Histological and microscopy FT-IR imaging study on the proliferative activity and angiogenesis in head and neck tumours

Micro Fourier transform infrared spectroscopy enables one to study small samples because of the high quality spectra that can be obtained.

Biochemical and morphological changes between control and pathological tissues of head and neck tumours have been monitored drawing three-dimensional chemical maps of the main vibrational modes in the regions of interest.

Comparison between spectral and histological data shows a satisfactory degree of accordance.

Among all, proliferating and regressive states of the tumours can be identified.


Vibrational spectroscopy can be considered as a fast tool to gain important information about the molecular and supramolecular structure of living tissues with no need of laborious and invasive sampling procedures.

In particular, microscopy FT-IR provides the possibility to perform a qualitative and quantitative analysis even at a molecular level.

In the last ten years, great effort has been made by numerous researchers to evaluate the possibility of using vibrational spectroscopy as a complementary technique in clinical diagnosis and prognosis.

However, the question of whether or not FT-IR spectroscopy can satisfy the requirements of accuracy, ease of use, low cost and early diagnosis, must still be demonstrated.

Nevertheless, many results obtained for various pathological states of human tissues are quite promising: changes in cell structure and composition can be ‘visualised’ in the vibrational spectrum of a biological fluid or tissue even at an early stage of the disease.

Micro FT-IR can join in vivo clinical diagnostics, immunocytochemical analysis and ex vivo histological determinations to set up a suitable protocol to study lesions in human tissues.1–6

To date, our contribution to the relatively abundant number of papers which have appeared on the topic during the last five years, concerns studies on carotid plaque characterisation and FT-IR imaging studies on carotid plaques and pathological thyroid tissues.7–9

Oral carcinomas account for about 5% of head and neck tumours.

The prevalence of tumours of the head and neck (HN) region is rapidly increasing and squamous cell carcinoma (SCC) represents the most frequent malignant tumour of the oral cavity.

The clinical behaviour of oral SCC is difficult to predict based on classical histopathological parameters.

Proliferating activity and angiogenesis provide important information on tumour behaviour that can be used to support traditional histopathology for more accurate tumour characterisation in diagnosis and prognosis.

Recently the identification of molecular markers that can accurately define those neoplasms that will manifest an aggressive clinical behaviour and worse prognosis is becoming particularly important.10

The aim of this study was to realise an experimental protocol, based mainly on immunohistochemical as well as microscopy FT-IR procedures to elucidate the expression of MIB1, CD34 and CD10 in normal mucosa and low and high grade tumours of the head and neck region.


Middle infrared and histological determinations were carried out on thin sections (4 µm thickness) of surgically removed tissues.

One face of each microtome was used for histopathological analysis and the other one for spectroscopic determinations.

FT-IR determinations

Spectral data were achieved with Perkin-Elmer Spectrum GX1 spectrometer equipped with a Perkin-Elmer Autoimage microscope.

For data handling the following software packages were used: Perkin-Elmer Spotlight v.100 and Spectrum v.303, Grams/32 Galactic Corp.

The spectral resolution was 4 cm–1.

The spatial resolution was 20 × 20 µm.

The absorption spectra are the results of 16 scans.

The microtomes were deposited on a steel support and reflectance spectra were collected.

Specific areas of interest were identified by means of the microscope television camera.

Baseline (polynomial line fit) was performed in all cases and curve fitting (Lorentzian character) was used to determine the ratio between bands of interest.

Attribution of the bands was done according to literature data7,9,11.

Histological determinations

(1) Squamous cell carcinoma of the gingival.

All surgical specimens were fixed in 10% neutral buffered-formalin and embedded in paraffin.

In each case 5 µm sections were cut and stained with hematoxylin-eosin (H&E) to define appropriately the histological grading of the different areas of the tumour.

The different areas were heterogeneous with grading to G2 to G3.

(2) Lymph node metastasis.

All surgical specimens were fixed in 10% neutral buffered-formalin and embedded in paraffin.

In each case 5 µm sections were cut and stained with hematoxylin-eosin (H&E) to define the different area of the metastatic lymph node.

(3) Sarcomatoid carcinoma.

The surgical specimen was fixed in 10% neutral buffered-formalin and embedded in paraffin.

In each case 5 µm sections were cut and stained with hematoxylin-eosin (H&E) to define appropriately the histological grading of the different area of the tumour.

The different areas were homogeneous.

In all cases additional staining was performed including CD 34, Mib-1 and CD10.

Results and discussion

A number of sections from 10 specimens for each type of clinical state (control, lymph node, neoplasia) were mapped to determine the composition, the morphology and hence the biochemical changes.

The reproducibility of spectral data was satisfactory even if the complexity of any biological tissue required investigation over a larger number of specimens (which is actually in progress) to ascertain the occurrence of characteristic patterns with a satisfactorily confident safe variance.

In the 1800–1500 cm–1 region the bands mainly arise from proteins, side-chains of amino acids and interfacial zones of phospholipids.

The vibration at 1735 cm–1, due to the CO stretching mode of phospholipids, appears as a single band unless hydrogen bonding, mainly with water molecules, takes place.

The amide I mode, found at 1650–1660 cm–1, mainly arises from protein contributions and can appear as a broad convoluted band.12–15

The overlap of this band with the OH bending modes of water, makes the assignment of secondary structures uneasy.

The amide II is present as a middle band at 1550 cm–1.

The presence of water can be related to a certain degree of inflammatory character of a tissue: if the ratio δOH/amide is much higher than 1 we can hypothesize the occurrence of a certain degree of inflammation.16

The bands at 1450 and 1400 cm–1 are attributable to the protein methyl δasym and δsym bending modes while the band at 1468 cm–1 comes from methylene bending of lipids.17

The bands centred at ca. 1240 and 1080 cm–1 mainly arise from asymmetric (νasymPO2) and symmetric (νsymPO2) vibrational modes of phosphodiester moieties, that are present in phospholipids and nucleic acids.

Peaks at 1340, 1280, 1204 cm–1, commonly present in these spectra, can help, together with the ones at 1083 and 1030 cm–1, to evaluate the presence of collagen.12,13,18

Both bands at 1083 and 1240 cm–1 are also assigned to C–O modes of carbohydrates blocks in collagen and nucleic acids.

The band at 970 cm–1 is attributed to the dianionic PO22– monoester of nucleic acids.

The band around 1165 is due to the C–O–(H) moiety in proteins.

Once again, it is important, to recall the well known finding, that, not only the frequency, but also the intensity and the shape of the band are important factors to be considered in the study of structural and morphological properties of any material and this holds mainly in biological systems.19,20

In this study we report the results on two tumours: the squamous cell carcinoma of the gingiva and the sarcomatoid carcinoma.

The data on the two malignant tissues are compared with the ones on normal tissues and with lymph nodes metastatic adiacent to the neoplasia.

Among all, it appeared important to analyse the intensity maps of representative bands of the main components of the tissue and, in particular: the ester band (1735 cm–1) for lipids, the amide I and II bands (1655 and 1550 cm–1) for proteins, the bands at 1026 cm–1 for collagen and the band at 970 cm–1 for DNA.

Control tissues

Two control tissues (named core and glisaliv) were chosen.

Core refers to the epithelium tissue coating the healthy mucosa: the intensity maps evidenced an homogeneous distribution of a consistent amount of proteins, DNA and small amounts of collagen and lipids.

Glisaliv, refers to the epithelium gland tissue located under the carcinomatous zone in patients with epitheliar tumour.

The tissue shows a substantial homogeneous distribution of proteins and cellular components.

Small amounts of lipids were present in this serosa gland together with few regressive aspects.

As confirmed by histopathology, there are limited zones with a dishomogeneous character both in composition and morphology with inflammation as supported (Fig. 1) by the occurrence of convoluted AI and AII protein bands (due to conformational changes in the secondary structure of tissue proteins) and CO of lipids (mainly due to H-bonding of the ester group with water molecules).

The inflammatory character of zones close to adipose cells, was argued from the δOH/AI ratio by performing the second derivative procedure on bands at 1655 cm–1 (AI) and 1640 cm–1 (bending OH from water).7,8,21

An appreciable amount cholesterol esters (bands at 3007 and 3038 cm–1) was also found.

Squamous cell carcinoma of the gingiva

Spectral findings on this neoplasia originating from the epithelium, evidenced segregated zones with the contemporary occurrence of high DNA, lipid and collagen content in an uniform distribution of protein over the whole section (Fig. 2).

In a pathological area, the presence of lipids and collagen in a cellular zone, is quite common because they can represent regressive states of neoplastic cellular tissues.22

In disagreement with our findings and histopathology concepts, some authors claim that the disappearance of lipids can represent a marker for malignant states in oral tissues23.

Metastatic lymph node of the gingival carcinoma

It exhibits gradients of a higher content of DNA with respect to proteins, and collagen.

In this section the dishomogeneous character of the tumour is evident: in fact, the cellular component is not distributed uniformly: zones with high protein content do not show a corresponding high DNA presence and vice versa.

The contemporary presence of a high content of DNA and collagen indicates the occurrence of proliferating and regressive states of the tumour, respectively.24

Representative recurrent spectra of control and pathological tissues in the case of gingival carcinoma are shown in Fig. 3.

It is conceivable that, in this region, changes in band shape and band intensity are mainly attributable to contributions from nucleic acids during the progression of the disease.24,25

In Fig. 4 infrared spectra of control and pathological sections are shown.

The profile around 1000 cm–1 evidences spectral differences that can be assumed as markers of the disease.

Sarcomatoid carcinoma

Even if it can be classified as a homogeneous tumour, some dishomogeneity comes from the presence of localised amounts of DNA, carbonates (1437, 874 cm–1) and lipids (1735 cm–1).

Various conformations can be present in proteins as evidenced by the broadening of the amide I band.

Representative spectra of the neoplastic region in the tissue are shown in Fig. 5, while the chemical and the contour maps of the main components are reported in Fig. 6.

The upper spectrum in Fig. 5 is the result of a pronounced dishomogeneous character due the contemporary presence of carbonates, lipids and nucleic acids.

A pronounced regressive character of the tumour is localised in the left upper side of the microtome.

This region evidences high contents of DNA, collagen, carbonates, lipids (viewed as a circled distribution) in a uniform content of proteins, that accounts for the existence of a highly infiltrating neoplastic character.

Inspection of the values of the absorbance ratio between the bands at 1460 and 1240, that allows one to evaluate the content of nucleic acids in the tissue, indicates an increase of nucleic acids on going toward malignant tissues.

In fact, the averaged values are: control (core) 0.92, control (glisaliv) 1.02, metastatic lymph node (lfn) 0.56, tumour (gingiva) 0.6 and tumour (sarcoma) 0.57.

The relative content of collagen with respect to nucleic acids, that can be estimated from the absorbance ratio 1026/1083, gave: control (core) ca. 1, control (glisaliv) ca. 1, metastatic lymph node (lfn) 1.4, tumour (gingival carcinoma) 1.7 and tumour (sarcomatoid carcinoma) 0.9.

This last unexpected value needs to be confirmed and explained with further determinations26,27.


The data reported are coherent and reproducible enough to encourage us to achieve data on a larger number of gingival carcinoma and sarcomatoid carcinoma and to open the application to additional pathological states of the oral cavity.

Agreement of spectroscopic and clinical results does not necessarily mean that imaging micro FT-IR is able to take the place of immunocytochemical and histological procedures, because of the complexity of components of biological tissues.

Nevertheless, we think that chemical mapping by FT-IR spectroscopy leads to an early and better understanding of chemical and morphological changes, because it is able to localize and characterize pathological regions at a level of a single cell.

The possibility to detect, even in apparently ‘normal tissues’, spectroscopic modifications due to subtle initial pathological changes, makes this technique a powerful tool for a correct early diagnosis and prognosis.