Introduction

Prostate-specific membrane antigen (PSMA; also termed glutamate carboxypeptidase II) is abundantly expressed in prostate cancer and is identical to the folate hydrolase I found in the jejunal brush border and the N-acetyl-alpha-linked-acidic-dipeptidase (NAALADase) in the nervous system [1].It is a type II transmembrane glycoprotein with an extracellular C-terminus and a cytoplasmic N-terminus such that the extracellular domain binds to N-acetyl aspartyl glutamate (NAAG) in the nervous system, where it hydrolyses NAAG to glutamate and N-acetyl aspartate, and to folyl-poly-γ glutamate in the jejunum, contributing to folate metabolism [1].

While PSMA is over-expressed on the epithelial tumor cells of prostate cancer, the expression is primarily on the endothelial cells of the neovasculature in many other solid tumors including colon, breast and adrenocortical cancers [2, 3].In a study of 68 patients with 91 samples, which included 37 patients with differentiated thyroid cancers (DTCs), PSMA expression was graded on a score of no expression (score 0), weak (score 1–3), moderate [4,5,6] or strong expression [2, 7]; all classic papillary thyroid cancers (PTCs) and follicular thyroid cancers (FTCs) stained positive for PSMA. Sixty-four percent of PTCs stained moderate or strong and 45% of FTCs stained moderate or strong [2]. PSMA expression was seen in 100% of distant metastases and 67% of lymph node metastases [2].

In another recently published study, PSMA expression in the neovasculature was significantly more frequent in malignant tumors compared to benign tumors (36/63; 57.1% vs. 5/38; 13.2%; p = 0.0001) [4]. These studies imply that agents targeting PSMA might be useful in diagnostic and therapeutic targeting of advanced DTCs that exhibit significant neoangiogenesis.

Small-molecule radiopharmaceuticals using 18F are of immense value due to their inherent suitability for medical imaging owing to abundant positron production allowing for positron emission tomography (PET) imaging, the ease of synthetic accessibility, salutary pharmacokinetics, capacity for introduction to clinical workflow, and favorable imaging characteristics. The agent 18F-DCFPyl (2-(3-{1-carboxy-5-[(6-[(18)F]fluoro-pyridine-3-carbonyl)-amino]-pentyl}-ureido)-pentanedioic acid) was developed at Johns Hopkins as a PSMA imaging agent. It displayed high affinity for PSMA (Ki 1.1 ± 0.1 nM), high radiochemical yield on synthesis (36–53%), and very favorable tissue distribution in both mice models and humans [5,6,7]. In 2011, a preclinical evaluation showed that 18F-DCFPyL had uptake in PSMA-positive prostate tumor xenografts and the uptake was significantly higher when compared to PSMA-negative tumors (358:1) at 2 h post injection [5]. In a 9-patient study, 18F-DCFPyL PSMA showed no safety issues in humans with highest dosimetry in kidneys and urinary bladder and substantial uptake in submandibular gland and liver [7].

No previous study has used the radiotracer 18F-DCFPyL for the imaging of DTC lesions prospectively. The aim of our study was to determine if 18F-DCFPyL PET/computed tomography (CT) improved the localization of DTC in patients with advanced thyroid cancer when compared to 131I post treatment scan and 2-deoxy-2-[18F]fluoro-D-glucose (18F-FDG) PET/CT as proof of principle. We also examined the degree and pattern of PSMA expression in stored histopathological samples of persons with advanced DTC.

Research design and methods

This study was approved by our local Johns Hopkins Institutional Review Board (IRB) and ethics committee (Office of Human Research). Imaging with 18F-DCFPyL PET/CT was carried out under a United States Food and Drug Administration Investigational New Drug application (IND 121,064). Patients who underwent imaging signed informed consent. In this pilot study, we first examined the expression of PSMA in 3 patients (shown in Table 1) with DTC and cervical lymph node metastatic disease. All the patients had undergone total thyroidectomy with neck dissection for metastatic lymph nodes.

Table 1 The characteristics of patients whose lymph node samples were stained for PSMA expression in the neovasculature
Full size table

In each of the patients, we reviewed the electronic medical record and abstracted information including: demographics (age, gender, ethnicity, body mass index), history of presentation, physical examination, first ultrasound (US) of the thyroid (nodule size, shape, vascularity, echogenicity, presence of sonographic high-risk factors, e.g., microcalcifications), thyroid function tests, other imaging findings (US, CT, PET/CT), fine-needle aspiration (FNA) cytopathology results (based on revised Bethesda classification), preoperative US of the soft tissue and neck, extent of lymph node involvement by sonography and/or FNA cytology, other imaging findings, surgery performed (lobectomy vs total thyroidectomy), extent of neck dissection, final histopathological diagnosis, and tumor stage (8th edition TNM classification), complications during surgery, postoperative thyroglobulin levels, diagnostic 123I results, dose of 131I administered, post treatment scan results, post treatment thyroglobulin (± thyroglobulin antibody trend), as well as serial TSH-stimulated and suppressed thyroglobulin level(s).

The stored lymph node samples were stained with the DAKO monoclonal PSMA antibody. Monoclonal anti-PSMA clone 3E6 recognizes an epitope present in the 57–134 amino acid region of the extracellular portion of PSMA [8]. Monoclonal mouse anti-PSMA has been shown to react in Western blotting with PSMA from cell lysates, seminal fluid, and with recombinant baculovirus expressed PSMA [8]. When compared to the anti-endothelial cell antibody against CD34, the anti-PSMA antibody 7E11 has shown a pattern of staining primarily consistent with the neovascular distribution within a wide range of tumors [9]. The DAKO monoclonal PSMA antibody (catalog# M3620) clone 3E6 is a commercially available and validated assay for detection of PSMA expression [8, 10].The reagent is supplied by the manufacturer in a liquid form as tissue culture supernatant in 0.05 mol/L Tris–HCl, pH 7.2 and 0.015 mol/L sodium azide, containing a stabilizing protein [8]. In a previously described method, thick sections (5 µm) were deparaffinized by Histo-clear (www.nationaldiagnostics.com), incubated in a target-retrieval solution for 30 min, treated with peroxidase block (to block endogenous peroxidase activity) for 5 min, and then incubated with PSMA antibody for 60 min [2]. If areas of activity were seen on the slides, those samples were considered positive.

Imaging protocol

Subsequently, we imaged 2 other patients with suspected DTC recurrence with 18F-DCFPyL. A bolus of less than or equal to 9 mCi (333 MBq) of 18F-DCFPyL was given intravenously (IV). The mass dose of 18F-DCFPyL administered was less than 3.98 µg per patient. PET/CT was performed on a Discovery DRX PET/CT scanner (GE Healthcare, Waukesha, WI, USA) or Biograph mCT PET/CT (Siemens Healthineers, Erlangen, Germany) operating in 3-dimensional emission acquisition mode and with attenuation correction via low dose CT. Patients were scanned in the supine position starting from the mid-thigh to the vertex of skull approximately 60 min after radiotracer injection. The scan included 7 to 9 fields-of-view, depending on patient height. An initial low dose x-ray CT transmission scan, preceding the PET emission acquisition, was used for tissue attenuation correction and anatomical correlation.

Results

The characteristics and findings of the three participants with histopathological examination is shown in Table 1. The three participants had findings of non-RAI (Radioiodine) avid residual disease. Patient A—had an FDG avid right cervical level 3 lymph node and a sub-centimeter-sized right lung nodule that did not show any radiotracer uptake on the iodine based WBS (Whole-Body Scan). Patient B—had uptake in the thyroid bed but no evidence of extrathyroidal uptake (especially in the resected lymph node sample that was compatible with non-RAI disease). Patient C- had multiple pulmonary nodules on CT scan of the chest that were not RAI avid (negative WBS) but likely represented advanced residual disease.

The pathology slides are shown in Fig. 1a–c. All the patients in Group I showed expression of PSMA in the neovasculature. However, none of the thyroid follicular cells stained positive for PSMA.

Fig. 1
figure 1

a–c The stored LN samples were stained with DAKO Monoclonal anti-PSMA Antibody clone 3E6 that recognizes an epitope present in 57–134 amino acid region. All the three LN samples stained positive for GCP-II (PSMA) shown by line arrows

Full size image

The characteristics of the participants (including age, subtype of DTC as well as unstimulated Thyroglobulin (Tg) at the time of imaging) who underwent the 18F-DCFPyL PET CT research scan is shown in Table 2. Of note, Tg levels in the patients imaged by the 18F-DCFPyL PET CT were significantly higher than in patients whose lymph nodes were stained with DAKO PSMA antibody (Tables 1 and 2).

Table 2 Characteristics of the patients who were imaged by 18F-DCFPyL
Full size table

The first imaged patient (ID # 1) had no uptake on the diagnostic 123I diagnostic whole-body scan (WBS) while the contrast-enhanced CT scan showed a retropharyngeal lymph node with abnormal enhancement (Table 3). This lymph node showed minimal, if any, uptake on the 18F-FDG PET/CT but more convincing increased uptake (although still only mild) on the 18F-DCFPyL PET/CT. The maximum standardized uptake value (SUVmax) was 3.1 with 18F-DCFPyL but only 1.8 on the 18F-FDG PET (Fig. 2).

Table 3 Summarizes the Imaging findings of the two patients
Full size table
Fig. 2
figure 2

Patient 1 – Molecular Imaging shows increased Radiotracer Uptake in the Retropharyngeal Lymph Node with an increased SUV. DCF-PyL PET Shows an SUV of 3.1 compared to FDG-PET that shows an SUV of 1.8

Full size image

The second patient (ID # 2), who underwent imaging, had mild elevation of Tg and was found to have left sided 0.9 cm lung nodule on CT chest that was radioiodine (RAI)-avid on the diagnostic 123I scan as well as the post treatment scan (Table 3). The 18F-FDG PET/CT had shown low level of uptake in the lung nodule. 18F-DCFPyL showed intense uptake in the anterior aspect of L3 (SUVmax 7.9) not seen on the post treatment 131I scan or the 18F-FDG PET/CT (SUVmax in this region was 2.2 and visibly the same as background blood pool or marrow). Magnetic resonance imaging (MRI) of the lumbar spine confirmed the presence of an 8-mm mixed sclerotic-lytic lesion at the same location, with enhancement and surrounding bone marrow edema, highly suggestive of metastatic disease (Fig. 3). Of note, the radioiodine-avid lung nodule did not have uptake of 18F-DCFPyL. Figure 4 shows the MIP images of 18F-FDG-PET CT and 18F-DCFPyL PET/CT of the two imaged patients. The cervical lymph node is not discernable in these images as this was best appreciated in the cross-sectional images (Patient 1). The small vertebral metastatic lesion is shown by the blue arrow (Patient 2).

Fig. 3
figure 3

(Patient 2) a Posterior planar whole-body 131I post-therapy scan shows evidence of lung metastases (red arrow) and physiologic bowel uptake, but no abnormal uptake in the region of the lumbar spine. b Sagittal CT image through the lumbar spine demonstrating subtle rarefaction of the anterior aspect of the L3 vertebral body (red arrowhead). c Sagittal fused 18F-DCFPyL PET/CT image shows that there is intense radiotracer uptake at the same location (red arrowhead). d Sagittal T2-weighted and e sagittal T1-weighted post-contrast magnetic resonance images show perilesional edema (red arrowhead in (d)) and subtle enhancement (red arrowhead in (e)), confirming the presence of the lesion seen on PET

Full size image
Fig. 4
figure 4

MIP images of FDG-PET CT and 18F-DCFPyL PET/CT of the two imaged patients. The cervical retropharyngeal lymph node is not discernable in these pictures as this was seen mostly in the cross-sectional images (Patient 1). The small vertebral metastatic lesion is shown by the blue arrow (Patient 2)

Full size image

Discussion

Most cases of DTC have excellent prognosis following lobectomy/ total thyroidectomy with and without radioiodine treatment and the surveillance usually involves measurement of Tg on levothyroxine, neck US and CT, and occasionally 18F-FDG PET/CT [11]. However, a small percentage of DTCs are aggressive with poor prognosis. Post treatment (surgery ± radioiodine therapy), unstimulated or stimulated Tg values greater than 10–30 ng/mL increase the likelihood of having persistent or recurrent disease, RAI-refractory disease, distant metastases, and increased mortality [12]. As was evident in our imaged patients, clinically localizing disease in patients with aggressive DTC poses a distinct challenge.

PSMA is a 100 kDa type II transmembrane protein (belonging to the M28 peptidase family) with folate hydrolase and NAALAdase activity [13]. FOLH1, the gene that encodes the protein, consists of 19 exons spanning approximately 60 kb of genomic DNA [13]. The genetic structure is shown in Fig. 5 (adapted from https://www.genecards.org). In a study by Silver et al. in 1997, PSMA expression in neovascular endothelium of tumors was a novel finding, while studying the immunohistochemical analysis performed on different extra-prostatic tissues, the implications of which was unclear at that time [14]. Its expression seems to be similar to another enzyme, CD13 (also known as Aminopeptidase N, a zinc metalloproteinase) [15, 16]. Chang et al., evaluated five different monoclonal antibodies against PSMA and showed that the protein is expressed in a wide range of tumors in the associated neovasculature (including renal cell carcinoma, colonic adenocarcinoma, and malignant melanoma) [9].

Fig. 5
figure 5

Source www.genecards.org)

Genomic Locations for FOLH1 Gene chr11:49,145,092–49,208,670, (GRCh38/hg38) (

Full size image

Conway et al. showed that PSMA is essential for endothelial invasion through in vitro studies that showed the protein participates in laminin-specific integrin signaling via the Rho GTPase effector molecule p21-activated kinase 1 (PAK-1) [16]. PSMA (in association with matrix metalloprotease-2) appears to produce small laminin peptide fragments (containing carboxy-terminal glutamate moieties) that increase endothelial cell adhesion, and activate integrin α6 β1 and focal adhesion kinase [17, 18]. These findings suggest the reason for PSMA expression in tumor neovascular endothelium. Regardless of the mechanism by which PSMA is expressed on the neovasculature of most solid tumors, its presence indicates a potential role for PSMA-based diagnostic and therapeutic (i.e. theranostic) agents in the management of a wide range of tumors targeting the critical angiogenesis component of tumor proliferation.

Recently, it was shown by Sollini et al., that in a multivariate regression model to predict RAI-refractoriness, the stage of the disease, the degree of PSMA expression if high (≥ 80%), and the interaction between moderate PSMA expression (11–79%) and stage were the most important covariates [19]. PSMA expression was strongly associated with tumor aggressiveness and poor patient outcomes [19].

131I Negative DTC was first detected by a PSMA-based PET radiotracer, 68 Ga-PSMA-11, which showed uptake in a cervical lymph node and pulmonary metastasis [20]. Subsequently, another patient with rising thyroglobulin and negative 131I whole-body scan had intense 68 Ga-PSMA-11 uptake in mediastinal and left supraclavicular lymph nodes, brain, pulmonary nodules, and skeletal sites [21]. There have also been reports of incidental thyroid uptake of 68 Ga -PSMA-11 in a follicular adenoma when the patient was being investigated for prostate cancer [22]. Beyond those studies, our results would suggest that PSMA expression in DTC tumor-associated neovasculature may be heterogenous, given that the second imaged patient had intense 18F-DCFPyL uptake in a spinal metastasis but lacked uptake in a lung metastasis that was positive on 131I post treatment scan. The exact utility of PSMA-targeted radiotracers and the ideal clinical scenarios in which they should be used for DTC will require further study. Furthermore, obtaining a tissue diagnosis post 18F-DCFPyL imaging is the next logical step in further consolidating evidence for this image modality.

PSMA-based radiotracers, especially those incorporating 18F, which has significant practical and image-quality advantages relative to 68 Ga, have not been comprehensively evaluated in advanced thyroid cancer [23, 24]. Our proof of principle study shows significantly increased neovascular PSMA expression in lymph nodes of DTC as well as the ability of 18F-DCFPyL to detect lesions that were previously unnoticed by RAI-based imaging as well as 18F-FDG PET/CT.

Given the above evidence of PSMA expression and imaging in advanced DTC, our research premise appears promising. In our patients, RAI-refractory disease was successfully imaged and detected by 18F-DCFPyL. Further prospective studies are needed in this regard.