A comparison of different breath sample collection approaches and its impact on results quality.

Analysis of exhaled breath is an exciting and fast developing field of
expertise. Many research groups focus nowadays on developing accurate,
effective and reproducible techniques to non-invasively diagnose patients with
all kinds of disorders. Different methodological approaches associated with
this topic may at the end cause a big variation in results as well as wrong
associations or conclusions while referring to work of others. Thus, techniques
and approaches should be compared to see if they generate relevant differences.
One of the issues on which research teams involved in this subject seem to
disagree is how to collect breath samples. Since the composition of exhaled
breath is influenced endogenously and exogenously many researches are trying to
come up with ways to limit exogenous sources of compounds as much as possible.
Below different ways of collection approaches are described.
Available literature gives an insight into many different approaches but not in
comparing them with each other. A common technique is to collect a background
sample of the ambient air at the same time as the subject who has reached
equilibrium with the room air donates a breath sample. The alveolar gradient is
then calculated [1, 2, 3]. In this case data interpretation is based on the
assumption that equilibrium exists for all VOCs between the body and the
ambient air which is not a known fact. Other teams give to theirs subjects a
clean air supply to reduce the background [4, 5, 6]. This technique is not only
reagent consuming but also compromises the portability of testing apparatus.
The issue is also whether dead space air should or should not be separated from
alveolar air. Alveolar air collection is thought to be most accurate to
characterize processes within the body, giving the best picture of blood born
volatiles. There are many techniques proposed to collect alveolar air, very
often including self-made equipment prepared for that purpose. Not unified way
of collection gives opportunity for creating not comparable results. The issue
if subjects should inhale normally through their nose or rather through their
mouth while using a nose clip to limit nasal contaminations is another question
present in breath research. Collection of mixed air, meaning total breath
including dead space air and alveolar air is a technique used by many teams
including ours [7,8]. Its simplicity reduces the mistakes introduced by more
complex techniques. Beside the fact, that lack of restrictions in a way of
exhalation is convenient and patient-friendly, not forced exhalation results in
a breath characterized by a normal state and behavior. No stress is induced
which can affect processes within the body. Even though there are clear
positives of using this approach there is still a lot of discussion if the
method is not negligible towards important factors.
All these questions have not been solved so far in the literature and they
bring us towards this study. So far the approach of our team is to sample mixed
breath (alveolar plus dead space) by exhalation into a 3 L Tedlar bag without
any special directives. The contents of the sampling bag is transferred to a
stainless steel carbon packed sorption tube which later on is thermally
desorbed and analysed using gas chromatography-time-of flight mass spectrometry
(TD-GC-tof-MS). Data-pre-processing, data-mining and statistical analysis is
used to separate the useful information in the breath samples from the
redundant information, thus being a tool that deals with problems that other
teams seems to deal with by putting more limitations during collection time.
In this study we intend to compare our sampling methodology with most common
used, more strict and complex ways of sampling to determine if the results are
different. The research question of the study will be: how garlic/fish oil
breath-o-print is extracted when using 4 different methodologies. Data will be
analyzed as two class problem, differences between Class 1 and Class 2 will be
established, using multilevel-PLS-DA in 4 set-ups * one for each collection
approach. The analysis takes into account the paired data structure underlying
the cross-over design. In total, 4 different multilevel-PLS-DA classification
models will be obtained. For each classification model a set of discriminatory
VOCs, sensitivity, specificity and overall correct prediction of Class 1 and
Class 2 will be received. Note, that sensitivity, specificity and overall
correct prediction will be calculated for the validation set.

To evaluate exhaled breath sampling approaches and their impact on results
quality. To establish if sampling procedures impact specificity and sensitivity
in extracting information of interest * in our case impact of garlic/fish oil
supplements on volatile organic compounds composition.

Randomised, cross-over study. Approximately 50 volunteers will be included in
the study and randomly assign to Group 1 or Group 2. Volunteers will be asked
to produce up to 4 breath samples using different collection methods. Group 1
will have four samples collected without any prerequisite with regard to diet.
Proceeding the sampling session, Group 2 will be asked to consume 2 garlic/fish
oil capsules each (1 in the evening a day before the collection and 1 - 3h
before the collection will occur). In the successive week Group 1 will have
four samples collected after consumption of garlic/fish oil capsules (1 in the
evening a day before the collection and 1 - 3h before the collection will
occur), while Group 2 will have four samples collected without any prerequisite
with regard to diet. For every variation in sampling method this will deliver a
series of samples of the same subjects with and without garlic/fish oil
supplementation.
The sensitivity and specificity obtained for the four different set-ups
determine which methodology performs best.
During 1 week:
Group 1: No garlic/fish oil supplementation *These samples will produce the
'normal' volatile organic compounds profiles.
Group 2.: At 10.00 p.m. of day -1 (prior to collection) and 3h before the
collection time on day 0 (collection day) each participant will take one
garlic/fish oil capsule.
These samples will show the *garlic/fish oil specific' volatile organic
compounds profile.

The following samples will be collected from both groups:
Sample 1: exhalation into a Tedlar bag, no special directions- our present way
of collecting
Sample 2: exhalation into a Tedlar bag with a nose clip
Sample 3: exhalation into a Teflon tube connected to a three way valve; the
first part of exhaled breath discarded as assumed to be a dead space air
(around 150ml of tidal volume) and second part collected into a Tedlar bag as
being alveolar air;
Sample 4: same as sample 3 but with a use of nose clip

During 1 week:

The groups switch responsibilities. Samples collected following the same
scheme.

Subject will be ask to consume 1 capsule of garlic/fish oli in the evening, a
day before sample collection and 1 capsule 3h before sample collection.

One of the advantages of exhaled breath analysis is its non-invasive and
patients/participants friendly nature. In comparison to other diagnostic tools,
breath collection is not known to cause troubles even when used for children or
sick subjects.
Participation in this study involves breathing in a 3L plastic bag during a few
minutes. In total maximally 8 bags will be collected: 4 in week 1 (no
supplementation) and 4 in week 2, after garlic/fish oil supplementation.
Consumption of garlic/ fish oil it is thought to have a beneficial influence
on a health but may cause some discomfort in more sensitive individuals. Since
those are only two tablets we do not expect to see this kind of reaction.