E-PSMA: the EANM standardized reporting guidelines v1.0 for PSMA-PET

[⁶⁸Ga]Ga-PSMA-11 was introduced in 2011 by the German Cancer Research Centre [7]. As the most used radioligand, its success can be explained by the relatively straightforward isotope production (germanium-gallium generator) and radiolabeling of the tracer [22]. [¹⁸F]DCFPyL [23] is another widely used tracer that was introduced as a successor to [¹⁸F]DCFBC [24]. Compared to [⁶⁸Ga]Ga-PSMA-11, ¹⁸F-labeled compounds have longer half-life, allowing imaging at later time point, a higher production capacity and a more centralized production [25, 26]. [¹⁸F]PSMA-1007 represents another ¹⁸F-tracer variant characterized by predominant hepatobiliary excretion [27], thus reducing the urinary excretion of the radiotracer. Several other PSMA ligands are available for PET imaging (e.g., [¹⁸F]rh-PSMA7, [⁶⁸Ga]Ga-PSMA I&T, [¹⁸F]JK-PSMA-7, [⁶⁸Ga]Ga-PSMA-R2), but few data in literature are available at present [28, 29].

For [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]DCFPyL, high PSMA uptake is noted in the cortex of the kidneys, parotid and submandibular salivary glands, and duodenum. Moderate median uptake (SUV_max >3) is noted in the spleen, liver, and lacrimal glands [30]. In comparison to [⁶⁸Ga]Ga-PSMA-11, the uptake of [¹⁸F]PSMA-1007 is higher in the liver, gallbladder, pancreas, and sublingual glands, and lower in the kidneys, bladder, and lacrimal glands [31]. Since daily repeatability on the PSMA-ligand uptake for each organ is essential for semi-quantitative analysis, knowledge on dosimetry and physiological uptake is important [32]. Moreover, knowledge about uptake in benign tissues is important, by enabling to calculate the thresholds that can classify potential malignant lesions when interpreting PET-scans [32]. Kidney, spleen, and salivary uptake are higher on [⁶⁸Ga]Ga-PSMA-11 compared to [¹⁸F]DCFPyL, while the liver shows slightly lower uptake. The blood pool (aorta) is similar in both [¹⁸F]DCFPyL and [⁶⁸Ga]Ga-PSMA-11 and can therefore be used as a benchmark in assessing lesions based on SUV_max [32]. One limitation of both [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]DCFPyL is their increased urinary excretion in the ureters and the bladder, which limits detection of local recurrence in the prostate bed. This excretion is absent for [¹⁸F]PSMA-1007 due to hepatobiliary clearance, and might improve local detection of PCa [31].

The expression of PSMA can predominantly be found in PCa, but benign and other malignant tissues are known to express PSMA and have extensively been described [33]. Here we present an array of pitfalls that can influence reader’s interpretations. Although the positive predictive value and specificity of PSMA-PET are known to be high [14, 34], cautious reading and knowledge of common pitfalls should be considered while interpreting PSMA-PET images and drafting the medical report.

The development of consensus guidelines for interpretation of PSMA-PET may contribute to provide more consistent reports in clinical practice. The standardization of PSMA-PET interpretation may also contribute to increasing the data reproducibility within clinical trials. Defined criteria for interpreting PSMA-PET images would help improving accuracy, precision, and repeatability of this diagnostic procedure, thus improving patients’ management and outcomes. Therefore, consensus interpretation is necessary to provide comparison between clinical trials and to meet upcoming clinical diagnostic needs. Consensus guidelines in PSMA-PET interpretation are also needed to communicate the exact location of findings to referring physicians, to support clinician therapeutic management decisions, as happens for MDT.

In view of these considerations, the E-PSMA standardized reporting guidelines, a document supported by the European Association of Nuclear Medicine (EANM), is aimed at providing consensus statements among a panel of experts in PSMA-PET imaging, to develop a structured report for PSMA-PET in PCa and to harmonize diagnostic interpretation criteria.

According to the abovementioned purposes, a panel of worldwide experts in PSMA-PET was established. Panelists were selected based on their expertise and publication record in the diagnosis or treatment of PCa, in their involvement in clinical guidelines and according to their expertise in the clinical application of radiolabeled PSMA inhibitors. The panelists involved are reported in Table 1.

Panelists were actively involved in all stages of a modified, non-anonymous, Delphi consensus process, as displayed in Fig. 1. In the first two rounds, panelists identified issues regarding PSMA-PET reporting and made proposals about possible criteria to harmonize and standardize the PSMA-PET reporting process. Finally, panelists gave their inputs to identify which criteria or parameters currently used in research studies might be implemented in clinical reports. According to comments, proposal, and suggestions made by panelists during the first and second round, a questionnaire (Fig. 2) composed by 16 questions has been generated. All panelists were asked to answer the questions as in favor or disagree. The inter-rater agreement was measured for each question using Fleiss’ kappa (0 poor agreement; 0.01–0.20 slight agreement; 0.21–0.40 fair agreement; 0.41–0.60 moderate agreement; 0.61–0.80 substantial agreement; 0.81–1.00 almost perfect agreement).

Fourteen out of sixteen questions reached an almost perfect concordance between panelists. Question nos. 13 (k = 0.61) and 14 (k = 0.55) reached only a moderate agreement between experts. At question nos. 1, 2, 3, 6, 7, 8, 9, 10, 11, and 16, the agreement was in favor of the topic proposed. Therefore, structured report, synoptic tables, technical information, type of acquisition, staging criteria, visual and quantitative PSMA expression, and incidental findings have been included in E-PSMA reporting guidelines.

At questions 4, 5, 12, and 15, the agreement was against the topic proposed. Therefore, reader experience, quality of the scan, TBR instead of SUV_max, and the use of sub-categories in the 5-point scales were not included in E-PSMA reporting guidelines. Since questions 13 and 14 did not reach consensus, PSMA-RADS criteria [20] were not included in the final report. However, considering comments and proposal made in rounds 1 and 2, all panelists agreed in the inclusion of a modified 5-point scale to reflect the likelihood of the presence of PCa-related lesions.

Many technical factors relating to methodology may affect the quality of a PET image acquisition. As these methodological aspects may influence the quality and interpretation of PSMA-PET images, they should be described in each PSMA-PET report. The head of the report should include synoptic tables to summarize PSMA-PET technical data (synoptic Table 1).

Tracer activity used should be reported in MBq, whether fixed (333 MBq for [¹⁸F]DCFPyL) or patient-specific (1.8–2.2 MBq/Kg for [⁶⁸Ga]Ga-PSMA-11 and 4 MBq/Kg for [¹⁸F]PSMA-1007). In case a different tracer dosage is used, this should be reported.

As tumor PSMA uptake does not plateau within commonly used intervals, uptake intervals may affect lesion-to-background contrast, detection rates, and quantitative reads. Therefore, the uptake interval between tracer injection and imaging should be reported. When used, type of diuretics and dosage should be reported.

The type of CT protocol (low dose vs. diagnostic) and use of oral or intravenous contrast should be reported. If intravenous contrast is used, consideration should be given to imaging in the urography phase to better delineate the ureters and bladder, in the staging or biochemical recurrence setting.

The standard acquisition should be from the vertex to the mid-thigh. In patients where there is known disease or concern for disease in the lower extremities, the acquisition can be extended (total-body acquisition). Total-body acquisition can be considered also in patients eligible for radioligand PSMA-based therapy.

The overall quality of the study should be assessed. This judgment is based on reader personal experience and it is aimed to evaluate the reproducibility of the scan. However, in consideration of experts’ recommendation, this information should not be included in the report.

An overview of all items necessary for reporting of findings on PSMA-PET that can be clinically used is provided in synoptic Table 2. Synoptic Table 2 should be included at the end of the clinical report as the final summary.

In primary staging, early detection of metastases is essential. Patients with proven metastatic disease are usually treated differently than patients with localized PCa. The detection of any additional lesion may change patient management and result in local radiotherapy, extended lymph node dissection, (oligo)metastases-directed therapy, or systemic (palliative) treatment. In a systematic review [66], high variation in sensitivity (33–92% on a per-lesion analysis) with overall optimal specificity (82–100% on a per-lesion analysis) was found in the detection of lymph node metastases, correlated by histopathological evaluation (extended pelvic lymph node dissection). Additionally, the primary tumor is nearly always detected by PSMA-PET, and PET metrics correlated with histologic grades (ISUP classification) [67, 68]. Analyses regarding the diagnostic accuracy of ¹⁸F-labeled PSMA-PET in primary staging are ongoing. In yet unpublished results of a prospective trial, a sensitivity of 30.6–41.9% of [¹⁸F]DCFPyL PET for the detection of lymph node metastases was determined [68]. Regarding [¹⁸F]1007-PSMA-PET, local staging appears a promising technique, considering the low urinary excretion of this radiotracer [69].

Recently, in an Australian, multi-center, randomized, phase III clinical trial (proPSMA) [70], [⁶⁸Ga]Ga-PSMA-11 PET provided greater accuracy in identifying nodal and distant metastases vs. conventional imaging (CT and bone scan) prior to curative-intent surgery or radiotherapy in high-risk PCa. Furthermore, [⁶⁸Ga]Ga-PSMA-11 PET vs. conventional imaging was associated with change in management in 28% vs. 15% of patients and was associated with a lower percentage of equivocal findings (7% vs. 23%). Finally, even if both imaging techniques involve exposure to radiation, the dose associated with [⁶⁸Ga]Ga-PSMA-11 was less than half that associated with conventional imaging (8.4 mSv vs. 19.2 mSv).

Regarding these considerations, PSMA-PET is a suitable replacement for conventional imaging, providing superior accuracy, to the combined findings of CT and bone scanning, in patients with high risk of nodal involvement [71], while patients at lower risk should be spared by this imaging procedure.

The detection rate of metastases (i.e., percentage positive scans) of PSMA-PET in patients with BCR has been studied intensively. A recent meta-analysis [11] showed an overall detection rate in patients with BCR of 76%. At low PSA values (<0.5 ng/mL), detection of metastases was 45%. Furthermore, a recently published large prospective study showed comparable results [14]. The positive predictive value (PPV) of [⁶⁸Ga]Ga-PSMA-11 PET has been calculated as 92% [14]. For ¹⁸F-labeled PSMA (namely [¹⁸F]DCFPyL and [¹⁸F]PSMA-1007), fewer results concerning the diagnostic accuracy in patients with BCR are available at present. For [¹⁸F]DCFPyL, detection rates ranging from 84.6 to 86.3% have been documented, with a detection at low PSA values (<0.5 ng/mL) of 60% [62, 72]. Preliminary results of a large prospective multicenter trial [73] showed a PPV of 84.5% for [¹⁸F]DCFPyL PET. Similar results have been described for [¹⁸F]PSMA-1007 [74, 75].

At present, EAU guidelines suggest performing PSMA-PET in any case of proven BCR [8]. However, the incidence of false-negative scans is not negligible. In recurrent setting, PSMA-PET detection rate is influenced by several factors as recently reported in some prediction models [55, 56, 58, 76]. PSA, as expression of tumor burden, is not the only influencing parameter. PSAdt and Gleason score as expression of tumor aggressiveness together with the administration of concurrent ADT are parameters able to influence the likelihood of a positive scan [76]. While reporting PSMA-PET in recurrent setting, also the clinical stage of the disease should be taken into consideration [55, 57]. Persistent disease after surgery (detectable PSA levels after surgery) [61] and BCR (undetectable PSA levels after surgery), while both represent an early recurrence, are two conditions with different outcome and different incidence of detectable metastatic disease [55, 57, 61]. Finally, the proper knowledge of potential pitfalls during PET image interpretation, while probably reducing PSMA-PET sensitivity, will increase its overall specificity [33].

The role of PSMA-PET in advanced setting presents less level of evidence compared to initial staging and BCR. Non-metastatic CRPC (nmCRPC) is a condition characterized by a rising PSA level, castrate testosterone levels, and no evidence of distant metastases by conventional bone scan and cross-sectional imaging of the chest, abdomen, and pelvis [8]. This clinical scenario became recently of high interest, since new androgen receptor-targeted therapies have been recently approved in this stage (SPARTAN, PROSPER, and ARAMIS trials). In the setting, PSMA-PET proved its ability to detect PCa locations in patients negative at conventional imaging (nmCRPC). Thus, as recently stated by the EAU consensus panel in advanced PCa [16], ASCO guidelines [15], and Advanced PCa Consensus Conference (APCC) [9], it should also be recognized that the majority of patients in clinical trials who benefited from the addition of next-generation ADT would probably have had positive PSMA-PET imaging results. It is uncertain whether stratification based on PSMA-PET would identify subgroups of patients (e.g., those with distant rather than local or loco-regional disease) that benefit most. Tumor heterogeneity is a key event in advanced PCa. Tumor cells exhibit different phenotypes and, accordingly, PSMA might not be over-expressed in all metastatic sites. This event should be taken into consideration while reporting PET scan in mCRPC, namely while evaluating the response to systemic therapy.

Regarding the response to therapy assessment, the Prostate Cancer Clinical Trials Working Group (PCWG) criteria include clinical and laboratory parameters, as well as conventional imaging modalities such as CT and bone scan findings but advanced molecular imaging techniques are not yet considered. PSMA-PET is not yet validated for response assessment, especially in the context of clinical trials [77]. Recently, consensus statement criteria for response evaluation using PSMA-PET were developed [16]. The statements regarded both the utility and the best time to perform PSMA-PET, as well as the optimal strategy to select patients who may benefit from treatments and criteria to be used for evaluation of response when using different types of PSMA tracers. Consensus was met on the utility of PSMA-PET for response assessment in patients with metastatic PCa, irrespective of the moment and type of treatment used (i.e., local or systemic), but solely in cases when clinical management is expected and with a 3-month interval after initiation of therapy in HSPC. Proposed criteria should only be adopted in the context of clinical trials, preferably by dividing patients in responders and non-responders. The category of responders is including the whole spectrum of patients presenting with stable disease, partial and complete response, while the non-responders are patients with progressive disease on PSMA-PET imaging. The pre-requisite of robust and reproducible interpretation of response to treatment when using PSMA-PET scans is adequate semi-quantitative evaluation. Consensus was reached on the use of SUV parameters for this purpose, by optimizing and harmonizing the protocols. Tools to estimate the total tumor burden represent a feasible alternative to reduce interobserver variability, being currently developed. However, one major issue is how to best define disease progression. Recently, PSMA-PET Progression (PPP) criteria have been proposed [78]. PPP defines PSMA treatment response in three different criteria: (1) appearance of 2 or more new PSMA-positive distant lesions, (2) appearance of 1 new PSMA-positive lesion plus consistent clinical and/or laboratory data and recommended confirmation by biopsy or correlative imaging within 3 months of PSMA-PET, and (3) increase in size or PSMA uptake of 1 or more existing lesions by 30% plus consistent clinical and/or laboratory data and/or confirmation by biopsy or correlative imaging within 3 months of PSMA-PET. These criteria should be taken into consideration while reporting PSMA-PET in patients undergoing systemic therapies.

Finally, the growing interest in PSMA-targeted therapies is not unnoticed. One phase 3 trial (VISION, NCT03511664) and one phase 2 trial (TheraP, NCT03392428) [79] are currently ongoing. Both trials use PSMA-PET to identify patients with high PSMA expression, who are suitable candidates for [¹⁷⁷Lu]Lu-PSMA-617 therapy, but they use different PET imaging thresholds to define suitability. In addition, the TheraP trial uses 2-[¹⁸F]FDG PET to assist in identifying sites of PSMA-negative disease that cannot be targeted with [¹⁷⁷Lu]Lu-PSMA-617. These patients have been shown to have a poor prognosis [80]. Finally, studies indicate that both intrapatient and interpatient PSMA expressions are highly heterogeneous in patients candidate for PSMA-targeted therapy, and that many of them express little or no PSMA. This consideration should be clearly stated and defined while interpreting PSMA-PET in advanced setting [81, 82].