Introduction

Psammoma bodies (PBs) is a type of pathological mineralization (calcification), which have been found in a great variety of neoplastic and non-neoplastic conditions. PBs are most commonly observed in papillary thyroid carcinoma (PTC), meningioma and papillary serous cystadenocarcinoma of ovary (Das 2009). In a review of worldwide literature (Fadare et al. 2004), 38 % of patients with PBs have an associated malignancy or ovarian borderline tumour. Specifically, PBs are confined mainly to serous papillary adenocarcinoma (Sorbe & Frankendal 1982), and 56 % ovarian cystic teratoma contains tooth or calcification (Buy et al. 1989). The 5- and 10-year survival rates were significantly different between the serous carcinomas with less than 5 % area of PBs and those at least 5 % area (Motohara et al. 2010).

Although PBs in ovarian tumours have been recognized for decades, the understanding of their formation mechanism remains poor. Until this century, PBs had been recognized as a reflection of dystrophic calcification resulting from cell necrosis (Ferenczy et al. 1977). In 2001, Kiyozuka et al. (2001) found the formation of PBs in ovarian cancer was closely related to BMP-2 and type-IV collagen. Silva et al. (2003) found hormone can affect the formation of PBs. Besides, nanobacteria may also get involved in the crystallization of PBs (Sedivy & Battistutti 2003; Hudelist et al. 2004). All previous studies indicated that PBs were not merely the consequence of dystrophic calcification.

Unfortunately, few studies were conducted to recognize the PBs in tumours from the viewpoint of mineralogy. We believe mineralogical studies may help uncover the nature of PBs and the relationship between PB formation and tumour generation, and do necessary help for medical diagnosis and treatment. However, it is difficult to conduct formal mineralogical measurements on PBs due to the small quantity and tiny size. Besides, the intergrowth of organic tissues and PBs had been a tough problem until the separation method was proposed (Yang et al. 2008). Xylene and sodium hypochlorite solution was used to remove the organics. Although mineralogical studies of ovarian serous cancer have been conducted with this procedure (Wang et al. 2009), only one sample was sufficient enough for traditional XRD test. Besides, few mineralogical researches have been conducted on calcification in teratoma up till now.

The purpose of this study is to establish mineralogical study of PBs in ovarian serous cancer and teratoma, to compare the mineralogical characteristics of PBs in these diseases, and to discuss the mechanism of PB formation and development.

Materials and methods

Sample collection and processing

Six serous carcinoma of ovary and two mature cystic teratoma samples, which were collected from Peking University Third Hospital from year 2002 to 2012, were successively processed by formalin-fixation and paraffin-embedding at Peking University Health Science Center (PUHSC). The age of the patients with serous ovarian cancer ranges from 32 to 70, while the teratoma patients were in twenties. The tissues were cut into serial sections by Leica Ultramicrotome and the thickness of the slices was either 4 μm or 10 μm, depending on the analysis methods. For polarization microscope (POM), 4 μm-thick sections were cut, stained with hematoxylin-eosin (HE) and observed under Nikon LV100 POL.

The separation of PBs samples from the organic tissues was based on Yang et al. (2008) and conducted as follows. Slices of the serial sections were put into a 1.5 mL centrifuge tube and treated with xylene for 24 h, 12 h and 5 h. Then graded alcohol of 100 %, 95 %, 90 %, 85 %, 70 %, 50 % and 30 %, was added to remove the xylene, 10 min for each grade. After that, the samples were washed with distilled water three times and digested in 5 % sodium hypochlorite solution in 37 °C water bath for 30 min, twice. Last, the remaining powders were washed by distilled water four times and dehydrated in graded alcohol of 50 %, 65 %, 75 %, 85 %, 95 %, 100 % and 100 % for 15 min, 25 min, 35 min, 15 min, 25 min, 30 min and 40 min, respectively. The powder samples were stored in alcohol under room temperature.

Methods for observation and analysis

Environmental scanning electron microscope (ESEM)

Ten-micrometre thick sections were collected on clean 20 mm × 20 mm silicon slice and heated in thermostatic oven at 50 °C for 40 min. Then all the tissues were treated with xylene twice for 24 h and 5 h, respectively, and alcohol for 10 min and three times. The in situ samples were investigated by Quanta 200FEG (School of Physics, Peking University, live time 50 s) and Quanta 650FEG (School of Earth and Space Sciences, Peking University, live time 10 s), both of which were equipped with energy dispersive X-ray spectrum (EDX) detector. The operating conditions were set at 15 kV and 110 Pa, at low vacuum (0.1-1.5 Torr).

Micro-fourier transform infrared spectroscopy (Micro-FT-IR)

Due to the tiny amount of the sample, the powder sample was transferred by a needle onto a diamond window after drying, and flattened with the diamond sheet before test. Micro-FT-IR spectra were collected on NICOLET iN10 MX Micro-FT-IR spectrometer (College of Chemistry and Molecular Engineering, Peking University) with a resolution of 4 cm-1 in the 600–4,000 cm−1 spectral range under transmission mode. Fitting procedure was conducted with Peakfit v4.12.

Transmission electron microscope (TEM)

Approximately 0.1 mL alcohol which contained powder samples after separation processing and ultrasonic dispersion was dripped onto a carbon-coated copper grid. The alcohol was absorbed by filter paper from the other side of the grid, thus the powders were attached to it. Then the prepared sample was loaded into the holder of TECNAIF 30 transmission electron microscope at School of Physics, PKU, which is equipped with an energy dispersive X-ray analysis (EDAX) detector, operating at 300 kV. The point resolution is 0.2 nm and the exposure time is 0.5 s. The acquisition time of EDAX is approximately 2 min. The data were processed and analyzed using Digital Micrograph version 3.6.5 (Gantan Ltd.).

Micro-area synchrotron radiation X-ray analysis

For micro-area synchrotron radiation X-ray powder diffraction (μ-SRXRD), about 0.1 mg powder samples were mounted between two pieces of 3 M films after alcohol volatilization. The tests were performed at BL15U, Shanghai Synchrotron Radiation Facility (SSRF) with wavelength of 0.6527 Å and spot size of 10 μm × 10 μm. Standard sample of hydroxyapatite (HA) was also tested for comparison. All of the diffractograms were calibrated with standard data of CeO2 and converted to one-dimensional information by using the software of Fit2D V12.077. Further processing and analysis was performed by Datalab and Jade 5.0.

For micro-area synchrotron radiation X-ray fluorescence (μ-SRXRF), the 3M-mounted samples were reused. Operation conditions were set to 17 keV and 4 μm × 4 μm in spot size. All fluorescence signals were collected for 20 s, and then the data were calibrated and fitted by using PyMca 4.4.0. HA was also tested as a standard.

For μ-SRXRF mapping, the 20 μm-thick sections on the silicon slices, which have been tested by ESEM, were reused by transferring them onto Kapton films for test. The energy was set to 10 keV for testing Ca, Zn, Fe, and the fluorescence signal of each testing point was also collected for 20 s. Data processing was done by pviewer at BL15U, SSRF.

Results

Distribution and morphology

All eight tumours contained numerous PBs with the diameter varying from several μm to 30-70 μm. The PBs presented a glassy appearance and constant extinction under cross-polarized light. These dark stained spherules were often encountered in the stroma including stromal histiocytes (Fig. 1a), fibrocytes (Fig. 1b) and cancer cell nest (Fig. 1c), whereas only a very few PBs occur around epithelial cells (Fig. 1d). As far as it could be judged from the nuclear structure and staining property, the cells around PBs did not appear to be degenerated. PBs were usually found in groups, especially in the stromal histiocytes where collagen fibers are abundant (Fig. 1a).

Fig. 1
figure 1

Psammoma bodies in ovarian serous cancer (HE). Thin arrows point to PBs. a PBs in stromal histiocytes and collagen fibers (thick arrow). b PBs in fibrocytes (thick arrow), c PBs in cancer cells (thick arrow). d PBs in epithelial cells (thick arrow)

Full size image

ESEM observations demonstrated the roughly spherical PBs, either concentric layered or clumpy, were usually covered with collagen fibers (Fig. 2a). Spherules with diameters of 40–130 nm were observed in the center (Fig. 2d) and periphery (Fig. 2b) of PBs. A single PB in ovarian serous cancer was no more than 35 μm in diameter, while PBs in teratoma could be as large as 70 μm. Besides, there were some confluent spherules with a diameter of 30-50 μm which had two crystalline cores (Fig. 2e). The layered PBs in teratoma (Fig. 2e) showed more complicated layers than that in ovarian serous cancer (Fig. 2c). However, PBs in both types of tumours could be distinguished with four layers. Adjacent to a cluster of PBs, blood red cells were encompassed by collagen fibers (Fig. 2f).

Fig. 2
figure 2

ESEM images of psammoma bodies in ovarian serous cancer (a–d) and teratoma (e–f). a clumpy PB covered with collagen fibers (thick arrow). b spherules with diameters of 40–130 nm on the periphery of PB (circle). c concentric layered PB with four layers. d spherules with diameters of 40–130 nm in the center of PB (circle). e confluent spherule with two crystalline cores. The layered PB showed more complicated layers than that in c. Collagen fibers (thick arrow) encompass the PB. f blood red cells (thin arrow) and collagen fibers (thick arrow) adjacent to a cluster of PBs (arrowhead)

Full size image

Mineral composition and fine structure

EDX analysis revealed the existence of Ca, P, C, O and a small amount of Mg and Na in PBs. The atomic ratio of Ca/P (At%) from center to periphery of PBs were summarized in Table 1. Although quantitative analysis was not achieved, EDX result of a single PB reflected subtle fluctuation between different layers.

Table 1 Ca/P (At%) ratio of each layer in concentric layered psammoma bodies
Full size table

The mineral phase of PBs in all cases was identified by μ-SRXRD (Fig. 3a) as hydroxyapatite (Ca5(PO4)3(OH), PDF#09-0432) or carbonate hydroxyapatite (Ca5[PO4,CO3]3(OH), PDF#19-0272), which have similar diffraction characteristics.

Fig. 3
figure 3

μ-SRXRD (a) and Micro-FT-IR (b) results of PBs in ovarian serous cancer (up) and teratoma (down). a the mineral phase of PBs was identified as hydroxyapatite (Ca5(PO4)3(OH), PDF#09-0432) or carbonate hydroxyapatite (Ca5[PO4,CO3]3(OH), PDF#19-0272). b Micro-FT-IR spectra confirmed the existence of carbonate hydroxyapatite. The strong peak around 1,040 cm−1 is associated with ν3(PO4 3−), while the peak at 963 cm−1 is due to ν1(PO4 3−). IR bands around 1,400–1,500 cm−1 and 880–870 cm−1 are assigned to the ν3, ν2 mode of the carbonate group, respectively. The broad bands between 2,800 and 3,600 cm−1 are attributed to stretching vibration of OH (~3,000–3,600 cm−1) and ν1(HPO4 2−)(~2,850–2,950 cm−1). Amide I and the H-O-H bending mode of water result in the band at ~1,662 cm−1

Full size image

Further, micro-FT-IR spectra confirmed the existence of carbonate hydroxyapatite (Fig. 3b). The strong peak around 1,040 cm−1 is associated with ν3(PO4 3−), while the peak at 963 cm−1 is due to ν1(PO4 3−). IR bands around 1,400–1,500 cm−1 and 880-870 cm−1 are assigned to the ν3, ν2 mode of the carbonate group, respectively. The broad bands between 2,800 and 3,600 cm−1 are attributed to stretching vibration of OH (~3,000-3,600 cm−1) and ν1(HPO4 2−)(~2,850-2,950 cm−1). Amide I and the H-O-H bending mode of water result in the band at ~1,662 cm−1. After fitting procedure in the 900-800 cm−1 spectral range, the intensity of 872 and 880 cm−1 was used to calculate the percentage of A type hydroxyapatite (replacement of CO3 2− for OH−) and B type hydroxyapatite (replacement of CO3 2− for PO4 3−) (Mangialardo et al. 2012). The results showed the predominance of B substitution in all cases with an average B substitution/A substitution ratio of 7.1 ± 3.3.

TEM analysis was performed on one case of ovarian serous cancer and one case of teratoma. In ovarian serous cancer, minerals with two different morphologies were observed. One morphology was columnar crystal (Fig. 4a), which was 10–15 nm in length and about 5 nm in width, and with a Ca/P (At%) ratio of 1.76. Its diffraction pattern (Fig. 4j) showed dispersed polycrystalline diffraction rings, and the lattice photograph (Fig. 4b) showed small crystalline domains, indicating the selected area is the aggregate of nanocrystals. A region of the lattice photograph was indexed after FFT, inverse FFT and calibration (Fig. 4c). Take the diffraction pattern (Fig. 4j) into consideration, the interplanar spacing d values of 0.35 nm and 0.28 nm can be assigned to the crystal faces of (002) and (112) of carbonate hydroxyapatite (ICDD, 19-0272), respectively. The single crystalline diffraction behavior of (002) indicated the preferred orientation on this face (Fig. 4j). The other morphology was granulates or dumbbells with 20-100 nm in diameter (Fig. 4d). The lattice photography (Fig. 4e) showed some areas of one-dimensional lattices (circles), which might be assigned to approximately 5 nm globules (arrow) in the center of granules. The d value of the one-dimensional lattices is approximately 0.35 nm, which can be attributed to the crystal face of (002) in either hydroxyapatite (3.44 Å) or carbonate hydroxyapatite (3.46 Å). Its Ca/P (At%) ratio was 1.87, which was also assigned to be carbonate hydroxyapatite. In teratoma, the morphology of minerals (Fig. 4g) is similar as the columnar crystals in ovarian serous cancer. But the size of single crystal was slightly smaller, which was 5-12 nm in length and about 5 nm in width. Although the lattice photograph (Fig. 4h, i) showed similar interplanar spacing d values with that in Fig. 4c, its diffraction pattern (Fig. 4k) showed more complicated diffraction rings and dots. The mineral phase in teratoma was identified as hydroxyapatite (ICDD, 09-0432) after calibration and indexed in Fig. 4k. It was distinguished from carbonate hydroxyapatite by visible split (circle) among (211) (2.81 Å), (112) (2.78 Å) and (300) (2.72 Å). Preferred orientation was also observed with the arching behavior of the (002) diffraction.

Fig. 4
figure 4

TEM morphology (a, d, g), lattice photograph (b, c, e, f, h, i) and diffraction pattern (j, k) of the minerals in ovarian serous cancer (a–f, j) and teratoma (g–i, k). a morphology of columnar crystal, which was 10–15 nm in length and about 5 nm in width. b lattice photograph of selected area in a. c magnified lattice photograph of selected area in b. The interplanar spacing d values of 35 nm and 28 nm can be assigned to the crystal faces of (002) and (112) of carbonate hydroxyapatite (ICDD, 19–0272), respectively. d morphology of granulates or dumbbells with 20–100 nm in diameter. Arrows point to approximately 5 nm globules in the center of granules. e lattice photograph of selected area in d. Areas of one-dimensional lattices were circled out., which might be assigned to 5 nm globules in d. f magnified lattice photograph of selected area in e. The d value of the one-dimensional lattices is 35 nm, which can be attributed to the crystal face of (002) in either hydroxyapatite or carbonate hydroxyapatite. g morphology of columnar crystals with slightly smaller size, which was 5–12 nm in length and about 5 nm in width. h lattice photograph of selected area in g. i magnified lattice photograph of selected area in h. Similar interplanar spacing d values with that in c was observed. j diffraction pattern of b and around. Dispersed polycrystalline diffraction rings indicated the selected area is the aggregate of nanocrystals. The single crystalline diffraction behavior of (002) indicated the preferred orientation on this face. Mineral faces of carbonate hydroxyapatite (ICDD, 19–0272) were labeled. k diffraction pattern of h and around. Mineral faces of hydroxyapatite (ICDD, 09–0432) were labeled after calibration. The visible split (circle) among 2.81 Å, 2.78 Å and 2.72 Å distinguished it from carbonate hydroxyapatite. Preferred orientation of (002) was also observed with the arching behavior of the (002) diffraction

Full size image

The μ-SRXRF results (Fig. 5) revealed the peaks of Ca, Zn Fe and Sr of the PBs in ovarian serous cancer (Fig. 5a) and teratoma (Fig. 5b) as well as the expected Ti and Br introduced by 3 M film (Fig. 5a). Compared with the curve of HA (Fig. 5b), PBs may contain certain amount of Zn, Fe and Sr. Further, we conducted μ-SRXRF mapping on a 80 μm × 80 μm region of a PB in ovarian serous cancer. The results (Fig. 6) revealed the distribution of Zn were generally associated with Ca. Although Fe was shown to be in the minerals by μ-SRXRF analysis, the mapping result showed a much lower counts of Fe in calcified region comparing with the nearby tissues. Therefore, it could not be excluded that Fe was introduced by trace amount of organic residual.

Fig. 5
figure 5

Fitting curves of μSRXRF results. Zn, Fe and Sr may exist in PBs. a 3 M matrix, b HA, c PBs in ovarian serous cancer, d PBs in teratoma

Full size image
Fig. 6
figure 6

μSRXRF mapping results of psammoma bodies in ovarian serous cancer. Zn2+ might substitute Ca2+ in the lattice of carbonate hydroxyapatite

Full size image

Discussions

Since few necrotic tissues were observed, our current study did not support the dystrophic mechanism of PBs. However, granules with 20-100 nm in diameter were observed by TEM even after separating procedures, which were previously reported to resemble membrane bound vesicles (Ferenczy et al. 1977; Yamashima et al. 1986; Credá-Nicolás 1992). However, the nature of the granules in amorphous state is obscure. The granules can also be amorphous calcium phosphate (Jiang et al. 2012), which is a common precursor of crystalline HAPs. It is not easy to judge whether they are inorganics or undissolved organics when they intergrowth with crystalline solid. Anyway, the lattice photograph (Fig. 4e) showed crystalline globules with diameter of 5 nm distributed among amorphous granules, indicating a tendency of crystallization. It suggested the globules with 0.35 nm d-spacing might be the precursors of PBs.

Since spherules in the centre and periphery of PBs were similar, in both morphology and composition, the two parts might be homologous and the layered PBs presumably resulted from coalescence and accumulation of the nano-spherules. The confluent spherules with two crystalline cores indicated the inside-out development sequence and the possibility of confluence of two PBs in later period.

The POM and ESEM of this study demonstrated the development of PBs is closely related to collagen fibers, considering their location and distribution density. Previously, many authors have claimed the importance of collagen fibers in the formation of the calcified structures in meningioma (Credá-Nicolás 1992; Tsuchida et al. 1996; Kubota et al. 1984). In their studies, collagen as well as elastic fibers, which laid adjacent and parallel to the crystals in the meningocytic whorls, oriented the precipitation during growth of PBs. However, these whorl structures are absent in ovarian serous cancer. And in our study, collagen fibers encompassed the large PBs optionally and irregularly, except for a case of teratoma where small calcification was surrounded by circular collagen fibers. Therefore, the morphologically controlling function of the collagen fibers might work in the initiation phase, regulating the preferred orientation on (002) face of (carbonate) hydroxyapatite (Rhee et al. 2001). In another word, the (carbonate) hydroxyapatite crystals preferred to grow along the crystallographic c-axis which was preferentially aligned in some certain directions. Blood cells might be a calcium source since some blood red cells presumably carrying Ca2+ could be encompassed by collagen fibers and voluminous PBs were observed in the vicinity. There could be coordination between carboxyl and carbonyl on the collagen fibers and Ca2+ and Zn2+ carried by blood or tissue fluid (Alberts et al. 1998; Rhee et al. 2000; Zhang et al. 2003). Greater odds of c crystal surface exposure facilitate the absorption of protein molecule onto (carbonate) hydroxyapatite through carboxyl (and phosphate) groups (Kawasaki et al. 1985).

The mineral phase was determined comprehensively by several techniques. According to the μ-SRXRD, PBs consist of either hydroxyapatite (Ca10(PO4)6(OH)2) or carbonate hydroxyapatite (Ca10[PO4,CO3]6(OH)2). Then with the help of micro-FT-IR, the presence of carbonate at both [PO4 3−] and [OH−] sites was verified. The existence of HPO4 2− was also revealed. Therefore, the dominant mineral phase of PBs in both ovarian serous cancer and teratoma can be identified as AB-type carbonate hydroxyapatite (Ca10[(PO4)6-x-y(CO3)x(HPO4)y][(OH)2−u(CO3)u] with 0 ≤ x,y,u ≤ 2), which is commonly found in normal and pathological calcification in the human body (LeGeros 2001; Rey et al. 2007). When it comes to TEM observation, our measurement of d spacing was not accurate enough to distinguish hydroxyapatite from carbonate hydroxyapatite, since they are similar to each other, especially considering the crystal face with 3.44 Å spacing in hydrxyapatite and 3.46 Å in carbonate hydroxyapatite. Nevertheless, the split among (211) (2.81 Å), (112) (2.78 Å) and (300) (2.72 Å) can reveal the characteristics of hydrxyapatite. In carbonate hydroxyapatite, 2.78 Å should be the only peak in this range. Thus, hydroxyapatite also existed in the PBs, at least in teratoma, although carbonate hydrxyapatite was the dominant. It can be inferred that the carbonate substitution was unevenly distributed and the selected area contained less carbonate. Therefore, detailed researches of carbonate distribution will be of great help to explain the development of the minerals.

Some differences of mineralogical characteristics between PBs in ovarian cancer and those in teratoma were found in our study. The PBs in ovarian serous cancer had the maximum diameter of 35 μm. They consisted of columnar crystal that was 10-15 nm in length and ~5 nm in width (Fig. 4a), and granulates or dumbbells with 20-100 nm in diameter. In teratoma, the PBs with the maximum diameter up to 70 μm consist mainly of smaller columnar crystals with 5-12 nm in length and ~5 nm in width. It is also noteworthy that the concentric layers in teratoma are more complicated than that in ovarian serous cancer. Therefore, the crystal size and concentric structure can be important indication for physiological and pathological changes in the ovary.

μ-SRXRF results revealed the existence of Zn and Sr in PBs, thus establishing more relevance between PBs and human bones which also contains additional ions. Specially, Zn plays an important role in initiation and progression of cancer (Ho 2004). Therefore, to figure out the interaction among (carbonate) hydroxyapatite, zinc and organic tissue will be helpful for oncology. According to μ-SRXRF mapping results, Zn2+ might substitute Ca2+ in the lattice of carbonate hydroxyapatite. Unfortunately, the possible substitution failed to be shown by μ-SRXRD results, owing to the low concentration of Zn.

Conclusions

Mineralogical investigations were conducted on the microcalcification in PB form within human ovarian serous cancer and teratoma. The optimized separating procedure proved to be feasible to pick out the pathological calcification from the human tissues, thus overcoming the interference from the organics when investigating the mineralogical characteristics of microcalcification. Micro-area analysis, such as micro-FT-IR, TEM, μ-SRXRD and μ-SRXRF, was of great importance in measuring the PBs which are often in small quantity and tiny size.

The dominant mineral phase of the PBs was identified as AB-type carbonate hydroxyapatite (Ca10[(PO4)6-x-y(CO3)x(HPO4)y][(OH)2−u(CO3)u] with 0 ≤ x,y,u ≤ 2) with trace elements of Na, Mg, Zn and Sr. Collagen fibers may not only be adjacent to the PBs, but also play a crucial role on the preferred orientation of (002) face and the inside-out growth pattern of PBs. The crystal size and concentric structure can be important indication for physiological and pathological changes in the ovary. The size of columnar crystals was smaller and the concentric layer was more complicated in teratoma. The reveal of extra cations and anionic groups in the microcalcification provides new clues for pathological diagnosis and detailed information of crystallization will help with either promotion or inhibition of the crystals.