Optimal MRI sequences for ⁶⁸Ga-PSMA-11 PET/MRI in evaluation of biochemically recurrent prostate cancer

After initial definitive treatment of prostate cancer with radical prostatectomy, brachytherapy, or external beam radiation, prostate-specific antigen (PSA) may rise due to local, regional, or systemic recurrence prior to the development of clinical symptoms. This rise in pre-clinical PSA is termed biochemical recurrence (BCR). After prostatectomy, BCR is defined by the American Urological Association as a PSA value of 0.2 ng/mL or higher on two separate tests [1]. In patients treated with radiation therapy, including external beam radiation and brachytherapy, the RTOG-ASTRO Phoenix Consensus Conference defined BCR as a rise in PSA by 2.0 ng/mL above the PSA nadir [2]. Patients with BCR may be assessed with a combination of technetium-99m bone scintigraphy and periodic CT scans [3]. However, CT scans do not accurately evaluate lymph nodes, especially those smaller than 5 mm [4], and bone scintigraphy has limited sensitivity and specificity for the detection of osseous metastases, due to radiotracer uptake by non-metastatic lesions, most commonly degenerative diseases.

Prostate-specific membrane antigen (PSMA) is a membrane protein that is overexpressed in prostate cancer cells in comparison to benign prostate cells and increases in advanced stage and androgen-independent prostate cancer [5]. Molecular imaging targeted to PSMA has been shown to have a high detection rate of lesions suspicious for local recurrence and metastatic disease [6, 7]. In two intra-patient comparisons, for example, ⁶⁸Ga-PSMA-11 PET had a higher sensitivity than fluorocholine PET, particularly in patients with a PSA less than 2.0 ng/mL [6, 8].

While PSMA PET is highly sensitive for suspected sites of prostate cancer recurrence, it does not provide precise localization of the PSMA-avid foci. Therefore, PET images need to be co-registered with images from an additional modality for localization, and this has typically been done with CT scans because of availability, ease of acquisition, and high spatial resolution. However, pairing PSMA PET with MRI, which has superior soft tissue resolution than CT, has the potential to provide additional valuable information for the evaluation of BCR patients.

Prior studies that investigated the use of PET/MRI in prostate cancer have focused on patients prior to treatment, given the ability of multiparametric MRI to evaluate primary tumor [9–11]. But combining the molecular, anatomical, and functional imaging data of PET and MRI may also allow for precise localization of disease in patients with BCR. Accordingly, we retrospectively evaluated the optimal MRI sequences to localize positive findings on ⁶⁸Ga-PSMA-11 PET of patients with BCR after prostatectomy or radiation therapy.

This study was approved by the institutional review board, and written informed consent was obtained from all patients. This study was performed under an Investigational New Drug approval from the Food and Drug Administration as part of a trial prospectively evaluating the accuracy of ⁶⁸Ga-PSMA-11 for the detection of prostate cancer (NCT02611882). Inclusion in the trial required a PSA doubling time of less than 12 months, and these patients have been reported as part of a change in management analysis [12]. Fifty-five consecutive patients who underwent ⁶⁸Ga-PSMA-11 PET/MRI for BCR from March 2016 to September 2016 were evaluated. All of the patients were previously reported in a study evaluating how ⁶⁸Ga-PSMA-11 PET changed management in patients with BCR [12]. This prior study surveyed the referring clinicians to determine the effect that the imaging results would have on patient management but did not compare the detection rate of different MRI and PET sequences for suspicious lesions, as was done in this study.

In this study, we demonstrated that T2-weighted imaging is the most effective MRI sequence for identifying anatomic correlates to ⁶⁸Ga-PSMA-11 PET-avid abdominopelvic lymph nodes, and DCE is most effective for identifying PSMA-avid foci in the prostatic bed. Also, longer PET acquisition times increased the detection sensitivity for subcentimeter PSMA-avid foci.

Early and accurate detection of sites of disease in patients with BCR allows for patient-specific treatments, for example, targeted external beam radiation to the prostate bed or sites of oligometastatic disease, or systemic therapy for more widespread disease [17]. It is important, therefore, to be able to precisely localize sites of persistent disease. ⁶⁸Ga-PSMA-11 PET/MRI may be especially important in patients with low serum PSA levels, who may present with subcentimeter nodes. The high incidence of PSMA-positive lymph nodes less than 1 cm in size indicates that MRI alone would not be as sensitive for detection of abnormal nodes.

The high sensitivity of PSMA in the study patient population (89.1%) mirrored that demonstrated in prior studies [6, 7]. The use of PET/MRI rather than PET/CT has the potential to detect even more lesions, especially in the prostate bed, where multiparametric MRI has been shown to be highly sensitive for detection of suspected local recurrence, including for tumor recurrence that is not evident on PET [18]. In particular, our results illustrate the utility of DCE MRI to detect PSMA-avid foci in the prostatic bed, with a localization rate of 87.5%. DCE MRI has previously been shown to have a high sensitivity: Haider et al. demonstrated a sensitivity of 72% of DCE for recurrence after external beam radiotherapy, and Rischke et al. demonstrated a sensitivity of 67% of DCE for recurrence after radical prostatectomy [19, 20], which is also reflected in a meta-analysis by Wu et al. [21]. The limitations that were observed in sensitivity of T2-weighted imaging and DWI are likely due to postoperative and radiation changes in the prostate as well as artifact from brachytherapy seeds [22–24]. Prior studies have suggested moderate to high accuracy of DWI for the detection of recurrent prostate cancer after radiation therapy [25, 26]. These results suggest that DCE should be an integral part of a PET/MRI protocol that aims to detect recurrent tumor in a background of treatment changes.

Previous work has shown that prostate MRI can be more sensitive than PSMA PET for suspected local recurrence, especially for lesions in close proximity to the bladder, due to severe scatter artifact associated with activities in the bladder [18]. This is a potential benefit of pairing PSMA PET with MRI instead of with CT. In our study, two additional lesions were detected, which were identified only on DCE MRI. Scatter did not appear to limit identification of PSMA-avid lesions on the PET sequences in our study. It is possible that time-of-flight reconstructions or furosemide administration resulted in lower artifact compared to what has been reported previously using non-time-of-flight PET/MRI systems [18]. With newer PSMA-targeted imaging agents that have more prominent biliary excretion, scatter artifact will likely become less of an issue [27, 28].

Although multiparametric MRI (mpMRI) alone or in conjunction with PSMA PET is highly sensitive for the detection of prostatic bed recurrence [18–21, 29], disease is frequently outside the prostatic bed. Thirty of 49 patients with PSMA-positive disease had no evidence of recurrent cancer in the prostatic bed, but had PSMA uptake elsewhere. Since sites of suspected recurrence may be found in many regions outside of prostatic bed, the combination of mpMRI of the prostate, whole-body MRI, and PSMA PET is promising for a comprehensive evaluation of patients with BCR. Although nearly all PSMA-avid lesions had an MRI correlate, these lesions would not have been interpreted as metastatic lesions without PSMA PET. Of note, we did not evaluate osseous lesions that were not PSMA avid, nor was our protocol optimized to characterize osseous lesions in the absence of PSMA avidity.

One of the unexpected benefits of simultaneous PSMA PET/MRI in this study was the added value of the 8-min PET acquisition compared to the 3-min acquisition for the detection of uptake within small lymph nodes. This is longer than the recommendation in the EANM/SNMMI imaging guidelines [30]. The increased acquisition time was initially implemented to take advantage of the time used to perform the simultaneous mpMRI. The 25 cm z-axis length of the detector in both available simultaneous PET/MRIs allows for doubling the field of view coverage compared to available PET/CTs, thereby allowing for the doubling of the PET acquisition time without changing the length of the study. Though increased uptake time is associated with increased detection sensitivity [31], the 8-min acquisition was performed before the 3-min acquisition, so this likely did not bias the results towards the longer acquisition. In future implementations, the 8-min acquisition will be combined into the whole-body protocol, allowing for a complete whole-body PET/MRI to be performed in 30 min.

This study has several limitations. First, this is a retrospective study, and there is no histopathological correlation of disease for the PSMA-positive foci that were detected. The majority of patients were treated according to the findings of their PET/MRI scans without biopsy confirmation. It is possible that there were false-positive PSMA PET-avid foci, although no pathology proven false-positive lesions are known. The goal of this study was to identify which MRI sequences should be used to co-localize suspicious findings, rather than estimating the diagnostic accuracy of PSMA PET/MRI. Additionally, only a small number of limited MRI sequences were obtained for localization of PSMA-positive lesions. It is possible that optimized, dedicated, MRI sequences would have a better diagnostic performance; yet, the goal of the study was not to estimate the diagnostic accuracy of MRI for the detection of PSMA PET-avid foci, but to identify the best sequences to co-localize suspicious findings.

A PSMA PET/MRI scan obtained for the assessment of men with BCR after treatment of prostate cancer should include small FOV T2 images of the abdomen and pelvis and DCE MRI of the pelvis to improve co-localization of suspicious findings identified on PSMA PET. In addition, a longer 8-min PET acquisition may improve the detection of small foci of a suspected recurrent cancer.