10: chloramphenicol, two aminoglycosides (gentamycin, and neomycin), and tigecycline, a tetracycline antibiotic. The aminoglycosides amikacin, gentamycin, kanamycin, neomycin, netilmycin, paromomycin, spectinomycin, streptomycin, and tobramycin inhibit protein synthesis in gram-negative bacteria and have bactericidal activity. Aminoglycosides rapidly cross the placenta. Aminoglycosides cause nephrotoxicity and ototoxicity. Congenital hearing loss has been documented in association with prenatal exposure to streptomycin and kanamycin, but not with the use of gentamycin and tobramycin. This distinction between the substances within the same class of antibiotics shows the importance of the toxicological and pharmacological evaluation in discriminating patient safety issues even with the same class of drug. In addition, aminoglycosides pose a potential risk for skin sensitization, which has been observed in patients who have applied the drug topically. Neomycin has an oral MinDD of 3000 mg/day and equally high oral PDE values. Due to its low oral bioavailability (less than 10%), the PDE intravenous is lower than oral; however, high maximal safe carry-over limits are expected to exceed other quality criteria in cleaning, such as visually clean.

	A number of antibiotics have effects other than just those on bacterial growth. For example, trimethoprim is a folate antimetabolite and thus has the potential for EFD effects. Chloramphenicol was also found to have R<10. Chloramphenicol is relatively toxic and can cause severe agranulocytosis. It readily crosses the placenta and can reach therapeutic concentrations in the fetus. It is genotoxic and has been classified by IARC in Group 2A, as probably carcinogenic to humans (13). Tetracyclines cross the placenta, and they bind strongly to calcium ions. From the sixteenth week of pregnancy, tetracyclines are strongly bound in this way in developing tooth and bone structures, causing brown permanent discoloration of deciduous teeth and inhibition of bone growth. They can cause liver toxicity in the mother and are contraindicated in pregnancy and in children younger than eight years of age.

	Fluoroquinolone antibiotics have a narrow range of doses between the minimal and maximal daily doses. For those substances, a larger factor is applied to the therapeutic dose to get from a lowest observed adverse effect level (LOAEL) to no observed adverse effect level (NOAEL). In addition, as they may cause hypersensitivity reactions, an additional factor for severity of effect is incorporated into the PDE calculation.

	Macrolide antibiotics are not considered high priority APIs, despite issues with erythromycin, which is considered a skin and respiratory sensitizer. Erythromycin is considered a safe and effective antibiotic during pregnancy. Other macrolides are also generally safe to be used in pregnancy.

	Polypeptide antibiotics are considered low in priority as they have not shown any effects on embryo-fetal development. Vancomycin has a high MinDD of 500 mg/day orally (indicated for the treatment of staphylococci enterocolitis and pseudomembranous colitis caused by Clostridium difficile in the gastrointestinal tract) and 1000 mg/day with intravenous administration (indicated for the treatment of systemic infections caused by gram-positive bacteria including staphylococci and streptococci). It is poorly absorbed from the gastrointestinal tract. The carryover values calculated for this substance are expected to be above the visually clean criteria. Teicoplanine has a very long elimination half-life and, therefore, the potential for accumulation, which is accounted for in the PDE calculation.

	For an innovator company, the PDE calculation for a drug is based on a review of the totality of clinical and nonclinical data to identify the NOAELs and LOAELs from the most relevant regulatory documents available, such as the investigator’s brochure or the core data sheet, supplemented by review of the literature for compounds with the same target. After adjustment, the lowest of these values termed the point of departure (POD), which is a dose that should not cause a significant pharmacological or toxicological effect, serves as the basis of the PDE calculation. For example, the lowest oral therapeutic dose of irbesartan is 75 mg/day, but its lowest pharmacologically active dose is 10 mg/day. There is a seven-fold difference between these two values.

	In the absence of the minimal pharmacologically active dose, there may be a different dose used in the initial therapy than in maintenance. For carbamazepine, the initial dose is 100 to 200 mg once or twice daily and the maintenance dose is 400 mg two to three times daily. The 12-fold difference between the lowest starting dose (100 mg/day) and highest maintenance dose (3x 400 mg/day = 1200 mg/day) simply reemphasizes one aspect of the problematic nature of the MinDD in the calculation of safe carryover limits.

	Also, when tabulating the MinDD values, it is important to include the route of administration. Toxicological documentation usually provides oral, parenteral (e.g., intravenous or intramuscular), and inhalation PDE values for each substance. These values may be significantly different because of differences in bioavailability. For example, zoledronic acid, ibandronate, and pamidronate are bisphosphonates indicated for the treatment and prevention of bone diseases, including hypercalcemia of malignancy, osteolytic bone metastases from solid tumors and osteolytic lesions of multiple myeloma, postmenopausal osteoporosis, and Paget’s disease. They have a low oral bioavailability of <10%. Therefore, selecting an intravenous MinDD when the next substance produced in the same equipment is dosed orally would give a conservative approach. Another example is some of the antibiotics. Neomycin was previously mentioned as an example of a drug that is indicated for different treatments with oral and intravenous administration. Selecting the appropriate route for MinDD is, therefore, crucial. The PDE value considers the pharmacokinetic and pharmacodynamics effects appropriately in the calculation of PDEs for various routes of administration.

	When substances are dosed with non-daily dosing regimen, the MinDD may not be an appropriate value without extrapolating to a daily dose. For example, the recommended dose of pamidronate for the treatment of predominantly lytic bone metastases and multiple myeloma is 90 mg administered as a single infusion once every four weeks. Extrapolation from 90 mg/28 days to daily dose is 3.2 mg/day pamidronate disodium. When the PDE is calculated, the extrapolated daily dose is taken as a point of departure (POD), given that the definition of the PDE is a safe dose for daily administration. It is doubtful whether the dosing regime has been routinely incorporated into the MinDD values used as a basis for safe carryover calculations. The non-daily dosing can be considered in the product-specific PDE values. The issues of establishing a product-specific PDE, when the following drug product is known, and methods to derive these values are discussed elsewhere (16).

	Lastly, the substances that are either dosed only for short treatment duration or substances that are dosed with non-daily treatment regime may have long elimination half-lives and a potential for accumulation. For example, netupitant, a neurokinine-1 (NK-1) receptor antagonist indicated as an antiemetic, has a mean elimination half-life of 86.3 hours. Accumulation of netupitant has been observed in a repeated-dose study with a once-per-day dosing frequency. Therefore, in the calculation of the PDE for netupitant, a factor of five was applied for accumulation.

	Data are presented comparing safe carryover limits derived from PDEs and 0.001 MinDD for 140 different APIs. Grouping these medications by therapeutic use or MOA differentiates between medications that would potentially pose a risk for patients. Antineoplastics have consistently had 0.001 MinDD values less conservative than PDE values. In addition, the adverse effect profile (EFD effects, genotoxicity, carcinogenicity, and other off-target effects), potential for accumulation, and non-daily dosing regimen of a drug will drive the PDE value potentially below the 0.001 MinDD. The MinDD in this study is useful in demonstrating how a pharmaceutical company could prioritize its drug portfolio to identify those requiring comprehensive assessments as a first priority. The results further show that there are limitations of the MinDD criterion and support the use of the PDE limit for cleaning validation.

	The authors would like to thank Michel Crevoisier for his valuable input.

EudraLex: The Rules Governing Medicinal Products in the European Union, Part I--Basic Requirements for Medicinal Products

. Volume 4. EU Guidelines for Good Manufacturing Practice for Medicinal Products for Human and Veterinary Use. Chapter 3: Premises and Equipment (Brussels, August 2014).
	2. European Commission, 

. Volume 4. EU Guidelines for Good Manufacturing Practice for Medicinal Products for Human and Veterinary Use. Chapter 5: Production (Brussels, August 2014).
	3. EMA, 

Guideline on setting health based exposure limits for use in risk identification in the manufacture of different medicinal products in shared facilities

 (EMA/CHMP/ CVMP/ SWP/169430/2012), (London, Nov. 2014).
	4. E. Lovsin Barle et al., 

 32, 18-23. (2014).
	5. G.L. Fourman and M.V. Mullen, 

., 17 (4) 54-60 (1993).
	6. M. Olson et al., 

. 79 Suppl 1:S19-27 (2016).
	7. A. Teasdale et al., 

., 40 (1) 58-62 (2016).
	8. J.P. Bercu et al

. 79 Suppl 1:S48-56 (2016).
	9. J. Gould et al., 

. 79 Suppl 1:S79-93 (2016).
	10. J.F. Reichard et al., 

. 79 Suppl 1:S67-78 (2016).
	11. R.G. Sussman et al

. 79 Suppl 1:S57-66 (2016).
	12. B.A. Chabner, “Chapter 60: General Principles of Cancer Chemotherapy” in 13. 

Goodman and Gilman’s The Pharmacological Basis of Therapeutics

, L. L. Brunton, B. A. Chabner, and B. C. Knollmann, Eds. (McGraw-Hill Professional, New York, 12th ed., 2011), pp. 1667-1675. 
	13. IARC, Agents Classified by the IARC Monographs, Volumes 1-112 (Alphabetical Order), April 7, 2015. Lyon, France: World Health Organization, International Agency for Research on Cancer (2015).
	14. Pfizer Canada Inc., Product Monograph: Adriamycin PFS* doxorubicin hydrochloride injection, 2 mg/mL (revised August 22, 2014).
	15. J.C. Berger and C.L. Clericuzio. 

, 146A (16) 2141-2144 (2008).
	16. J.P. Bercu, R. Sharnez, and D.G. Dolan. 

	Ester Lovsin Barle, PhD*, is head of Health Hazard Assessment at Global Health Safety and Environment and Business Continuity Management, Novartis AG, 

; Camille (Bossard) Jandard, Pharm.D, MSc, is toxicology project leader and occupational toxicologist at Biologie Servier, 905 route de Saran, 45520 GIDY, France, 

; Markus Schwind, PhD, is HSE manager at Sanofi-Aventis Deutschland GmbH, HSE Germany, Industriepark Hoechst, Bldg. H831, Room A.04.72.2, D-65926 Frankfurt am Main, 

; Gregor Tuschl, PhD is principal scientist, Early Non-Clinical Safety, Merck KgaA, Frankfurter Str. 250, U009/101, 64293 Darmstadt, Germany, 

; Claudia Sehner, PhD, is principal scientist, Nonclinical Drug Safety, Boehringer Ingelheim Pharma GmbH & Co. KG, 

; and David G. Dolan, PhD, is a senior manager, Occupational, Environmental & Quality Toxicology, Product Stewardship, Environment, Health, Safety and Sustainability (EHSS), Amgen Inc., 

	*To whom all correspondence should be addressed.

	When referring to this article, please cite it as E. Barle et al., “Comparison of Permitted Daily Exposure with 0.001 Minimal Daily Dose for Cleaning Validation," 

Articles

In this study, the authors investigated the relationship between the 0.001 MinDD and the PDE values for 140 drug substances as an attempt to identify high-risk groups of products for patient safety. This comparison can serve as a method for prioritization of APIs for development of PDEs.
