1. Introduction
In November 2014, the European Medicines Agency (EMA) published the Directive EMA/CHMP/ CVMP/SWP/169430/2012⁽¹⁾ on the establishment of exposure limits based on health criteria during the manufacture of different drugs in shared installations.
In January 2015, the European Commission also revised Chapters 3 and 5 of the EU GMP Guidelines, updating the sections on the prevention of cross-contamination. These updates entered into force in March of the same year. Further, Annex 15 of the EU GMP Guidelines (item 10.6) stipulates that “The residual contamination limits of the product must be based on a toxicological assessment”. The dates for implementation of the new EMA Guidelines were June 2015 for new products and December 2015 for existing products (with deferral of these deadlines for a year for veterinary products).
Up until then, the limit values of 10 ppm or 1/1000 of the lowest clinical dose were used for cleaning validations. The use of conventional limits translated, on the one hand into excessively restrictive limits for low toxicity products, far exceeding the limits necessary to guarantee patient safety. And on the other hand, the conventional limits did not offer enough protection for products considered to be highly toxic.
Up until then, the categories with the highest risks concerned “certain hormones”, “certain cytotoxic agents”, etc., with no specific criteria.
Release of the EMA Directive implies determining exposure limits in accordance with toxicological criteria, and on the basis of characteristics inherent to each substance. The PDE (Permitted Daily Exposure) is calculated from pharmacological, toxicological and pharmacokinetic data and is similar to the ADE value (Acceptable Daily Exposure) described in the IPSE (International Society for Pharmaceutical Engineering) RiskMaPP⁽²⁾. The two values represent the maximum daily dose of a substance that is not likely to cause adverse effects in an individual who is exposed to this dose or to a lower dose, every day throughout their life.
In order to facilitate the implementation of these new GMP regulations in the pharmaceutical industry, the Azierta team of toxicology experts launched a project to calculate PDE values for Active Pharmaceutical Ingredients (“API”), as external service provider in January 2015.

With an experience that is unique in Europe as a result of having performed the toxicological assessment and preparation of 1200 PDE monographs, we conducted an in-depth meta-analysis of the data obtained. A classification of PDE values was established, identifying 5 groups and allocating a danger level to each of them. The results were analyzed against the ATC classification (anatomical, therapeutic, chemical) therapeutic groups.

2. Methodology

2.1 Calculation of the Permitted Daily Exposure (PDE)
One hundred and fifty international pharmaceutical laboratories ordered a PDE report for their API (Active Pharmaceutical Ingredient) according to their needs. 1,200 active substances in total were assessed by a team of AETOX/EUROTOX experts. The PDE Value was determined by following the procedures described in the following reference guidelines:

• ICH Topic Q3C (R4): Impurities – Guideline for residual solvents (CPMP/ICH/283/95)(3)
• VICH GL18(R): Impurities- Residual solvents in new veterinary products, active substances and excipients (EMA/CVMP/VICH/502/99-Rev.l)⁽⁴⁾
• ICH Q3D: Elemental impurities (Sept. 2015)⁽⁵⁾

The toxicological evaluation was performed by a review of the literature, identifying both the dangers and critical effects associated with the substance. From these reviews, a point of departure (POD) was selected to calculate the PDEs.

Using the information available the starting value most appropriate to each case was determined on an individual basis. The POD values used were the following:

• No-observed-adverse-effect level “NOAEL”,
• No-observed-effect level “NOEL”,
• Lowest-observed-adverse-effect level “LOAEL”,
• Lowest-observed-effect-level “LOEL”,
• Threshold of toxicological concern “TTC”.

Depending on the POD selected, and taking account of both the preliminary study and the toxicological data found for the substance concerned, the safety factors (F1, F2, F3, F4, F5) of the equation, described in the literature⁽³⁾, were determined for calculation of the PDE. The PDE values obtained were expressed in mg/day.

2.2 ATC Classification
To conduct the study of PDE values in accordance with therapeutic groups, each API was allocated its ATC code (Anatomical, Therapeutic, Chemical code)⁽⁶⁾.
In this classification system, active substances are divided into different groups depending on the target organs on which they act and their therapeutic, pharmacological and chemical properties. In the first level, the APIs are distributed between the 14 main groups which are then subdivided into five levels. For this study, only classifications up to the third level were taken into account.

The study did not include products limited to veterinary use.

2.3 PDE Categories
On the basis of the different categories described in the Occupational Health for OEL values (Occupational Exposure Limits) by Safebridge⁽⁷⁾ and Naumann⁽⁸⁾, we established different categories of PDE to analyze the results obtained. Different levels of dangerous substances were defined depending on their PDE value, 5 groups (1 to 5) differentiated by increasing toxicity(Table 1).

Group 1 comprises PDE values > 1 mg/day and is associated with a very low danger level. Group 2 comprises a PDE range between 1 and 0.1 mg/day associated with a low danger level. Group 3 has a PDE range between 0.1 and 0.01 mg/day with a moderate API danger level. Level 4 has a PDE range between 0.01 mg/day and 1 μg/day and is associated with a high danger level. The final level, Group 5, comprises all PDE values below 0.001 mg/day (1 μg/day) associated with a very high danger level (Table 1).

3. Results and interpretation

3.1 Point of departure values (POD)
Figure 1 represents the different types of values that were used as points of departure to determine the PDE. In most cases, NOAEL – NOEL values taken from toxicological studies (41% and 15% respectively) were used to calculate the values. Owing to an absence of studies giving these values or a lack of reliability of the latter, the therapeutic dose, such as LOEL, was used in 40% of cases. When no NOAEL value was available, the TTC value (Threshold of toxicological concern) was used as a point of departure for genotoxic substances (1%).

3.2 Types of substances (API)
In Figure 2, we can see the distribution of substances according to the different ATC therapeutic classes. Most of these fall within class N (nervous system) and class A (alimentary tract and metabolism) with 16.3% and 15.6% of the products total respectively. Approximately 5% of all substances are intended exclusively for veterinary use and were excluded from these investigations.

3.3 Categorization of PDEs
Figure 3 represents the distribution of substances according to the PDE value and the classification system applied (5 categories).

• 33% of substances assessed belong to group 1, which means that they have a PDE >1 mg/day, associated with a very low danger level.
• 29% of the APIs assessed can be included in group 2, with a PDE range between 1 and 0.1 mg/day. These substances are considered to have a low danger level.
• 17% of substances have PDE values between 0.1 and 0 .01 mg/day and are considered as having a moderate danger level.
• Group 4 comprises 12% of APIs in the study with a PDE range between 0,01 mg/day and 1 µg/day associated with a high danger level.
• The last level, group 5, comprises all PDE values below 0,001 mg/day (<1 µg /day). These are substances with a very high danger level which represent 9% of the total APIs studied.

Figure 4 presents the distribution of the different therapeutic classes according to the PDE categories for groups 1, 2, 3 and 4.

A more in-depth examination of the group of substances presenting a very high danger level (level 5, PDE <1 µg/day) reveals that it is not solely represented by the expected substances, particularly hormones and cytotoxic agents (ATC classes H, G, L), but by substances that belong to other therapeutic groups, especially N (nervous system), S (sensory organs) R (respiratory system) and A (alimentary tract and metabolism)(Figure 5).

We therefore analyzed the ATC groups that contributed the most to the group 5 PDE classification individually (G, H, L, S, N, R).

As expected, the ATC groups H, G and L, which contain hormones and cytotoxic agents, have a higher percentage of products with PDE values <1 µg/day, associated with a very high level of toxicity 5 (group 5). These results isolated by therapeutic group are represented in Figure 6.

• 23% of group H substances (specialist products based on systemic hormones) have PDE values below 1 μg/day and are equivalent to products in the HO1 classes (hypothalamic hormones, particularly octreotide) or H03 (thyroid hormone). 32% of H substances are in group 4. In the hormonal preparations, 9% of substances have a PDE value between 1 and 0.1 mg/day and 36% between 0.1-0.01 mg /day.
• 18% of group G substances (genito-urinary system and sex hormones) have PDE values below 1 μg/day and they represent the sex hormones in this class (ATC G03).
• When the group L therapeutic group is analyzed, 25% of these products have PDE values below 1 μg/day and they correspond to immunosuppressive agents (fingolimod) or antineoplastic agents (mercaptopurine, docetaxel, etc.). In this group, 13% of substances have PDE values higher than 1 mg/day.
• On examining group S (sensory organs) it is observed that 27% of substances in this class fall within group 5 (PDE below 1 μg/day). These products are ophthalmic decongestive agents, class S01G (particularly naphazoline), anti-glaucoma agents, class SOIE (latanoprost), ophthalmic anti-allergic agents, class S01G (azelastine) or ophthalmic anticholinergics such as atropine.
• Analysis of nervous system products (ATC class N) shows that only 7% have PDEs below 1 μg/day, corresponding principally to antipsychotic substances such as benperidol, benzodiazepine derivatives (nitrazepam) or anesthetics (sufentanil).
• 12% of substances in group R (respiratory system) have PDE values below 1 μg/day and they correspond to decongestants or adrenergic agents in inhalants such as salbutamol.

Conclusion
The specific pharmacological and toxicological properties of each Active Substance must be duly assessed before establishing exposure limits based on health criteria for drugs manufactured in shared installations according to the criteria of the directive EMA/CHMP/CVMP/SWP/169430/2012. On the basis of the individual properties of Active Substances, an in-depth scientific study enabled the establishment of a concrete exposure limit in the event of inspection by the health authorities which is binding for the auditing of cleaning method validations.

Our analysis of 1200 PDE Active Substances monographs revealed that it was not possible to estimate the danger level or toxicity of Active Substances in advance solely on the basis of their therapeutic group. Substances with a high toxicity level, even very high, do not necessarily represent the expected groups, such as those of hormones or cytotoxic agents, and not all Active Substances in these therapeutic groups display in reality a high exposure level in risks to patients linked to cross-contamination. It is then effectively relevant to carry out an individual toxicological and pharmacological assessment for each Active Substance, regardless of the group to which they belong.