Multiparametric MRI and 18F-DCFPyL PET/CT for detection of locally recurrent prostate cancer after low dose rate brachytherapy

Background and purpose: Both multiparametric magnetic resonance imaging (mpMRI) and prostate specic membrane antigen (PSMA)-targeting positron emission tomography (PET) imaging have shown promise in early localization of prostate cancer (PCa) recurrence after primary external beam radiotherapy. Detecting recurrence after brachytherapy for PCa using MRI is signicantly hampered by susceptibility artifacts secondary to brachy seeds. Here, we compare the ecacy of 18 F-DCFPyL (2-(3-{1-carboxy-5-[(6- 18 F-uoro-pyridine-3-carbonyl)-amino]-pentyl}-ureido)-pentanedioic acid) PET/CT versus mpMRI for detecting sites of local recurrence after low dose rate (LDR) brachytherapy for prostate cancer. Materials and methods: A total of 155 patients with a history of recurrent PCa who underwent mpMRI at 3 Tesla and 18 F-DCFPyL PET/CT were retrospectively reviewed. Patients who underwent LDR brachytherapy for PCa and had subsequent biochemical recurrence (BCR) followed by mpMRI and 18 F-DCFPyL PET/CT were included in this study. mpMRI was performed on a 3T scanner with endorectal and surface coils and images were prospectively read by a single expert radiologist. The 18 F-DCFPyL PET/CT scan was prospectively interpreted by one of two nuclear medicine physicians. Patients underwent targeted biopsy when deemed clinically necessary and specimens were interpreted by an expert GU pathologist. Positivity rates (PR) from mpMRI and 18 F-DCFPyL PET/CT were compared, and pathology results were used to calculate the positive predictive value (PPV) of each imaging modality for detecting PCa recurrence. Results: 15 patients who underwent LDR brachytherapy and had subsequent biochemical failure were imaged with mpMRI and 18 F-DCFPyL PET/CT. 18 non-lymph node lesions were identied on at least one imaging modality. The PR for detection of intraprostatic lesions was 60% (9/15) for both PET/CT and

For the detection of intraprostatic lesions, 18 F-DCFPyL PET/CT and mpMRI had a PPV of 87.5% and 71.4%, respectively.
Conclusion: This case series analysis suggests that 18 F-DCFPyL PET/CT may detect local and regional recurrent PCa after LDR brachytherapy at a higher rate than mpMRI and that it has the potential to detect suspicious pelvic lymph nodes at a much higher rate. Further studies are needed to validate these ndings in larger cohorts.

Background
Prostate Cancer (PCa) is the most frequently diagnosed cancer in American men, and is the second leading cause of their cancer related deaths with an estimated 174,650 new cases and 31,620 deaths in 2019 (1). There are several established treatment modalities for localized PCa including surgery, external Page 4/18 beam radiation therapy [EBRT], brachytherapy, and focal therapies. LDR brachytherapy delivered alone or in combination with EBRT for PCa is selected as initial de nitive therapy in up to 40% of patients with localized prostate cancer (2).
Although e cacy of modern brachytherapy approaches is excellent, biochemical failure occurs in 6% to 15% of patients receiving monotherapy with LDR brachytherapy at 10 years (3)(4)(5)(6)(7). Early localization of sites of recurrence in these patients is critical to planning further treatment (8). Although patients with biochemical recurrence after de nitive radiotherapy were historically managed with temporizing androgen deprivation therapy, numerous potentially curative therapeutic options now exist for patients with isolated local recurrence after radiotherapy, such as focal therapies, reirradiation, HIFU, prostatectomy, and cryotherapy (9)(10)(11). These potentially curative salvage therapies carry a substantial risk of treatment related side effects. Thus, accurate localization of recurrence is critical to avoid unnecessary treatment related toxicity of local therapies in patients with regional or distant disease. Additionally, de ning patterns of failure after LDR brachytherapy may provide valuable insights into optimal patient selection and seed implantation techniques (12).
Multiparametric magnetic resonance imaging (mpMRI) consisting of T2-weighted images, apparent diffusion coe cient (ADC) maps, high b-value images, and dynamic contrast enhanced (DCE) imaging, has improved the capacity to identify clinically signi cant malignant lesions within the prostate (13). mpMRI has been successfully used for the detection of local recurrence following radiotherapy, though it is often avoided in the post-LDR brachytherapy setting due to morphological changes and signi cant susceptibility artifacts generated by the implanted brachytherapy seeds (14)(15)(16). Despite its limitations, few recent studies have demonstrated that mpMRI can still play an important role in the detection of local recurrence after LDR brachytherapy (17).
Recently, novel PET radiotracers targeting prostate speci c membrane antigen (PSMA) have shown promise in detecting early biochemical recurrence (18). When combined with CT or MRI, it offers a valuable combination of metabolic and anatomic information for cancer localization. PSMA, a transmembrane protein that is overexpressed in PCa cells but not benign tissue, has been successfully targeted and these PSMA-targeting radiotracers are currently the most sensitive for detection of early recurrence. Several investigational PSMA tracers including PSMA-11 ( 68 Ga-PSMA) and 18 F-DCFPyL have shown to be highly sensitive for detecting sites of local recurrence and metastasis even at low PSA values (19)(20)(21).
While mpMRI and 18 F-DCFPyL PET/CT have been studied for detecting sites of PCa recurrence in the general prostate cancer population and those who have undergone prostatectomy, there is a lack of published data on the utility of these imaging modalities following treatment with LDR brachytherapy, a setting in which anatomic imaging such as MRI may be compromised due to artifacts (22,23). The objective of this study is to compare the e cacy of mpMRI versus 18 F-DCFPyL PET/CT in the localization of prostate cancer recurrence in patients with biochemical failure after LDR brachytherapy.

Methods
Patient selection: Patients with a history of recurrent PCa who underwent 18 F-DCFPyL PET/CT and mpMRI between 2015-2019 were retrospectively reviewed. Inclusion criteria were as follows: a) history of PCa and primary treatment with LDR brachytherapy, b) followed by biochemical recurrence (BCR) (according to the Phoenix criteria (24)), and c) subsequent imaging with both mpMRI of the prostate and 18 F-DCFPyL PET/CT with <45 days between scans. Patients were excluded if they had primary treatment with highdose rate brachytherapy. Patients were not excluded for receiving external beam radiation therapy in addition to LDR brachytherapy ( Figure 1).
Prostate mpMRI and 18  Presence or absence of cancer was assessed using a six-grade Likert system for likelihood of recurrence (grade 1, very low; grade 2, low; grade 3, low-moderate; grade 4, moderate; grade 5, moderate-high; grade 6, high likelihood for recurrent prostate cancer). Any prostate lesion assigned a grade was considered positive on mpMRI in our analysis. Only lesions within the prostate bed were given a grade. Nodal, softtissue, and bone ndings suspicious for metastasis were noted on report, but not graded. For attenuation correction and co-registration, low-dose CT scans were obtained (120 kV, 60 mAs). By using MIM Software, regions of interest were placed around lesions that were highly suspicious for malignancy based on focal uptake greater than background not due to physiologic activity that fused to a CT abnormality. Standardized uptake values from these regions of interest were not used in this qualitative comparative analysis because of the different modalities. PET/CT images were interpreted by one of two board-certi ed nuclear medicine physicians with over 14 years of experience interpreting PET/CT scans.
Biopsy technique and pathology analysis: Biopsy for pathologic con rmation of PCa recurrence was performed in 9 of 15 patients using transrectal ultrasound (TRUS)-guided 12 core systematic biopsy and/or MRI/TRUS fusion guided biopsy. Biopsies were performed by one of two physicians (one urologist and one interventional radiologist), each with over ten years of experience in systematic and fusion guided prostate biopsy. Biopsy specimens were evaluated by a single expert genitourinary pathologist.

Results
A total of 15 patients with history of PCa who were treated with brachytherapy alone (n=7) or with supplemental external beam radiation therapy (EBRT) (n=8) were imaged with both mpMRI and 18 F-DCFPyL PET/CT secondary to BCR according to Phoenix criteria. Patient demographics are summarized in Table 1. The median age of the cohort was 69 years (range 56-78), and the median time to BCR following brachytherapy was 4.56 years (range 0.9-13.2). The median (range) PSA values prior to treatment and at the time of imaging were 7.20ng/ml (4.2-49.5) and 4.47ng/ml (2.13-28.9), respectively. Gleason grade group prior to brachytherapy ranged from 1-5 with a majority of patients (93%, 14/15) having Gleason grade group 1-3 cancer at baseline.
All patients had at least one positive nding on one or both imaging modalities ( Fig. 1-3). Table 2 summarizes individual patient clinical characteristics, sites of recurrence, imaging ndings, and associated biopsy results when available. 9 patients (60%) underwent biopsy of suspicious lesions, resulting in corresponding pathology results for 15 local pelvic lesions (9 prostate lesions, 3 SV lesions, 1 penile bulb lesion, 1 bladder wall lesion, and 1 pelvic lymph node).
A total of 11 intraprostatic lesions were identi ed on imaging in 10 patients. The positivity rate for intraprostatic lesions was 60.0% (9/15) for both 18 F-DCFPyL PET/CT and mpMRI. Eight of the 11 identi ed intraprostatic lesions were positive on both mpMRI and PET/CT, 1 was positive on mpMRI only (pathology proven false positive), and 2 were positive on PET/CT only (pathology proven true positive). In the group that did not receive EBRT, 5/7 patients (71%) had an intraprostatic lesion identi ed and all lesions were positive on both imaging modalities. Of the 8 patients who received EBRT, the PR for detecting intraprostatic lesions was 50% for both PET/CT and mpMRI. Table 3 summarizes the positivity rates of each imaging modality for intraprostatic lesions, SV lesions, and other positive ndings.
The PR for SV lesions was also similar between the two imaging modalities, with a PR of 27% (4/15) on PET/CT and 20% (3/15) on mpMRI. 3 SV lesions were positive on both mpMRI and PET/CT, and 1 SV lesion was positive on PET/CT alone. Associated pathology results (available for 3 out of 4 of the SV lesions) revealed cancer in both the MRI(-)/PET(+) SV lesion and the three MRI(+)/PET(+) lesions. (Table  4) 18 F-DCFPyL identi ed a colonic lesion and a penile bulb lesion in separate patients which were both negative on mpMRI. The colonic lesion was located in the ascending colon; however, it was not biopsied as the patient declined. The penile bulb lesion was biopsied with pathology con rming prostatic adenocarcinoma. 18

Discussion
We compared the e cacy of mpMRI and 18 F-DCFPyL PET/CT for the detection of recurrent PCa lesions in patients who were treated with LDR brachytherapy and subsequently experienced biochemical failure. In our study, overall positivity rates were similar among the two modalities for intraprostatic and seminal vesicle lesions, however, biopsy and histopathology analysis revealed a higher PPV for 18 F-DCFPyL PET/CT. Additionally, 18 F-DCFPyL PET/CT had a higher positivity rate for the detection of pelvic and extra-pelvic lymph nodes than mpMRI, which can be attributed to a greater dependence on size criteria (>1cm) in evaluation of lymph nodes by mpMRI and the limited coverage of MRI compared to PET/CT.
There are limited studies in the literature evaluating or comparing these two imaging modalities in patients with recurrence after LDR brachytherapy. Unlike other treatment modalities, LDR brachytherapy results in a substantial degree of artifact from the numerous permanently implanted brachytherapy seeds within the prostate. Although these seeds can result in mild CT scatter, they substantially impact the quality of MRI. In several studies, PET/CT was found to perform better than mpMRI for the identi cation extraprostatic disease. In one recent study, Liu et al demonstrated that 18 F-DCFPyL PET revealed extraprostatic cancer foci in twice as many men with radio-recurrent prostate cancer compared with conventional imaging, which included pelvic MRIs. 18 F-DCFPyL PET detected intraprostatic recurrence in 87% of men compared with 67% with MRI. However, this study included multiple types of radiation therapy including EBRT and brachytherapy; moreover there was no histopathologic validation for the intraprostatic foci suspicious for recurrence (26).
mpMRI has played a large and increasing role in localized prostate cancer management over the past couple of decades due to its ability to provide exceptional anatomic detail. However, in patients who have undergone LDR brachytherapy, the diagnosis of local recurrence on mpMRI is challenging secondary to the susceptibility artifacts related to implanted seeds and therapeutic glandular changes (i.e. decrease in glandular size and diffusely heterogeneous prostatic signal intensity pattern in MRI) (15,27). This is re ected in the relatively poor performance of mpMRI in this study which demonstrated an intraprostatic PPV of only 71.43%Biopsy results revealed cancer in 2 lesions which were MRI(-) and benign tissue in 2 out of 9 mpMRI(+) lesions which lowered the PPV.
In contrast, PET/CT is a molecular imaging study which relies less on anatomic features and more on cellular biology and is thus less impacted by seed implantation. For intraprostatic lesions, 18  Multiple studies have con rmed a high detection rate of recurrent PCa with various PSMA-targeting radiotracers even in biochemical recurrence patients with low PSA levels, and usually exceeds mpMRI (28)(29)(30). Some studies suggest a strong correlation between PSA levels and diagnostic performance of PSMA-targeting PET tracers (28)(29)(30)(31)(32). Our small sample size cohort did not allow for such a correlation, however in a future study with a larger cohort, imaging results should be correlated with PSA levels to test whether this holds true in patients who have undergone brachytherapy.
Previous studies have commented on the bene t of combined use of mpMRI and PSMA PET/CT for more accurate diagnosis of foci of local recurrence after radical prostatectomy or primary radiotherapy (12,33,34). In our cohort, we observed a similar bene t. While MRI was able to deliver high resolution anatomic scans, PSMA PET/CT illustrated cancer speci c uptake. This combination resulted in a radiologic diagnosis of recurrence in most of our patients. An important advantage of PSMA PET/CT was the higher positivity rate to identify lymph node involvement. The nding of disease in extraprostatic sites changes the potential treatments appropriate for the patient.
Biochemical recurrence after radiation therapy has been traditionally managed by use of conventional staging and imaging methods such as computed tomography and bone scan with subsequent initiation of androgen deprivation therapy (35). Use of more novel and dedicated imaging techniques such as mpMRI combined with PSMA PET/CT has been documented to better depict and map the disease burden in prostate cancer patients with BCR (36). This robust imaging approach can also aid delivering personalized salvage treatment instead of standalone systematic androgen deprivation. Few groups have reported promising results on utility of using PSMA PET/CT guided salvage radiation therapy for patients with BCR (37, 38). Although we did not perform a formal analysis, several of the patients included in our study were found to have isolated local recurrence and were found to be eligible to enroll on trials of salvage local therapy.
There are several limitations to this study. First, thorough statistical methods were impossible with the small study population, but relevant clinical analysis was performed. Secondly, there was a lack of uniformity in the treatment history of our cohort, with 8/15 patients receiving both brachytherapy and EBRT. Additionally, not all patients had corresponding biopsy results, and lymph nodes were not biopsied uniformly. Lymph node biopsy can often be challenging in this setting given the proximity to critical structures, lack of enlargement of the suspected lymph node, and the con nes of the bony pelvis.
Additionally, biopsy was not performed if the results would not affect disease management. The ndings could also be confounded by the referral bias. Salvage systemic treatment is usually commenced early in patients with rapid PSA doubling time after onset of biochemical failure as these patients have a higher predilection for distant metastasis (39, 40). No patient in this study was on ADT at the time of imaging, therefore, it is likely that the majority of our patients had relatively slower PSA kinetics which could affect the preponderance of local failure in our study. Finally, all imaging data were interpreted by experts with signi cant experience in mpMRI and PSMA PET which could limit the reproducibility of results in other centers.
In conclusion, we demonstrate that 18 F-DCFPyL PET/CT may be superior to mpMRI for identifying sites of recurrence in patients with suspected PCa recurrence after LDR brachytherapy with or without external beam radiotherapy. Further studies with larger patient populations and more robust histopathological sampling are necessary to validate our ndings.

Declarations
Ethics approval and consent to participate: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent was obtained from all individual participants included in the study.
Consent for publication: All co-authors con rm that they agree with publication of this manuscript.
Availability of data and material: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Funding: This research is funded by intramural research program of NIH, Pelvic Lymph Node(s) 8 2 (25%) 0 (0%) 6 (75%) Extra-Pelvic Lymph Node(s) 6 0 (0%) 0 (0%) 6 (100%) Table 4. Identified lesions stratified by positivity on each imaging modality for each anatomic location. Figure 1 Patient accrual ow chart selection criteria.  57-year-old patient with history of biochemical recurrence of prostate cancer 3.9 years following brachytherapy; PSA at time of imaging was 9.12 ng/mL. mpMRI reveals a hypointense lesion in the midline distal apical peripheral zone (A) which is more readily seen on ADC map (B) and has associated early focal enhancement on DCE imaging (D) (arrows). The prostate lesion demonstrates increased uptake on 18F-DCFPyL PET/CT (E) (arrow). Additionally, two left common iliac nodes are seen on DCFPyL PET/CT with avid uptake (F) (arrow) which were not prospectively identi ed on MRI (C). Figure 4 65-year-old patient with history of biochemical recurrence of prostate cancer 6.65 years following brachytherapy; PSA at time of imaging was 2.86 ng/mL. mpMRI reveals a hypointense lesion in the midline to right apical-mid anterior transition zone (A) which is hypointense on ADC map (B) and has associated early focal enhancement on DCE imaging (C) (arrows). The prostate lesion does not demonstrate increased uptake on 18F-DCFPyL PET/CT (D). Targeted biopsy revealed benign prostatic tissue with treatment effect.