EBE Position Paper

Draft: Final:
Author: EBE BioManufacturing WG
Date of Publication: 13.09.2016

EBE Position Paper

A Risk-Based Approach to Setting Sterile Filtration

Bioburden Limits
13 September 2016 Version 1

Executive Summary

The rationale behind the EMA recommended bioburden limit before sterile filtration of NMT 10
CFU / 100 ml (EMA Guidelines of 1996 (1), 2012 (2) and Draft EMA Guideline 2016 (3)) has no
clear origin. The limit has been taken from the pharmacopoeial specification for ’water for
injection to produce bulk’, but has no scientific basis when applied to drug product. Overall, we
would instead propose a "Risk-based Approach" to bioburden control which justifies alternative
bioburden specifications that take the product manufacturing process into consideration, in
contrast to defaulting to the historical 10 CFU / 100 ml limit.

This risk-based approach considers the two main risks at the sterile filtration step of drug product
manufacture: (a) capability of the microbiological bioburden method and, (b) microbial
breakthrough during sterile filtration. These risks have been assessed for their interaction by
means of statistics and possible risk-mitigating measures are described.

Table of Contents
1. Risk: The bioburden determination method .................................................................... 2
2. Risk: Bacteria breakthrough during sterile filtration ................................................... 2
3. Combination of both main risks ........................................................................................... 3
4. The holistic view considering a portfolio of risk-mitigating measures ................ 4
5. Conclusions .................................................................................................................................. 6
6. References ..................................................................................................................................... 7
7. Addendum ...................................................................................................................................... 7

Page 1 of 12

www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

1. Risk: The bioburden determination method

As Jornitz et al. (5) described, it is impossible due to the inherent variability of microbiological analyses,
to determine an exact bioburden in the order of 10 CFU / 100 ml. The true bioburden could likely be
below or above and thus the batch incorrectly accepted or falsely rejected. It is therefore more accurate
to specify the acceptable limit of 10 CFU plus a margin of error. This inaccuracy in microbiological
analyses is taken into account in USP Chapter <61> and also in the Ph.Eur. (Chapters 2.6.12 and 5.1.4):
an acceptance criterion for microbiological quality of 10 CFU would therefore be corresponding to a
maximum acceptable count of 20 CFU when the variability of analysis is considered.

We have modeled the sensitivity of the Ph.Eur. 2.6.12. bioburden assay method for an acceptance
criterion of 10 CFU / 100 ml (6) while varying the test sample volume. The Poisson model assumes a
uniform distribution of the bacteria in the solution prior to sterile filtration and is therefore not
considered a suitable model. Bacteria can clump together to produce a non-uniform dispersion which
leads to either overestimation or underestimation of detected bioburden when determined by the
number of counts in the solution. This over-dispersion can be described by a negative binomial
distribution with a corresponding probability density function and an over-dispersion factor.

Based on a negative binomial model with an over-dispersion factor of 2, it can be estimated that using a
limit of 10 CFU / 100 ml, a solution with a true bioburden level of 10 CFU / 100 ml would have a 41.2%
probability of batch rejection, although the specification would still be met. Conversely, a true
bioburden of 11 CFU / 100 ml would still lead to a probability of batch acceptance of 50% (see figure 2,
Addendum).

The probability curve also demonstrates that - given an acceptable risk bound of 5% - a batch passing
with a measured bioburden of not greater than 10 CFU/100 mL could reveal a true bioburden of up to
20 CFU/100 mL (for a 100 mL test sampling plan). The probability curve shifts to the right as the volume
tested decreases such that a 10 mL test volume may have up to 63 CFU/100 mL actual bioburden for the
same 5% probability of false negatively passing and the same control limit. This increase in acceptable
actual bioburden can be compensated by adjusting parameters for Risk 2, as described in the next
section.

We conclude that the specification of the bioburden at the level 10 CFU / 100 ml is too tight, leading to
unreasonably high rejection probability of acceptable batches. Furthermore, the microbiological quality
of the solution prior to sterile filtration should be considered holistically and the total risk composed of
many individual risks should be considered in order to avoid falsely accepting unacceptable batches due
to the limited sensitivity of the test.

2. Risk: Bacteria breakthrough during sterile filtration

According to the FDA guideline (4) and industry standards, filters used for the final filtration should be
validated to reproducibly remove microorganisms from a carrier solution containing bioburden of a high
concentration of at least 107 CFU/cm2 of effective filter area (EFA). The validation should be conducted
under the worst-case production conditions for the material to be filtered, and challenge experiments
should result in no passage of the challenge microorganism. Thus the retention capacity of a validated
sterilizing-grade filter with an EFA of A (cm2) is at least A x 107 CFU. However, the currently used 0.22
µm sterile filter membranes can withhold much higher microbial challenges (unpublished company
results for 0.22 µm polyvinylidene fluoride membranes a bacterial challenge concentration as high as
109 CFU/cm2 can be validated).

In practice, accumulation of bioburden on the final filter might even cause partial clogging. As a result, a

Page 2 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

bioburden CFU that reaches the filter early in the process may have a higher probability to penetrate
the filter than those that reach a partly clogged filter later in the process. Therefore it is conservative to
assume that all bioburden CFUs in the solution have the same probability to go through the final filter.

Similar to terminal sterilization methods, for sterile filtration a Sterility Assurance Level (SAL = degree of
assurance with which the process renders a population of items sterile) can be calculated based on the
bacteria retention capacity of the filter. Here, worst-case assumptions have been made, for example
during the validation using a small microorganism (typically Brevundimonas diminuta).

Since the pre-sterile filtration bioburden is determined for each lot and not only as part of a validation
procedure, the SAL can also be calculated as a LOT-SPECIFIC parameter from the actual, measured
bioburden load (with a factor included for variability), the applied total filter area and the validated
effective bacteria retention capability. Thus, the decisive factor of pre-filtration bioburden is known
(within assay variability) and not assumed.

For products subjected to terminal sterilization, the true bioburden is not determined (the Draft EMA
Guideline on the Sterilization of the Medicinal Product, Active Substance, Excipient and Primary
Container suggests a maximum bioburden limit of 100 CFU / 100 ml or 100 g before) as the
manufacturing process does not take place under aseptic conditions, and this must be compensated by
incorporating appropriate safety factors. Since the filling process is run under class C conditions, a
relatively high bioburden must be assumed which then has to be killed by the terminal sterilization
procedure.

In contrast, the sterile filtration process prior to aseptic filling can be considered a closed system,
followed by manufacture under class A conditions. Finally, the bacteria are mechanically retained
instead of being inactivated, and therefore do not enter the final product.

In summary, for sterilization of product through sterile filters SAL values lower than used for terminal
sterilization (<10-6) should be acceptable e.g. 10-4 or 10-5.

3. Combination of both main risks

The two types of main risks associated with the sterile filtration process can be statistically described
(6):
• Risk 1 (“pre-sterile filtration risk”): risk due to bioburden test method insensitivity (risk of false
negative = passing a batch with unacceptable bioburden), i.e. the drug solution with an
unacceptable bioburden level before sterile filtration passes the pre-sterile filtration bioburden
test, either due to inherent test method variability or a sampling method that does not have
sufficient statistical power to detect drug solutions with unacceptable levels of bioburden
• Risk 2 (“post-sterile filtration risk”): risk of breakthrough of bioburden through the final sterile
filter with ≥ 1 CFU entering the sterile filtered solution due to an inappropriate process, i.e.
process-related risks and microbial breach across sterile filter

These two risks are inter-dependent as a high pre-sterile filtration risk would require a more stringent
control of the post-sterile filtration risk and vice versa. Therefore an effective overall control strategy
should take into account this inter-correlation.

Yang et al (6) linked maximum batch size, sample volumes of 10 ml, 30 ml and 100 ml (negative binomial
model, specification limit 10 CFU / 100 mL and acceptance limit of pre-filtration bioburden 1 CFU/10 mL,
3 CFU/30 mL, 10 CFU/100 mL) to the Risk 1 probability of incorrectly accepting a batch (5%, 1% and
0.1%) and the Risk 2 probability of a breakthrough of ≥ 1 CFU in one of 10,000 cases (0.01% probability,

Page 3 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

10-4) or 1 of 100,000 cases (0.001% probability, 10-5). A validated bacterial retention capability of the
sterile filter of 107 CFU per cm2 was used, for a 1000 cm2 filter, to calculate the maximum volume to be
sterile filtered. A modified version of the risks-linking table is attached to this position paper (see table
1, Addendum). By example, a 10 mL sample size using an action limit of 1 CFU/10 mL, with a 5% risk of
passing a batch through a false negative result and a filter breakthrough risk of 10-4, allows a maximum
filtration batch volume of 424 L when using a 1000 cm2 (0.1 m2) filter.

Note that when e.g. δ0 = 5% (pre-filtration risk, i.e. risk of bioburden testing insensitivity) and
δ = 10-4 (post-filtration risk, i.e. breakthrough risk of bioburden through the sterile filter), there is a 95%
probability that the pre-sterile filtration bioburden test (for any given test scheme) will successfully
detect unacceptable levels of bioburden and only 1 batch out of 10,000 would have bioburden
breakthrough (see Table 1). This implies that in a facility producing 100 batches a year, this event would
only occur every 100 years. By limiting the maximum filtered batch size, accepted post-sterile filtration
risk levels can achieve the same level of sterility quality assurance, for any given post-filtration risk (δ),
as the bioburden testing with 100 mL samples and a 10 CFU/100 ml acceptance limit. Indeed, it is
proposed that a Risk 1 probability of ≤ 5% should be acceptable.

Alternatively, to manipulating the filtration batch size, other parameters may also be varied to reduce
risk, e.g. increasing filter area or validated maximum CFU. The relationship between sample volume and
the batch size varies according to filter area, as illustrated in Figure 3 of the Addendum. A 10 mL sample
size using a 1000 cm2 filter allows up to 424 L to be filtered to retain a ≤ 5% probability of passing a
batch with bioburden exceeding the acceptance limit and filter penetration of less than 1 in 10,000
batches. For a 2000 cm2 filter the filtered volume could be increased to 849 L at the same risk level (see
table 1). The relationship between filtration batch size and filter area is tabulated and provides the
Sponsor with the opportunity to select the sterile filtration, minimum filter area according to filtration
batch size to provide a desired level for Risk 1 (0.1% to 5%) and Risk 2 (10-4 to 10-5).

Therefore, for a given risk bound, the pre-filtration risk can be controlled with a properly selected batch
size or the post-filtration risk can be controlled through filter area and filter validation. The bioburden
level D0 may be even greater than 10 CFU/100 ml since its impact on the final post-filtration risk can be
controlled through placing a limit on the size of a batch for the sterile filtration S0 or adjusting the filter
area or the validated range so as to ensure the total bioburden in the batch would not exceed the
retention capacity of the final sterilizing filter.

The presented calculations support use of bioburden test sample volumes ≥ 10 ml using a filter area of ≥
1000 cm2, a maximum filtration batch size of 400 L and an action limit of up to 2 CFU/10 ml that
includes a 3-fold safety factor for the bioburden determination precision (calculated from 63 CFU/100
ml). These limits ensure a pre-filtration bioburden risk of 5% and a post-filtration risk of 10-4. Further risk
mitigation is built into the product manufacturing process as summarized in the next section.

4. The holistic view considering a portfolio of risk-mitigating measures

Mitigating risk factors are essential for the final bioburden risk assessment of the sterile filtration
process. This includes examination of the related process steps and understanding of the product
characteristics and potential product related risk factors like viscosity, pH or growth promoting
properties of the formulation.
For example, if the formulated product has growth promoting properties, this constitutes a risk factor
per se.

It is also important to evaluate available bioburden data, including the manufacturing facility historical

Page 4 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

bioburden detection trends for the product and across products when a multi-product site is used.
Historical bioburden trend data are often indicative of the effectiveness of the overall bioburden
control, and need to be considered in identification of risk factors.

The focus of this paper is on parenteral biotech drugs that cannot be terminally sterilized in the final
container because they are thermo-labile. For such products, however, the testing of the bioburden
prior to sterile filtration and the subsequent sterile filtration are not the only measures to control the
microbiological status of the solutions for container filling (see figure 2, addendum):

• The upstream and downstream processes are likely to have multiple 0.2 micron or nano-filtration
steps for reduction of adventitious microbial, particulates or viral contamination.
• The fermentation itself takes place in closed systems subject to a CIP / SIP process and under the
addition of filtered or HTST-treated or even sterile media.
• The final formulation steps are carried out in a Grade C environment and also use steam-
decontaminated or sterilized containers into which the solution is being transferred via 0.2 micron
filter.
• In the event of extended storage of the formulated solution, storage is usually at low temperatures,
typically -20, -40 or -80 ° C. In the frozen state bacterial growth is extremely unlikely.
• Interim storage in the liquid state is time-limited for reasons of product stability alone and
performed mostly also at reduced temperatures (2 – 8°C), further bacterial growth is hereby also
limited. Storage and hold times of bulk after final formulation or thawing are evaluated in studies
and eventually validated. Room temperature storage is kept at a minimum. Monitoring the
container integrity during such interim storage is often done by superimposing pressure (nitrogen).
Furthermore, interim bulk storage after thawing may be (re-)validated using growth-promoting
media in order to uncover microbiological weaknesses (as part of the regular media fills).
• Increasingly, upstream, downstream and fill/finish processing use pre-sterilized single-use materials
and aseptic connectors (“disposables processing components”). These are often sourced pre-
assembled as manifolds, pre-sterilized and aseptically connected. In addition, these components
represent closed systems, which reduce the risk of bioburden entering by non-aseptic connections.
Through increasing automation of manufacturing processes, individual personnel errors are further
mitigated.
• The commercially available sterile filters have a larger bacteria retention capacity than the
underlying minimum expectation of 107 CFU / cm2.
• For early clinical phase manufacture, the ratio between filter surface area and batch size is typically
relatively high compared to commercial scale manufacture.
• Endotoxin content (in case e.g. pyrogens are produced by bioburden) and sterility testing is part of
the Drug Product release process
• An analysis of data from the companies involved in this paper has shown that it is a very rare event
to find bioburden over the limit set by the EMA before the sterile filtration. Often it is an artifact of
sampling or in the analysis, but both are each difficult to prove with 100% certainty, so finally the
batch is often still discarded which is very costly and may endanger the product development
timelines. Each bioburden found, even if below the limit of 10 CFU / 100 mL may trigger
identification of the microbial contaminant, so that regardless of a limit set, investigation and
corrective measures are initiated.

Page 5 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

For this reason, the authors are of the opinion that the processing of biotechnological active ingredients
is typically under ample control for bioburden, which allows the determination and justification of lower
sample volumes and wider pre-defined levels of bioburden before sterile filtration, using risk-based
strategies.

5. Conclusions
Biotechnology-derived drug products are typically labile to methods of terminal sterilization such as
irradiation, chemical treatment or autoclaving, and are manufactured through aseptic processing which
typically employs sterile filtration to remove microbial contaminants. High bioburden in the drug
solution prior to sterile filtration may increase the chance of breaching the sterilizing filter and cause
product safety and quality issues. As a result, pre-filtration bioburden test is recommended by
regulatory guidelines. The current EMA guidelines stipulate that a maximum acceptable bioburden level
must be established before the sterile filtration step and that in general no more than 10 CFU/100 mL
would be acceptable. However, a sample volume of 100 mL often represents a significant proportion of
the batch and a high cost for biotech products, particularly taking into account that sufficient material
would typically be drawn for replicate testing, assay controls, overage and reserve samples. Therefore it
is desirable to explore alternate sample volumes and test methods that warrant the same or higher
level of quality assurance as the EMA-recommended method. The guidelines allow the Sponsor to
justify alternative sampling procedures though no guidance is provided on what might constitute an
acceptable justification. Furthermore, the EMA guidelines do not provide scientific rationale for limit 10
CFU/100 mL, thereby compounding the difficulty to justify a test volume smaller than 100 mL or other
limits.
In this paper, a risk-based method is proposed to provide a strategy and scientific methodology for
justifying alternate (smaller) bioburden test volumes than 100 mL and alternate specifications (>10
CFU/100 mL or >1CFU/10 mL).. By modeling bioburden in the solution prior to sterile filtration, the
relationship(s) between pre-sterile filtration risk (probability of detecting bioburden in the solution prior
to sterile filtration) and post-sterile filtration risk (probability of bioburden breaching the filter), their
combined risk as well as the associated risk factors such as test sample volume or batch size are
established. Such relationships allow for quantitative evaluation of the impact of the risk factors on the
development of effective risk-based control strategies, which include acceptable selection of sample
test volume and maximum pre-filtration bioburden level. The risk-based method is in accordance with
the quality by design (QbD) principles in ICH Q8 that enable manufacturers to define a manufacturing
process design space that consistently produces high-quality products through increased understanding
and knowledge of the product and process. It is also consistent with ICH quality initiative, Quality Risk
Management, which achieves product greater quality assurance through risk identification, analysis, and
control.
The risk-based approach also justifies use of action limits for control of the pre-sterile filtration
bioburden limit through implementation of an effective Quality Management System (QMS) that
describes the actions followed in the eventuality of any detected bioburden. The conditions that initiate
investigation of bioburden, determination of likely root cause and any resulting CAPA should be
available to the agency and may be verified through GMP inspection. Sponsors may also employ tighter
internal criteria per the QMS. Ultimately, the agency should be assured that any batch with a significant
bioburden breach potentially impacting product quality or safety would be rejected. The onus should
remain with the Sponsor to determine the definition of ‘significant’.

Page 6 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

Key findings of the paper include:

o Pre-filtration and post-filtration risks are inter-correlated. A holistic approach needs to be taken
in development of risk mitigation strategies so that a seemingly high risk in one process step can
be mitigated through controlling risk factors in other step.
o Risk factors are inter-dependent and should be considered jointly when evaluating their impact
of the pre-filtration and post filtration risks. This includes selection of sample test volume,
acceptable bioburden limits, sterilizing filter area, in an integrated assessment.
o A bioburden level higher than 10 CFU/ 100 mL in the unfiltered drug solution may not incur
unnecessary risk as its impact can be mitigated through effective sterilizing filtration.
o Sample volumes less than 100 mL and acceptance limits different from 10 CFU/100 mL can be
justified, through controlling other risk factors such as batch size or filter area, without
increasing the risk of bioburden breakthrough in the final filtration.
o Testing of bioburden prior to final filtration and the sterile filtration are not the only measures
to control the microbial status and to generate sterile product. Holistic concepts may be
established that reduce risks prior to the final bioburden testing and the sterile filtration.
o Enhanced understanding of the manufacturing process and product attributes is key to
successful bioburden risk management.

6. References
1. EMA (1996). CPMP Notes for Guidance on Manufacture of Finished Dosage Form.
2. EMA (2012). EMA Guideline on the Requirements for Quality Documentation Concerning
Biological Investigational Medicinal Products in Clinical Trials.
(EMA/CHMP/BWP/534898/2008), currently under revision
3. Draft Guideline of EMA on the sterilization of the medicinal product, active substance,
excipient and primary container, April 11, 2016
4. U.S. FDA Guidance for Industry: Sterile Drug Products Produced by Aseptic Processing—
Current Good Manufacturing Practice, 2004.
5. Jornitz MW, Akers JE, Agalloco JP, Madsen RE, and Melzer TH (2003). Considerations in
sterile filtration. Part II: the sterilizing filter and its organism challenge: a critique of
regulatory standards. PDA Journal of Pharmaceutical Science and Technology. March/April.
Vol. 57, No. 2, p 88 – 96.
6. Yang H, Li N and Chang S (2013). A risk-based approach to setting sterile filtration bioburden
limits. PDA J. of Pharm. Science and Technology. Vol. 67: 601-609

7. Addendum

Manufacture of drug products (chemical synthesis or from cell/microbial cell cultures) generally follow a
similar set of process steps illustrated in Figure 1. Each of the nine steps includes control of either
bioburden or sterility. They also include several filtration and chromatographic steps that may deplete
any adventitious bioburden from the product. The product contact process materials are cleaned and
validated to minimize risk on contamination.

Page 7 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

No. of Reduction/
Source of material process Process Step Bioburden
control of
step bioburden checks
Material generated by
biopharmaceutical Sterile conditions
1 Check for foreign
process (protein, Mab, Fermentation (closed systems, sterile
organisms
feeds)
DNA)

Pre- and
CFF (Cell free filtrate) Depth filtration or
2 postfiltration
after harvest Centrifugation
check

Nanofiltration, column
resins trapping micro-
Purification (Cation organisms (+ cleaning
exchange, Anion cycles at unfavourable Pre- and
3 postfiltration
exchange, HIC, conditions and validated
checks
nanofiltration, others) re-use of columns/
filter), qualified hold
times

Material generated by Final Formulation after 0.2 micron filtration into

4
chemical synthesis buffer exchange e.g. sterilized containers, Pre- and post-
(peptide, siRNA, small by UF/DF Class C environment filtration check
molecule)
Storage temperature
5 DS storage cold or frozen slowing down or impeding
in various containers microbial growth, Bioburden limit
(specification)
(bags, cryovessels, qualified DS hold time
bottles, cans) (expiry) including
bioburden

0.2 micron filtration (Pre- and)

6 post-filtration
(optional), limited
Thawing of DS (if frozen) check
and qualifiied hold
time (optional)

Mixing/pooling and
7 dilution/addition of Class C environment, (Pre-) and post-
excipients or full 0.2 micron filtration filtration check
(optional)
compounding

8 Limited and qualified Pre-sterile

Bulk for sterile filtration hold time filtration check

Sterile Filtration (inline Sterility check of

9
during filling or off-line Sterile filtration final drug product
before filling) and (one or two and of sterile
dispensing into vials, filters) filtered bulk (if
applicable)
syringes, cartridges etc.


Figure 1. Typical process flow diagram for manufacture of liquid biological drug product sterilised by
filtration.

Page 8 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

Figure 2: Performance characteristics of bioburden testing using 100 ml samples and 10 CFU/100 ml
acceptance limit, based on negative binomial distribution with a dispersion factor of 2 (6); dotted line:
5% acceptable risk bound.

The Ph.Eur test for bioburden does not have the capability to determine an accurate bioburden count in
a solution as detection depends on the volume tested. The probability of passing a batch for bioburden
is best determined using a negative binomial distribution to account for the clumping properties of
microbes. Figure 2 illustrates such a probability curve relationship with the actual bioburden level when
100 ml is tested. There is a significant probability of failing and rejecting a ‘good’ batch or passing a
‘bad’ batch on the basis of any single bioburden test when considered in isolation and applying an
acceptance criterion of 10 CFU/100 ml. A holistic approach to bioburden risk is recommended that
accommodates the sterile filtration process and associated risk.

Page 9 of 12 www.ebe-biopharma.org
 european biopharmaceutical enterprises Position Paper: A Risk-Based Approach to Setting
Sterile Filtration Bioburden Limits
13 September 2016 Version 1

Figure 3: Design Space for bioburden test scheme (here: probability of at least 1 CFU < 10-4)
dependent on filtered volume and filter area following a risk-based approach

The relationships demonstrate the breakthrough risk following bioburden determination can be
controlled through the batch volume filtered and/or filter area. The Sponsor may determine the
bioburden control strategy that best fits the process and facility within the design space. A test sample
volume of 10 ml from a 424 L filtration batch size is illustrated (red circle) as acceptable when using a
1000 cm2 filter.

Page 10 of 12 www.ebe-biopharma.org