Prostate Specific Membrane Antigen (PSMA)-Targeted Photoacoustic Imaging of Prostate Cancer In Vivo

Abstract

A sensitive, non-invasive method to detect localized prostate cancer, particularly for early detection and repetitive study in patients undergoing active surveillance, remains an unmet need. Here we propose a molecular photoacoustic (PA) imaging approach by targeting the prostate-specific membrane antigen (PSMA), which is over-expressed in the vast majority of prostate cancers. We performed spectroscopic PA imaging in an experimental model of prostate cancer, namely, in immunocompromised mice bearing PSMA+ (PC3 PIP) and PSMA− (PC3 flu) tumors through administration of the known PSMA-targeted fluorescence agent, YC-27. Differences in contrast between PSMA+ and isogenic control tumors were observed upon PA imaging, with PSMA+ tumors showing higher contrast in average of 66.07-fold with five mice at the 24-hour post-injection time points. These results were corroborated using standard near-infrared fluorescence imaging with YC-27, and the squared correlation between PA and fluorescence intensities was 0.89. Spectroscopic PA imaging is a new molecular imaging modality with sufficient sensitivity for targeting PSMA in vivo, demonstrating the potential applications for other saturable targets relevant to cancer and other disorders.

Graphical abstract

A sensitive, non-invasive method to detect localized prostate cancer, particularly for repetitive study an patients undergoing active surveillance, remains an unmet need. We propose a photoacoustic (PA) imaging approach by targeting the prostate-specific membrane antigen (PSMA), which is over-expressed in the vast majority of prostate cancers. Through in vivo demonstration on an experiment with mice model, spectroscopic PA imaging can visualize the targeted agent, YC-27, that binds on PSMA+ tumor.

1. Introduction

Prostate cancer is the second leading cause of cancer-related death among men in the United States. In 2016, 180,890 men were diagnosed with the prostate cancer, and 26,120 men died from this disease [1]. Prostate cancer is known for its high survival rate when it is localized because of its slow tumor growth, but the survival rate drops with the onset of metastasis and subsequent accelerated tumor growth. Although screening blood level of prostate specific antigen (PSA) has been served as a non-invasive early detection practice, recent clinical studies revealed that aggressive treatments based on the PSA screening might not benefit all patients [2]. Detecting the aggressive forms of prostate cancers at its early state with precision is critical, when they are likely to be easier to treat. For this, a transrectal or transperineal needle biopsy is performed for definite diagnosis, but it carries high chances of side effects including pain, bleeding and infection [3,4]. In addition, only a small minority of patients present with metastases from prostate cancer, and localized disease can be associated with such a benign course as to enable active surveillance. Active surveillance currently involves periodic, generally annual, biopsy, which is non-trivial and fraught with a variety of possible complications. One way of minimizing the impact is to add tumor specific visualization during biopsy, in that conventional transrectal ultrasound (TRUS) guided biopsy is more systematic rather than lesion-specific, and has high error rate due to random sampling. On the other hand, there have been an alternative long-term approach to avoid such issues; molecular diagnosis of cancer aggressiveness and lethality have been extensively researched for non-invasively detection and staging of several cancer types [5]. Biophotonic imaging method is being regarded as the most sensitive imaging modality with high sensitivity and easiness in labeling to specific structures, but suffers from its limited imaging depth (~few mm) [6]. Therefore, there are clinical needs on novel molecular imaging modality with both tumor-specific contrast and deep sensing depth.

Photoacoustic (PA) imaging is a noninvasive hybrid imaging modality that combines the high sensitivity inherent to optical techniques, and real-time capability of ultrasound imaging with the high spatial resolution. By using both near-infrared (NIR) laser excitation and ultrasound concurrently, PA imaging minimizes absorption and scattering of light through tissue to enable imaging at a multi-centimeter thick tissue in millimeters or sub-millimeters axial spatial resolution [7]. In particular, multi-wavelength spectroscopic PA imaging has been highlighted as an emerging molecular technique to effectively differentiate the specific molecules in high sensitivity and specificity. Several exogenous PA contrast agents have been evaluated for tumor visualization and characterization [8-9].

Prostate-specific membrane antigen (PSMA) is a type II integral membrane protein with expression within normal prostate and over-expression on the surface of prostate cancer cells [10-11]. It is an excellent target for imaging cancer [10], particularly prostate cancer, because of its limited expression in normal (non-malignant) tissues. Levels of the PSMA expression have also been positively correlated with a degree of malignancy. PSMA has been extensively leveraged with a wide variety of imaging agents, including those for radionuclide-based techniques, magnetic resonance imaging and near-infrared fluorescence (NIRF) imaging [12-15]. Recently, Dogra et al. reported the feasibility of PSMA-targeted PA imaging with YC-27, a known PSMA-targeted NIR fluorescence imaging agent [12], in a study contrasting PSMA-expressing tumor cells from non-expressive tumor cells in vitro (i.e., PSMA+ C4-2 and PSMA− PC3, respectively) [15]. The authors further demonstrated that IRDye800CW, the dye contained within YC-27, has advantages for PA imaging over four other non-targeted dyes commercially available.

In this letter, we present PSMA-targeted spectroscopic PA imaging for characterizing prostate cancer in vivo. In particular, we evaluate the YC-27 in in vivo differentiation of PSMA-expressing tumor from non-expressive tumor type (i.e., PSMA+ PC3 and PSMA− PC3, respectively).

2. Materials and Methods

2.1. PSMA-targeted photoacoustic contrast agent

Properties of the known NIRF PSMA-targeted agent, YC-27, have been described in [12]. Briefly, this is a high-affinity (The reporter Ki value of YC-27 is 0.37 nM with 95% confidence intervals from 0.18 to 0.79 nM), urea-based agent that is functionalized with the NIR dye, IRDye800CW. Fig. 1A shows a chemical structure of the proposed YC-27 as the PA imaging modality, and its PA spectroscopic characteristics in PBS show the near-infrared absorbance peak (i.e., 780 nm, Fig. 1B). Also, Fig. 1C presents a linear relationship between PA intensity and molar concentration of YC-27 in PBS (0.5, 1, 5, 10, 25, and 50 μM, R² = 0.99). Note that the sample was loaded into transparent tubes with 1.27-mm diameter (AAQ04133, Tygon®, Saint-Gobain Corp.) located at 30 mm depth in the water tank. The identical PA imaging system introduced in Section 2.4 was used for this phantom study.

Figure 1.

The proposed PSMA-targeted PA contrast agent, YC-27: (a) chemical structure, (b) spectroscopic PA spectrum, and (c) its linear relationship between PA intensity and the concentration of YC-27.

2.2. Animal preparation

Sublines of the androgen-independent PC3 human prostate cancer cell line, originally derived from an advanced androgen independent bone metastasis, were used. These sublines have been engineered to express high level of PSMA (PC3 PIP) and to maintain no expression of PSMA (PC3 flu). These cells were generously provided by Dr. Warren Heston (Cleveland Clinic). PSMA-expressing (PC3 PIP) and non-expressing (PC3 flu) PCa cell lines were grown in RPMI 1640 medium (Corning Cellgro, Manassas, VA) containing 10% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO) and 1% penicillin–streptomycin (Corning Cellgro, Manassas, VA). PC3 PIP cells were grown under 20 μg/mL of puromycin to maintain PSMA expression [12]. All cell cultures were maintained in an atmosphere of 5% carbon dioxide (CO₂) at 37.0°C in a humidified incubator. Animal studies were undertaken in compliance with the regulations of the Johns Hopkins Animal Care and Use Committee. Six- to eight-week-old male, non-obese diabetic (NOD)/severe-combined immunodeficient (SCID) mice (Charles River Laboratories, Wilmington, MA) were implanted subcutaneously with 1×10⁶ PSMA+ PC3 PIP and PSMA− PC3 flu cells on the lower back near right- and left-posterior flanks. Two tumors were placed adjacent to one another to enable imaging side-by-side in PA and ultrasound imaging (Fig. 2A). Mice were shaved and imaged when the tumors reached 2.5–5 mm in diameter. Inhalational anesthesia (isoflurane) was applied to the mice through a nose cone attached to the imaging bed. A set of PA and ultrasound images were acquired from both PSMA+ PC3 PIP and PSMA− PC3 flu tumors before the administration of the proposed PSMA-targeted PA contrast agent. The PA agent (50 nmol in PBS) was injected via tail vein, and tumors were imaged at 24-hour for five mice with both PC3 PIP and PC3 flu tumors. One mouse underwent an imaging session at 2-hour post-injection.

Figure 2.

In vivo experiment setup. (A) Xenograft tumor preparation. Xenografts (PSMA+ PC3 PIP and PSMA− PC3 flu) were placed on opposite sides of the lower back of SCID mice. (B) Photoacoustic (PA) imaging setup. Laser was illuminated through bar-type fiber bundles, and PA signals were collected by a linear array transducer.

2.3. Ultrasound imaging

Ultrasound was used to localize tumors at baseline. A clinical ultrasound scanner (SonixCEP, Ultrasonix Corp., Canada) was employed. A linear array transducer (L14-5/38, Ultrasonix Corp.) with 128 elements and a pitch of 0.3 mm was used. Mice were anesthetized and placed on a fixation stage in dorsal view (Fig. 2). Ultrasound gel was used to couple the ultrasound probe and mice with tumors. Ultrasound B-mode images were used to position the transducer and to maximize the interrogated tumor cross-section in PA imaging.

2.4. Photoacoustic imaging

After determining the ultrasound slice position, PA images in the same image field were collected consecutively (Fig. 1B). A Q-switched Nd:YAG laser integrated with an Optical Parametric Oscillator (OPO) (Phocus Inline and MagicPrism, Opotek Inc., United States) was used as the tunable light source. The emitted light was coupled with a 4-cm bar-shape optical fiber bundle, which enabled the co-registered emission with respect to the linear array transducer. The spectral responses for NIR wavelengths (700–850 nm at 10-nm intervals) were scanned with 20-Hz pulsed repetition frequency and 5-nm pulse duration. At each wavelength, multiple PA datasets were collected with 64 pulsed laser excitations for frame averaging. Raw PA signals from mice were received by the ultrasound probe. A PA image was reconstructed and post-processed from the received channel data, and a collection of PA images corresponding all multiple wavelengths were used for spectroscopic analysis.

2.5 Material decomposition

The contrast of the reconstructed PA images depends upon a combination of multiple absorbers including the dye, blood, and other tissues. The spectroscopic decomposition process of each source material from the raw PA intensity is necessary to specifically quantify the exogenous contrast agent intensity [16-17]. Assuming that the PA signals are composed of a linear combination of multiple signal sources, the unknown dye and other contrast m are calculated as

$$\underset{m_{1,2\dots M}}{\arg min}\Vert {\displaystyle \sum _w\left(p-{\displaystyle \sum _{i=1}^M{m}_i{\mu }_{a,i}}\right)}\Vert ,$$

where p is the measured PA spectrum, i represents the contrast number, μ is the absorption spectrum of the contrast i, M is the number of absorbers, and w is the wavelengths used for the proposed material decomposition. We assumed three components to the contrast within the PA image: oxygenated and deoxygenated hemoglobin, and YC-27. The spectrum of hemoglobin was referred from the literature [18], and the PA spectrum of YC-27 was obtained experimentally (Fig. 1B).

2.6. Fluorescence imaging

Fluorescence imaging was performed using the Pearl Imager (LI-COR Biosciences), which employs a CCD camera with the field-of-view of 11.2 by 8.4 cm at the surface of the imaging bed. The scan time was less than 30 s to complete white-light imaging and near-infrared image acquisitions through 700- and 800-nm channels optimized for IRDye800CW detection. The output images were displayed in pseudo-color with a corresponding scale to depict relative signal intensities. All images were acquired at the same parameter settings and are scaled to the same maximum values. Imaging bed temperature was maintained at 37°C. Animal preparation and the protocol of YC-27 administration were precisely performed as described in Section 2.2. Fluorescence images were taken right after the PA image acquisition on the same tumor and dose conditions.

2.7. Data analysis

The tumor volume size, V, was estimated by the measured tumor diameter d, which is given by

$$V=\frac{4}{3}\pi d^3.$$

The PA and fluorescence intensities of YC-27 under the various conditions were quantified by measuring the averaged intensity at the tumor region. The contrast enhancement at different time points was calculated to quantify the PA agent intensity change, which is defined by

$$\mathit{Contras}{t}_{\mathit{Enh}-n}=\frac{In{t}_{\mathit{After}-n}}{In{t}_{\mathit{Before}-n}},$$

where N is the number of mice; Int_After-n and Int_Before-n are the PA intensity measured from n^th mouse before and after the PA agent administration, respectively. Similarly, the contrast difference is the metric indicating the difference of the PA agent intensity between PSMA+ and PSMA− tumors at the same time point, given as

$$\mathit{Contras}{t}_{\mathit{Diff}-n}=\frac{In{t}_{\mathit{PIP}-n}}{In{t}_{\mathit{flu}-n}},$$

where Int_PIP-n and Int_flu-n are the measured intensity of PC3 PIP and PC3 flu tumors for n^th mouse, respectively. Quantified values from multiple samples were presented as mean ± standard deviation. Statistical significance was determined by the unpaired Student t-test.

3. Results and Discussion

PA and ultrasound imaging of PSMA− PC3 flu and PSMA+ PC3 PIP tumors before and after the YC-27 administration are shown in Fig. 3. Fig. 3A shows the images of one mouse under different time points, and with and without material decomposition, respectively. The wavelength for the representative single-wavelength PA images was chosen as it gives peak absorbance of YC-27 (i.e., 774 nm), and it yielded the most significant contrast in PA imaging. Though non-dye specific PA signals were observed, material decomposition succeeded to isolate the blood contrasts in the spectroscopic PA imaging results. A clear YC-27 contrast was depicted in both PIP and flu tumors in after 2-hour images, and this is attribute to the remaining dye inside the blood circulation around the tumors. After 24 hours from injection, YC-27 in the blood was washed out, so that the contrast from the flu tumor was substantially decreased, while a strong uptake of the PSMA-targeted agent was observed in the PIP tumor. The same contrast difference and change was observed in fluorescence results (Fig. 3B). The PA intensities in the tumor regions were quantified in Fig. 3C, and same trend was observed in the fluorescence results shown in Fig. 3D (the ratios between pre-injection and 2-hour post-injection for PSMA+ PIP and PSMA− flu tumors were 0.48 and 0.44 in PA and fluorescence imaging, respectively).

Figure 3.

PA and ultrasound imaging of PSMA− PC3 flu and PSMA+ PC3 PIP tumors, and the quantitative comparison with fluorescence (FL). (A) PA and ultrasound images of the tumors before the dye injection, and after 2-hour and 24-hour injection. (B) FL images after 2-hour and 24-hour injection. (C) Quantified the PA intensity of flu and PIP tumors. (D) Quantified the fluorescence (FL) intensity of PSMA− PC3 flu and PSMA+ PC3 PIP tumors. (E) PA intensities of five mice at pre-injection and at 24-hour post-injection of YC-27. Open circles represent data from each mouse. (F) Comparison of fluorescence intensities of five mice at 24-hour post-injection. Open circles represent data from each mouse. (G) Comparison of PA and fluorescence, and its correlation. The asterisk denotes P-value < 0.02.

To statistically evaluate the specificity of PSMA targeted PA imaging, five mice were imaged in both PA and fluorescence imaging, and the intensity of YC-27 at the tumor region was compared under different conditions (Figs. 3E–G). As a result, the quantified PA intensities of YC-27 before injection were 0.01±0.001 and 0.005±0.005 at the PSMA+ PC3 PIP and PSMA− PC3 flu tumors, respectively. After 24-hour injection, the PA YC-27 intensities measured at the PIP and flu tumors were 0.56±0.36 and 0.02±0.02, respectively. The YC-27 contrast showed a statistically significant difference by comparing the intensities of pre-injection and 24-hour post-injection on PSMA+ PC3 PIP tumors (p < 0.02), as well as comparing PSMA− PC3 flu and PSMA+ PC3 PIP tumors at 24-hour’s time point (p < 0.02) (Fig. 3E). The PA agent contrast enhancement on the PSMA+ tumor comparing before injection and after 24-hour was 54.10 ± 49.53-fold, and the PA agent contrast difference comparing PSMA+ and PSMA− tumors at 24-hour post-injection was 66.07 ± 65.61-fold. The quantified fluorescence intensities at 24-hour post-injection point were 1.63 ± 0.74 and 0.25 ± 0.13 for PIP and flu tumors, respectively (Fig. 3F). Note that PSMA+ tumor on one mouse presents a very weak PA agent intensity. This may attribute to many factors including the small tumor size (65.42 mm³ for this small tumor vs. 204.19 ± 149.37 mm³ in the average from PSMA+ tumor on the other mice), and it is also confirmed in fluorescence that this tumor shows the weakest signal (1 vs 1.78±0.76 in the average from PSMA+ tumor on the other mice). We compared the squared correlation between PA and fluorescence intensity using the all data points, including ten data points at 24-hour post-injection (Fig. 3E) and two data points at 2-hour post-injection (Fig. 3C) measured from both PIP and flu tumors (Fig. 3G). The PA and fluorescence intensities were strongly correlated, and this suggests that PA imaging can be used as a complemental imaging modality of fluorescence imaging. Having said that, these two modalities have images different image-plane, in which PA imaging provides a slice-like tomographic view of tissue, while fluorescence imaging delineates the superficial distribution. Thereby, PA imaging holds its unique advantage of deep tissue imaging that visualizes prostate cancer in a more non-invasive manner.

5. Conclusions

Here we demonstrated the principle of PSMA-targeted PA imaging of prostate cancer in vivo. Results were correlated with standard NIRF imaging with YC-27, an imaging agent containing the NIR dye IRDye800CW that we have previously validated to target PSMA in experimental models. As a hybrid technique leveraging both optical and ultrasound modalities to provide superior depth of penetration and molecular specific imaging enhancement, PA imaging is a promising new way to characterize prostate tumors for early detection and active surveillance. It can be further effective for other indications at a localized prostate tumor and other cancerous diseases that express PSMA such as breast cancer, melanoma, and non-small cell lung cancer [19-20].