Journal of Cranio-Maxillo-Facial Surgery xxx (2014) 1e3

Contents lists available at ScienceDirect

Journal of Cranio-Maxillo-Facial Surgery journal homepage: www.jcmfs.com

Fractal dimension of time-resolved autofluorescence discriminates tumour from healthy tissues in the oral cavity Jan Klatt a,1, Carola E. Gerich b,1, Alexander Gröbe a, Jörg Opitz b, Jürgen Schreiber b, Henning Hanken a, Georg Salomon c, Max Heiland a, Lan Kluwe a, d, *, Marco Blessmann a a

Department of Oral and Maxillofacial Surgery, University Medical Center Hamburg-Eppendorf, Hamburg, Germany Fraunhofer Institute for Non-Destructive Testing, Laboratory of Optical Diagnostics, Dresden, Germany Martiniclinic, Prostate Cancer Center, University Medical Center Hamburg-Eppendorf, Hamburg, Germany d Department of Neurology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany b c

a r t i c l e i n f o

a b s t r a c t

Article history: Paper received 15 May 2013 Accepted 17 December 2013

Early detection and complete resection of oral carcinomas is of crucial importance for patient survival. This could be significantly improved by developing a non-invasive, sensitive and real-time detection technique. Time-resolved autofluorescence measurement is state-of-the-art technology originally developed for non-destructive inspection of material. In this study, we measured time-resolved autofluorescence in tumours and healthy tissues of the oral cavity ex vivo and calculated the corresponding fractal dimension which was significantly higher in tumours than in healthy tissues (1.8 vs. 1.6, P < 0.001, unpaired t-test) with non-overlapping 95% confidential intervals 1.88e1.84 and 1.57e1.69, respectively. Very high specificity (86%) could be reached at 100% sensitivity. The area under the curve was 99%, further suggesting the superior prediction potential of fractal dimension based on time-resolved autofluorescence spectra. Ó 2013 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.

Keywords: Real-time cancer cell detection Oral squamous cell carcinoma Surgical margins Fractal dimension Time-resolved fluorescence spectra

1. Introduction Oral squamous cell carcinoma (OSCC) is among the most common malignancies in humans with an annual incidence of approximately 10,000 in Germany and 400,000 world-wide (Thiele et al., 2011; Hertrampf et al., 2012). Prognosis for OSCC patients is generally poor with a five-year survival rate of slightly above 50% which has been hardly improved over the past 30 years (Scully and Bagan, 2009). Chemo- and radiotherapies have only limited effects. Early detection and total resection of oral carcinomas are the most effective measures to reduce mortality (Feichtinger et al., 2010). There has been scarcely any progress in the past 20 years, with most patients diagnosed at an advanced stage of the disease. A sensitive and non-invasive method for real-time detection of tumour tissue would change this unsatisfactory situation and timeresolved autofluorescence measurement may be one such method.

* Corresponding author. Laboratory for Tumor Genetics, Clinical Neuroscience, O48, Room 418, University Medical Centre Hamburg-Eppendorf, Martinistr aße 52, D-20246 Hamburg, Germany. Tel.: þ49 40 7410 58267; fax: þ49 40 7410 59665. E-mail address: [email protected] (L. Kluwe). 1 Shared first authorship: these authors contributed equally to this work.

Originally, time-resolved autofluorescence measurement was developed as a non-destructive method for detecting damages and micro-damages in material in the field of material science. Recently, this measurement has also been assessed for its potential in discriminating lesions from healthy tissues in various tumours (Butte et al., 2005; Yong et al., 2006; Meier et al., 2010; Farwell et al., 2010). Most of these studies used peak position and intensity of autofluorescence as the measuring parameters, which vary largely depending on surface conditions (moisture, temperature) of the tissues and on distance between tissue and the excitation light source. Furthermore, previous studies aimed to follow decay of single or several fluorophores including collagen, NADH and porphyrin which have different decay wavelength ranges (Butte et al., 2005; Yong et al., 2006; Meier et al., 2010; Farwell et al., 2010). We have developed a special algorithm to describe the nonexponential decay of global time-resolved autofluorescence, instead of those of single fluorophores. This algorithm is based on a model for fractal structures which reflects collective molecular interaction and metabolism of living tissues (Gerich, 2008; Gerich et al., 2009). In this study, we evaluated the potential of this calculated fractal dimension (DF) of time-resolved autofluorescence in discriminating tumours from healthy tissues of the oral cavity.

1010-5182/$ e see front matter Ó 2013 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jcms.2013.12.004

Please cite this article in press as: Klatt J, et al., Fractal dimension of time-resolved autofluorescence discriminates tumour from healthy tissues in the oral cavity, Journal of Cranio-Maxillo-Facial Surgery (2014), http://dx.doi.org/10.1016/j.jcms.2013.12.004

J. Klatt et al. / Journal of Cranio-Maxillo-Facial Surgery xxx (2014) 1e3

A total of 37 specimens were included in this study: 22 healthy tissues from the oral cavity and 15 oral squamous cell carcinomas. The specimens were taken during tumour resection at the Department of Maxillofacial Surgery of the University Medical Center Hamburg-Eppendorf. A fragment was cut out from the centre of a tumour. Healthy tissue was taken from oral mucosa. The specimens were immediately shock frozen and stored in liquid nitrogen to minimize tissue-degradation. All tumours were histologically examined for malignancy. The study protocol was approved by the institutional review board; all patients were informed about the purpose of the study and then gave their written consent. The specimens were shipped on dry ice to the Fraunhofer Institute for Non-Destructive Testing in Dresden where the measurements were carried out. Time-resolved fluorescence was measured using a mobile laser-induced multi-emission spectrometer (set-up with modules from Lasertechnik, Berlin, Germany). A high-pressure pulsed nitrogen laser (MNL 200, Lasertechnik) at 337.1 nm was applied as the excitation source, and the detecting module consisted of a monochromator and an intensifier. Spectra of autofluorescence were recorded in the visible range. For the measurement, the specimens were placed on a cold plate, and autofluorescence was initially measured at two points on the upper side. The specimens were then turned upside down for two repeated measurements of the other side. The whole procedure took approximately 3 min for each specimen. A total of 155 time-resolved fluorescence measurements were carried out. The obtained data sets were evaluated with a special algorithm developed at Fraunhofer IZFP using the program MathCad which calculates the non-exponential decay behaviour of autofluorescence in a complex molecule collective model for time-timecorrelation: I(t, l) ¼ I0(l)B(l)*(tt0)a 1. Fractal dimension can be calculated as DF ¼ 2a/2 whereas a can be obtained from the above equation. DF has a value between 1 and 2. For each specimen with multiple adequate data sets, mean DF was used for subsequent analysis. DF of healthy and neoplastic specimens was compared using unpaired t-test (two-tailed). Area under the curve was calculated by plotting the sensitivity against 1specificity at various DF.

A

2,0

tumor

1,9

Fractal dimension

2. Material and methods

1,8 1,7 1,6

healthy tissue

1,5 1,4 1,3

Samples

B

1,9

1,8

1,7

1,6

1,5

Healthy tissues

C

Tumor

1,2 1,0

Sensitivity

2

0,8 0,6 0,4

3. Results 0,2

A total of 155 measurements of time-resolved autofluorescence were carried out on 22 healthy and 15 tumour specimens. Among them, 100 spectra from 14 healthy tissues and 11 tumours were of adequate quality for subsequent calculation of fractal dimension DF. The remaining 55 measurements had poor signal/noise ratio and thus could not be analysed. The number of adequate spectra (and thus the calculated DF) pro sample varied from 1 to 4. For specimens with multiple DF, the mean and maximum coefficient of variation were 5.8% and 14.5%, respectively, indicating good reproducibility of the measurement. Clearly separated clusters of DF values for healthy and tumour tissues are evident (Fig. 1A,B), with means of 1.63 and 1.8, respectively. The 95% confidential intervals (1.57e1.69) and (1.81e1.84) did not overlap with each other. Unpaired t-test also revealed significantly (P < 0.001, two-tailed) higher DF in the tumours than in the healthy tissues. A DF threshold of 1.75 identified 11/11 tumours and 2/14 healthy tissues as tumour, corresponding to 100% sensitivity and 86% specificity. The area under the curve was 99%, again suggesting superior prediction power of DF (Fig. 1C).

0,0

0,0

0,2

0,4

0,6

0,8

1,0

1,2

1-specificity Fig. 1. Calculated fractal dimension (DF) of time-resolved autofluorescence. A: Mean and standard deviation of DF of each of the 14 healthy (open circles) and 11 neoplastic (filled circles) specimens. Data points without standard deviation represent samples from each which data of adequate quality were obtained from less than 3 measurements. The horizontal broken line indicates the DF threshold value of 1.75 which identifies all tumours (100% sensitivity) and only 2/14 (86% specificity) of the healthy tissues as tumours. B: Box plot of DF of healthy and tumourous tissues. The horizontal lines are median and the height of the boxes are 25e75% intervals. The vertical lines mark the 95% confidence intervals and outliers were marked with circles. C: ROCcurve: Area under the curve of DF.

A minor shift of peak wavelength was observed between healthy and neoplastic tissues which were not significant. No difference was observed in the intensity of autofluorescence between tumour and healthy tissues (data not shown).

Please cite this article in press as: Klatt J, et al., Fractal dimension of time-resolved autofluorescence discriminates tumour from healthy tissues in the oral cavity, Journal of Cranio-Maxillo-Facial Surgery (2014), http://dx.doi.org/10.1016/j.jcms.2013.12.004

J. Klatt et al. / Journal of Cranio-Maxillo-Facial Surgery xxx (2014) 1e3

4. Discussion In the present study, we demonstrated a promising potential of single autofluorescence measurement with subsequent fractaldimension-calculation in discriminating neoplastic tissues from healthy tissues in the oral cavity. We did not observe any significant difference in peak position and intensity of autofluorescence between tumour and healthy tissues in our study, in contrast to some previous studies which reported potential value of these parameters (Butte et al., 2005; Yong et al., 2006; Meier et al., 2010; Farwell et al., 2010). Peak position and intensities of autofluorescence are generally not reliable parameters because they vary largely depending on surface conditions (moisture, temperature) of the tissues and on the distance between tissue and the excitation light source, and are further influenced by intensity fluctuation of the light source. In contrast, fractal dimension describes global molecular interaction of the tissue and is thus more robust. Nevertheless, both form and size of the specimen influence the measurement of time-resolved autofluorescence and also DF. In the present study, these parameters varied to a large extent, probably explaining why approximately 1/3 of the measurements lacked adequate quality. Intensity of the excitation light optimized for good signal/noise ratio for a certain specimen or for a certain area on a specimen can be too weak and too strong for other specimens or other areas on the same specimen if they are thicker or thinner, resulting in poor signal/noise ratio. Improving uniformities of form and size is therefore a key point for improving sensitivity and specificity of the method. Recently, we applied a biopsy needle to sample specimen from prostate carcinoma for autofluorescence measurement. In this way, the obtained specimens have defined form and fixed diameter. Similarly, a biopsy punch can be used for oral carcinomas in future studies. Another draw-back of the present study is the using of frozen specimen which does not reflect the physiological metabolic condition. However, for ex vivo measurement, the specimens have to be kept frozen to suppress degradation which starts as soon as the tissue is excised. In future studies, one strategy to circumvent this problem is to use fresh specimen from preclinical animal models. We are also planning a study where a special device for oral samples will be placed next to the operation room and autofluorescence measurement will be carried immediately on the biopsy or resected samples. A technical development is currently in progress toward endoscopic measuring, live and directly in the oral cavity of the patients. Very high sensitivity and specificity were obtained for DF in discriminating tumour from healthy tissues in this study. However,

3

it should be kept in mind that only well defined tumour specimens with >80% tumour cells were used for the measurements. Therefore, despite the exciting and promising results, we are still at the very beginning of a long process towards the goal of sensitive detection of tumour tissues. 5. Conclusion Fractal dimension based on time-resolved autofluorescence spectra has promising potential in real-time detection of tumour cells in the oral cavity. Conflict of interest statement All authors declare that they do not have conflict of interest to disclose. Acknowledgement We thank Ms. Jessica Knoblauch for her technical assistant. References Butte PV, Pikul BK, Hever A, Yong WH, Black KL, Marcu L: Diagnosis of meningioma by time-resolved fluorescence spectroscopy. J Biomed Opt 10: 064026, 2005 Farwell DG, Meier JD, Park J, Sun Y, Coffman H, Poirier B, et al: Time-resolved fluorescence spectroscopy as a diagnostic technique of oral carcinoma: validation in the hamster buccal pouch model. Arch Otolaryngol Head Neck Surg 136: 126e133, 2010 Feichtinger M, Pau M, Zemann W, Aigner RM, Kärcher H: Intraoperative control of resection margins in advanced head and neck cancer using a 3D-navigation system based on PET/CT image fusion. J Craniomaxillofac Surg 38: 589e594, 2010 Gerich CE: Untersuchung zum nicht-exponentiellen Abklingverhalten von Laser induzierter AutoeFluoreszenz zur Diagnose von Prostata-Carcinom. Hall/Tirol, Eduard Wallnöfer Zentrum 1: UMIT Private Universität für Gesundheitswissenschaften, Medizinische Informatik und Technik, 2008 Master’s Thesis Gerich CE, Opitz J, Schreiber J: Machbarkeitsstudie Zeitaufgelöste Fluoreszenz zur Krebsdiagnose. Fraunhofer Institut für zerstörungsfreie Prüfverfahren, 2009 Tech. Rep. Hertrampf K, Wenz HJ, Koller M, Grund S, Wiltfang J: Early detection of RadespielTröger M, Meyer M, Fenner M. Geographic differences and time trends of intraoral cancer incidence and mortality in Bavaria, Germany. J Craniomaxillofac Surg 40: e285ee292, 2012 Meier JD, Xie H, Sun Y, Sun Y, Hatami N, Poirier B, et al: Time-resolved laser-induced fluorescence spectroscopy as a diagnostic instrument in head and neck carcinoma. Otolaryngol Head Neck Surg 142: 838, 2010 Scully C, Bagan J: Oral squamous cell carcinoma: overview of current understanding of aetiopathogenesis and clinical implications. Oral Dis 15: 388e399, 2009 Thiele OC, Freier K, Bacon C, Flechtenmacher C, Scherfler S, Seeberger R: Craniofacial metastases: a 20-year survey. J Craniomaxillofac Surg 39: 135e137, 2011 Yong WH, Butte PV, Pikul BK, Jo JA, Fang Q, Papaioannou T, et al: Distinction of brain tissue, low grade and high grade glioma with time-resolved fluorescence spectroscopy. Front Biosci 11: 1255e1263, 2006

Please cite this article in press as: Klatt J, et al., Fractal dimension of time-resolved autofluorescence discriminates tumour from healthy tissues in the oral cavity, Journal of Cranio-Maxillo-Facial Surgery (2014), http://dx.doi.org/10.1016/j.jcms.2013.12.004

Fractal dimension of time-resolved autofluorescence discriminates tumour from healthy tissues in the oral cavity.

Early detection and complete resection of oral carcinomas is of crucial importance for patient survival. This could be significantly improved by devel...
249KB Sizes 0 Downloads 0 Views