Measuring Hydrogen Peroxide Residuals and Product Exposure On Isolator and Rabs Filling Lines

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

MEASURING HYDROGEN PEROXIDE RESIDUALS AND

PRODUCT EXPOSURE ON ISOLATOR AND RABS FILLING


LINES
Posted on November 3, 2015
By: Aaron Hubbard, Engineer I
Kirk Eppler, Principal Engineer
Genentech, a Member of the Roche Group
Introduction & Purpose
Vapor Phase Hydrogen Peroxide (VPHP) is the preferred method for
sanitizing aseptic pharmaceutical filling isolators and Restricted Access
Barrier Systems (RABS). In this article, the term “RABS” will be used to
designate any barrier system such as an isolator or Restricted Access
Barrier. Genentech has found that some protein therapeutics are very
sensitive to oxidation by hydrogen peroxide, which quickly absorbs from
the RABS atmosphere into liquid drug product (DP). This fast absorption is
driven by an extremely high partition coefficient for hydrogen peroxide
moving from the vapor phase into liquid solution. As a result, hydrogen
peroxide that absorbs into liquid DP does not come into equilibrium with
the air, but instead, continually increases in concentration while the
product is exposed to the environment (e.g. during operations, fill line
interruptions, etc.). Therefore, residual levels of atmospheric hydrogen
peroxide resulting from insufficient aeration or less than ideal materials of
construction choices can produce liquid hydrogen peroxide concentrations
high enough to impact product quality.
To prevent product impact, Genentech establishes hydrogen peroxide
uptake limits, which are product- and formulation-specific and often fall
below 100 ng/mL. In order to stay below these uptake limits, we must set
aeration and hold times that mitigate the risk of exposure during
operations. However, long aeration times and short allowable hold times
impose difficult operation restrictions, especially if they are set too
conservatively. Consequently, it is important to properly identify and
quantify sources of hydrogen peroxide uptake in the filling process and
accurately set product-specific hydrogen peroxide uptake limits. With
sufficient data from both the process and formulation perspectives, we can
set procedures to mitigate product impact during filling while still allowing
flexibility in operations.
This article details a number of the tests to characterize VPHP uptake in a
filling environment, and also reviews some VPHP monitoring technologies
currently under exploration to better understand and control drug product
filling processes. To better portray the applicability of these studies, this
article will follow a hypothetical mAb product going through a typical
transfer into a RABS filling line. The data presented is all generated in a
small-scale test isolator, which is programmed to continually dose
hydrogen peroxide, resulting in a constant VPHP level. This is the same
methodology and equipment Genentech would use to support a
commercial transfer of any real drug product.
Tests
Determining Hydrogen Peroxide Uptake Limit
When initially characterizing the effect of hydrogen peroxide on a product,
it is necessary to a set an exposure or uptake limit. The standard method
for determining an uptake limit is to perform a spiking study where
hydrogen peroxide is spiked at various levels from 0 ng/mL (negative
control) up to 1000 or 5000 ng/mL (positive control) into the liquid DP. The
uptake limit is set two levels below the lowest spike level that shows
product impact. For the hypothetical mAb product, we will assume that the
peroxide spiking study showed protein oxidation at the 200 ng/mL spiking
level, and the 2 spiking levels below 200 ng/mL are 150 and 100 ng/mL. This
sets the uptake limit for the hypothetical mAb at 100 ng/mL hydrogen
peroxide.
Liquid Vial Uptake
During filling operations in a RABS, drug product is exposed to many
potential sources for hydrogen peroxide uptake. These different exposures
add up to a cumulative hydrogen peroxide uptake level in any single
product vial. The first and most unavoidable location of uptake is in liquid
drug product vials (or syringes) that have been filled, but not yet stoppered.
At this point in the process, the liquid surface is directly exposed to any
residual VPHP left in the RABS atmosphere after decontamination and
aeration. In normal production, a vial only spends 5-10 seconds between
filling and stoppering and hydrogen peroxide uptake is not a concern.
However, when the filling line stops for an interruption longer than 10-15
minutes, the hydrogen peroxide concentration in the vial begins to rise. If
the product is going to be lyophilized, the stopper is placed in the vented
position and absorption will continue until the vial is moved into the
lyophilizer. The rate of absorption depends on the particular vial size and
fill volume; a low fill volume in a relatively large vial has a large liquid surface
area, allowing for fast uptake that quickly drives up the hydrogen peroxide
concentration as the liquid volume is too small to provide significant
dilution. The following datasets show uptake results from a 0.90 mL fill in a
6cc vial exposed at 20 ppb VPHP in a test isolator. These vials are exposed
in two conditions: with no stopper and with a lyo stopper in the vented
position.
Given the hypothetical product’s 100 ng/mL uptake limit, it is apparent that
these vials will easily absorb too much hydrogen peroxide if held
unstoppered for more than a few hours in this fill configuration. If the
product will be lyophilized, it must be loaded into the lyophilizer and quickly
removed from exposure to residual VPHP, with few interruptions in
production. Even when the vial is in the lyophilizer, there is a chance of
continued hydrogen peroxide uptake if airflow moves from the RABS
through the lyophilizer door. In this configuration and at this exposure
level, we would suggest limiting open vial hold times to <60 minutes to
allow for longer holds up to 6 hours while loading partially stoppered vials
into the lyophilizer without going above the uptake limit in vials. If the
product were filled into a 2cc vial or a syringe, these uptake rates would be
significantly lower due to the difference in geometry. A taller and narrower
primary package creates a smaller air/liquid interface, reducing hydrogen
peroxide absorption. However, the reduction in headspace would shorten
the path length for hydrogen peroxide to diffuse from the atmosphere to
the liquid surface, so the reduction is not linear with surface area.
Silicone Tubing Uptake and Comparisons
One of the most common and problematic points of hydrogen peroxide
ingress has been silicone tubing. Almost all DP filling lines use platinum-
cured silicone tubing in the product flowpath, in a variety of sizes, lengths,
and holdup volumes. When this silicone tubing is exposed to VPHP
residuals in a RABS atmosphere, it absorbs and transmits hydrogen
peroxide to the liquid product inside during production. While this problem
is particularly apparent in isolators, where silicone tubing is in-place during
decontamination, it is also a concern with RABS systems where tubing may
be set up after decontamination. Much like open liquid vials in a RABS,
silicone tubing will absorb a large amount of hydrogen peroxide from
residual VPHP, and then transmit it into the liquid product once it is primed.
During production, the tubing is constantly transmitting hydrogen peroxide
into the product.
To explore the rate of hydrogen peroxide transmission across silicone
tubing walls, we have performed a series of experiments in a small-scale
test isolator. In these experiments, a loop of tubing is run into the chamber
and then back out through tri-clamp ports as seen in Figure 3, simulating
the product flow path from a surge vessel to a fill needle.

The isolator is held at a constant hydrogen peroxide injection rate to


maintain a desired VPHP concentration. The silicone tubing can be primed
and held for varying amounts of time, and then the liquid is flushed out
serially as individual samples. The sample volume is correlated to a length
of tubing based on the inner diameter (ID). In this manner, samples from
the length of tubing inside the isolator environment can be distinguished
from samples outside the isolator wall. The results from one such test using
water in 1.6 mm ID tubing are found in Figure 4. Using a larger ID tubing
gives a slower rise in hydrogen peroxide concentration, but the holdup
volume is much larger and large tubing sizes are not always appropriate for
filling processes.
The transmission rate of hydrogen peroxide across silicone tubing walls is
clearly dependent on tubing geometry and VPHP concentration, but it can
also depend on the tubing manufacturer and model. For example,
Genentech has explored a variety of silicone tubing types to reduce wear in
peristaltic pump heads. However, one tubing model selected for its
improved wear characteristics showed higher hydrogen peroxide
transmission rates, as seen in Figure 5. Due to these differences, detailed
information including materials of construction for the product flowpath is
required whenever a product is transferred to a new line.

In contrast to the uptake tests with silicone tubing, PTFE tubing shows no
hydrogen peroxide transmission. While this suggests that PTFE tubing can
solve the issue of hydrogen peroxide uptake in flowpath tubing, silicone is
key for use in peristaltic pump heads or time-pressure fillers. As a result, all
lengths of silicone tubing should be as short as possible to reduce hydrogen
peroxide uptake.
Returning to the hypothetical product, it is apparent that silicone tubing will
be a problem, before a vial is even filled. Depending on the brand of silicone
tubing on the fill line, a stoppage of 30 minutes can introduce enough
hydrogen peroxide to be over the product’s limit if the RABS has not been
aerated well below 100 ppb. This is especially common at the beginning of
a fill immediately after line release, when VPHP residuals are still high and
the production line may have been sitting idle. Based on the 0.9 mL fill
volume, every meter of 1.6 mm ID tubing will hold approximately one dose
of material. With this information, an appropriate purge can be calculated
and tested to clear any potentially contaminated product from the line.
Filling Needle Uptake
Many pharmaceutical companies are targeting lower fill volumes for a
variety of products. As a result, it is important to pay close attention to areas
of the flowpath impacting small volumes of product, which would be
insignificant with a larger fill volume. One such area is the end of the filling
needle, where a small amount of liquid DP is exposed to the RABS
atmosphere and can absorb hydrogen peroxide. To test filling needles, a
variety of needle sizes were attached to PTFE tubing and placed inside the
test isolator. These tests showed that the shape and size of the exposed
liquid surface could dramatically increase or decrease hydrogen peroxide
uptake at this location. The shape of this surface is the largest factor in
determining hydrogen peroxide uptake rate, more than filling needle size
or even length of exposure.

In addition to demonstrating the high hydrogen peroxide concentrations


that can develop at the liquid surface, this experiment showed that
hydrogen peroxide does not readily travel up the filling needle and into the
flowpath. As a result, a purge of a single fill cycle can clear this hydrogen
peroxide ingress. This purge can be necessary even with a well-controlled
filling cycle that avoids exposing liquid as a droplet, since liquid creep at the
tip of the filling needle can be inevitable in a long filling interruption. Again,
following the hypothetical mAb product, a brief 15-minute interruption will
produce a high amount of hydrogen peroxide if material is left at the needle
tip. This hydrogen peroxide will not significantly dilute with a 0.9 mL fill
volume, requiring a single purge cycle for all filling heads.
Empty Glass Uptake
Another common source of hydrogen peroxide ingress is via empty glass
vials. Empty vials can stay in a filling RABS for an extended period of time
before filling and stoppering, and VPHP in the atmosphere quickly adsorbs
to the vial surface. Additionally, hydrogen peroxide on a glass surface stays
in equilibrium with the atmospheric concentration as it rises or falls.
Combining data collected from multiple vial sizes with their interior
geometry, we are able to determine the amount of hydrogen peroxide that
adsorbs to a specific glass surface area. Since the quantity of hydrogen
peroxide adsorbed to the glass is in equilibrium with the atmosphere, we
assume that adsorption is linear against atmospheric concentration. The
resulting adsorption ratio can be expressed in terms of ng/cm2/ppb.

In the case of the hypothetical mAb product, a standard 6cc vial with
interior area of 2.4 cm2 is assumed. At the beginning of a fill, assuming the
RABS is aerated to 100 ppb, approximately 0.05 ng/cm2/ppb will adsorb to
the vial surface. This results in 12 ng of hydrogen peroxide inside each vial,
or 13 ng/mL in a 0.9 mL fill. If the product changes to a typical 2cc vial with
an interior area of 1.6 cm2, the amount of hydrogen peroxide in the vial
drops to 8 ng. Even though these numbers are not large, products that have
a fill volume of much less than 1 mL must also be considered, as this will
severely concentrate small quantities of hydrogen peroxide.
Sensor Comparison
Due to the sensitivity of various biologic products to hydrogen peroxide, it
is extremely important to monitor the residual concentrations of VPHP in
the RABS atmosphere after decontamination and aeration. Many of the
filling RABS in the industry are equipped with on-line monitoring, but very
few have the sensitivity or resolution to accurately measure below 0.1
ppmv (100 ppbv) VPHP. To ensure the safe filling of therapeutic protein
products, effective monitoring must be accurate well below this point,
down to approximately 10 ppbv (0.010 ppmv). To achieve this precision, we
are working with a number of vendors to develop GMP-quality sensors with
various capabilities and price points. Figure 11 shows 3 of these high-
resolution sensors, along with a standard sensor with 0.1 ppmv resolution.
All are reading the same test isolator in parallel, as it is manipulated to
maintain a variety of VPHP levels.

These four sensors monitor VPHP based on three different mechanisms.


Sensors 1 and 4 both use electrochemical sensors, but sensor 1 can be
calibrated to a much lower range with 0.001 ppmv resolution, a feature not
offered by sensor 4. Sensor 2 utilizes a direct spectroscopic reading of a gas
sample from the isolator atmosphere, which is continuously drawn by a
vacuum pump. The unit achieves a path length long enough to quantify the
concentration of atmospheric hydrogen peroxide by reflecting its laser
pulse around a set of mirrors in the sample chamber. This sensor is very
easy to start up and use, but is large enough to require a cart or rack mount.
Finally, sensor 3 advertises the highest resolution, down to part-per-trillion
levels of VPHP. It strips gaseous hydrogen peroxide into a liquid solution
and measures the concentration with a continuous horseradish peroxidase
reaction. However, this sensor is not very suitable for continuous online use
in a GMP facility, as it requires a set of temperature controlled liquid
reagents to be made weekly.
One additional hurdle in the path of getting one of these high-resolution
sensors qualified and into GMP usage is difficulty of calibration. Since
hydrogen peroxide is an inherently unstable molecule, it is impossible to
make a calibration container of low-concentration H2O2 gas. Some
manufacturers calibrate against SO2 as a surrogate gas, and some
manufacturers offer a calibration that is only valid for very high percent
error. Neither of these options is adequate to be used as a GMP quality
monitor at the VPHP levels that are of interest. However, sensor 3 measures
hydrogen peroxide in the liquid phase using a continuous horseradish
peroxidase reaction. This is an assay that Genentech regularly qualifies for
GMP usage, making sensor 3 the only one of the lot that can be calibrated,
while it is also the one least suited for continuous monitoring in a GMP
environment.
Conclusions
Throughout this article, a hypothetical mAb product was followed through
an evaluation of hydrogen peroxide uptake on a new filling line. While this
has been strictly a hypothetical and mathematical exercise based on lab-
scale data, we have followed this process in numerous technology transfers
to support filling of new and existing products on different manufacturing
lines. This process can apply to any manufacturing facility regularly
decontaminated by hydrogen peroxide, most often isolators and open or
closed style RABS.
While examining each area of potential hydrogen peroxide uptake,
potential hold times and applicable VPHP levels have been considered. For
example, uptake across silicone tubing and at the tip of a filling needle will
only occur during a filling interruption when exposed product is not
continuously flushed with fresh material. While an interruption lasting
minutes to hours can occur at any time during a fill, special attention is
required at the beginning of a fill when VPHP levels are highest and line
release may include mandatory stoppages. On the other hand, when
looking at uptake into filled vials, the entire duration of the fill must be
taken into account. A single vial of product may be filled, then held on the
line for an extended interruption before having a stopper applied in the
vented position. Once the stopper is in position, hydrogen peroxide uptake
continues while the vial is staged for loading into the lyophilizer. These
different uptake times are cumulative, and may be adding onto tubing or
needle uptake that occurred before filling. With these considerations in
mind, it is clear that high concentrations of hydrogen peroxide can develop
quickly in liquid product and this problem must be considered in any facility
decontaminated by hydrogen peroxide.
In order to prevent residual hydrogen peroxide in a filling environment
from impacting product quality in protein therapeutics, both formulation
and manufacturing perspectives must be considered to mitigate potential
problems. On the formulation side, we ensure that our spiking studies are
well designed to determine exactly what concentration of hydrogen
peroxide will begin to oxidize our drug product. In manufacturing, we
ensure that the filling environment is well characterized and monitored
whenever possible to have an understanding of how quickly hydrogen
peroxide will move into the product and what areas of the process must be
more carefully examined or improved. Between these two approaches, we
are able to develop operational constraints such as maximum hold times,
minimum aeration times, and required purge cycles to safely and efficiently
process sensitive protein therapeutics on a variety of RABS and isolator
filling lines.
Author Biographies
Aaron Hubbard is a Process Engineer at Genentech, working in the
Pharmaceutical Processing and Technology Development group. His work
focuses on RABS-based filling processes, particularly hydrogen peroxide
mechanics, and he supports a number of Genentech products in both
clinical development and commercial manufacturing. He holds a Bachelor
of Science in Chemical Engineering and a Master of Engineering in
Biomedical Engineering, both from Cornell University.
Kirk Eppler is a Principal Engineer in the Pharmaceutical Processing and
Technology Development group at Genentech. In his 17+ years at
Genentech, he has been a Filling Operations Support Engineer, and a
Process Development Engineer.
He has spent the last nine years focusing on VHP aeration in Isolators and
RABS. He holds a Bachelor’s Degree in Physics from Occidental College.

You might also like