Breath Testing in Laryngeal Cancer- Comparing in Situ Cancer and Advanced Cancer
Worldwide there are 130 000 new larnx cancers diagnosed annually resulting in 82 000 deaths
.Survival after diagnosis of larynx cancer depends on initial stage. For T3N0Mo laryngeal
cancers 5-year survival ranges from 59 to 66%. Patients survivals are as follows: receiving
either chemoradiation (59.2%), irradiation alone (42.7%) ,patients after surgery with
irradiation (65.2%) and surgery alone (63.3%)  By contrast in early stage larynx cancer
survivals range from 90-100%. Tamura et al reported therapeutic outcomes of 130 cases with
laryngeal cancer treated at Kyoto University Hospital between 1995 and 2004 In all, 121
males and 9 females were involved. Their ages ranged from 40 years to 92 years (average 66
years). All tumors were squamous cell carcinoma - arising at the glottis in 111 cases, the
supraglottis in 18, and the subglottis in 1 case. Most glottic cancers (77.5%) were
classified as stage I or II, while most supraglottic cancers (77.8%) were at stage III or
IV. Stage I/II cancers were basically treated by conventional radiotherapy (60-66 Gy) and
twice-daily hyperfractionated radiotherapy (70-74 Gy), respectively, attempting to preserve
the larynx. Total laryngectomy with neck dissection was performed in the treatment of stage
III/IV cases. Five-year disease-specific survival rates were 100%, 96%, 100%, and 68% for
stage I, II, III, and IV, respectively. Five-year laryngeal preservation rates were 98%,
100%, 86%, 0%, and 0% for T1a, T1b, T2, T3, and T4 of glottic cancer, respectively. Local
recurrence occurred in five cases of stage I/II glottic cancer, which was successfully
Chera et al  reported excellent treatment outcomes of definitive radiotherapy (RT) for
early-stage squamous cell carcinoma (SCCA) of the glottic larynx. The median follow-up for
survivors was 12 years. Five-year Local Control rates were as follows: T1A, 94%; T1B, 93%;
T2A, 80%; and T2B, 70%. Multivariate analysis revealed that overall treatment time greater
than 41 days (p = 0.001) and poorly differentiated histology (p = 0.016) adversely affected
LC. Five-year rates of ultimate LC with laryngeal preservation were: T1A, 95%; T1B, 94%,
T2A, 81%; and T2B, 74%. Overall therefore there is a high cure rate.
Endoscopic laser resection can also have an excellent outcome in early stage larynx cancer.
Schrivers et al reported survival analysis on 100 patients with T1a glottic carcinoma
treated with CO(2) laser surgery (n = 49) or radiotherapy (n = 51). No significant
differences in local control and overall survival were found. Ultimate 5-year laryngeal
preservation was significantly better in the CO(2) laser surgery group (95% vs 77%, p =
.043). Patients with T1a glottic carcinoma with normal/diminished mucosal wave treated with
CO(2) laser surgery had a significantly better laryngeal preservation rate than patients
treated with radiotherapy.
Staging The larynx is divided into three anatomic regions: 1. Supraglottis (suprahyoid
epiglottis, infrahyoid epiglottis, aryepiglottic folds (laryngeal aspect), arytenoids, and
ventricular bands (false cords) )2, Glottis ( true vocal cords, including anterior and
posterior commissures) and 3. Subglottis ( subglottis, extending from lower boundary of the
glottis to the lower margin of the cricoid cartilage) Volatile organic compound (VOC) breath
testing in cancer detection The concept for VOC testing is that VOCs, mostly alkanes and
aromatic compounds, are preferentially produced and exhaled by cancer patients and can be
used as accurate markers of malignancy[14,15]. As early as 1971, testing on normal breath
identified more than 100 volatile organic compounds In the 1980s Gordon and Preti used
mass spectroscopy and gas chromatography to identify specific alterations in the profile of
volatile organic compounds in the breath of lung cancer patients. In two papers in 1999
and 2003, Phillips further refined this original data to identify a group of 9 volatile
organic compounds which were highly sensitive and specific for the presence of lung cancer
[13,17]. The concentration of these alkane and methylalkane oxidative stress products was
reduced in the breath of lung cancer patients. The mechanism of this alteration in breath
volatile compound profile in lung cancer is unknown. One hypothesis is that lung cancer
patients have accelerated clearance of VOCs generated by oxidative stress, and that this is
due to heightened production of cytochrome p450 as a result of exposure to tobacco smoke
components in genetically predisposed individuals. Whatever the mechanism, the potential of
VOC breath testing in early case finding warrants further investigation.
The detection of VOCs in breath for the purpose of diagnosis has a long history. Ancient
Greek physicians already knew that the aroma of human breath could provide clues to
diagnosis. The astute clinician was alert for the sweet, fruity odor of acetone in patients
with uncontrolled diabetes; the musty, fishy reek of advanced liver disease; the urine-like
smell that accompanies failing kidneys; and the putrid stench of a lung abscess . Modern
breath analysis started in the 1970s when researchers, using gas chromatography (GC),
identified more than 200 components in human breath.
In terms of study methods, breath testing has been the focus of both cross-sectional and
longitudinal studies [18,19]. Cross-sectional studies have investigated exhaled biomarkers
as a function of disease, both as biomarkers of disease state and as predictive markers. In
cross-sectional studies, a control group is compared with a patient or diseased group, and
breath markers are analyzed to identify qualitative or quantitative differences between the
two groups. The differences established in this way should be large enough to enable
clinically relevant predictive use of breath markers Oxidative stress is a condition in
which cells are damaged as the result of a chemical reaction with oxidative agents such as
oxygen-derived free radicals. Free radicals damage components of cell membranes, proteins,
or genetic material by "oxidizing" them—the same chemical reaction that causes iron to rust.
Reactive oxygen species (ROS), such as the superoxide anion (O2-) or the hydroxyl radical
(OH-), act physiologically as defense mechanisms against microbial attack [20,21]. Under
healthy conditions, ROS activity is restricted to limited regions of external attack or
inflammation and is well balanced by antioxidant protection of the body. However, in some
diseased states, the balance between ROS activity and protection may be impaired when
antioxidant systems are overwhelmed or exhausted . Whenever ROS activity takes place in
an uncontrolled manner, the organism itself will be damaged by oxidative stress.
Phillips and coworkers  investigated alveolar gradients (i.e., the abundance in breath
minus the abundance in room air) of C4 to C20 alkanes and monomethylated alkanes in the
breath as tumor markers in primary lung cancer. They concluded that a breath test for C4 to
C20 alkanes and monomethylated alkanes provided a rational new set of markers that
identified lung cancer in a group of patients with histologically confirmed disease. The
analytical methodology was described in 2003 , where it was reported that amongst
smokers and ex-smokers there was a sensitivity for malignancy of 86% (55/64) and a
specificity of 83% (19/23). This compared with sensitivity and specificity in non smokers of
66% (2/3) and 78% (14/18). Overall therefore the VOC breath test was not affected by
Changes in breath VOC patterns are independent of the size of the lung cancer in that T1
tumours (<3cm) have a similar breath pattern of abnormality to T4 tumours , raising the
possibility that VOC abnormalities may even be detectable at the preneoplastic (severe
dysplasia or carcinoma in situ) stage. It describes a comparison between 212 controls
without lung cancer and 195 patients with primary lung cancer. The breath test was as likely
to be abnormal in stage 1 disease as in stage 4 disease. This implies firstly that as a
screening tool VOC breath testing has potential to detect operative curable cases. Secondly,
it implies that oxidative changes leading to altered breath VOCs are an early feature of
lung cancer development, and that the method may therefore detect the presence of
preneoplastic lesions in the bronchial tree.
The strength of VOC breath testing is the simplicity of methodology and specimen collection.
Patients breathe into a portable collection apparatus tube for 5 minutes. This requires
only tidal breathing and therefore presents no difficulty even for patients with pulmonary
In a 2006 review Lam and Shaipanich  looked forward to the possible role of breath
testing as this tool: "For example using a biomarker with a sensitivity of 85% and a
specificity of 81%,( breath testing- my insert) at a disease prevalence of 2.7%, instead of
screening every person in the cohort with spiral CT and fluorescence bronchoscopy, only 21%
of the cohort needs to have the CT and fluorescence bronchoscopy."
Breath testing in Laryngeal cancer In a recent article Hakim et al described for the
first time that Head and Neck cancer can be identified by breath testing.
Alveolar breath was collected from 87 volunteers (HNC and LC patients and healthy controls)
in a cross-sectional clinical trial. The discriminative power of a tailor-made Nanoscale
Artificial Nose (NA-NOSE) based on an array of five gold nanoparticle sensors was tested,
using 62 breath samples. The NA-NOSE signals were analysed to detect statistically
significant differences between the sub-populations using (i) principal component analysis
with ANOVA and Student's t-test and (ii) support vector machines and cross-validation. The
results showed breath testing could clearly distinguish between (i) HNC patients and
healthy controls, (ii) LC patients and healthy controls, and (iii) HNC and LC patients. The
GC-MS analysis showed statistically significant differences in the chemical composition of
the breath of the three groups.
The Cyranose / Enose VOC testing with the eNose allows groups of patients to be tested for
differences or similarities of breath signal [25-27]. A single expired breath is collected
in a sample bag then a pump draws the sample into the device where it passes over 32
electronic sensors. Over 400 possible chemicals affect these sensors in different ways, ad a
pattern of electronic signals is generated. It is the distribution of the electric signals
across the 32 sensors which gives the pattern. Software within the device determines which
of the 32 sensors is giving the strongest signal in each test, and uses these sensor results
in a combination result called a factor. This is known as Principal Component analysis. When
comparing 3 groups of patients the software will generate 2 factors for each breath sample
and plot these on a graph. Where a group of patients has a distinctive signal the factor
analysis will clump that group together, at a certain "distance" on the graph from the other
group. The greater the distance t(Mahalobinus distance) the more different the groups are.
Numerous authors have published data on this type of analysis for a variety of disease
states, particularly lung cancer. This approach is very easy technically and leads to
further study of the individual VOCs which are responsible for the signal. It is likely
however based on results from other tumours that a combination of VOCs are present in
different amounts in cancer patients as opposed to a single VOC. The ENose approach has not
been applied in Head and Neck cancer patients and nor has there been any report of detection
of in situ cancer.
Because of the step wise development of squamous cell cancer it is quite possible that In
situ cases would be clumped together with advanced cases of Squamous cell carcinoma, and
that both would be different to smoking controls. Alternatively it may be the signal in the
early cases is different from later stages but different from controls as well, so that both
early and advanced cases could be diagnosed from breath testing.
It is known that both CT and VOC breath test can detect stage 1 cancer of the lung which has
at least a 50% cure rate. There is potential however that VOC can detect even earlier
stages of lung cancer, such as in-situ-carcinoma which when properly staged and treated has
over 95% long term cure rate. The goal of any screening study is to find cancers at a
curable stage. VOC breath testing combined with fluorescence/NBI bronchoscopy and CT could
perhaps achieve this desired goal.
It is possible that VOC testing will ultimately be used in larynx cancer screening either as
the first step (high negative predictive value) or as a second line test to further evaluate
equivocal results of screening low dose CT chest. Also, we have expertise in NBI and
fluorescence bronchoscopy and our focus is on the management of the type of early lesions
found by this approach.
Summary of background:
Detecting Laryngeal cancer at an in-situ or T1 stage allows ablative treatment (Transoral
Laryngeal surgery or Radiotherapy) with excellent long term outcomes.
Smoking is the main risk factor, and whilst some early Laryngeal cancer patients have
symptoms many do not.
The possibility of screening heavy smokers would be useful particularly if it detected
cancers at a pre-clinical stage.
Breath testing has been shown to detect a range of cancers including lung and breast cancer,
by detecting a signature pattern of exhaled volatile organic compounds (VOC).
Importantly with lung cancer, the VOC signal is the same across all TNM stages of disease,
(I though IV).
- VOC breath testing of patients with early stage (Tis/T1) Larynx cancer will be the same
as that for stage 3 or 4 Larynx cancer
- both Early and Advanced Larynx Cancer VOC signals will be different from smoking
Methods Breath testing will be done using the Cyranose ENose in Thoracic Medicine
Established protocol for testing from Lung Cancer study, x 2 single expirations into a
collection bag Ideally this would be best done when a lesion has been seen by ENT surgeon
but BEFORE it is biopsied (to avoid confounding effects on VOCs of tissue disruption by the
biopsy) The ENose software allows comparisons of 3 groups of 10 subjects each - 10 Tis/T1,
10 advanced Larynx Ca, 10 smoking controls with demonstrated normal ENT and tracheobronchial
Patients would have a routine panendoscopy before treatment with NBI to exclude
concommittant second primary disease either in head and neck or Bronchial tree
Potential Significance Proof of principal of screening detecting highly treatable lesions
Supportive data for similar tumours, particularly Squamous cell carcinoma of the bronchus,
viz the benefits of early detection
Procedures All will be done in the Thoracic Mediine department Breath test sampling for
VOCs: A portable breath collection apparatus will capture VOCs in a slow vital capacity
exhalation breath sample , using Standard Operation Procedure process already in place. Two
samples are taken, with the patient breathing gently on a mouthpiece with a nose clip on for
5 minutes each time. Patients should be
1. Nil by mouth
2. No smoking for 12 hours
3. No alcohol for >24 hours Breath will be processed by 1. The Enose and 2. Gas
Time Perspective: Prospective
Difference in breath test signal for diagnosis
Statistical differences can be obtained using software in the Enose- Mahlobinis distance after Principle component analysis of breath signals to separate controls from in situ cancer and from advanced cancer.
David I Fielding, FRACP MD
Royal Brisbane and Womens Hospital
Australia: Human Research Ethics Committee