Raman Spectroscopic Analysis Identifies Testicular Microlithiasis as Intratubular Hydroxyapatite

0022-5347/04/1711-0092/0 THE JOURNAL OF UROLOGY® Copyright © 2004 by AMERICAN UROLOGICAL ASSOCIATION

Vol. 171, 92–96, January 2004 Printed in U.S.A.

DOI: 10.1097/01.ju.0000101948.98175.94

RAMAN SPECTROSCOPIC ANALYSIS IDENTIFIES TESTICULAR MICROLITHIASIS AS INTRATUBULAR HYDROXYAPATITE B. W. D. DE JONG,*,† C. A. DE GOUVEIA BRAZAO,*,† H. STOOP, K. P. WOLFFENBUTTEL, J. W. OOSTERHUIS, G. J. PUPPELS, R. F. A. WEBER, L. H. J. LOOIJENGA AND D. J. KOK From the Departments of Paediatric Urology (BWDDJ, KPW, DJK), Andrology (CADGB, RFAW), Urology (CADGB, RFAW), Pathology/ Laboratory for Experimental Patho-Oncology/Josephine Nefkens Institute (CADGB, HS, JWO, LHJL) and General Surgery (GJP), Erasmus Medical Center, University Medical Center, Rotterdam, The Netherlands

ABSTRACT

Purpose: As diagnosed by ultrasonography, testicular microlithiasis is associated with various benign and malignant conditions. The molecular constitution of these microliths is largely unknown. Raman spectroscopy provides detailed in situ information about the molecular composition of tissues and to our knowledge it has not been applied to gonadal microliths. We analyzed the molecular composition of gonadal microlithiasis and its surrounding region using Raman spectroscopy in malignant and benign conditions. Materials and Methods: Multiple microliths from 6 independent samples diagnosed with gonadal microlithiasis by ultrasound and histologically confirmed were investigated by Raman spectroscopy. The samples included 4 testicular parenchyma samples adjacent to a germ cell tumor (4 seminomas), a gonadoblastoma of a dysgenetic gonad and testicular biopsy of a subfertile male without malignancy. Results: Raman spectroscopic mapping demonstrated that testicular microliths were located within the seminiferous tubule. Glycogen surrounded all microliths in the samples associated with germ cell neoplasm but not in the benign case. The molecular composition of the 26 microliths in all 6 conditions was pure hydroxyapatite. Conclusions: Microliths in the testis are located in the seminiferous tubules and composed of hydroxyapatite. In cases of germ cell neoplasm they co-localize with glycogen deposits. KEY WORDS: testis; lithiasis; spectrum analysis, Raman; durapatite

Testicular microlithiasis (TM) or microcalcification is a rare urological condition characterized by calcifications (microliths) within the seminiferous tubules. Microliths are visualized in situ by imaging techniques and in tissue sections by microscopy as foci of calcification. The exact composition of these microliths is largely unknown. Ultrastructural analysis shows a central calcified core surrounded by concentric layers of connective or stratified collagen fibers. This finding suggests that they originate from degenerating intratubular cells debris.1 The current model is that microlith formation starts by an accumulation of cellular debris, followed by the deposition of glycoprotein material around the core and subsequent calcification of this material.2 TM is associated with various pathological conditions, including testicular torsion and atrophy, cryptorchidism, gonadal dysgenesis (which carries a high risk for germ cell tumor) and epididymal cysts. In all of these conditions disturbance of the testicular milieu may lead to mineral formation. In addition, TM is associated with male infertility, testicular germ cell tumors (TGCTs), that is seminomas and nonseminomas, Klinefelter’s syndrome, hypogonadism, male pseudohermaphroditism, varicocele and nonHodgkin’s lymphoma. In these conditions it is not as obvious whether mineralization follows, concurs or precedes it, or if microliths indicate a specific disease stage. TM has been found in 5.2% to 45% of men with invasive TGCT.3–5 Interestingly TM is

also found in 15% of the contralateral testes of patients with a diagnosed unilateral TGCT6 and carcinoma in situ (CIS), the precursor of all TGCTs.7 The prevalence in healthy young men is 5.6%.8 Information concerning the exact composition of microliths in the various benign and malignant conditions may provide a clue concerning the underlying mechanism of microlith formation and, thereby, may shed light on the initiating pathological process. To date knowledge of the molecular composition of microliths is limited. Raman spectroscopy has evolved rapidly into a technology for analyzing the molecular composition of normal and pathological tissues. For frozen sections of bladder tissue it was recently shown that Raman mapping clearly identifies the separate bladder structures, urothelium, lamina propria and muscle layer.9 We hypothesized that Raman spectroscopic mapping of tissue containing microlithiasis might reveal the molecular composition of the microlith and surrounding tissue. For this purpose Raman analysis was performed on gonadal microliths of different origins, ie the testis of a subfertile male, and gonadoblastoma and testicular parenchyma samples adjacent to a TGCT. The obtained spectra were compared to a data bank of known compounds. In addition to molecular investigation of the microliths, Raman spectroscopic mapping performed at a higher magnification can reveal tissue structure on the molecular level. It may provide novel insight into the genesis of microliths.

Accepted for publication August 8, 2003. * Correspondence: Departments of Paediatric Urology and Andrology/Urology, Erasmus Medical Center, Room Be 362b and Room Ba 146, Dr. Molewaterplein 50, 3015 GE, Rotterdam, The Netherlands (telephone: 0031-104087690 and 0031-104635767; FAX: 0031104089386 and 0031-104635763; e-mail: [email protected] and [email protected]). † Equal study contribution.

Tissue preparation. Freshly obtained tumor specimens in this study included 1 gonadoblastoma, 4 testicular parenchyma samples of patients with testicular seminoma and 1 testicular biopsy of a subfertile male. All seminoma samples

MATERIAL AND METHODS

92

RAMAN SPECTROSCOPY OF TESTICULAR MICROLITHIASIS

also contained CIS in the parenchyma. These specimens were collected in collaboration with the departments of urology and pathology at our institution. Samples were obtained by surgery prior to chemotherapy or irradiation. Microlithiasis was confirmed by histology based on a hematoxylin and eosin stained frozen section. Parallel representative tissue sections were used for investigation. Therefore, for each of the 4 patient seminoma samples a frozen section of 20 ␮m was cut and mounted on a calcium fluoride window for Raman spectroscopic mapping. Raman spectroscopic mappings were performed to measure the molecular composition of tissue around microliths. Adjacent slides of each sample were formalin fixed and stained with periodic acid-Schiff (PAS) staining. Single point Raman measurements were performed on 3 microliths within the biopsy of the subfertile man, 3 in the sample of the patient with gonadoblastoma and a total of 20 in the 4 different samples of the patients with seminoma. Raman spectroscopy. Figure 1 shows the setup with which Raman spectroscopic data were acquired. Detailed information on this setup and data acquisition has been previously described.9, 10 Air-dried histological sections were placed on a computer controlled sample stage of a DM-RXE (Leica, Cambridge, United Kingdom) microscope. To collect Raman spectra near infrared laser light (approximately 845 nm) was coupled into the microscope via a single mode optical fibre. An 80⫻ NIR-plan (Olympus, Tokyo, Japan) optimized objective was used to focus laser light onto the tissue sample and collect light that was scattered by the sample. Via an optical fiber and a chevron type dielectric short pass filter set the Raman signal was guided into a System 100 (Renishaw, Wotton under Edge, United Kingdom) spectrometer equipped with a thermo-electrically cooled, deep depletion charge coupled device (CCD) camera. The CCD camera was connected to a personal computer, in which Raman spectral data were stored. Raman spectra measured on microliths and all chemicals used for reference analysis were taken in a measurement time of 10 seconds each. For the Raman map the spectra were taken in a measurement time of 20 seconds each. All spectral data processing software was programmed in Matlab 6.1

FIG. 1. Raman mapping system. Argon ion, argon ion laser. Ti: Sapphire, tunable titanium sapphire laser set at approximately 845 nm. OF1, single mode optical fiber. OF2, 100 ␮m diameter core optical fiber. SPF, short pass filter less than approximately 700 nm. CF, chevron filter of 2 dielectric short pass filters less than 850 nm. MO, 80⫻ microscope objective. MS, computer controlled x-y-z microscope stage. S, spectrometer with back depletion CCD. PC, computer. DC, digital video camera.

93

(Mathworks, Natick, Massachusetts) using the PLS (Eigenvector Research Inc., Manson, Washington) toolbox. Wavelength calibration and correction for wavelength dependent signal detection efficiency was done as described previously.11 Single point Raman spectra from microliths were compared with reference spectra. Raman map data were processed with principal component analysis for data reduction and orthogonalization. Principal components analysis scores were analyzed by cluster analysis. Cluster analysis is used to find groups of spectra that have similar spectral characteristics. In this study it enabled identification of compositional features of microliths and surrounding tissue. We applied the hierarchical Ward’s cluster method with euclidean distances.12 Clusters that were formed with the hierarchical cluster method were assigned a color. Each grid element of the Raman map was then assigned the color of the particular cluster to which its spectrum belonged. Spectra obtained from grid elements with the same color showed high spectral similarity, implying high similarity in molecular tissue composition. Raman maps were then compared with bright field images of the same tissues after PAS staining, which was performed after Raman measurements. For reference purified glycogen and collagen (Sigma, Zwijndrecht, The Netherlands) were used. RESULTS

In addition to single point Raman measurements of 26 microliths from 6 independent samples, Raman spectroscopic maps were constructed of a part of each parenchyma sample adjacent to seminoma/CIS containing microliths. Figure 2, A shows an example of such a Raman spectroscopic map on tissue surrounding a microlith. The image shows a seminiferous tubule with a microliths and surrounding tissue. Four maps were constructed, that is 1 per seminoma/CIS patient sample. The results of all maps were identical and, therefore, only 1 representative example is discussed. Each spectrum was taken on a 9 ␮m2 square section of the map with a total area of 153 ⫻ 150 ␮m, resulting in 2,550 spectra (51 ⫻ 50 map sections). According to hierarchical Ward’s clustering the obtained spectra were clustered based on their spectral characteristics. Figure 2, B shows a map in which 5 clusters provided the optimum separation of morphological features. The outlines of these 5 clusters followed the morphology of the measured tissue section that was stained with the PAS stain after Raman mapping measurement (fig. 2, C). Comparing figure 2, B and C clearly shows that the Raman map meticulously matched histological findings, although the microlith was lost during staining. PAS staining performed on formalin fixed, consecutive histological slides showed similar results (data not shown). The averaged spectra per patient cluster were plotted (fig. 2, D). The red cluster clearly followed the microlith outline. The black cluster surrounding the crystal was characterized by a clear collagen contribution to the spectrum. As confirmed by the stained section, it represented the basal membrane of the seminiferous tubule. Because the basal membrane completely enclosed the crystal, the microlith was located within the tubule. Parts of the black (collagen) cluster also appeared within the extratubular region, representing supportive tissue. The cluster of collagen (basal membrane) surrounded the blue and green clusters. These clusters were characterized by a strong glycogen component in the Raman spectrum. The difference between these 2 clusters was purely the concentration of glycogen in each map section with the blue cluster having the highest concentration. Because tissue sections were not treated before Raman measurement, glycogen, which is usually washed out during tissue preparations, was still abundantly present. It was present in all 4 seminoma/ CIS samples. The remaining intratubular and extratubular

94

RAMAN SPECTROSCOPY OF TESTICULAR MICROLITHIASIS

FIG. 2. A, unstained histological section of testicular parenchyma with microlith (M) adjacent to seminoma. (Scale bar represents 30 ␮m). B, Raman spectroscopic map of unstained section composed of 7 clusters. C, tissue section, stained after Raman mapping. Microlith was lost during staining. Dotted line indicates location. Scale bar represents 30 ␮m. D, averaged spectra of clusters (fig. 2, B). Each spectrum color matched color of corresponding cluster. Specific band positions were glycogen at 481, hydroxyapatite at 959, and lycopene at 1,157 and 1,523 cm⫺1.

tissue spectra were assigned to 1 cluster by cluster analysis (yellow). It indicated that intratubular and extratubular tissues had similar spectra in these experimental conditions. One of the components present in this cluster was a carotenoid, most likely lycopene, which was present in all seminoma/CIS samples. The bands at 1,157 and 1,523 cm⫺1 were characteristic for lycopene (fig. 2, D). More detailed analysis of the difference in molecular composition between different clusters followed from the calculation of difference spectra. When the averaged spectra of the black cluster (collagen) (fig. 2, B) was subtracted from the averaged spectra of the glycogen and collagen clusters, the difference spectra provided detailed information on the presence of any components other than tissue (cells). When using the yellow cluster for this purpose, the difference spectrum was less accurate due to the high lycopene contribution, although slight lycopene features were still present. The reference glycogen spectrum matched with the characteristics of the difference spectra from the 2 glycogen clusters, although some small bands of protein-like influences marked by asterisks were also present. The difference between collagen cluster and the remaining tissue with lycopene was purely collagen since it perfectly matched the collagen reference spectrum. It was seen all 4 maps. Figure 3 shows how these different spectra compared with reference spectra of commercially available purified glycogen and collagen. The single point measured microliths spectra measured on

each sample were averaged and compared to a data base of reference Raman spectra of crystals and stones measured in house, isolated from various tissues. In addition, spectra of the microlith cluster measured within the Raman map of the example were also compared to the reference data base. Figure 4 shows that the spectra of microliths of each sample perfectly matched the reference spectrum of hydroxyapatite. Band positions at 432, 590, 960 and 1,072 cm⫺1 were present in all spectra in equal contributions. However, in the spectra of the microliths bands at 1,448 cm⫺1 (CH2 bending mode) and in the amide I region between 1,662 and 1,672 cm⫺1 were found, which were not seen in the hydroxyapatite reference spectrum. This observation suggests that organic, most likely protein-like material was captured within the microliths. DISCUSSION

A number of groups have reported the origin and composition of testicular microlithiasis. Most of these studies were based on immunohistochemical analysis. One investigation used x-ray diffraction and transmission electron microscopy to study the composition of microliths.13 One testicular sample of an azoospermic patient was analyzed, showing infrequent microliths composed of calcium phosphate (hydroxyapatite) and no chemical differences were found between the core and peripheral regions of the microliths. Transmission electron microscopy revealed an acellular composition resem-

RAMAN SPECTROSCOPY OF TESTICULAR MICROLITHIASIS

95

FIG. 3. Difference spectra compared with glycogen and collagen references. A, difference spectra of black cluster spectrum subtracted from blue cluster spectrum (fig. 2, B). B, black cluster spectrum subtracted from green cluster spectrum. Asterisks indicate band positions not assigned to glycogen, eg lycopene. C, glycogen reference spectrum. D, yellow cluster spectrum subtracted from black cluster spectrum. E, collagen reference spectrum.

FIG. 4. Averaged Raman spectra. A, microliths in gonadoblastoma. B, microliths in testicular parenchyma of subfertile patient. C to F, microliths in seminoma. G, Raman spectroscopic map red cluster (fig. 2, B). H, hydroxyapatite reference spectrum.

bling hydroxyapatite calcification in the core and a high degree of collagen-like material within the external shell.14 A similar composition was found using Raman spectroscopic analysis for microliths of the breast, although calciumoxalate was found in addition to hydroxyapatite.15 In this report we describe the molecular composition of

microliths found in different types of benign and malignant gonadal pathologies. To our knowledge no previous studies have been performed analyzing multiple and mutually compared microlithiasis samples. Raman spectra of all samples, including the microlithiasis cluster of the Raman map, were compared with the reference spectrum of pure hydroxyapa-

96

RAMAN SPECTROSCOPY OF TESTICULAR MICROLITHIASIS

tite and they showed absolute homology. A putatively interesting observation made by Raman analysis was the presence of a carotenoid, most likely lycopene,16 in the tissues of the seminoma/CIS cases, of which the meaning is unclear to date. The structures (clusters) obtained by Raman mapping closely followed the morphology visible in PAS stained tissue. The spectra that characterize these clusters first of all revealed the clear presence of glycogen surrounding the microlith located in each testicular sample adjacent to TGCT and CIS as well as gonadoblastoma. This finding may point toward a pathogenetic role of the precursor cells of germ cell neoplasm in the gonads because they contain a large amount of glycogen. This observation is in line with our previous finding that bilateral TM identifies infertile men at risk for TGCT.17 Raman mapping showed indisputable evidence of a glycogen presence around microlithiasis peripheral regions in all malignant samples. Notably it was not found in the only benign lesion investigated. Two clusters containing signal contributions of glycogen were detected and the difference between these clusters was mainly the concentration. It indicates that Raman spectroscopy might be used for semiquantitative measurements in tissues. The precise etiology of microlithiasis and the mechanism of formation remain unclear because reports suggesting an extratubular18 and an intratubular19 origin are available. Recently it was suggested that microliths are located outside of the seminiferous tubule and they originate during the early stage of testicular development.18 It was described earlier as extracellular stromal calcification with microvesicles and collagen fibers,1 and as extracellular calfication.20 However, our findings support the model that microliths originate within the seminiferous tubule. It is known that testicular microlithiasis is a common finding in asymptomatic men that may not be related to testicular cancer,8 although it is more frequent in pathological conditions, including neoplasm. With this study we conclude that the main composition of TM in our limited number of samples was hydroxyapatite. We cannot conclude that hydroxyapatite would be the only microlithiasis present in gonadal tissue. Therefore, a larger study, as described by Peterson et al,8 combined with Raman spectroscopy may be applied. TM is located within the seminiferous tubule in the testis in benign and malignant conditions. In addition, microliths found in gonadoblastoma had the same composition, which might be related to their common pathogenetic mechanism. Further studies, including Raman spectroscopic mapping, might reveal novel information about the process involved. REFERENCES

1. Vegni-Talluri, M., Bigliardi, E., Vanni, M. G. and Tota, G.: Testicular microliths: their origin and structure. J Urol, 124: 105, 1980 2. Nistal, M. Paniagua, R. and Dı´ez-Pardo, J. A.: Testicular microlithiasis in 2 children with bilateral cryptorchidism. J Urol, 121: 535, 1979 3. Hobarth, K., Szabo, N., Klinger, H. C. and Kratzik, C.: Sonographic appearance of testicular microlithiasis. Eur Urol, 24: 251, 1993

4. Backus, M. L., Mack, L. A., Middleton, W. D., King, B. F., Winter, T. C., 3rd and True, L. D.: Testicular microlithiasis: imaging appearances and pathologic correlation. Radiology, 192: 781, 1994 5. Miller, R. L., Wissman, R., White, S. and Ragosin, R.: Testicular microlithiasis: a benign condition with a malignant association. J Clin Ultrasound, 24: 197, 1996 6. Bach, A. M., Hann, L. E., Shi, W., Giess, C. S., Yoo, H. H., Sheinfeld, J. and Thaler, H. T.: Is there an increased incidence of contralateral testicular cancer in patients with intratesticular microlithiasis? AJR Am J Roentgenol, 180: 497, 2003 7. von Eckardstein, S, Tsakmakidis, G., Kamischke, A., Rolf, C. and Nieschlag, E.: Sonographic testicular microlithiasis as an indicator of premalignant conditions in normal and infertile men. J Androl, 22: 818, 2001 8. Peterson, A. C., Bauman, J. M., Light, D. E., McMann, L. P. and Costabile, R. A.: The prevalence of testicular microlithiasis in an asymptomatic population of men 18 to 35 years old. J Urol, 166: 2061, 2001 9. de Jong, B. W. D., Bakker Schut, T. C., Wolffenbuttel, K. P., Nijman, J. M., Kok, D. J. and Puppels, G. J.: Identification of bladder wall layers by Raman spectroscopy. J Urol, 168: 1771, 2002 10. de Jong, B. W. D., Bakker Schut, T. C., Coppens, J., Wolffenbuttel, K. P., Kok, D. J., Puppels, G. J.: Raman spectroscopic detection of changes in molecular composition of bladder muscle tissue caused by outlet obstruction. Unpublished data 11. Wolthuis, R., Bakker Schut, T. C., Caspers, P. J., Buschman, H. P. J., Ro¨mer, T. J., Bruining, H. A. et al: Raman spectroscopic methods for in vitro and in vivo tissue characterization. In: Fluorescent and Luminescent Probes for Biological Activity. Edited by W. T. Mason. San Diego: Academic Press, chapt. 32, p. 433, 1999 12. Jain, A. K. and Dubes, R. C.: Algorithms for Clustering Data. Englewood Cliffs: Prentice Hall, 1988 13. Smith, G. D., Steele, I., Barnes, R. B. and Levine, L. A.: Identification of seminiferous tubule aberrations and a low incidence of testicular microliths associated with the development of azoospermia. Fertil Steril, 72: 467, 1999 14. Garvin, A. J., Pratt-Thomas, H. R., Spector, M., Spicer, S. S. and Williamson, H. O.: Gonadoblastoa: histologic, ultrastructural, and histochemical observations in five cases. Am J Obstet Gynecol, 125: 459, 1976 15. Haka, A. S., Shafer-Peliter, K. E., Fitzmaurice, M., Crowe, J., Dasari, R. R. and Feld, M. S.: Identifying microcalcifications in benign and malignant breast lesions by probing differences in their chemical composition using Raman spectroscopy. Cancer Res, 62: 5375, 2002 16. Stahl, W., Schwarz, W., Sundquist, A. R. and Sies, H.: Cis-trans isomers of lycopene and beta-carotene in human serum and tissues. Arch Biochem Biophys, 294: 173, 1992 17. De Gouveia Brazao, C. A., Pierik, F. H., Oosterhuis, J. W., Dohle, G. R., Looijenga, L. H. J. and Weber, R. F. A.: Bilateral testicular microlithiasis predicts the presence of the precursor of testicular germ cell tumors in subfertile men. Unpublished data 18. Drut, R. and Drut, R. M.: Testicular microlithiasis: histologic and immunohistochemical findings in 11 pediatric cases. Pediatr Dev Pathol, 5: 544, 2002 19. Nistal, M., Martinez-Garcia, C. and Paniagua, R.: The origin of testicular microliths. Int J Androl, 18: 221, 1995 20. Anderson, H. C.: Calcific diseases. A concept. Arch Pathol Lab Med, 107: 341, 1983