|
1.IntroductionRenal cell carcinoma (RCC), the most common type of kidney cancer, is increasing in incidence and is the most lethal genitourinary cancer.1 Surgical resection of RCC is the benchmark for long-term cure of this disease. In recent years, enhanced employment of different imaging techniques has increased the number of patients with small renal tumors. Nephron-sparing surgery is the preferred operative modality for small renal masses because it offers equivalent oncologic efficacy and improved renal function outcomes compared with complete nephrectomy.2 Additionally, minimally invasive surgical approaches, including laparoscopic and robot-assisted partial kidney resection, have become more and more popular due to better cosmetic results and faster recovery compared with open operations. Despite the advantages of nephron-sparing resections, the surgical technique is difficult to perform and has a steep learning curve. Preoperative diagnostic examinations by computer tomography or magnetic resonance imaging and ultrasound provide borders for the safe tumor resection. However, intraoperative detection of tumor margins is challenging and, up till now, is merely based on palpation and visual inspection, resulting, frequently, in incomplete tumor resection. Intraoperative rapid pathological microscopic examination, known as a frozen section procedure, is time consuming, is of lower quality than fixed tissue processing, highly depends on the investigator’s skills, and may be unreliable.3 Therefore, rapid and reliable technology for intraoperative identification of cancer cells is urgently needed for tumor margin controlled surgery. Recently, several attempts have been made to apply bioanalytical methods for an intraoperative classification of tissue. In particular, vibrational spectroscopy has the potential for identification of tumor cells under intraoperative conditions.4 During the last decades, many studies have demonstrated that Fourier transform infrared (FT-IR) spectroscopy is a tool for quick and label-free characterization of biological materials.5–10 FT-IR spectroscopy can be used to detect very small variations in spectral fingerprints. Different techniques are applied for diagnosis of tissue pathologies:11 IR microspectroscopy,6,8,9 Raman spectroscopy, and fluorescence spectroscopy.12 FT-IR and Raman spectroscopic techniques have been applied to a broad spectrum of biochemical samples, such as thyroid gland,13 brain, and even the primary tumor of brain metastases,14,15 breast,16 skin, and cervical cancer,17 and many others, such as prostate, stomach, kidney, and lung tumors. The majority of these studies are based on the investigation of cells.18,19 Recent evidence indicates that tumor cells interact through direct cell–cell as well as cell–matrix interactions controlled by the production of specific extracellular matrix (ECM) components and the secretion of grow factors.20 It is well known that the ECM plays an important role in sustaining tissue cells and alterations may indicate the transformation of normal tissue into tumor tissue. Of the many available techniques for characterization of ECM, FT-IR spectroscopy is gaining attention as it provides label-free insights into the molecular composition.21–23 For example, glioma cells are known to produce an ECM-like chemical fingerprint,24 which can be detected by FT-IR spectroscopy.25 In this study, we demonstrate that FT-IR spectroscopy can be used for an analysis of ECM to classify normal and tumor kidney tissue. 2.Experimental2.1.Sample PreparationThe samples for FT-IR spectroscopic studies were prepared from freshly resected human kidney tissue. Tumor and control specimens were obtained from 31 adult patients, who underwent kidney tumor surgery at the Urology Department of Vilnius University (9 female, 22 male, with an age of 42 to 83 years). The protocol of this study was approved by the Vilnius regional bioethics committee (approval no. 158200-12-131-056LP6, 05 05 2009). Figure 1 illustrates the preparation of ECM films graphically. A small volume of freshly resected tissue is deposited on slide and removed after a few seconds. ECM on the slide was dried at room temperature. 2.2.FT-IR Spectroscopic ImagingFT-IR spectroscopic imaging was performed in transmission mode using an FT-IR spectrometer Vertex 70 coupled with an infrared microscope Hyperion 3000 (both from BrukerOptik GmbH, Ettlingen, Germany) and a mercury cadmium telluride (MCT) focal plane array detector. The 15-fold Cassegrainian objective with a numerical aperture of 0.4 imaged a sample area of . Composition of individual infrared images was captured. Pixel binning of was applied for the larger composition in order to reduce the amount of spectra. Pixel binning reduces the lateral resolution to . The background spectroscopic image was recorded before the ECM films were investigated from the pure window. A total number of 16 interferograms (scans) were co-added. The interferograms were Fourier transformed applying Blackman–Harris 3-Term apodization and a zero filling factor of 2. Spectra at a resolution of of the sample image were ratioed against the spectra of the background image and transformed to absorbance values. This spectral resolution was chosen in order to improve the signal-to-noise ratio, to reduce the size of the spectral data set, and to ensure that all prominent bands, even those with medium intensity, appear clearly in the spectrum. 2.3.FT-IR MicrospectroscopyFor each sample, FT- IR spectra were acquired using the Vertex 70 instrument coupled with an infrared microscope Hyperion 3000 and a single-point MCT detector. The aperture size was set to . One hundred and twenty eight interferograms were recorded at a resolution of , apodized (Blackman–Harris 3-Term), zero filled (two times), and phase corrected (Mertz). After Fourier transformation, the single-beam sample spectra were ratioed against a background spectrum obtained from the pure window and converted to absorbance units. 2.4.Data EvaluationEvaluation of all spectral data was performed using the MATLAB® Package (Version 7, Math Works Inc., Natick, Massachusetts). The so-called fingerprint region between 950 and was considered. Data preprocessing for both spectroscopic image and single-point measurements involves the removal of outliers, a baseline correction, and a normalization of each absorbance value of a spectrum to the integral absorbance. Outliers are spectra that are obviously not associated to tissue or spectra with a maximum absorbance value or . The baseline of each spectrum was corrected by using the msbackadj function of the Statistics Toolbox of the MATLAB® package. Afterward, the spectra were area-normalized to eradicate spectral differences due to sample thickness or variation in the density of cellulose fibers. The eig function of the basic MATLAB® package was used for principal component analysis (PCA) calculation of the spectroscopic image. 3.ResultsFigure 2 shows the hematoxylin and eosin (H&E) stain for a section of renal tumor tissue, adjacent to a piece that was used for preparation of ECM. The H&E section shows areas of different cellular density and structure as well as necrotic tissue. Small dark blue dots evenly distributed in the H&E image represent the nuclei of tumor cells. Such features are associated with an undifferentiated type of RCC. In a first experimental step, FT-IR imaging was used to quantitatively investigate the homogeneity of the prepared ECM films. The objective of this initial study was to examine whether single-point measurements of ECM films are meaningful to distinguish between normal and tumor samples. Figure 3(a) shows the FT-IR image of the raw data set. Normal tissue was pressed on the left side and tumor tissue on the right side of the window. All absorbance values in the range from 950 to of each individual spectrum are summarized and color-coded in the rainbow scale. Baseline correction was applied to the absorbance spectra. Spectra that obviously do not correspond to the ECM film are detected as outliers and were removed from the data set. Figure 3(b) shows the FT-IR image of the preprocessed data set. Similar to Fig. 3(a), absorbance values are summarized and color-coded in the rainbow scale. White pixels indicate spectra that were removed from the data set. The ECM film of normal tissue exhibits stronger absorbance values than the ECM film of tumor tissue. Normal tissue is denser and less wet than tumor tissue. Consequently, after drying, the ECM film of tumor tissue is thinner than the film from normal tissue. As the FT-IR spectroscopic imaging has generated a complex data set containing thousands of spectra, a multivariate approach has been chosen for the analysis. The investigation of the spectral signals was performed by PCA. This method is known to provide a rapid way to assess variation within the spectral data set and it assumes no explicit statistical model underlying the variance of the original spectra. Principal components (PC) are formed by loading plots and score maps. The loading plots correspond to spectral bands where the variation is the highest and weight the signals in the positive and negative directions. The score map reveals the weight of the loading plot for each pixel of the image. PCA is applied on the preprocessed spectra. Loading vectors and score maps of the first five PC are represented in Fig. 4. The first PC exhibits a variance of 91%. It represents the mean spectrum of the samples. The amide I and II bands, located at 1542 and ,26 respectively, are the major bands in the loading plot. The signal at arises from modes of lipids or esters, and the signal located at is assigned to of fatty acids.26 Additional absorption features here are due to at and the amide III mode at . A broad signal between 1000 and is the result of the , , and modes.26 The score map of the second PC shows a clear difference between normal and tumor samples. Chartreuse pixels mainly indicate positive score values for the ECM film of normal tissue, whereas the ECM film of tumor tissue is dominated by green and blue pixels and negative score values. The corresponding loading plot displays three pronounced negative signals, which are assigned to of esters (), , and modes. Vibrations of fatty acids may also contribute to the signal at . The signals indicate that the ECM of tumor tissue contains more esters and fatty acids than the ECM of normal tissue. It should be noted that the signal intensity between 1730 and can be affected by variations of the ester content as well as by pH. A lower extracellular pH is an important feature of tumor tissue.27 The extracellular acidification of tumor tissue is considered to be due to lactate secretion from anaerobic glycolysis.28 An acidic pH leads to an increase of the protonated acid form and, consequently, to an enhanced signal in the aforementioned spectral range. Three small areas of blue pixels in the ECM film of normal tissue may occur due to blood clots present in the sample of the intercellular fluid. Nevertheless, the score map clearly indicates that the ECM from both, normal and tumor tissues, form nearly homogeneous films, which allow point measurements. The third PC also shows slight differences between both types of ECMs. However, a detailed discussion of the loading plot is hardly possible. Weaker and slightly shifted signals compared to the loading plot of the second PC and, most notably, the appearance of a broad signal with a maximum at as well as a weak negative amide I signal at points to complex variations of biochemical composition. The fourth and higher PCs contribute only slightly to the measured variance; hence, they mainly represent noise signals without any further relevant information. Since the PCA reveals significant spectral differences between ECM films of normal and tumor tissues as well as a comparatively homogeneous distribution of the different spectral features, further spectral analysis was performed by the faster single-point measurement mode. For each sample, five randomly chosen points were measured and then averaged. This reduces variations in the spectra due to inhomogeneities of the ECM. Figure 5(a) shows the mean spectra of the ECM films prepared from normal tissue. Although the overall absorbances fluctuate from sample to sample, the spectral pattern is very similar. This becomes much more pronounced when the spectra are area normalized, as shown in Fig. 5(b). Shown in Figs. 5(c) and 5(d) are the raw and normalized spectra of ECM films from tumor tissue. All the tumor ECM films exhibit a stronger signal between 1730 and , which arises from carbonyl groups of fatty acids. The broad signal between 1000 and is dramatically altered as well, splitting into several separate bands at , 1080, and . These bands are known to be due to carbohydrates, phosphate groups, and glycolipids, which, in the tumor samples, have a weak but discernible increase in absorption. This finding is supported by a recently published Raman spectroscopic study which shows that kidney tumor tissue exhibits increased phosphate stretching modes.29 Obviously, the carbonyl stretching band, centered at , and the broad signal between 1000 and can both be used as spectral markers to distinguish the ECM of tumor from those of normal tissue. The ECM of tumor tissue exhibits a significantly higher content of fatty acids, carbohydrates, phosphate groups, and glycolipids than the ECM of a normal sample. This general trend is observed for all 31 patients. The scatter plot of the integrated intensity between 1000 and versus the integrated intensity between 1710 and is depicted in Fig. 6. The variance for the ECM of tumor tissue is higher than the ECM of normal tissue. Tumor tissue varies significantly between individuals, as expected. RCC and urothelial cell carcinoma are known for displaying a high amount of interindividual variance. The absorbance values corresponding to the ECM of normal and tumor tissues can be precisely separated by two defined thresholds as indicted in Fig. 6. Spectra that have an integrated absorbance in the spectral range from 1710 to and a higher integrated absorbance of 0.07 in the range from 1000 to are classified as tumor. The ECM sample, marked with a star in the plot, exhibits a much higher intensity of the carbonyl band than that of the other ECM samples of normal tissue. A detailed inspection of this ECM film indicated very low absorbance values and, consequently, a low signal-to-noise ratio. This led, obviously, to a calculated stronger intensity of the carbonyl band. Low absorbance values can be explained by the very low thickness of the dried film, produced by not applying enough pressure during the stamping of the tissue on the substrate. For the first time, we have demonstrated an optical marker-free technology to identify kidney tumor tissue. Although intraoperative FT-IR spectroscopy is still far from clinical routine applications, this study demonstrates that freshly resected tissue can be analyzed quickly and accurately to obtain diagnostic information in the operating room. Moreover, the evaluation of ECM enables an intraoperative analysis of questionable tissue without surgical resection. 4.ConclusionsIn this article, we demonstrate that FT-IR spectroscopy based on the characterization of ECM is useful for a label-free differentiation of kidney tumor from normal tissue. Reproducible differences are apparent between spectra of tumor and normal ECM in the stretching region (1710 to ) and in the range between 1000 and . These spectral regions can be used as markers for the identification of the presence of the tumor cells in tissue. Importantly, FT-IR spectroscopy, in particular, makes possible a label-free and rapid assessment of questionable kidney tissue during the operative resection of the tumor. Larger prospective studies are needed to confirm these preliminary data. AcknowledgmentsFunding from the European Community’s social foundation under Grant Agreement No. VP1-3.1-ŠMM-08-K-01-004/KS-120000-1756 is acknowledged. ReferencesJ. N. Ebleet al., Pathology and Genetics of Tumors of the Urinary System and Male Genital Organs, World Health Organization(2004). Google Scholar
M. Crepelet al.,
“Nephron-sparing surgery is equally effective to radical nephrectomy for T1BN0M0 renal cell carcinoma: a population-based assessment,”
Urology, 75
(2), 271
–275
(2010). http://dx.doi.org/10.1016/j.urology.2009.04.098 0090-4295 Google Scholar
S. VenigallaG. WuH. Miyamoto,
“The impact of frozen section analysis during partial nephrectomy on surgical margin status and tumor recurrence: a clinicopathologic study of 433 cases,”
Clin. Genitourin Cancer, 11
(4), 527
–536
(2013). http://dx.doi.org/10.1016/j.clgc.2013.05.003 CGCLA7 1558-7673 Google Scholar
G. SteinerM. Kirsch,
“Optical spectroscopic methods for intraoperative diagnosis,”
Anal. Bioanal. Chem., 406
(1), 21
–25
(2014). http://dx.doi.org/10.1007/s00216-013-7401-4 ABCNBP 1618-2642 Google Scholar
R. K. SahuS. Mordechai,
“Fourier transform infrared spectroscopy in cancer detection,”
Future Oncol., 1
(5), 635
–647
(2005). http://dx.doi.org/10.2217/14796694.1.5.635 FOUNBN 1479-6694 Google Scholar
G. SrinivasanR. Bhargava,
“Fourier transform–infrared spectroscopic imaging: the emerging evolution from a microscopy tool to a cancer imaging modality,”
Spectroscopy, 22
(7), 40
–51
(2007). SPIJDZ 0712-4813 Google Scholar
B. R. Woodet al.,
“An investigation into FTIR spectroscopy as biodiagnostic tool for cervical cancer,”
Biospectroscopy., 2
(3), 143
–153
(1996). http://dx.doi.org/10.1002/(SICI)1520-6343(1996)2:3<143::AID-BSPY1>3.0.CO;2-9 BIOSFS 1075-4261 Google Scholar
B. R. Woodet al.,
“Fourier transform infrared (FTIR) spectral mapping of the cervical transformation zone, and dysplastic squamous epithelium,”
Gynecol. Oncol., 93
(1), 59
–68
(2004). http://dx.doi.org/10.1016/j.ygyno.2003.12.028 GYNOA3 0090-8258 Google Scholar
I. W. LevinR. Bhargava,
“Fourier transform infrared vibrational spectroscopic imaging: integrating microscopy and molecular recognition,”
Annu. Rev. Phys. Chem., 56 429
–474
(2005). http://dx.doi.org/10.1146/annurev.physchem.56.092503.141205 ARPLAP 0066-426X Google Scholar
R. K. Dukor,
“Vibrational spectroscopy in the detection of cancer,”
The Handbook of Vibrational Spectroscopy, 3335
–3361 John Wiley & Sons Ltd., Chichester
(2002). Google Scholar
C. Krafftet al.,
“Disease recognition by infrared and Raman spectroscopy,”
J. Biophotonics, 2
(1–2), 13
–28
(2009). http://dx.doi.org/10.1002/jbio.v2:1/2 JBOIBX 1864-063X Google Scholar
R. Manoharanet al.,
“Raman spectroscopy and fluorescence photon migration for breast cancer diagnosis and imaging,”
Photochem. Photobiol., 67
(1), 15
–22
(1998). http://dx.doi.org/10.1111/php.1998.67.issue-1 PHCBAP 0031-8655 Google Scholar
Y. Liuet al.,
“Detection of cervical metastatic lymph nodes in papillary thyroid carcinoma by Fourier transform infrared spectroscopy,”
Br. J. Surg., 98
(3), 380
–394
(2011). http://dx.doi.org/10.1002/bjs.v98.3 BJSUAM 0007-1323 Google Scholar
N. Bergneret al.,
“Tumor margin identification and prediction of the primary tumor from brain metastases using FTIR imaging and support vector machines,”
Analyst, 138
(14), 3983
–3990
(2013). http://dx.doi.org/10.1039/c3an00326d ANLYAG 0365-4885 Google Scholar
C. Krafftet al.,
“Identification of primary tumors of brain metastases by SIMCA classification of IR spectroscopic images,”
BBA, 1758
(7), 883
–891
(2006). http://dx.doi.org/10.1016/j.bbamem.2006.05.001 BBALAJ 0067-2734 Google Scholar
A. Benardet al.,
“Discrimination between healthy and tumor tissues on formalin-fixed paraffin-embedded breast cancer samples using IR imaging,”
Spectroscopy, 24
(1–2), 67
–72
(2010). http://dx.doi.org/10.1155/2010/376095 SPIJDZ 0712-4813 Google Scholar
S. MordechaiK. R. SahuZ. Hammody,
“Possible common biomarkers from FTIR microspectroscopy of cervical cancer and melanoma,”
J. Microsc., 215
(1), 86
–91
(2004). http://dx.doi.org/10.1111/j.0022-2720.2004.01356.x JMICAR 0022-2720 Google Scholar
J. P. Couapelet al.,
“Optical spectroscopy techniques can accurately distinguish benign and malignant renal tumors,”
BJU Int., 111
(6), 865
–871
(2013). http://dx.doi.org/10.1111/j.1464-410X.2012.11369.x BJINFO 1464-410X Google Scholar
S. Stewartet al.,
“Distinguishing between renal oncocytoma and chromophobe renal carcinoma using Raman molecular imaging,”
J. Raman Spectrosc., 45
(4), 274
–280
(2014). http://dx.doi.org/10.1002/jrs.v45.4 JRSPAF 0377-0486 Google Scholar
C. M. Vicenteet al.,
“Syndecan-2 is upregulated in colorectal cancer cells through interactions with extracellular matrix produced by stromal fibroblasts,”
BMC Cell Biol., 14
(25),
(2013). http://dx.doi.org/10.1186/1471-2121-14-25 BCBMAY 1471-2121 Google Scholar
R. Noreenet al.,
“FTIR spectro-imaging of collagen scaffold formation during glioma tumor development,”
Anal. Bioanal. Chem., 405
(27), 8729
–8736
(2013). http://dx.doi.org/10.1007/s00216-013-7337-8 ABCNBP 1618-2642 Google Scholar
S. Kumaret al.,
“Change in the microenvironment of breast cancer studied by FTIR imaging,”
Analyst, 138
(14), 4058
–4065
(2013). http://dx.doi.org/10.1039/c3an00241a ANLYAG 0365-4885 Google Scholar
N. Hassleret al.,
“Real-time spectroscopic analysis of extracellular matrix produced by MC3T3-E1 preosteoblastic cells cultured under dynamic conditions,”
Appl. Spectrosc., 66
(1), 40
–47
(2012). http://dx.doi.org/10.1366/11-06282 APSPA4 0003-7028 Google Scholar
I. A. Edwardset al.,
“Effect of brain-tumor derived connective tissue growth factor on glioma invasion,”
J. Natl. Cancer Inst., 103
(15), 1162
–1178
(2011). http://dx.doi.org/10.1093/jnci/djr224 JNCIEQ 0027-8874 Google Scholar
A. L. Stellinget al.,
“Infrared spectroscopic studies of cells and tissue: triple helix proteins as a potential biomarker for tumors,”
PLOS One, 8
(3), 358332
(2013). http://dx.doi.org/10.1371/journal.pone.0058332 1932-6203 Google Scholar
G. Socrates, Infrared and Raman Characteristic Group Frequencies, J. Wiley & Sons Ltd., West Sussex, England
(2001). Google Scholar
L. E. GerweckK. Seetharaman,
“Cellular pH gradient in tumor versus normal tissue: potential exploitation for the treatment of cancer,”
Cancer Res., 56
(6), 1194
–1198
(1996). CNREA8 0008-5472 Google Scholar
Y. Katoet al.,
“Acidic extracellular microenvironment and cancer,”
Cancer Cell Int., 13 89
(2013). http://dx.doi.org/10.1186/1475-2867-13-89 CCIACC 1475-2867 Google Scholar
Z. Zhengfeiet al.,
“Study of molecule variations in renal tumors based on confocal micro-Raman spectroscopy,”
J. Biomed. Opt., 18
(3), 031103
(2013). http://dx.doi.org/10.1117/1.JBO.18.3.031103 JBOPFO 1083-3668 Google Scholar
|