- Email: [email protected]
Prostatic Zn determination for prostate cancer diagnosis
Talanta 70 (2006) 914–921
Prostatic Zn determination for prostate cancer diagnosis夽 S.Sh. Shilstein a , M. Cortesi a , A. Breskin a , R. Chechik a,∗ , D. Vartsky b , G. Raviv d , N. Kleinman d , J. Ramon d , G. Kogan f , V. Gladysh e , E. Moriel f , M. Huszar e , A. Volkov c , E. Fridman c a
Department of Particle Physics, Weizmann Institute of Science, POB 26, Rehovot 76100, Israel b Soreq NRC, Yavne 81800, Israel c Department of Pathology, Sheba Medical Center, Tel Hashomer 52624, Israel d Department of Urology, Sheba Medical Center, Tel Hashomer 52624, Israel e Department of Pathology, Kaplan Medical Center, POB 1, Rehovot, Israel f Department of Urology, Kaplan Medical Center, POB 1, Rehovot, Israel Available online 22 June 2006
Abstract We present our studies of prostatic Zn concentration measurements, carried out in the light of a novel prostate cancer (CAP) diagnosis method proposed by us. The method is based on in vivo prostatic Zn mapping by XRF trans-rectal probe. We report on the extensive clinical studies, intended to assess the validity of the novel proposed diagnostic method. Zn content was measured in vitro in needle-biopsy samples from several hundreds of patients, and was correlated with histological findings and other patient parameters. For this purpose, a technique of absolute Zn content determination in ∼1 mm3 fresh tissue samples by XRF was developed. The experimental details and the main clinical-evaluation results are presented. We further outline the suggested design of the XRF trans-rectal probe for an efficient in vivo detection and mapping of the Zn fluorescence radiation from the prostate through the rectal wall. Laboratory phantom studies, a preliminary design concept and its expected performance are also reported. © 2006 Elsevier B.V. All rights reserved. PACS: 78.70.En.81.70.Jb.87.19.Xx Keywords: Prostate; Cancer diagnosis; Zinc; XRF; Biopsy
1. Introduction The diagnostic process for prostate cancer (CAP) usually starts with positive findings in digital rectal examination and/or in elevated prostate-specific antigen (PSA) serum levels, and ends with trans-rectal ultrasound-guided needle biopsy, in which the histological examination is the decisive diagnostic factor. This diagnostic process is costly and not very efficient, due the inadequate specificity and sensitivity of both digital rectal examination and PSA initial tests. In this context it is noteworthy that the needle-biopsy analysis, in addition to side effects, has a nonnegligible false-negative rate. It is thus understandable that new
夽 ∗
Presented at the Colloquium Spectroscopic International XXXIV. Corresponding author. Tel.: +972 8 934 4966; fax: +972 8 934 2611. E-mail address: [email protected] (R. Chechik).
0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2006.05.053
methods for CAP diagnosis are constantly being proposed and their efficacy investigated. In a previous article [1] we discussed the possibility of exploiting the Zn content in the prostate posterior part (peripheral zone) as a sensitive indicator for CAP development in this region. Many works were published in the past [2,3], reporting on high concentrations of Zn, of about 100–150 g/g wet weight (ppm) in normal human prostate tissue; this is several times higher than the Zn content in most other human tissues [4]. Furthermore, numerous in vitro studies indicated that the average Zn content in the prostate is reduced by about 2–5 times in cancerous tissue as compared to benign one [5–9]. The high Zn content in normal prostate tissue and the reported significant decrease in cancerous one can be measured using the XRF technique, which has adequate sensitivity of a few ppm and has been implemented on humans [10]. If the Zn content decrease is indeed proven to be a sufficiently sensitive and specific indica-
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
915
in the Zn content data in both malignant and non-malignant patient groups [1,5,9], it is clear that a large body of data is required to permit drawing statistically sound conclusions. Our extensive study has been carried out on several hundreds of patients undergoing needle-biopsy examination. This study is the first reported in the literature to encompass such a large number of patients (previously published works usually cite a few tens of cases [1,5–8]) and its statistical validity is thus unprecedented. The approach of using needle-biopsy samples rather than extracted organs allowed access to a large number of patients from all of whom the samples are systematically extracted according to a fixed protocol. It also permitted precise correlation of the Zn content data with histological examination and other patient data. Moreover, since for each patient typically about six or eight samples were measured, and in about half of them the samples were further scanned along the core, we are able to portray a broad picture of the Zn distribution within the prostate. 2.1. The XRF systems for needle-biopsy sample analysis
Fig. 1. The proposed XRF trans-rectal probe concept for prostate cancer diagnosis.
tor for CAP diagnosis, then the development of a Zn detection method would be perfectly adequate and highly attractive. We have proposed to do this using an XRF trans-rectal probe capable of mapping the absolute prostatic Zn content in the peripheral zone, behind the rectal wall (see Fig. 1). In this article we present two main aspects of the studies we have carried out to confirm this postulate: First, we describe an extensive clinical study on the relevance of prostatic Zn for CAP diagnosis, performed in vitro on needlebiopsy samples taken from several hundreds of patients. This unprecedentedly extensive study is essential for a reliable validation of the relation of Zn content to CAP. This study required the development of techniques for fresh tissue sample handling and absolute determination of the Zn content in very small samples, of ∼1 mm3 . Second, we present our conceptual design of a trans-rectal XRF probe for non-invasive in vivo mapping of prostatic Zn and, in particular, we report on the laboratory feasibility study of its detection through the rectal wall, with acceptable radiation exposures. 2. Clinical study of prostatic Zn content using needle-biopsy samples We have launched a systematic clinical study that aims at the evaluation of the diagnostic value of prostatic Zn. This was necessary since existing data fail to provide a clear picture of the spatial and temporal correlation between the malignancy process and the Zn content variation. Moreover, due to the large spread
The measurements of Zn content in prostate needle-biopsy samples were carried out, in parallel, in two medical centers, with the help of two tabletop XRF systems. The first system, used at the Kaplan Medical center (Rehovot, Israel), was a locally assembled system based on a commercially available air-cooled X-ray generator (XTF5011/75 TH Oxford Instruments, UK), and a 25 mm2 Peltier-cooled X-ray PIN diode detector (XR-1-CR Amptek). The X-ray generator incorporated a Mo target and the primary X-ray beam was filtered by a 50 m Mo foil. The PIN diode had a 3 mm in diameter entrance aperture made of pure aluminum. Up to six needle-biopsy samples per patient, each being a 0.5 mm in diameter and 15–20 mm long tissue cylinders folded in two, were placed at 8 mm distance on the sample-table; the latter was made of a 2.5 m Mylar foil stretched on a rectangular plastic frame. Wet sponges and water drops were placed at the table edges; the samples were covered with an identical Mylar foil (see Fig. 2a). This procedure was found to ensure minimal scattering from the table constituents; it also prevented sample drying during the measurement (measured to be ∼ =1%). The sample-table was linearly translated manually, introducing one sample at a time into the measurement site. The irradiated area on the sample-table was 4 mm × 9 mm, defined by a rectangular aperture on the primary beam. Signals from the X-ray detector, after proper amplification and time-shaping, were analyzed by a PC-borne multi-channel analyzer; the stored spectra were then analyzed and the intensities of the Zn fluorescence peak as well as the Compton-scattering peak were recorded, which permitted deriving the absolute Zn content as explained below. The measurements, carried out over a fixed time duration of 120 s per sample, yielded the average Zn content per sample with a precision (defined by the number of counts within the peaks) ranging from 30% to 5%, for 30 and 500 ppm, respectively. Due to the manual handling of sample displacement and data acquisition, and the relatively long measurement time per sample (dictated
916
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Fig. 2. The needle-biopsy sample support used with the home assembled (a) and the commercial (b) XRF systems. In (a) wet sponges are included next to the six samples to avoid drying; in (b) each sample is secured between the two foils closing the plastic and metal cups (right), when these are tightened together (left).
by the limited X-ray generator flux), each patient measurement took 20–30 min. The second system, used at the Sheba Medical Center (TelHashomer, Israel), was a commercial unit (EX-2600 bench top XRF analyzer manufactured by Jordan Valley Applied Radiation, Israel), custom-modified for our application. With its optimized assembly geometry, collimators and filters (150 m Mo filter on the primary X-ray beam from a Mo target), it had about 20-fold higher X-ray flux and superior spectrum quality compared to the XRF instrument described above. In this system, up to eight needle-biopsy samples, aligned in the tangential direction, were mounted on a rotating table. Each sample was placed on a 2.5 m Mylar foil fixed on standard plastic cups (inner diameter 24 mm) which, in turn, were tightly fitted into the Al cups placed on the instrument table; the Al cups were sealed by an identical Mylar foil. (To eliminate contribution from traces of Zn in the plastic cups material, a thin Al layer was inserted in these cups, see Fig. 2b.) Thus each sample was individually sandwiched between the two Mylar foils, which prevented them from drying. The sample loading procedure took only a couple of minutes per patient; the table movement, the data acquisition and the analysis were performed automatically. This XRF system was designed to scan the samples and measure the Zn content within four different segments, each 5 mm long, along the sample. For this purpose, the irradiated area was limited to 2 mm × 5 mm and four different spectra were acquired and analyzed for each sample. The duration of each acquisition point lasted 20 s, with a total of 90 s per sample and typically 12 min per patient. The resulting precision was 15% to 5% for levels of 50 to 200 ppm, respectively.
2.2. Absolute Zn content and calibration standards The goal of these measurements was to determine the absolute Zn concentration in the needle-biopsy samples. We assumed that the Zn amount is proportional to the intensity of the Zn K␣ line, I(Zn)sample , which we obtained from the X-ray spectra by integrating the intensity under the peak and subtracting the background. The sample mass is unknown and could not be directly measured; instead, we used the Compton-scattering intensity, I(Cs), obtained from the same X-ray spectrum by integrating the intensity under the Compton Mo K␣ peak (see Fig. 3). From this quantity we subtracted the contribution I(Cs)empty from the empty table or empty cup (typically amounting to less than 30% of I(Cs)), and calculated I(Cs)sample = I(Cs) − I(Cs)empty . The relation between I(Cs)sample and the actual sample weight was derived from calibration standards made for this purpose. The calibration standard is a sample having a known Znst (ppm) concentration and designed to have a geometry and composition similar to that of a tissue sample. A measurement of the calibration standard yielded the respective I(Zn)st and I(Cs)st intensities (the latter is the value after correction for the empty table or cup), from which the calibration coefficient K was calculated: K=
I(Cs)st × Znst (ppm) I(Zn)st
(1)
The Zn concentration in the sample is given by Znsample (ppm) =
K × I(Zn)sample I(Cs)sample
(2)
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
917
the average results of all calibration measurements. Another stable standard was produced from a 0.5 mm thick rectangular Lucite bar, on which a drop of Zn-loaded varnish was deposited; it was used to inter-calibrate both XRF systems, ensuring that absolute Zn concentration values were consistent in both sets of data. Examples of X-ray spectra measured at Sheba Medical Center from biopsy samples and from an empty cup are shown in Fig. 3. 2.3. Measurements in clinic
Fig. 3. XRF spectra recorded from an empty cup and biopsy samples at Sheba Medical Center. (a) An empty cup and a sample with 48 ppm Zn; (b) an empty cup and a sample with 233 ppm Zn.
The calibration coefficient K was determined from measurements performed with various standard samples. It is most important for the standard sample to have a similar composition to the tissue, because both the self-absorption effect and the Compton scattering are very strongly dependent on the composition. In our geometry the scattering strongly depends also on sample shape and size; we therefore prepared standards of similar dimensions to those of the tissue samples. We have used samples extracted with a needle-biopsy instrument, from a hardboiled egg-white substance, previously loaded with a known amount of Zn. Only very short measurements could be performed on the egg samples, which tend to dry rapidly. A more practical standard sample was made of a 0.5 mm inner diameter and ∼0.25 mm thick wall polyethylene tube, filled with tissueequivalent solution of known Zn concentration. In this case the I(Cs)empty was measured with an empty tube, to account for the tube wall contribution to the Compton peak, and the wall attenuation correction was measured independently from the X-ray fluorescence of a thin Cu wire placed inside this plastic tube. Another very convenient standard sample was made of Vaseline loaded with a known Zn concentration, and placed on the sample holder foil with a syringe to obtain the correct size and shape. Sandwiched between two foils, this standard was stable over many months; it is routinely used to check the system calibration stability, together with a thin Cu wire used to verify the energycalibration stability. The calibration factor K was derived from
The above-described tabletop XRF systems were located at the ultrasound-guided biopsy clinics in both medical centers. The patient-extracted samples were immediately placed on their respective supports and covered with a temporary, water-loaded cover, while waiting to be measured. Following the measurement described above, the rectal end of all samples was marked with a black tissue-ink (Black India, No. 4418, Item # 44204 by Stanford, USA) and they were stored in Formaldehyde, in separate tubes; track was kept of the extraction prostate site. The samples were then processed in a routine way at the pathology departments: embedding in paraffin wax, slicing into 4 m thick slices, and staining with hematoxylin and eosin. The slides were examined by pathologists to provide the diagnosis and the %gland, namely the fraction of surface occupied by the glandular tissue. At Sheba Medical Center this analysis was done in the four sections along the sample, to match the four sections Zn measurement. The diagnosis is adenocarcinoma (CAP), benign prostatic hyperplasia (BPH), prostatic intraepithelial neoplasia (PIN), atypical small acinar proliferation (ASAP) or granulomatous inflammation (GRAN) [11–13]. The measured Zn concentration (e.g. per measurement point, averaged per sample or averaged per patient) was correlated with various parameters from the histological examination and from the patient’s files. Some of the results are presented below. 2.4. Results The currently ongoing clinical study in both medical centers has so far resulted in samples examined from 310 patients. As explained above, the data can be analyzed to show various correlations of measured Zn levels with patient’s examination parameters. In the data analysis we applied three categories of Zn content averaging: I. no averaging, using single measurement points (SMP) at Sheba Medical Center (i.e. one quarter of a full needlebiopsy core); II. per-sample Zn concentration data (averaging four SMPs at Sheba Medical Center or using per-sample data at Kaplan Medical Center); III. per-patient Zn concentration data (averaging all samples of a given patient). A general conclusion from our analysis is that category III data (the per-patient data) has limited value for cancer diagno-
918
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Table 1 Zinc, %glands and diagnosis from 31 single measurement points of a single patient Sample no.
L1 L2 L3 L4 R5 R6 R7 R8
Fourth quarter
Third quarter
Second quarter
First quarter (rectal end)
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
35 39 16 29 190
CAP CAP CAP CAP BPH
60 30 10 80 5
342 343
BPH BPH
10 10
45 41 17 21 140 302 194 269
CAP CAP CAP CAP BPH BPH BPH BPH
40 30 15 80 10 10 10 10
39 43 20 19 145 225 215 256
CAP CAP CAP CAP BPH BPH BPH BPH
20 20 15 80 15 10 15 10
40 58 25 26 129 195 230 248
CAP CAP CAP CAP BPH BPH BPH BPH
20 10 5 10 15 10 15 10
Average zinc/sample
40 45 19 24 151 241 245 279
This case presents a perfect match between low Zn and CAP.
sis, with low prospects of yielding better diagnosis than PSA. On the other hand, category I (the single measurement point) Zn concentration data exhibit very good prospects for yielding a diagnostic tool that is much superior to PSA, with diagnostic power approaching that of biopsy (see below). Category II data yielded an intermediate result. This indicates a local Zn-decrease process (at least in the first stage of the malignancy development), the effect of which is diluted when averaged over a large tissue volume. This conclusion is, of course, very important for the design of the trans-rectal XRF probe, discussed below. Table 1 provides the measured Zn concentration and %gland values as well as the histological diagnosis in 32 SMPs from a single patient. In this particular case, an exceptionally clear match is exhibited between low Zn concentration values and the occurrence of cancer. Such clear-match cases are however, rare. In the majority of cases, as is typical for biological data, the fluctuations in Zn concentration are large, and the clinical value of the Zn level may only be evident statistically. This is well demonstrated in Fig. 4: it is based on SMPs and shows the normalized distribution of CAP and BPH diagnosis rate versus zinc concentration. It is obvious from this figure that the two groups have different distributions, peaking at different
Fig. 4. Normalized distributions of BPH and CAP diagnosis vs. zinc concentration, based on ∼5000 single measurement point data. The two distributions peak at different Zn values but they are both broad and largely overlap.
Zn values; however, they are both broad and have significant overlap. Fig. 5a shows the average zinc content in SMPs diagnosed as cancer or benign, selected by their location along the sample;
Fig. 5. (a) The average zinc concentration, based on ∼5000 single measurement point data, in the four sections along the sample core, for CAP and BPH. The distribution is flat. (b) The occurrence of cancerous tissue samples in the four sections along the sample core.
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Fig. 6. The average zinc concentration vs. %gland, based on ∼5000 single measurement point data of Zn level, %glands and diagnosis. The error bars represent the standard deviation of the data in each data group. In spite of the large spread in the Zn values, it is possible to delineate the two diagnosis groups within clearly distinct limits.
the response is flat, in the sense that there is no dependence on the location along the sample. Fig. 5b is similarly based on SMPs and presents the relative occurrence of CAP-diagnosed tissue sections selected by their location along the sample; there is no dependence on location except for a slight drop at the most distant location. The %glands obtained from histology carries further information on the malignant process, as it significantly increases due to the excessive multiplication of epithelial cells. The malignant process involves the loss of cells’ ability to accumulate zinc and a change in their metabolism. The average zinc concentration versus %gland, based on ∼5000 SMPs data, is plotted in Fig. 6. The error bars represent the standard deviation of the data in each data group. In spite of the large spread in the Zn values, it is possible to delineate the two diagnosis groups within distinct limits. The diagnostic power of a given criterion (i.e. the value of a given parameter or set of parameters) is best judged by the curve depicting the relation between true positive to false positive rate variation with this criterion; the larger the surface under the curve the higher its diagnostic power. In Fig. 7 we show such curves for PSA, Zn and Zn content normalized to %glands, based on SMP data. Judging by the area under the curves, Zn content analysis has diagnostic value superior to PSA, whereas Zn content normalized to %glands is far better than either. The detailed discussion of these results will be given elsewhere. 3. Feasibility study of in vivo measurement of prostatic Zn content 3.1. Laboratory phantom studies The general scheme of the proposed apparatus for CAP diagnosis by detection of Zn content loss is shown on Fig. 1: A
919
Fig. 7. Curves of sensitivity vs. specificity for PSA, Zn (SMP) and Zn normalized to %glands (single measurement points). An ideal diagnostic tool will have 100% sensitivity and 100% specificity, namely a step function with area 100%.
trans-rectal probe containing an X-ray source, or X-ray guiding system connected to an X-ray generator, irradiates the prostate through the rectal wall. An array of X-ray detectors generates the spectrum of the scattered and fluorescence radiation emitted from the tissue. The most severe impediments to designing such a probe are the low penetrability of the characteristic ∼8.64 keV Zn X-rays, and the necessity of obtaining precise quantitative data under conditions of appreciable scattering from the surrounding tissue. Moreover, it is absolutely mandatory to eliminate any contribution (Zn fluorescence and scattering) from the rectal wall tissue, situated between the probe and the examined tissue. The wall is typically 2–3 mm thick, and unless special precaution is taken, the scattered and fluorescence radiation from this tissue will dominate the spectrum. To study this problem we have used a laboratory phantom, made of two compartments, a thin one representing the rectal wall and a thick one representing the prostate, filled with tissue-equivalent solutions loaded with various concentrations of Zn. We have demonstrated [14] that the above-discussed impediments may be surmounted by using a collimation scheme that strictly limits the solid angle of both primary and fluorescence + scatterred X-ray beams with an appropriate collimator. We have shown that the Zn concentration can be reliably measured within the focal volume defined by the intersection of the two collimated beams, and when this focal volume is placed in the “prostate compartment” of the phantom, it provides data free of contributions from the “rectal wall compartment”. This method should permit eliminating the spectral contributions from the rectal wall tissue, reliably measuring the Zn concentration in the prostate tissue at depths up to several millimeters. The rectal wall will nevertheless cause considerable attenuation of the primary and fluorescence beams, and give rise to some scattering into the detector. Consequently, the main factor limiting the radiation intensity is the dose absorbed in the
920
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
4. Summary
Fig. 8. A conceptual scheme of the probe: an array of confocal thin collimators, drilled in an 8 mm metal plate, is placed between the tissue and the annular, cooled detector. It protects the detector from scattered radiation and defines the examined tissue element.
rectal wall. Using the laboratory phantom data, we have calculated the integrated fluence of incident radiation required to obtain evaluation of the Zn content in the prostate. Assuming an annular X-ray detector (see Figs. 1 and 8) of area ∼200 mm2 and including a 3 mm thick rectal wall attenuation effect, we have determined the fluence required to measure Zn levels of 200–50 ppm to a precision of 10%. This resulted in an incident fluence of 4 × 109 to 2 × 1010 photons/cm2 , for the 200 and 50 ppm, respectively. These fluences are equivalent to radiation exposures of 0.8–4 R. These exposures are of the same order of magnitude as the typical skin entrance exposure values in mammography, which range between 0.5 and 5 R for thin and thick breasts, respectively [15]. Calculations of the absorbed dose using Monte Carlo simulations are in progress at present. A large detection solid angle is beneficial for improving the data statistics and reducing the dose. 3.2. A probe scheme The main components in the proposed probe are the X-ray source, of a narrow energy spectrum centered at ∼17.4 keV energy, an annular-shape ∼20 mm diameter array of spectroscopic X-ray detectors, possibly cooled, and a multi-collimator system, arranged to have the focal volume within the examined tissue. The latter is realized with a few mm thick metallic steel or aluminum plate, perforated with confocal holes of small diameter (0.1–0.2 mm), pointing at one focal volume within the prostate; the holes may be cylindrical or of a slightly conical profile (Fig. 8). The primary beam, focused into the gland, passes through a central collimating aperture in the multi-collimator plate, and through a corresponding opening in the annularshaped X-ray detector, placed next to the collimator plate. The intersection of the primary beam with the focal volume of the multi-collimator plate defines the target volume within the prostate; the fluorescence radiation from this volume will be detected by the annular detector that subtends a large solid angle. In this design the detector will be well shielded from scattered radiation.
We have presented the idea of a novel prostate cancer diagnosis tool, based on in vivo mapping of the Zn concentration in the posterior prostate tissue, where most malignant tumors develop. The Zn examination is non-invasive and will be performed on patients prior to the biopsy examination, to classify them better prior to this invasive and costly test. To this end we launched a clinical study aiming at evaluating the diagnostic value of the prostatic Zn content measurement. The study, currently ongoing in two medical centers, is based on Zn concentration measurements in needle-biopsy samples from several hundreds of patients; the data analysis is performed in correlation with the patient’s histological diagnosis and with other patient parameters. Two tabletop XRF systems were developed and adapted for this study and special care was taken to develop tools and protocols for system calibration and for handling of the XRF measurement on such fragile, small-volume samples. These new XRF measurement protocols, performed in between the biopsy procedure and the histological analysis, were shown not to interfere with standard clinical procedures and not damage the biopsy samples. The unprecedentedly large body of data on prostatic Zn content, accumulated within the framework of this clinical study, is important for evaluating the sensitivity and specificity of the newly proposed prostate cancer diagnosis concept. We have presented some data showing the clear difference in Zn levels in BPH (benign) and CAP (cancer) tissue, and the problem of overlap in the distributions from these two patient groups. The sensitivity vs. specificity curves based on single measurement points (SMP) data accumulated so far summarize the diagnostic power and show that Zn concentration is indeed a more indicative parameter than PSA: for example, for detecting 80% of cancer points the specificity of PSA is ∼25% and the specificity of Zn is 45% (almost a factor of 2!). It indicates that as a pre-biopsy criterion Zn could be much superior to PSA and using it may reduce the rate of non-malignant cases that are currently referred to biopsy. (This rate today is typically 75–78% worldwide.) Such a procedure will render the costly biopsy test more efficient and cost effective, and will permit extending the biopsy test to a broader population, e.g. younger age-group, lower PSA, etc. Moreover, the proposed CAP diagnostic method involves mapping of the Zn concentration, namely providing detailed information on the spatial distribution of Zn; this will provide further possibilities for distinguishing between random local fluctuation in Zn content (noise) or a systematic Zn reduction over an extended region, supposedly related to cancer lesion. The aspect of spatial correlations between Zn and CAP is the subject of another study, to be reported elsewhere. In view of the promising preliminary results from the clinical study, and the conclusion that Zn should be measured locally and not averaged over a large tissue volume, we have developed the concept of a trans-rectal XRF probe, which measures and maps the prostatic Zn content through the rectal wall. A conceptual probe design was outlined, and the results of some laboratory tests with phantoms were reported. It is claimed that, with an appropriate collimation of the primary and flu-
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
orescence + scattered beams, it should be possible to reliably measure the Zn content in a small volume of the prostate tissue, in vivo, without the interference of the rectal wall tissue. The proposed probe is estimated to yield the prostatic Zn content within 10–20 s per measuring point, and within a permissible absorbed dose in the rectal wall, of 0.13–0.5 mGy. Acknowledgments We would like to thank Dr. H. Gutman and Dr. A. Gvirtzman of Jordan Valley Applied Radiation for a very fruitful collaboration, Y. Asher, M. Klin and R. Lozovsky for their technical assistance and. Y. Gil and Y. Telner for their skillful handling of the biopsy samples. We are indebted to the medical staff in both medical centers for their patience and collaborative spirit. The work was partially supported by the Horowitz Fund of the Weizmann Institute of Science. A. Breskin is the W.P. Reuther Professor of Research in the Peaceful use of Atomic Energy. References [1] D. Vartsky, S.Sh. Shilstein, A. Bercovich, M. Huszar, A. Breskin, R. Chechik, S. Korotinsky, S.D. Malnick, E. Moriel, J. Urol. 170 (2003) 2258–2262.
921
[2] R. Hoare, G.E. Delory, D.W. Penner, Cancer 9 (1956) 721–726. [3] M. Shirakawa, Acta Urol. Jpn. 3 (1961) 352. [4] R.S. Cotran, V. Kumar, T. Collins, Robbins Pathologic Basis of Disease, 6th ed., W.B. Saunders Co., Philadelphia, 1999, p. 1029. [5] F. Gyorkey, K.W. Min, J.A. Huff, P. Gyorkey, Cancer Res. 27 (1967) 1348–1353. [6] F.K. Habib, M.K. Mason, P.H. Smith, S.R. Stitch, Br. J. Cancer 39 (1979) 700–704. [7] R. Lahtonen, Prostate 6 (1985) 177–183. [8] J.O. Ogunlewe, D.N. Osegbe, Cancer 63 (1989) 1388–1392. [9] V.Y. Zaichick, T.V. Sviridova, S.V. Zaichick, Int. Urol. Nephrol. 29 (1997) 565–574. [10] L.J. Somervaille, D.R. Chettle, M.C. Scott, Phys. Med. Biol. 30 (1985) 929–943. [11] Christopher, in: S. Foster, D.G. Bostwick (Eds.), Pathology of the Prostate, 1st ed., W.B. Saunders Company, 1998. [12] P.A. Humphrey, Prostate Pathology, Illustrate edition, American Society Clinical Pathology, 2003. [13] J.N. Ebele, G. Sauter, J.I. Epstein (Eds.), WHO Classification of Tumors: Pathology and Genetics of Tumors of the Urinary System and Male Genital Organs (Who/IARC Classification of Tumors) International Agency for Research on Cancer, 2004. [14] S.Sh. Shilstein, A. Breskin, R. Chechik, G. Feldman, D. Vartsky, Phys. Med. Biol. 49 (2004) 485–499. [15] Xizenh Wu, in: G.T. Barnes, G.D. Frey (Eds.), Screen Film Mammography-Imaging Considerations and Medical Physics Responsibilities, Medical Physics Publishing, Madison, WI, 1991, pp. 159–175.
Prostatic Zn determination for prostate cancer diagnosis夽 S.Sh. Shilstein a , M. Cortesi a , A. Breskin a , R. Chechik a,∗ , D. Vartsky b , G. Raviv d , N. Kleinman d , J. Ramon d , G. Kogan f , V. Gladysh e , E. Moriel f , M. Huszar e , A. Volkov c , E. Fridman c a
Department of Particle Physics, Weizmann Institute of Science, POB 26, Rehovot 76100, Israel b Soreq NRC, Yavne 81800, Israel c Department of Pathology, Sheba Medical Center, Tel Hashomer 52624, Israel d Department of Urology, Sheba Medical Center, Tel Hashomer 52624, Israel e Department of Pathology, Kaplan Medical Center, POB 1, Rehovot, Israel f Department of Urology, Kaplan Medical Center, POB 1, Rehovot, Israel Available online 22 June 2006
Abstract We present our studies of prostatic Zn concentration measurements, carried out in the light of a novel prostate cancer (CAP) diagnosis method proposed by us. The method is based on in vivo prostatic Zn mapping by XRF trans-rectal probe. We report on the extensive clinical studies, intended to assess the validity of the novel proposed diagnostic method. Zn content was measured in vitro in needle-biopsy samples from several hundreds of patients, and was correlated with histological findings and other patient parameters. For this purpose, a technique of absolute Zn content determination in ∼1 mm3 fresh tissue samples by XRF was developed. The experimental details and the main clinical-evaluation results are presented. We further outline the suggested design of the XRF trans-rectal probe for an efficient in vivo detection and mapping of the Zn fluorescence radiation from the prostate through the rectal wall. Laboratory phantom studies, a preliminary design concept and its expected performance are also reported. © 2006 Elsevier B.V. All rights reserved. PACS: 78.70.En.81.70.Jb.87.19.Xx Keywords: Prostate; Cancer diagnosis; Zinc; XRF; Biopsy
1. Introduction The diagnostic process for prostate cancer (CAP) usually starts with positive findings in digital rectal examination and/or in elevated prostate-specific antigen (PSA) serum levels, and ends with trans-rectal ultrasound-guided needle biopsy, in which the histological examination is the decisive diagnostic factor. This diagnostic process is costly and not very efficient, due the inadequate specificity and sensitivity of both digital rectal examination and PSA initial tests. In this context it is noteworthy that the needle-biopsy analysis, in addition to side effects, has a nonnegligible false-negative rate. It is thus understandable that new
夽 ∗
Presented at the Colloquium Spectroscopic International XXXIV. Corresponding author. Tel.: +972 8 934 4966; fax: +972 8 934 2611. E-mail address: [email protected] (R. Chechik).
0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2006.05.053
methods for CAP diagnosis are constantly being proposed and their efficacy investigated. In a previous article [1] we discussed the possibility of exploiting the Zn content in the prostate posterior part (peripheral zone) as a sensitive indicator for CAP development in this region. Many works were published in the past [2,3], reporting on high concentrations of Zn, of about 100–150 g/g wet weight (ppm) in normal human prostate tissue; this is several times higher than the Zn content in most other human tissues [4]. Furthermore, numerous in vitro studies indicated that the average Zn content in the prostate is reduced by about 2–5 times in cancerous tissue as compared to benign one [5–9]. The high Zn content in normal prostate tissue and the reported significant decrease in cancerous one can be measured using the XRF technique, which has adequate sensitivity of a few ppm and has been implemented on humans [10]. If the Zn content decrease is indeed proven to be a sufficiently sensitive and specific indica-
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
915
in the Zn content data in both malignant and non-malignant patient groups [1,5,9], it is clear that a large body of data is required to permit drawing statistically sound conclusions. Our extensive study has been carried out on several hundreds of patients undergoing needle-biopsy examination. This study is the first reported in the literature to encompass such a large number of patients (previously published works usually cite a few tens of cases [1,5–8]) and its statistical validity is thus unprecedented. The approach of using needle-biopsy samples rather than extracted organs allowed access to a large number of patients from all of whom the samples are systematically extracted according to a fixed protocol. It also permitted precise correlation of the Zn content data with histological examination and other patient data. Moreover, since for each patient typically about six or eight samples were measured, and in about half of them the samples were further scanned along the core, we are able to portray a broad picture of the Zn distribution within the prostate. 2.1. The XRF systems for needle-biopsy sample analysis
Fig. 1. The proposed XRF trans-rectal probe concept for prostate cancer diagnosis.
tor for CAP diagnosis, then the development of a Zn detection method would be perfectly adequate and highly attractive. We have proposed to do this using an XRF trans-rectal probe capable of mapping the absolute prostatic Zn content in the peripheral zone, behind the rectal wall (see Fig. 1). In this article we present two main aspects of the studies we have carried out to confirm this postulate: First, we describe an extensive clinical study on the relevance of prostatic Zn for CAP diagnosis, performed in vitro on needlebiopsy samples taken from several hundreds of patients. This unprecedentedly extensive study is essential for a reliable validation of the relation of Zn content to CAP. This study required the development of techniques for fresh tissue sample handling and absolute determination of the Zn content in very small samples, of ∼1 mm3 . Second, we present our conceptual design of a trans-rectal XRF probe for non-invasive in vivo mapping of prostatic Zn and, in particular, we report on the laboratory feasibility study of its detection through the rectal wall, with acceptable radiation exposures. 2. Clinical study of prostatic Zn content using needle-biopsy samples We have launched a systematic clinical study that aims at the evaluation of the diagnostic value of prostatic Zn. This was necessary since existing data fail to provide a clear picture of the spatial and temporal correlation between the malignancy process and the Zn content variation. Moreover, due to the large spread
The measurements of Zn content in prostate needle-biopsy samples were carried out, in parallel, in two medical centers, with the help of two tabletop XRF systems. The first system, used at the Kaplan Medical center (Rehovot, Israel), was a locally assembled system based on a commercially available air-cooled X-ray generator (XTF5011/75 TH Oxford Instruments, UK), and a 25 mm2 Peltier-cooled X-ray PIN diode detector (XR-1-CR Amptek). The X-ray generator incorporated a Mo target and the primary X-ray beam was filtered by a 50 m Mo foil. The PIN diode had a 3 mm in diameter entrance aperture made of pure aluminum. Up to six needle-biopsy samples per patient, each being a 0.5 mm in diameter and 15–20 mm long tissue cylinders folded in two, were placed at 8 mm distance on the sample-table; the latter was made of a 2.5 m Mylar foil stretched on a rectangular plastic frame. Wet sponges and water drops were placed at the table edges; the samples were covered with an identical Mylar foil (see Fig. 2a). This procedure was found to ensure minimal scattering from the table constituents; it also prevented sample drying during the measurement (measured to be ∼ =1%). The sample-table was linearly translated manually, introducing one sample at a time into the measurement site. The irradiated area on the sample-table was 4 mm × 9 mm, defined by a rectangular aperture on the primary beam. Signals from the X-ray detector, after proper amplification and time-shaping, were analyzed by a PC-borne multi-channel analyzer; the stored spectra were then analyzed and the intensities of the Zn fluorescence peak as well as the Compton-scattering peak were recorded, which permitted deriving the absolute Zn content as explained below. The measurements, carried out over a fixed time duration of 120 s per sample, yielded the average Zn content per sample with a precision (defined by the number of counts within the peaks) ranging from 30% to 5%, for 30 and 500 ppm, respectively. Due to the manual handling of sample displacement and data acquisition, and the relatively long measurement time per sample (dictated
916
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Fig. 2. The needle-biopsy sample support used with the home assembled (a) and the commercial (b) XRF systems. In (a) wet sponges are included next to the six samples to avoid drying; in (b) each sample is secured between the two foils closing the plastic and metal cups (right), when these are tightened together (left).
by the limited X-ray generator flux), each patient measurement took 20–30 min. The second system, used at the Sheba Medical Center (TelHashomer, Israel), was a commercial unit (EX-2600 bench top XRF analyzer manufactured by Jordan Valley Applied Radiation, Israel), custom-modified for our application. With its optimized assembly geometry, collimators and filters (150 m Mo filter on the primary X-ray beam from a Mo target), it had about 20-fold higher X-ray flux and superior spectrum quality compared to the XRF instrument described above. In this system, up to eight needle-biopsy samples, aligned in the tangential direction, were mounted on a rotating table. Each sample was placed on a 2.5 m Mylar foil fixed on standard plastic cups (inner diameter 24 mm) which, in turn, were tightly fitted into the Al cups placed on the instrument table; the Al cups were sealed by an identical Mylar foil. (To eliminate contribution from traces of Zn in the plastic cups material, a thin Al layer was inserted in these cups, see Fig. 2b.) Thus each sample was individually sandwiched between the two Mylar foils, which prevented them from drying. The sample loading procedure took only a couple of minutes per patient; the table movement, the data acquisition and the analysis were performed automatically. This XRF system was designed to scan the samples and measure the Zn content within four different segments, each 5 mm long, along the sample. For this purpose, the irradiated area was limited to 2 mm × 5 mm and four different spectra were acquired and analyzed for each sample. The duration of each acquisition point lasted 20 s, with a total of 90 s per sample and typically 12 min per patient. The resulting precision was 15% to 5% for levels of 50 to 200 ppm, respectively.
2.2. Absolute Zn content and calibration standards The goal of these measurements was to determine the absolute Zn concentration in the needle-biopsy samples. We assumed that the Zn amount is proportional to the intensity of the Zn K␣ line, I(Zn)sample , which we obtained from the X-ray spectra by integrating the intensity under the peak and subtracting the background. The sample mass is unknown and could not be directly measured; instead, we used the Compton-scattering intensity, I(Cs), obtained from the same X-ray spectrum by integrating the intensity under the Compton Mo K␣ peak (see Fig. 3). From this quantity we subtracted the contribution I(Cs)empty from the empty table or empty cup (typically amounting to less than 30% of I(Cs)), and calculated I(Cs)sample = I(Cs) − I(Cs)empty . The relation between I(Cs)sample and the actual sample weight was derived from calibration standards made for this purpose. The calibration standard is a sample having a known Znst (ppm) concentration and designed to have a geometry and composition similar to that of a tissue sample. A measurement of the calibration standard yielded the respective I(Zn)st and I(Cs)st intensities (the latter is the value after correction for the empty table or cup), from which the calibration coefficient K was calculated: K=
I(Cs)st × Znst (ppm) I(Zn)st
(1)
The Zn concentration in the sample is given by Znsample (ppm) =
K × I(Zn)sample I(Cs)sample
(2)
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
917
the average results of all calibration measurements. Another stable standard was produced from a 0.5 mm thick rectangular Lucite bar, on which a drop of Zn-loaded varnish was deposited; it was used to inter-calibrate both XRF systems, ensuring that absolute Zn concentration values were consistent in both sets of data. Examples of X-ray spectra measured at Sheba Medical Center from biopsy samples and from an empty cup are shown in Fig. 3. 2.3. Measurements in clinic
Fig. 3. XRF spectra recorded from an empty cup and biopsy samples at Sheba Medical Center. (a) An empty cup and a sample with 48 ppm Zn; (b) an empty cup and a sample with 233 ppm Zn.
The calibration coefficient K was determined from measurements performed with various standard samples. It is most important for the standard sample to have a similar composition to the tissue, because both the self-absorption effect and the Compton scattering are very strongly dependent on the composition. In our geometry the scattering strongly depends also on sample shape and size; we therefore prepared standards of similar dimensions to those of the tissue samples. We have used samples extracted with a needle-biopsy instrument, from a hardboiled egg-white substance, previously loaded with a known amount of Zn. Only very short measurements could be performed on the egg samples, which tend to dry rapidly. A more practical standard sample was made of a 0.5 mm inner diameter and ∼0.25 mm thick wall polyethylene tube, filled with tissueequivalent solution of known Zn concentration. In this case the I(Cs)empty was measured with an empty tube, to account for the tube wall contribution to the Compton peak, and the wall attenuation correction was measured independently from the X-ray fluorescence of a thin Cu wire placed inside this plastic tube. Another very convenient standard sample was made of Vaseline loaded with a known Zn concentration, and placed on the sample holder foil with a syringe to obtain the correct size and shape. Sandwiched between two foils, this standard was stable over many months; it is routinely used to check the system calibration stability, together with a thin Cu wire used to verify the energycalibration stability. The calibration factor K was derived from
The above-described tabletop XRF systems were located at the ultrasound-guided biopsy clinics in both medical centers. The patient-extracted samples were immediately placed on their respective supports and covered with a temporary, water-loaded cover, while waiting to be measured. Following the measurement described above, the rectal end of all samples was marked with a black tissue-ink (Black India, No. 4418, Item # 44204 by Stanford, USA) and they were stored in Formaldehyde, in separate tubes; track was kept of the extraction prostate site. The samples were then processed in a routine way at the pathology departments: embedding in paraffin wax, slicing into 4 m thick slices, and staining with hematoxylin and eosin. The slides were examined by pathologists to provide the diagnosis and the %gland, namely the fraction of surface occupied by the glandular tissue. At Sheba Medical Center this analysis was done in the four sections along the sample, to match the four sections Zn measurement. The diagnosis is adenocarcinoma (CAP), benign prostatic hyperplasia (BPH), prostatic intraepithelial neoplasia (PIN), atypical small acinar proliferation (ASAP) or granulomatous inflammation (GRAN) [11–13]. The measured Zn concentration (e.g. per measurement point, averaged per sample or averaged per patient) was correlated with various parameters from the histological examination and from the patient’s files. Some of the results are presented below. 2.4. Results The currently ongoing clinical study in both medical centers has so far resulted in samples examined from 310 patients. As explained above, the data can be analyzed to show various correlations of measured Zn levels with patient’s examination parameters. In the data analysis we applied three categories of Zn content averaging: I. no averaging, using single measurement points (SMP) at Sheba Medical Center (i.e. one quarter of a full needlebiopsy core); II. per-sample Zn concentration data (averaging four SMPs at Sheba Medical Center or using per-sample data at Kaplan Medical Center); III. per-patient Zn concentration data (averaging all samples of a given patient). A general conclusion from our analysis is that category III data (the per-patient data) has limited value for cancer diagno-
918
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Table 1 Zinc, %glands and diagnosis from 31 single measurement points of a single patient Sample no.
L1 L2 L3 L4 R5 R6 R7 R8
Fourth quarter
Third quarter
Second quarter
First quarter (rectal end)
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
Zinc (ppm)
Diagnosis
%gland
35 39 16 29 190
CAP CAP CAP CAP BPH
60 30 10 80 5
342 343
BPH BPH
10 10
45 41 17 21 140 302 194 269
CAP CAP CAP CAP BPH BPH BPH BPH
40 30 15 80 10 10 10 10
39 43 20 19 145 225 215 256
CAP CAP CAP CAP BPH BPH BPH BPH
20 20 15 80 15 10 15 10
40 58 25 26 129 195 230 248
CAP CAP CAP CAP BPH BPH BPH BPH
20 10 5 10 15 10 15 10
Average zinc/sample
40 45 19 24 151 241 245 279
This case presents a perfect match between low Zn and CAP.
sis, with low prospects of yielding better diagnosis than PSA. On the other hand, category I (the single measurement point) Zn concentration data exhibit very good prospects for yielding a diagnostic tool that is much superior to PSA, with diagnostic power approaching that of biopsy (see below). Category II data yielded an intermediate result. This indicates a local Zn-decrease process (at least in the first stage of the malignancy development), the effect of which is diluted when averaged over a large tissue volume. This conclusion is, of course, very important for the design of the trans-rectal XRF probe, discussed below. Table 1 provides the measured Zn concentration and %gland values as well as the histological diagnosis in 32 SMPs from a single patient. In this particular case, an exceptionally clear match is exhibited between low Zn concentration values and the occurrence of cancer. Such clear-match cases are however, rare. In the majority of cases, as is typical for biological data, the fluctuations in Zn concentration are large, and the clinical value of the Zn level may only be evident statistically. This is well demonstrated in Fig. 4: it is based on SMPs and shows the normalized distribution of CAP and BPH diagnosis rate versus zinc concentration. It is obvious from this figure that the two groups have different distributions, peaking at different
Fig. 4. Normalized distributions of BPH and CAP diagnosis vs. zinc concentration, based on ∼5000 single measurement point data. The two distributions peak at different Zn values but they are both broad and largely overlap.
Zn values; however, they are both broad and have significant overlap. Fig. 5a shows the average zinc content in SMPs diagnosed as cancer or benign, selected by their location along the sample;
Fig. 5. (a) The average zinc concentration, based on ∼5000 single measurement point data, in the four sections along the sample core, for CAP and BPH. The distribution is flat. (b) The occurrence of cancerous tissue samples in the four sections along the sample core.
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
Fig. 6. The average zinc concentration vs. %gland, based on ∼5000 single measurement point data of Zn level, %glands and diagnosis. The error bars represent the standard deviation of the data in each data group. In spite of the large spread in the Zn values, it is possible to delineate the two diagnosis groups within clearly distinct limits.
the response is flat, in the sense that there is no dependence on the location along the sample. Fig. 5b is similarly based on SMPs and presents the relative occurrence of CAP-diagnosed tissue sections selected by their location along the sample; there is no dependence on location except for a slight drop at the most distant location. The %glands obtained from histology carries further information on the malignant process, as it significantly increases due to the excessive multiplication of epithelial cells. The malignant process involves the loss of cells’ ability to accumulate zinc and a change in their metabolism. The average zinc concentration versus %gland, based on ∼5000 SMPs data, is plotted in Fig. 6. The error bars represent the standard deviation of the data in each data group. In spite of the large spread in the Zn values, it is possible to delineate the two diagnosis groups within distinct limits. The diagnostic power of a given criterion (i.e. the value of a given parameter or set of parameters) is best judged by the curve depicting the relation between true positive to false positive rate variation with this criterion; the larger the surface under the curve the higher its diagnostic power. In Fig. 7 we show such curves for PSA, Zn and Zn content normalized to %glands, based on SMP data. Judging by the area under the curves, Zn content analysis has diagnostic value superior to PSA, whereas Zn content normalized to %glands is far better than either. The detailed discussion of these results will be given elsewhere. 3. Feasibility study of in vivo measurement of prostatic Zn content 3.1. Laboratory phantom studies The general scheme of the proposed apparatus for CAP diagnosis by detection of Zn content loss is shown on Fig. 1: A
919
Fig. 7. Curves of sensitivity vs. specificity for PSA, Zn (SMP) and Zn normalized to %glands (single measurement points). An ideal diagnostic tool will have 100% sensitivity and 100% specificity, namely a step function with area 100%.
trans-rectal probe containing an X-ray source, or X-ray guiding system connected to an X-ray generator, irradiates the prostate through the rectal wall. An array of X-ray detectors generates the spectrum of the scattered and fluorescence radiation emitted from the tissue. The most severe impediments to designing such a probe are the low penetrability of the characteristic ∼8.64 keV Zn X-rays, and the necessity of obtaining precise quantitative data under conditions of appreciable scattering from the surrounding tissue. Moreover, it is absolutely mandatory to eliminate any contribution (Zn fluorescence and scattering) from the rectal wall tissue, situated between the probe and the examined tissue. The wall is typically 2–3 mm thick, and unless special precaution is taken, the scattered and fluorescence radiation from this tissue will dominate the spectrum. To study this problem we have used a laboratory phantom, made of two compartments, a thin one representing the rectal wall and a thick one representing the prostate, filled with tissue-equivalent solutions loaded with various concentrations of Zn. We have demonstrated [14] that the above-discussed impediments may be surmounted by using a collimation scheme that strictly limits the solid angle of both primary and fluorescence + scatterred X-ray beams with an appropriate collimator. We have shown that the Zn concentration can be reliably measured within the focal volume defined by the intersection of the two collimated beams, and when this focal volume is placed in the “prostate compartment” of the phantom, it provides data free of contributions from the “rectal wall compartment”. This method should permit eliminating the spectral contributions from the rectal wall tissue, reliably measuring the Zn concentration in the prostate tissue at depths up to several millimeters. The rectal wall will nevertheless cause considerable attenuation of the primary and fluorescence beams, and give rise to some scattering into the detector. Consequently, the main factor limiting the radiation intensity is the dose absorbed in the
920
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
4. Summary
Fig. 8. A conceptual scheme of the probe: an array of confocal thin collimators, drilled in an 8 mm metal plate, is placed between the tissue and the annular, cooled detector. It protects the detector from scattered radiation and defines the examined tissue element.
rectal wall. Using the laboratory phantom data, we have calculated the integrated fluence of incident radiation required to obtain evaluation of the Zn content in the prostate. Assuming an annular X-ray detector (see Figs. 1 and 8) of area ∼200 mm2 and including a 3 mm thick rectal wall attenuation effect, we have determined the fluence required to measure Zn levels of 200–50 ppm to a precision of 10%. This resulted in an incident fluence of 4 × 109 to 2 × 1010 photons/cm2 , for the 200 and 50 ppm, respectively. These fluences are equivalent to radiation exposures of 0.8–4 R. These exposures are of the same order of magnitude as the typical skin entrance exposure values in mammography, which range between 0.5 and 5 R for thin and thick breasts, respectively [15]. Calculations of the absorbed dose using Monte Carlo simulations are in progress at present. A large detection solid angle is beneficial for improving the data statistics and reducing the dose. 3.2. A probe scheme The main components in the proposed probe are the X-ray source, of a narrow energy spectrum centered at ∼17.4 keV energy, an annular-shape ∼20 mm diameter array of spectroscopic X-ray detectors, possibly cooled, and a multi-collimator system, arranged to have the focal volume within the examined tissue. The latter is realized with a few mm thick metallic steel or aluminum plate, perforated with confocal holes of small diameter (0.1–0.2 mm), pointing at one focal volume within the prostate; the holes may be cylindrical or of a slightly conical profile (Fig. 8). The primary beam, focused into the gland, passes through a central collimating aperture in the multi-collimator plate, and through a corresponding opening in the annularshaped X-ray detector, placed next to the collimator plate. The intersection of the primary beam with the focal volume of the multi-collimator plate defines the target volume within the prostate; the fluorescence radiation from this volume will be detected by the annular detector that subtends a large solid angle. In this design the detector will be well shielded from scattered radiation.
We have presented the idea of a novel prostate cancer diagnosis tool, based on in vivo mapping of the Zn concentration in the posterior prostate tissue, where most malignant tumors develop. The Zn examination is non-invasive and will be performed on patients prior to the biopsy examination, to classify them better prior to this invasive and costly test. To this end we launched a clinical study aiming at evaluating the diagnostic value of the prostatic Zn content measurement. The study, currently ongoing in two medical centers, is based on Zn concentration measurements in needle-biopsy samples from several hundreds of patients; the data analysis is performed in correlation with the patient’s histological diagnosis and with other patient parameters. Two tabletop XRF systems were developed and adapted for this study and special care was taken to develop tools and protocols for system calibration and for handling of the XRF measurement on such fragile, small-volume samples. These new XRF measurement protocols, performed in between the biopsy procedure and the histological analysis, were shown not to interfere with standard clinical procedures and not damage the biopsy samples. The unprecedentedly large body of data on prostatic Zn content, accumulated within the framework of this clinical study, is important for evaluating the sensitivity and specificity of the newly proposed prostate cancer diagnosis concept. We have presented some data showing the clear difference in Zn levels in BPH (benign) and CAP (cancer) tissue, and the problem of overlap in the distributions from these two patient groups. The sensitivity vs. specificity curves based on single measurement points (SMP) data accumulated so far summarize the diagnostic power and show that Zn concentration is indeed a more indicative parameter than PSA: for example, for detecting 80% of cancer points the specificity of PSA is ∼25% and the specificity of Zn is 45% (almost a factor of 2!). It indicates that as a pre-biopsy criterion Zn could be much superior to PSA and using it may reduce the rate of non-malignant cases that are currently referred to biopsy. (This rate today is typically 75–78% worldwide.) Such a procedure will render the costly biopsy test more efficient and cost effective, and will permit extending the biopsy test to a broader population, e.g. younger age-group, lower PSA, etc. Moreover, the proposed CAP diagnostic method involves mapping of the Zn concentration, namely providing detailed information on the spatial distribution of Zn; this will provide further possibilities for distinguishing between random local fluctuation in Zn content (noise) or a systematic Zn reduction over an extended region, supposedly related to cancer lesion. The aspect of spatial correlations between Zn and CAP is the subject of another study, to be reported elsewhere. In view of the promising preliminary results from the clinical study, and the conclusion that Zn should be measured locally and not averaged over a large tissue volume, we have developed the concept of a trans-rectal XRF probe, which measures and maps the prostatic Zn content through the rectal wall. A conceptual probe design was outlined, and the results of some laboratory tests with phantoms were reported. It is claimed that, with an appropriate collimation of the primary and flu-
S.Sh. Shilstein et al. / Talanta 70 (2006) 914–921
orescence + scattered beams, it should be possible to reliably measure the Zn content in a small volume of the prostate tissue, in vivo, without the interference of the rectal wall tissue. The proposed probe is estimated to yield the prostatic Zn content within 10–20 s per measuring point, and within a permissible absorbed dose in the rectal wall, of 0.13–0.5 mGy. Acknowledgments We would like to thank Dr. H. Gutman and Dr. A. Gvirtzman of Jordan Valley Applied Radiation for a very fruitful collaboration, Y. Asher, M. Klin and R. Lozovsky for their technical assistance and. Y. Gil and Y. Telner for their skillful handling of the biopsy samples. We are indebted to the medical staff in both medical centers for their patience and collaborative spirit. The work was partially supported by the Horowitz Fund of the Weizmann Institute of Science. A. Breskin is the W.P. Reuther Professor of Research in the Peaceful use of Atomic Energy. References [1] D. Vartsky, S.Sh. Shilstein, A. Bercovich, M. Huszar, A. Breskin, R. Chechik, S. Korotinsky, S.D. Malnick, E. Moriel, J. Urol. 170 (2003) 2258–2262.
921
[2] R. Hoare, G.E. Delory, D.W. Penner, Cancer 9 (1956) 721–726. [3] M. Shirakawa, Acta Urol. Jpn. 3 (1961) 352. [4] R.S. Cotran, V. Kumar, T. Collins, Robbins Pathologic Basis of Disease, 6th ed., W.B. Saunders Co., Philadelphia, 1999, p. 1029. [5] F. Gyorkey, K.W. Min, J.A. Huff, P. Gyorkey, Cancer Res. 27 (1967) 1348–1353. [6] F.K. Habib, M.K. Mason, P.H. Smith, S.R. Stitch, Br. J. Cancer 39 (1979) 700–704. [7] R. Lahtonen, Prostate 6 (1985) 177–183. [8] J.O. Ogunlewe, D.N. Osegbe, Cancer 63 (1989) 1388–1392. [9] V.Y. Zaichick, T.V. Sviridova, S.V. Zaichick, Int. Urol. Nephrol. 29 (1997) 565–574. [10] L.J. Somervaille, D.R. Chettle, M.C. Scott, Phys. Med. Biol. 30 (1985) 929–943. [11] Christopher, in: S. Foster, D.G. Bostwick (Eds.), Pathology of the Prostate, 1st ed., W.B. Saunders Company, 1998. [12] P.A. Humphrey, Prostate Pathology, Illustrate edition, American Society Clinical Pathology, 2003. [13] J.N. Ebele, G. Sauter, J.I. Epstein (Eds.), WHO Classification of Tumors: Pathology and Genetics of Tumors of the Urinary System and Male Genital Organs (Who/IARC Classification of Tumors) International Agency for Research on Cancer, 2004. [14] S.Sh. Shilstein, A. Breskin, R. Chechik, G. Feldman, D. Vartsky, Phys. Med. Biol. 49 (2004) 485–499. [15] Xizenh Wu, in: G.T. Barnes, G.D. Frey (Eds.), Screen Film Mammography-Imaging Considerations and Medical Physics Responsibilities, Medical Physics Publishing, Madison, WI, 1991, pp. 159–175.