Endogenous Fluorescence Analysis under Deep UV Excitation to

Discriminate Human Brain Tumor Tissue

Difference between Glioblastoma and Healthy Control Tissue

F. Poulon

, F. Jamme

, A. Ibrahim

, C. Métais

, P. Varlet

4,5

, M. Juchaux

, B. Devaux

3,5

M. Refregiers

and D. Abi Haidar

1,6

IMNC Laboratory, UMR 8165-CNRS/ IN2P3, Paris-Saclay University, 91405 Orsay, France

DISCO beamline, Synchrotron SOLEIL, Gif-sur-Yvette, France

Neurosurgery Department, Sainte-Anne Hospital, Paris, France

Neuropathology Department, Sainte-Anne Hospital, Paris, France

Paris Descartes University, Paris, France

Paris Diderot University, Sorbonne Paris Cité, F-75013, Paris, France

Keywords: Endomicroscopy, Brain Tumours, Deep UV, Synchrotron SOLEIL, Spectroscopy, Wide-field Imaging.

Abstract: In order to build a multimodal nonlinear endomicroscope to image brain border during operation, our group

is building an optical database on brain biopsy tissues analysis collected with excitation panning from deep

UV to near infrared. This paper focuses on the results from deep UV excitation of endogenous fluorescence

from glioblastoma and control human brain samples. The samples were imaged and spectrally analysed. The

excitation wavelengths were tuned from 275 nm to 340 nm. Two promising indicators to discriminate

tumorous tissue from the control were found. A preliminary correspondence between fluorescence images

and histological H&E staining open a huge door to confirm results with a medical expertise.

1 INTRODUCTION

Cancer is one of the major causes of death

worldwide, in 2012, there were 14 million new cases

and 8.2 million cancer-related deaths(“Cancer

Statistics,” n.d.). Brain tumour may only be the 17

most common cancer in the world, recent studies in

the US showed that it is one of the most dangerous

one, inducing the most important number of cancer

related death in the population aged between 15 to

39 years (“Brain Tumor Statistics | American Brain

Tumor Association,” n.d., “Worldwide data | World

Cancer Research Fund International,” n.d.).

Improving survival rate and recovery from such

tumours is a constant key challenge for the medical

community. One of the major issues to tackle in

brain surgery is the extent of resection. Indeed, most

of the brain tumours tend to infiltrate quickly the

surrounding healthy area, and if actual technologies,

such as scanner and IRM give a clear margin around

necrosis, they do not give any information on the

rate of infiltration from the surrounding. The

surgeons today follow the rule of maximum possible

resection; stopping before touching any vital

functional part of the brain(Sanai and Berger, 2008).

The aim of our project is to give new optical contrast

and information on the tissue during the operation in

order to identify this infiltrated region. To succeed in

such project we are building a new endomicroscopic

tool combining different imaging modalities to give

the more precise answer to the surgeon. In parallel to

the instrumental development, it was crucial to

discriminate tissue nature in order to confirm the

power of multimodal optical analysis. For that

purpose, we are creating a database of optical

signature from the different tumour tissue and their

corresponding control using different optical

contrast. This database will help to give a discrimi-

natory answer on the nature of the tissue and to

define different indicators. This database groups

different optical response: (i) fluorescence imaging,

(ii) spectroscopy, (iii) lifetime fluorescence analysis

and (iv) second harmonic generation imaging. These

data were acquired using an excitation ranging from

the deep UV to the near infrared.

In this article we will focus on the results from the

deep UV excitation of tissues. Under this excitation

window we were able to make full field imaging and

152

Poulon F., Jamme F., Ibrahim A., MÃl’tais C., Varlet P., Juchaux M., Devaux B., Refregiers M. and Abi Haidar D.

Endogenous Fluorescence Analysis under Deep UV Excitation to Discriminate Human Brain Tumor Tissue - Difference between Glioblastoma and Healthy Control Tissue.

DOI: 10.5220/0006103601520157

In Proceedings of the 5th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2017), pages 152-157

ISBN: 978-989-758-223-3

precise spectral measurements. Emission-collection

matrixes of different endogenous molecules were

measured and analysed. Some molecules of interest

were spectrally followed using different excitation

wavelength. Fluorescence images were compared to

the hemotoxylin and eosin (H&E) staining standard

of the Sainte-Anne hospital. Interesting structures

were identified in optical images and matched finely

with anatomopathologist diagnostic indicator.

2 MATERIELS AND METHOD

2.1 Imaging and Spectroscopic

Measurements

The data have been acquired on the DISCO

beamline at the Synchrotron SOLEIL. Two

microscopic set-ups are available on the beamline

and were used in this study.

The samples are excited with the continuous

emittance from the DISCO beamline bending

magnet at 275nm and between 310 and 340 nm.

First the samples were analysed under a full-field

microscope (Zeiss Axio-observer Z-1) with a x40

glycerine immersion objective (Zeiss Ultrafluar, NA

0.6). Emitted fluorescence was collected via a PIXIS

1024 BUV camera (Princeton, USA) through four

bandpass filters: 307-323nm, 323-357nm, 408-

438nm and 435-455nm (Semrock), with an

integration time of 10s. This set-up is completely

controlled by an open source microscopy software

Micro-Manager (Edelstein et al., 2014). Spectral

measurements were recorded with an inverted

Olympus IX71 microscope stand with homemade

DUV lenses. Light detection was collected through a

DUV lens and an adjustable pinhole. Then, the

fluorescence emission spectrum was projected onto

a −70◦C peltier-cooled iDus CCD detector (Andor)

of 1024 × 256 pixels with a 26 × 26 μm pixel size

and a 26.6 × 6.7 mm detector size.(Jamme et al.,

2013)

2.2 Samples

A strong collaboration has been established with the

departments of anatomopathology and neurology of

the Saint-Anne Hospital (Paris, France). The result

of this collaboration was the access to a large cohort

of human biopsy samples. The protocol of experi-

mentation was approved by the Institutional Review

Board of Sainte Anne Hospital (Ref CPP S.C.3227).

For this specific study, ten samples have been

selected, five of them were from glioblastoma

tumour and the other five were control sample from

epileptic surgery. The DISCO beamline microscopic

set-up requiring 10 microns slices of each sample, a

very strict process was put in place. The biopsies

were fresh samples conserved at -80°C. A

specialized transport service (360°, France) brought

the selected cohort to the IMNC lab. Samples were

then conserved at -80°C. These samples were kept

under -20° freezer 24 hours before being cut with a

cryostat (Leica CM 1950). Then 10 microns from

samples were deposited on quartz coverslips. Serial

10 microns slices were cut for different colorations.

Once on the coverslips or glass slides the samples

were fixed at hundred percent alcohol solution. The

quartz coverslips were brought to the Synchrotron in

a box dedicated to microscopic slides transportation

2.3 Histological Staining

One of the slices was used to perform the gold

standard H&E staining. Staining protocol was

provided by Sainte-Anne hospital anatomopatho-

logical staff. The sample has to go through

hydratation, hematoxylin and Eosin coloration,

deshydratation and toluene fixation. The stained

samples were then imaged in a slide scanner. A

comparison between UV results and H&E staining

was possible and performed thanks to the expertize

of the Sainte-Anne anatomopathologists.

2.4 Data Analysis

Spectral data were acquired with the Labspec

software, on each image a square of 12 by 12 points

with 5 microns between each points was selected, a

spectrum was acquired for every point with a 10s

integration time. Preliminary treatments were

applied to suppress noise and detector dead pixel

using a Matlab script developed by the DISCO team.

The treated data are then fitted with a homemade

Matlab script. This script was already used and

published in the visible (Haidar et al., 2015) and

near infrared range and readapted here for the deep

UV spectral analysis. Four molecules fluorescence

emission were fitted: Tyrosine, tryptophan, collagen

and NADH. These molecules were chosen through a

complete review of the literature and the expertise of

the DISCO scientists (Croce and Bottiroli, 2014;

Jamme et al., 2013). The first three components

were fitted by a Gaussian curve, the NADH

component was fitted by an experimental curve

established during previous studies of the group

(Haidar et al., 2015).

Endogenous Fluorescence Analysis under Deep UV Excitation to Discriminate Human Brain Tumor Tissue - Difference between

Glioblastoma and Healthy Control Tissue

153

3 RESULTS

The deep UV excitation range excites four different

molecules: Tyrosine, Tryptophan, Collagen and the

NADH (Nicotinamide adenine dinucleotide) the

acquired emission spectra were fitted in regards of

these four molecules. An example of a fitted

spectrum is given in the figure 1a. It shows that the

main contributing fluorescent molecule in the deep

UV range is the tryptophan, its response is two times

higher than the other identified molecules. The mean

of the fitted curve in each group (glioblastoma and

control) were calculated and plotted on the figure 1b,

in order to compare the two types of sample. The

first thing we observe is that the intensity is 20%

lower in Glioma tissue, this phenomenon seems to

be in adequacy with some of the literature (Butte et

al., 2005; Palmer et al., 2003).

Figure 1: (a) Example of experimental data fitted with our

homemade script. (b) Comparison of the mean spectra for

the two groups: Glioblastoma and Control samples.

In the figure1b, we can hint that the ratio

between tryptophan and collagen change with the

type of tissue. To precise the tryptophan-collagen

ratio, for each type of tissue, we measured it in all

the spectra recorded for the five samples. The results

were then presented in a boxplot using the software

R, the ratio was calculated from the max of emission

obtained in the Matlab fit and rearranged in boxplot,

see figure 2b. A significant difference appeared

between the mean of the two boxplots, to confirm

this hypothesis a Mann-Whitney analysis was

performed on the ratio data. The hypothesis of

identical groups, gave a p value of 3.6.10

-4

which is

lower than 5.10

-2

and rejected the hypothesis. This

statistical result gives more confidence to confirm

that the collagen-tryptophan ratio can discriminate

glioblastoma from control samples.

A spectral analysis was accomplished using

different excitation wavelengths. This study was

performed from 310 to 340 nm by step of 5 nm. We

could follow the behavior of different endogenous

molecules through the excitation and go further in

the exploration of the characteristics of the tissue.

From the spectra, we obtained a closer look to the

maximum emission of NADH component to look for

the most appropriate excitation wavelength. Results

are shown in Figure 3.

Figure 2: (a) Illustrate the chosen ratio for a statistical

analysis of the tissues. (b) Box plots of the ratio.

This graph shows that the NADH component

increases with the wavelength. This results is

PHOTOPTICS 2017 - 5th International Conference on Photonics, Optics and Laser Technology

154

explained by an optimal excitation wavelength at

345nm(Jamme et al., 2013), nevertheless here in

tumorous tissue, the maximum of emission is at an

excitation at 325nm, this shift could be a new

indicator of change in cancerous cells.

Figure 3: Variation of NADH maximum emissions as a

function of the excitation wavelength.

Our collaboration with the Sainte-Anne hospital

gives us access to a histological analysis. The deep

UV wide-field images have been compared to the

H&E staining, figure 4. Deep UV fluorescence

images as presented in the figure 4 are the

combination of four channels: Tyrosine for red

channel, Tryptophan for Green channel, Collagen

for Blue channel and the NADH for yellow channel,

all combined with an identical coefficient. While

pointing out specific structures in each type of tissue

is difficult, we were still able to extract some

information; indeed the area showing high density of

cells on the H&E images appears darker on the deep

UV images. The tryptophan being the main

component (green images) these darker areas could

correspond to a loss of tryptophan, but also to an

increase of blue channel, corresponding to collagen.

Figure 4: Comparison between H&E staining images, (a,c)

and wide-field fluorescence (b,d) in control tissue (a,b)

and in glioblastoma (c,d). The scales are 250 (a), 200

(b,c), 100 nm (d).

4 DISCUSSION

This study was accomplished thanks to the

straightforward collaboration between physicist,

neurosurgeon and anatomopathologists team of

Sainte-Anne hospital and DISCO beamline Staff of

Synchrotron SOLEIL.

In this paper we performed a preliminary study

on two groups of biopsy samples: a healthy control

group and glioblastoma tissues. For this preliminary

study, we choose a tumorous group in an advanced

stage that shall present major differences compared

to the control. This choice shall allow us to find

indicators of discrimination more easily. Thereafter

we will be able to analyze infiltrated tissues, which

have a lower density of cancerous cells, with

analysis tools strengthened by this preliminary work.

Tissues were analyzed under deep UV excitation,

through imaging and spectroscopic techniques. Deep

UV is a well-known energy range to look at

endogenous fluorescence molecules of interest to

discriminate the nature of tissue. However, the

literature lack of analysis on human tissues and the

different types of tumor. It will be completed by an

analysis at visible and near-infrared excitation to

cover the whole spectrum and establish relationships

between the endogenous molecules.

In the paper we were able to highlight two

indicators that discriminate the nature of tissue: the

fluorescence intensity and the ratio collagen-

tryptophan.

The fluorescence intensity was already found

useful in our previous study on the grade of

meningioma in the visible and near infrared (Zanello

et al., 2016) and appears so as one of the promising

way to find an optical signature on tumorous brain

tissues. However, the spectral response is controlled

by numerous environmental and experimental

factors such as molecules concentration or laser

power, creating a huge experimental uncertainty and

needing as much parameters as possible into

consideration to make a realistic comparison

between signals intensities. In the literature a lot of

groups have already started to work with a spectral

response of endogenous fluorescence(Chorvatova

and Chorvat, 2014; Croce and Bottiroli, 2014),

looking for changes in the quantity of molecules

present between types of tissues. In this paper and

our future works we want to correlate the spectral

data to other modalities such as lifetime

measurement or second harmonic generation to

build a matrix of optical characteristics for each type

of tissue.

Endogenous Fluorescence Analysis under Deep UV Excitation to Discriminate Human Brain Tumor Tissue - Difference between

Glioblastoma and Healthy Control Tissue

155

The second indicator presented in this article is a

ratio on two major components of the tissue, the use

of a ratio eliminates all the previous bias enumerated

before and will bring more reliable results. Finding a

change in Tryptophan-Collagen ratio seems realistic

in tumorous tissues, knowing that tryptophan is

linked to vascular region and the tumorous tissue

present increased vascularity. Necrosis tissues in

glioblastoma are poor in collagen such as control

tissues and the change in tryptophan can be

highlighted through this ratio, however in other

tumorous tissues such as metastasis, a collagen

matrix spreads to organize cells migration inducing

both an increase of tryptophan and collagen,

therefore this ratio will not change as significantly as

in glioblastoma, giving false negative. It will be

interesting to look at other ratio and increase the

number of tumor types in the cohort.

The data showed that in this excitation range we

were able to fit the NADH component, this molecule

plays an essential role in metabolism, as a coenzyme

in redox reactions. And appears in the literature as a

major indicator in endogenous fluorescence. Its

behavior in tissues under visible and two photon

excitation has been well documented (Huang et al.,

2002; Skala et al., 2007). Articles looked at its cross

section over the excitation range or the redox ratio.

Knowing the role of this component we followed it

in deep UV, with a study over the excitation

wavelength from 310 to 340nm. The maximum in

excitation for the NADH in solution is at 345nm, so

we should get a curve increasing with the

wavelength. In our result we noticed a decrease at

340nm in the glioblastoma tissue. It could be either a

new indicator for cancerous tissue or just an

experimental artifact, due to the fact that this

measurement has been done on a very small number

of sample. Increasing the statistic of this analysis

could give significance to this result and highlight an

important phenomenon in tumorous metabolism.

Wide field images of endogenous fluorescence

allow us to correlate an area on each sample to the

H&E staining, the gold standard in histology to

validate the tumorous nature of a sample. This

correlation was possible thanks to the help of

anatomopathologists from Sainte-Anne hospital.

Correlation allowed us to demonstrate that high cells

density area in H&E images correspond to darker

area in wide-field images. Green channel in images

represent Tryptophan filter, these areas could

correspond to a loss of tryptophan or an increase if

the other channels, especially the collagen one.

Correlation with two-photon imaging could be an

interesting way to find more information.

All this promising resulfts encourage to increase

the cohort in order to have a better statistic on the

results.

ACKNOWLEDGEMENTS

This Work as a part of the MEVO and IMOP project

was supported by “Plan Cancer” program founded

by INSERM (France), by CNRS with “Défi

instrumental” grant, and the Institut National de

Physique Nucléaire et de Physique des Particules

(IN2P3).

We would like to thank Synchrotron SOLEIL for

beamtime under project #20160206. Thanks also to

PIMPA Platform partly funded by the French

program “Investissement d’Avenir” run by the

“Agence Nationale pour la Recherche” (grant

“Infrastructure d’avenir en Biologie Santé – ANR –

11-INBS-0006”).

REFERENCES

Brain Tumor Statistics | American Brain Tumor

Association [WWW Document], n.d. URL

http://www.abta.org/about-us/news/brain-tumor-

statistics/ (accessed 9.6.16).

Butte, P. V., Pikul, B. K., Hever, A., Yong, W. H., Black,

K. L., Marcu, L., 2005. J. Biomed. Opt. 10, 064026.

Cancer Statistics [WWW Document], n.d. Natl. Cancer

Inst. URL http://www.cancer.gov/about-cancer/under

standing/statistics (accessed 9.6.16).

Chorvatova, A., Chorvat, D., 2014. Tissue fluorophores

and their spectroscopic characteristics, in: Marcu, L.,

French, P., Elson, D. (Eds.), Fluorescence Lifetime

Spectroscopy and Imaging. CRC Press, pp. 47–84.

Croce, A. C., Bottiroli, G., 2014. Eur. J. Histochem. 58.

Edelstein, A. D., Tsuchida, M. A., Amodaj, N., Pinkard,

H., Vale, R. D., Stuurman, N., 2014. J. Biol. Methods

1, 10.

Haidar, D. A., Leh, B., Zanello, M., Siebert, R., 2015.

Biomed. Opt. Express 6, 1219–1233.

Huang, S., Heikal, A. A., Webb, W. W., 2002. Biophys. J.

82, 2811–2825.

Jamme, F., Kascakova, S., Villette, S., Allouche, F., Pallu,

S., Rouam, V., Réfrégiers, M., 2013. Biol. Cell 105,

277–288.

Palmer, G. M., Keely, P. J., Breslin, T. M., Ramanujam,

N., 2003. Photochem. Photobiol. 78, 462–469.

Sanai, N., Berger, M. S., 2008. Neurosurgery 62,753–766.

Skala, M. C., Riching, K. M., Gendron-Fitzpatrick, A.,

Eickhoff, J., Eliceiri, K. W., White, J. G., Ramanujam,

N., 2007. Proc. Natl. Acad. Sci. 104, 19494–19499.

Worldwide data | World Cancer Research Fund

International [WWW Document], n.d. URL

PHOTOPTICS 2017 - 5th International Conference on Photonics, Optics and Laser Technology

156

http://www.wcrf.org/int/cancer-facts-figures/world

wide-data (accessed 9.6.16).

Zanello, M., Poulon, F., Varlet, P., Chretien, F.,

Andreiuolo, F., Pages, M., Ibrahim, A., Pallud, J.,

Dezamis, E., Abi-Lahoud, G., Nataf, F., Turak, B.,

Devaux, B., Abi-Haidar, D., 2016. J. Biophotonics

n/a-n/a.

Endogenous Fluorescence Analysis under Deep UV Excitation to Discriminate Human Brain Tumor Tissue - Difference between

Glioblastoma and Healthy Control Tissue

157