MICROSCOPY RESEARCH AND TECHNIQUE 72:67–75 (2009)

Microimaging FTIR of Head and Neck Tumors. IV

CARLA CONTI,1 PAOLO FERRARIS,1 MARIA GARAVAGLIA,2 ELISABETTA GIORGINI,1
CORRADO RUBINI,3 SIMONA SABBATINI,1 AND GIORGIO TOSI1*
1
Dipartimento di Scienze e Tecnologie Chimiche, Ancona, Italy
2
Perkin-Elmer, Italia, Monza, Italy
3
Dipartimento di Neuroscienze, Università Politecnica delle Marche, Ancona, Italy

KEY WORDS infrared microspectroscopy; pattern recognition analysis; cluster analysis;

cancer; salivary glands neoplasia
ABSTRACT On continuing our studies on head and neck neoplasia, specimens from salivary
gland tumors have been explored by using infrared microimaging spectroscopy to discern healthy
from neoplastic tissues. Samples with Warthin tumor, epithelial displasia, marginal B-cell lym-
phoma, low-grade adenocarcinoma, and adenoid cystic carcinoma pathologies have been investi-
gated by using conventional light sources. Changes were monitored at the molecular level, probing
spectral markers such as Amide I and II, phosphate, nucleic acids, and carbohydrates vibrational
modes. In all cases, supervised and unsupervised spectral analyses resulted in satisfactory agree-
ment with histopathological findings. Microsc. Res. Tech. 72:67–75, 2009. V 2008 Wiley-Liss, Inc. C

INTRODUCTION and defining characteristic spectral profiles in salivary

The potential role of IR spectroscopy in biomedical
sciences has been exploited to distinguish different bio-
molecules by probing chemical bond vibrations and
using these molecular and submolecular patterns to
define and differentiate pathological from healthy sam-
ples (Jackson and Mantsch, 2002; Jackson et al., 1997;
Kneipp et al., 2000; Lasch and Naumann, 1998).
One of the advantages of FTIR microspectroscopy is
that it can monitor pathological variations even at an
early stage, and it is particularly suitable to characterize
tissue structures and identify tissue pathologies, owing
to the capability of generating maps or images of samples
areas (Toyran et al., 2006). IR spectroscopy provides a
spectral signature of the intensity and a spatial location
of the chemical components, thus highlighting biochemi-
cal changes. The use of multivariate data and artificial
neural network analysis can afford an unsupervised
valid classification of biochemical components (Lasch
et al., 2002b, 2004). Studies on biological specimens
(fluids, cells and tissues) have been carried out either
with instruments employing conventional light sources
or, more recently, with spectrometers using multidetec-
tor devices or synchrotron radiation sources; in the two
latter cases, the spectral quality was improved and the
acquisition time drastically reduced. These tools have
opened new potential frontiers in biomedical research
affording results unavailable by standard histopatholog-
ical analysis (Burattini et al., 2007; Diem et al., 2004;
Romeo et al., 2006). Obviously, due to the complexity of
these systems, a large number of experiments are neces-
sary in order to acquire reliable statistical evidences.
Since the mid 1990s, we have applied infrared spectros-
copy to study pathological states in breast, colon, and,
mainly, oral cavity tissues characterizing various biochem-
ical components with the aim to detect pathological states
(Conti et al., 2005, 2007a,b, 2008; Bruni et al., 2004).
The present contribution aims to further prove the
potentiality of infrared spectroscopy in better isolating
glands attributable to the following pathologies:
1. Warthin tumor: the second most common tumor of
salivary glands. It is a benign tumor composed of
glandular (with solid pattern) and often cystic struc-
tures; the stroma contains a variable amount of
lymphoid tissue with germinal center.
2. Epithelial displasia: an architectural disturbance
accompanied by cytologic atypia.
3. Lymphoma: a low-grade B-cell lymphoma arising in
mucosa-associated lymphoid tissue, usually devel-
oping in the setting of lymphoepithelial sialadenitis.
4. Polymorphous low-grade adenocarcinoma: a malig-
nant epithelial tumor with histologic heterogeneity
and an infiltrative growth pattern. Foci of oncocytic
or mucous cells may be found.
5. Adenoid cystic carcinoma: a basaloid tumor consist-
ing of epithelial and myoepithelial cells in variable
morphologic configurations.
Additionally, we attempted to discriminate ‘‘normal’’
connective tissue from the desmoplastic stroma of the
neoplastic process.
MATERIALS AND METHODS
Sample Preparation
Thirty three samples (tumor bank of the Ospedale
Regionale ‘‘Umberto I’’ of Ancona) from patients with a
diagnosed salivary glands pathology were analyzed.
References ‘‘Bruni et al., 2004,’’ ‘‘Conti et al., 2005,’’ and ‘‘Conti et al., 2007’’ are
part I–III.
*Correspondence to: Giorgio Tosi, Dipartimento di Scienze e Tecnologie
Chimiche, Facoltà di Ingegneria, via Brecce Bianche, 60131 Ancona, Italy.
E-mail: g.tosi@univpm.it
Received 16 May 2008; accepted in revised form 18 August 2008
DOI 10.1002/jemt.20644
Published online 29 September 2008 in Wiley InterScience (www.interscience.
wiley.com).

C
V 2008 WILEY-LISS, INC.
68 C. CONTI ET AL.
The samples were grouped as follows: Warthin tumor
(A), oral epithelium with dysplasia (B), lymphoma (C),
polymorphous low-grade adenocarcinoma (D), adenoid
cystic carcinoma (E), and its corresponding healthy tis-
sue (F).
Tissues were cut into sections of 5 lm thickness by
using a cryo-microtome. Three adjacent sections were
used for both histopathological and FTIR analysis. The
first section was stained with hematoxylin-eosin
(H&E) for histological examination to identify various
regions, whereas the second and the third were depos-
ited on steel and silicon supports for infrared analysis
(Conti et al., 2007b).
Data Processing for IR Microspectroscopy
Spectral data were achieved with the following spec-
trometers: Perkin Elmer Spectrum One and Perkin
Elmer Spotlight 400, equipped with a Perkin-Elmer
Autoimage microscope (with a photoconductive
HgCdTe, MCT, array detector, operating at liquid nitro-
gen temperature which covers the entire IR spectral
range from 4,000 to 700 cm21). The spectral resolution
was 4 cm21. The spatial resolution, depending on the
heterogeneity, ranged from 50 3 50 to 6.25 3 6.25 lm2.
The spectra were the result of 256 scans. Background
scans were acquired and ratioed against the sample
spectrum. On the thin sections, deposited on steel or
silicon supports, reflectance and transmission spectra
were obtained, respectively. Specific areas of interest
were selected by means of the microscope television
camera.
Baseline (polynomial line fit) was performed in all
cases, whereas Second Derivative (9 points), Fourier
Self Deconvolution and Curve Fitting (Gaussian char-
acter) procedures were used to determine the absorb-
ance ratio between bands of interest. All spectra were
scaled for equal intensity in the Amide II absorption.
Attribution of the bands was done according to litera-
ture data (Jackson and Mantsch, 2002; Steudel, 1995).
Spectral images were produced by Perkin Elmer
Auto IMAGE 5.0.1 and by Spectrum IMAGE 1.6.1.
For data handling, the following software packages
were used: Spectrum v.6.3.1 (Perkin-Elmer), Grams AI
(Galactic Corp.), and Pirouette 4.0 (Infometrix Corp.)
for multivariate analysis. Pirouette has already been
used in spectral data treatment of spectra from oral
cavity tissues (Conti et al., 2007b; Tobin et al., 2004).
Hyper View Images Perkin-Elmer software was used to
generate pseudo color maps.
On a number of sections, supervised analysis was
undertaken following histopathological suggestions,
whereas on the remaining specimens blind tests (unsu-
pervised analysis) were performed. For each whole sec-
tion, a number of spectra from 10,000 to 40,000 could
be extracted. The spectra from selected areas were sub-
mitted to hierarchical clustering analysis in the region
1,800–900 cm21 where the main absorptions of biomo-
lecules are located. For the HCA procedure all the spec-
tra were converted in absorbance, baseline linear fit-
ted, smoothed (5 points) vector normalized (Wood et al.,
2004).
The suitability of this method was controlled by run-
ning the HCA on some sections by using second deriva-
tive spectra (Savitzky-Golay, 9 points) and comparing
both clustering results.
SIMCA correlations (by using classes and parame-
ters for HCA procedures) enabled us to discharge out-
liers spectra. The remaining spectra were resubmitted
to HCA and the final grouping was used to extract rep-
resentative spectra of each subcluster by means of
Standard Normal Variate Correction. These spectra
were compared with those extracted point–point from
selected regions of the map (by following the sugges-
tions of the pathologist); with a satisfactory compari-
son, the average spectra were used for the spectral
analysis.
On bands of interest, frequency and absorbance val-
ues were obtained and analyzed statistically.
RESULTS AND DISCUSSION
IR microspectroscopic imaging generates molecular
tissue maps providing a spectral signature of the inten-
sity and a spatial location of the chemical components
of diseased tissues. In the following, the spectral analy-
sis of a representative section (well featuring the spec-
tral behavior of the remaining sections) will be
described in detail for each group of specimens.
The microphotograph of a salivary gland tissue inter-
ested by Warthin tumor (A), H&E stained, is shown in
Figure 1.
Following histological assessments, two different
zones were examined where the H&E staining revealed
a diversified histological composition: a yellow-signed
zone with cystic component, lymphoid tissue and focal
necrosis and a red signed zone with mainly solid tumor.
By the use of HCA (not reported) and PCA analysis,
the 873 and 2,219 spectra, collected from cystic (yellow
square area with neoplastic lesions, CK and CK0 ) and
solid (red rectangular area, SK) regions, were nicely
grouped in subclusters (Figs. 2a,b).
In both regions, necrosis (Ne), lymphoid tissue (Ly),
and vessels (V) were identified too. Representative
spectra are reported in Figures 3a,b.
In the region 3,000–2,800 cm21, differences are evi-
dent in the spectral profiles of the six components
(bands normalized to the intensity of the phosphate
stretching mode at
1,080 cm21) (Fig. 3a): (i) the bands
shapes of SK and CK are more or less the same and
resemble the one exhibited by cellular components; in
both spectra, the lower masymCH3/msymCH2 intensity ratio
with respect to the CK0 one, can be ascribed to a hypo-
methylation process during carcinogenesis (Table 1)
and confirmed also by the same intensity ratio between
SK and the healthy tissue F (Fig. 4); (ii) a discrete pres-
ence of cellular component in CK0 and, mainly in Ne,
can be argued by band broadening; (iii) the lymphoid
tissue (Ly) shows the mCH2sym mode red shifted to 2,858
cm21; (iv) the vessels show proper spectral profiles
where the mCH3sym is hidden by the broad mCH2sym mode
(Conti et al., 2005, 2008; Wong et al., 1991; Yamada
et al., 2002).
It is known that cholesterol, phospholipids and also
creatine are important cellular metabolites with bands
in the region 3,100–2,800 cm21. In particular, the val-
ues of masymCH2/masymCH3 ratio are related to phospholi-
pids peroxidation after oxidative stress and, hence, can
be considered cancer targets for carcinogenesis
(Petibois and Déléris, 2006; Petibois et al., 2007). Also
in our case, the values of the above ratio in tumoral
Microscopy Research and Technique
 MICROIMAGING FTIR OF TUMORS 69

Fig. 1. Photo of a H&E-stained section of a salivary gland tissue

(Warthin tumor, A).

Fig. 3. Representative spectra of the components of tissue A to-

gether with control F in the region 3,000–2,800 cm21 (a) and in the
region 1,800–900 cm21 (b).

TABLE 1. masymCH3/msymCH2 and masymCH2/masymCH3 intensity ratio

on going from SK (representative spectra of Warthin tumor)
Fig. 2. PCA analysis of spectra of cyst (a) and solid cancer
to F (representative spectra of the corresponding
areas (b).
healthy tissue). Overall SD 6 0.1–0.3
SK CK CK0 B C D E F
masymCH3/msymCH2 1.4 1.2 1.6 1.6 1.6 1.5 1.6 1.7
regions SK and CK are higher than in CK0 and, obvi- masymCH2/masymCH3 1.6 1.5 1.3 1.4 1.5 1.2 1.5 1.2
ously, in the spectrum of healthy F (Table 1).
In the spectrum of the solid tumor SK, AI, and AII
vibrational modes exhibit component bands at 1,656
and 1,550 cm21 characteristic of the a-helix conforma- ent only in the spectra of lymphoid tissue, can be
tion of proteins with a ratio %2 and weak bands at attributed to the OH bending vibration of water mole-
1,681, 1,663 and 1,540, 1,513 cm21 attributable to the cule and to the random-coil structure of proteins (Bruni
b-sheet structure. On the contrary, in the normal tis- et al., 1998; Stuart, 2004). The mCO band of the carbonyl
sues (F), the b-sheet content prevails on a-helix (Figs. moiety at 1,744 cm21 (interfacial region of lipids) is
4a,b) (Ganim et al., 2008; Hollosi et al., 1994; Wong almost absent in SK and CK, whereas it is clearly visi-
et al., 1993b). This spectral behavior could arise from a ble in Ly, Ne, and V. All these spectral differences in
change in the secondary structure of proteins during the position of components bands are clearly detected
carcinogenesis (Conti et al., 2007a; Ismail et al., 1992; by second derivative results (Fig. 5a). In SK and CK,
Parker, 1971; Rigas and Wong, 1992) the bands at 1,458 (methyl bending) and 1,398 cm21
On going from SK to CK and CK0 , AI band gradually (methylene bending and carboxylate stretching) have a
broadens and falls at lower frequencies (1,662 and proper profile, where the higher intensity of the latter
1,656 cm21). Broadening and red shift of amide bands mode is again in favor of a hypomethylation process
are also evident in Ly, Ne, and V with a AI/AII ratio along with carcinogenesis (Conti et al., 2008; Hollosi
% 1. The relevant component band at 1,647 cm21, pres- et al., 1994; Rigas et al., 1990; Wong et al., 1993a,b). In

Microscopy Research and Technique

70 C. CONTI ET AL.
Fig. 4. Peak fitting in the region 1,800–1,450 cm21 (absorbance in arbitrary units, A.U.) of: (a) SK
(Warthin tumor, tissue A) and (b) control (tissue F).
Fig. 5. Second derivative of representative spectra of the six com-
ponents of A in the region 1,750–900 cm21. [Color figure can be
viewed in the online issue, which is available at www.interscience.
wiley.com.]
the region 1,250–800 cm21, the bands are mainly at-
tributable to phosphodiester groups of nucleic acids
and phospholipids in tissues and cells (Fig. 5b). If, as in
our case, the mCO band of the carbonyl moiety at
1,740 cm21 is lacking and if the secondary structure of
proteins is mainly a-helical, the bands at
1,240 and
at
1,083 cm21, arising frommasymPO2- and msymPO2-
stretching modes, can be attributed mainly to nucleic
acids (Wong et al., 1991). In this region, all the bands
of SK, CK and, at some extent, CK0 , are blue shifted
with respect to those of Ly, Ne, and V; in particular, the
blue shift of C Oasym stretching mode (from 1,163 to
1,172 cm21) can be due to a pathological phosphoryla-
tion process (Ci et al., 1999; Jackson and Mantsch,
2002). It is not surprising that in tumor spectra, the ra-
tio between the band at 1,085 (mainly phosphate
stretching modes from nucleic acids and proteins) and
1,055 cm21 (carbohydrate C O H stretching mode
like mucin) is higher than one as a consequence of a
consumption of mucin and a substitution of C O H
by CO P (Lasch et al., 2002a; Pacifico et al., 2003).
Fig. 6. Infrared spectra of SK, CK, CK0 , and F in the region
1,100–1,020 cm21. [Color figure can be viewed in the online issue,
which is available at www.interscience.wiley.com.]
The decrease of C O H moieties results in a reduced
H-bonding, thus explaining the blue shift of bands in
the spectra of tumor samples (Fig. 4b). The position
and, mainly, the shape of the downfield C OP band
from DNA can be the consequence of different confor-
mational features (Pacifico et al., 2003).
In previous articles, the value ratio 1,030/1,080 has
been hypothesized as a conceivable spectral marker of
neoplastic changes in the sense that, during proteins
phosphorylation, there is a consumption of glycogen,
and hence, the ratio decreases (Gazi et al., 2004). In our
case, the glycogen band is detectable (DII 5 points) only
in CK0 that can signify a complete consumption in CK
and SK. The band of carbohydrates centered at
1,050 cm21 is sharp in CK0 and CK, blue shifted at 1,055
cm21 in F and is a doublet at 1,055 and 1,047 cm21 in
SK. On these spectral features, which depend on glucide
composition, some contribution probably comes from S 5
Microscopy Research and Technique
 Fig. 7. Microphotograph of limited areas of solid tumor (a) and cyst (b–d).

Fig. 8. Image from cluster of cystic (a) and solid tumor area (b): green (SK), red (CK and CK0 ,
mainly), blue (Ne and Ly), dark green (holes).

Fig. 9. Photo of an H&E-stained section of the C tissue (lymphoma) and relative PCA analysis: green
(connective), red (tumor), and blue (lipids).
72 C. CONTI ET AL.
O groups, from various compositions of proteins, and
from modifications of disulphide links (Fig. 6).
It is reported that an excessive increase in lactic acid
content can be related to glucose consumption by ma-
lignant cells to supply oxygen decrease during cells car-
cinogenesis. This behavior can be proved also in our tis-
sues, where the presence of the band of lactic acid at
1,127 cm21 is evident only in CK and SK and increases
from the former to the latter (John, 2001; Petibois
et al., 2007).
A visual inspection on high power 50 3 50 slides of
limited areas of cyst and solid tumor (Fig. 1), suggested
that not only biochemical changes but also tissue mor-
phology and packing are responsible for the spectral
differences on going from Ly to CK0 , CK and, finally, to
the homogeneous and dense SK neoplasia (Figs. 7a–d).
In Figures 8a,b, the pseudo color maps are reported,
in which each pixel (6.25 3 6.25 lm2) is characterized
by a proper spectrum; by using HCA analysis, similar
spectra are grouped in subclusters (shown in the maps
by different colors) assigning a specific zone for various
components and reproducing the histological biochemi-
cal distribution and architecture.
In the case of marginal B-cell lymphoma (C), the
spectral and subsequent HCA (not reported) and PCA
analysis classified tumor, connective, and lipids compo-
nents (Fig. 9).
The spectrum of this neoplasia (see Fig. 12) resulted
similar to the one of the solid tumor (A) as regard to
the profile and the absorption of bands in almost the
entire spectral region, except for the position of AI and
AII (found 9 and 7 cm21 to lower frequencies), for the
peak intensity ratio of CH2 and CH3 at 2,960/2,854 and
1,452/1,397 cm21 and for a red shift of 6 cm21 of DNA
band at 970 cm21 (Fig. 12 and Table 2).
Also in D tissues (polymorphous low-grade adenocar-
cinoma), HCA and PCA (not reported) showed three
clusters assigned to the cancer, to the connective and to
the cancer with a significant presence of mucus. With
respect to the solid SK, the spectrum of the cancer (see
ahead, Fig. 10), showed different 2,960/2,854. AI/AII
and 1,453/1,398 intensity ratios (Table 2).
In addition, Amide bands appeared broad for the
presence of components bands to lower frequencies.
The presence of mucine is argued by the bands found
at 1,376, 1,317, 1,120, 1,072, 1,035 cm21. The DNA
band at 970 cm21 is practically absent (Fig. 12) (Lasch
et al., 2002b, 2004).
All the sections from patients with an adenoid cystic
tumor (E) resulted constituted by muscular, connec-
tive, necrotic, lipidic, and tumoral components (multi-
variate analysis not reported). The representative
spectrum of the tumor resulted identical to the one of
the solid tumor, SK, with some minor dissimilarities
mainly for the lower intensity of phosphate bands of
nucleic acids (1,240, 1,080 e 970 cm21) probably due to
the presence of cells with a lower mitotic activity (Fig.
12) (Lasch et al., 2002a; Tobin et al., 2004).
To point out spectral differences between eosino-
philic and ialine connective tissue, produced during
carcinogenesis, high-resolution spectra (6.25 3 6.25
lm2) were taken into consideration (Fig. 10). Contrary
to that found in eosinophile connective tissue, analo-
gies with the tumoral spectrum were found in ialine
connective tissue spectrum: a-helix prevalent on b-
TABLE 2. dasymCH3/dsymCH3 and AI/AII intensity ratio
in the healthy tissue (F) and tumoral tissues (A–E).
Overall SD 6 0.1–0.3
A B C D E F
dasymCH3/dsymCH3 0.8 0.8 0.9 1.5 0.8 1.2
AI/AII 2.0 2.2 2.5 1.9 2.2 1.7
sheet conformation (a-helix/b-sheet 5 2.9 6 0.2 and 0.8
6 0.1 in ialine and eosinophile connective tissue,
respectively), the profile of CH2 and CH3 bending
modes, the C OH band found at 1,172 (due to a
change of C OH into C OP), a higher intensity of the
component bands on the right side of the convoluted
band centered at 1,080 cm21. In Figure 10, the spec-
trum of the healthy tissue F is also reported. Spectral
differences of F with the tumoral spectrum are clearly
evident: the vibrational mode C OH of sugars found,
as expected, at 1,164 cm21; a 1,080/1,240 ratio <1 and
the DNA band at 960 cm21 almost absent (Fig. 10). (Ci
et al., 1999; Conti et al., 2005; Gazi et al., 2004, 2007).
From tissues B, containing oral displastic epithelium
together with healthy zones characterized by glands
with alveolar shape, two main average spectra were
isolated: one well resembling the spectrum of the solid
tumor, SK (with differences in Amide I and II band
shapes, only) and the other with a high content of gly-
coproteins (mucine) (Figs. 11a,b).
For a comparison among the spectra representing
various neoplasia, some spectral windows resulted rel-
evant: (i) the ratios between methyl and methylene
vibrational modes the position; (ii) the profile and the
peak ratio of Amide I and II; (iii) the presence of the
band at 1,171 cm21; (iv) the shape of the band centered
at about 1,080 cm21 and its intensity ratio with the one
at 1,240 cm21; (v) the intensity of the band at 970 cm21
(Fig. 12). The comparison of spectral with histological
classification resulted more than satisfactory.
To further verify the reliability of our classification,
all the spectra of SK were mixed with those representa-
tive of the other oral cavity tumors, always isolated
from the corresponding clusters, obtaining a new HCA
in the spectral region (Fig. 13).
From the dendogram, the various classes of neopla-
sia resulted well grouped and in a conceivable
sequence. The nearness of D with A subclusters is not
strange due, as also stated by pathologists, to the anal-
ogous heterogeneity of the two displasia. The existence
of a unique subcluster for CK and CK0 is in favor of a
very smooth difference in biochemical composition and,
probably, of a prevalent incidence of the morphological
assembly to the spectral profile. Owing to the similar-
ity of the representative spectra, the grouping of E
with B is also feasible.
CONCLUSIONS
The results from salivary gland specimens, pre-
sented here, once again demonstrate the capability of
infrared microspectroscopy imaging, in combination
with multivariate data analysis, to point out even
subtle biochemical and morphological changes, distin-
guishing various kinds and grades of neoplasia in tis-
sues from a specific region of the human body. To study
Microscopy Research and Technique
 MICROIMAGING FTIR OF TUMORS 73

Fig. 10. Comparison between representative spectra in the region 1,800–900 cm21 of neoplastic and
connective zones of E tissue (adenoid cystic carcinoma) with the healthy tissue F. [Color figure can be
viewed in the online issue, which is available at www.interscience.wiley.com.]

Fig. 12. Representative spectra in the region 1,800–900 cm21 of

different pathologies: Warthin tumor A (1), low-grade adenocarci-
noma D (2), epithelium with displasia B (3), adenoid cystic carcinoma
E (4), lymphoma C (5), and the corresponding healthy tissue F (6).
[Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.]

Fig. 11. (a) Microphotograph of a section of B tissue (oral displas-
tic epithelium) where the displastic (blue panel) from the healthy
(black panel) epithelium can be distinguished and (b) their represen-
tative spectra. [Color figure can be viewed in the online issue, which
is available at www.interscience.wiley.com.]
biological systems, a rapid images acquisition at a
high-spatial resolution is requested that can be fully
satisfied with the introduction of new conventional
(thermal) light sources. For example, Spotlight system

Microscopy Research and Technique

74 C. CONTI ET AL.
Fig. 13. HCA analysis in the region 1800–900 cm21 of tumoral spec-
tra. [Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.]
approaches spectral performances of synchrotron light
sources, operating at and down the diffraction limit,
with excellent spectral quality and very short acquisi-
tion time (Diem et al., 2004).
On considering this and our previous reports, the po-
tentiality of spectral analysis in the study of various
kinds of head and neck neoplasia can be outlined. Obvi-
ously, the complexity of human tissues requires more
experimental efforts and statistical evidences to unam-
biguously formulate a diagnosis of a pathology at a mo-
lecular level, especially in the early stage and hence, to
constitute a valid and synergic support to the classical
histopathological screening.
REFERENCES
Bruni P, Alò F, Giorgini E, Tosi G. 1998. Fourier transform infrared
spectroscopy. In: Proceedings of the 11th ICOFTS International
Conference Fourier Transform Spectroscopy, Am Inst of Phys.
Athens, Georgia. pp. 298–301.
Bruni P, Conti C, Giorgini E, Pisani M, Rubini C, Tosi G. 2004. Histo-
logical and microscopy FT-IR imaging study on the proliferative ac-
tivity and angiogenesis in head and neck tumours. Faraday Discuss
126:19–26.
Burattini E, Malvezzi-Campeggi F, Chilosi M, Conti C, Ferraris P,
Monti F, Sabbatini S, Tosi G, Zamò A. 2007. FPA micro spectral
imaging of non-Hodgkin lymphomas. J Mol Struct 834–836:170–
175.
Ci YX, Gao TY, Feng J, Guo ZQ. 1999. Fourier transform infrared
spectroscopic characterization of human breast tissue: Implications
for breast cancer diagnosis. Appl Spectrosc 53:312–315.
Conti C, Giorgini E, Pieramici T, Rubini C, Tosi G. 2005. FT-IR mi-
croscopy imaging on oral cavity tumours. II. J Mol Struct 744–
747:187–193.
Conti C, Ferraris P, Giorgini E, Sabbatini S, Tosi G, Anastassopou-
lou J, Arapantoni P, Boukaki E, Konstadoudakis S, Theophanides
T, Valavanis C. 2007a. Micro-FT-IR spectroscopic studies of breast
tissues. In: Tsakanov V, Wiedemann H, editors. Brilliant Light in
Life and Material Science. Yrevene, Armenia: Springer. pp. 273–
278.
Conti C, Ferraris P, Giorgini E, Pieramici T, Possati L, Rocchetti R,
Rubini C, Sabbatini S, Tosi G, Mariggiò A, Lo Muzio L. 2007b.
Microimaging FT-IR of oral cavity tumours. III: Cells, inoculated
tissues and human tissues. J Mol Struct 834–836:86–94.
Conti C, Ferraris P, Giorgini E, Sabbatini S, Tosi G, Anastassopoulou
J, Arapantoni P, Boukaki E, Konstadoudakis S, Theophanides T,
Valavanis C. 2008. FT-IR microimaging spectroscopy: A comparison
between healthy and neoplastic human colon tissues. J Mol Struct
881:46–51.
Diem M, Romeo M, Matthäus C, Miljkovic M, Miller L, Lasch P. 2004.
Comparison of Fourier transform infrared (FTIR) spectra of indi-
vidual cells acquired using synchrotron and conventional sources.
Infrared Phys Tech 45:331–338.
Ganim Z, Chung HS, Smith AW, Deflores LP, Jones KC, Tokmakoff A.
2008. Amide I two-dimensional infrared spectroscopy of proteins.
Acc Chem Res 41:432–441.
Gazi E, Dwyer J, Lockyer N, Gardner P, Vickerman JC, Miyan
J, Hart CA, Brown M, Shanks JH, Clarke N. 2004. The com-
bined application of FTIR microspectroscopy and ToF-SIMS
imaging in the study of prostate cancer. Faraday Discuss 126:
41–59.
Gazi E, Dwyer J, Lockyer N, Gardner P, Shanks JH, Roulson J, Hart
CA, Clarke N, Brown M. 2007. Biomolecular profiling of meta-
static prostate cancer cells in bone marrow tissue using FTIR
microspectroscopy: A pilot study. Anal Bioanal Chem 387:1621–
1631.
Hollosi M, Majer ZS, Ronai AZ, Magyar A, Medzihradszky K, Holly S,
Perczel A, Fasman GD. 1994. CD and Fourier transform IR spectro-
scopic studies of peptides. II. Detection of b-turns in linear peptides.
Biopolymers 34:177–185.
Ismail AA, Mantsch HH, Wong PTT. 1992. Aggregation of chymotryp-
sinogen: Portrait by infrared spectroscopy. Biochim Biophys Acta
1121:183–188.
Jackson M, Mantsch HH. 2002. Pathology by infrared and Raman
spectroscopy. In: Chalmers JM, Griffiths PR, editors. Handbook of
vibrational spectroscopy, Vol. 5. Chichester: Wiley. pp. 3227–
3245.
Jackson M, Sowa MG, Mantsch HH. 1997. Infrared spectroscopy: A
new frontier in medicine. Biophys Chem 68:109–125.
John AP. 2001. Dysfunctional mitochondria, not oxygen insufficiency,
cause cancer cells to produce inordinate amounts of lactic acid: The
impact of this on the treatment of cancer. Med Hypothesis 57:429–
431.
Kneipp J, Lasch P, Baldauf E, Naumann D. 2000. Detection of patho-
logical molecular alterations in scrapie-infected hamster brain by
Fourier transform infrared (FT-IR) spectroscopy. Biochim Biophys
Acta 1501:189–199.
Lasch P, Naumann D. 1998. FT-IR microspectroscopic imaging of
human carcinoma thin sections based on pattern recognition tech-
niques. Cell Mol Biol 44:189–202.
Lasch P, Boese M, Pacifico A, Diem M. 2002a. FT-IR spectroscopic
investigations of single cells on the subcellular level. Vibration
Spectrosc 28:147–157.
Lasch P, Haensch W, Lewis EN, Kidder LH, Naumann D.
2002b. Characterization of colorectal adenocarcinoma sections
by spatial resolved FT-IR microspectroscopy. Appl Spectrosc
56:1–9.
Lasch P, Haensch W, Naumann D, Diem M. 2004. Imaging colorectal
adenocarcinoma using FT-IR microspectroscopy and cluster analy-
sis. Biochim Biophys Acta 1688:176–186.
Pacifico A, Chiriboga LA, Lasch P, Diem M. 2003. Infrared spectros-
copy of cultured cells. II. Spectra of exponentially growing, se-
rum-deprived and confluent cells. Vibration Spectrosc 32:107–
115.
Parker FS. 1971. Application of Infrared Spectroscopy in Biochemis-
try, Biology and Medicine. New York: Plenum Press.
Petibois C, Déléris G. 2006. Chemical mapping of tumor progression
by FT-IR imaging: Towards molecular histopathology. Trends
Biotechnol 24:455–462.
Petibois C, Drogat B, Bikfalvi A, Déléris G, Moenner M. 2007. Histo-
logical mapping of biochemical changes in solid tumors by FT-IR
spectral imaging. FEBS Lett 581:5469–5474.
Rigas B, Wong PTT. 1992. Human colon adenocarcinoma cell lines dis-
play infrared spectroscopic features of malignant colon tissues.
Cancer Res 52:84–88.
Rigas B, Morgello S, Goldman IS, Wong PTT. 1990. Human colorectal
cancers display abnormal Fourier-transform infrared spectra. Proc
Natl Acad Sci USA 87:8140–8144.
Romeo M, Mohlenhoff B, Diem M. 2006. Infrared micro-spectroscopy
of human cells: Causes for the spectral variance of oral mucosa
(buccal) cells. Vibr Spectrosc 42:9–14.
Schrader B. 1995. Infrared and Raman Spectroscopy. Methods and
applications. In: Schrader B, editor. VCH: Weinheim.
Stuart B.H. 2004. Infrared Spectroscopy: Fundamentals and Applica-
tions. Chichester, England: Wiley.
Microscopy Research and Technique
 MICROIMAGING FTIR OF TUMORS
Tobin MJ, Chesters MA, Chalmers JM, Rutten FJM, Fisher SE, Symonds
IM, Hitchcock A, Allibone R, Dias-Gunasekara S. 2004. Infrared mi-
croscopy of epithelial cancer cells in whole tissues and in tissue culture,
using synchrotron radiations. Faraday Discuss 126:27–39.
Toyran N, Lasch Peter P, Naumann D, Turan B, Severcan F. 2006.
Early alterations in myocardia and vessels of the diabetic rat heart:
An FTIR microspectroscopic study. Biochem J 397:427–436.
Wong PTT, Papavassiliou ED, Rigas B. 1991. Phosphodiester stretch-
ing bands in the infrared spectra of human tissues and cultured
cells. Appl Spectrosc 45:1563–1567.
Wong PTT, Wong RK, Fung Kee Fung M. 1993a. Pressure-tuning
FTIR study of human cervical tissue. Appl Spectrosc 47:1058–1063.
Microscopy Research and Technique
75
Wong PTT, Lacelle S, Yazdi HM. 1993b. Normal and malignant
human colonic tissues investigated by pressure-tuning FT-IR spec-
troscopy. Appl Spectrosc 47:1830–1836.
Wood BR, Chiriboga L, Yee H, Quinn MA, McNaughton D, Diem M.
2004. FTIR mapping of the cervical transformation zone, squamous
and glandular epithelium: Search for the elusive infrared cancer
markers. Gynecol Oncol 93:59–68.
Yamada T, Miyoshi N, Ogawa T, Akao K, Fukuda M, Ogasawara
T, Kitagawa Y, Sano K. 2002. Observation of molecular changes
of a necrotic tissue from a murine carcinoma by Fourier-Trans-
form Infrared Microspectroscopy. Clin Cancer Res 8:2010–
2014.