Comparing the diagnostic performance of radiotracers in prostate cancer biochemical recurrence: a systematic review and meta-analysis

Objectives To systematically assess the early detection rate of biochemical prostate cancer recurrence using choline, fluciclovine, and PSMA. Methods Under the guidance of the Preferred Reporting Items for Systematic reviews and Meta-Analysis Diagnostic Test Accuracy guidelines, literature that assessed the detection rates (DRs) of choline, fluciclovine, and PSMA in prostate cancer biochemical recurrence was searched in PubMed and EMBASE databases for our systematic review from 2012 to July 15, 2021. In addition, the PSA-stratified performance of detection positivity was obtained to assess the DRs for various methods, including fluciclovine, PSMA, or choline PET/CT, with respect to biochemical recurrence based on different PSA levels. Results In total, 64 studies involving 11,173 patients met the inclusion criteria. Of the studies, 12, 7, and 48 focused on choline, fluciclovine, and PSMA, respectively. The pooled DRs were 24%, 37%, and 44%, respectively, for a PSA level less than 0.5 ng/mL (p < 0.001); 36%, 44%, and 60% for a PSA level of 0.5–0.99 ng/mL (p < 0.001); and 50%, 61%, and 80% for a PSA level of 1.0–1.99 ng/mL (p < 0.001). The DR with 18F-labeled PSMA was higher than that with 68Ga-labeled PSMA, and the DR was 58%, 72%, and 88% for PSA levels < 0.5 ng/mL, 0.5–0.9 ng/mL, and 1.0–1.99 ng/mL, respectively. Conclusion The DRs of PSMA-radiotracers were greater than those of choline-radiotracers and fluciclovine-radiotracers at the patient level. 18F-labeled PSMA achieved a higher DR than 68Ga-labeled PSMA. Key Points • The DRs of PSMA-radiotracers were greater than those of choline-radiotracers and fluciclovine-radiotracers at the patient level. • 18F-labeled PSMA achieved a higher DR than 68Ga-labeled PSMA.

Introduction this stage of the disease, it is still essential to define the location and extent of metastasis and initial recurrence to help urologists make further treatment plans [5]. The detection of subtle or occult recurrence and metastasis after treatment continues to pose a challenge [6]. In this setting, PET/CT is particularly superior to conventional imaging modalities such as CT (computed tomography)/MRI (magnetic resonance imaging) because of its higher detection rate (DR) for low-volume metastatic or locally initial recurrence [7,8].
To date, of the radiolabels that have been evaluated, 11Ccholine and 18F-fluciclovine were approved by the Food and Drug Administration (FDA) in 2012 and 2016 [9]. Prostatespecific membrane antigen (PSMA), a cell surface protein, is highly expressed in the majority of PCa cells and in PCa recurrence [10,11]. 68 Ga-and 18 F-labeled PSMA are promising new radiotracers for detecting the BCR of PCa and radionuclear therapy. [ 68 Ga]Ga-PSMA-11 was the first PSMA PET tracer that was approved by the FDA [6]. Moreover, 18 F-labeled PSMA agents have also been employed in clinical practice.
Previous evidence-based studies have compared the diagnostic performance of choline, fluciclovine, and PSMA PET/ CT in PCa patients with BCR, and particularly at a PSA level less than 2 ng/mL 6 . However, they only concentrated on longhalf radionuclides as 18 F-labeled PET tracers and compared their diagnostic performance in detecting patients with BCR. To our knowledge, a comprehensive comparative metaanalysis of choline, fluciclovine, and PSMA for detecting PCa patients with BCR and low PSA levels has not been performed. Furthermore, several studies have shown that the higher image resolution of 18 F, as a longer half-life nuclide, is slightly better than that of 68 Ga [12,13]. However, evidencebased data based on 18 F-labeled and 68 Ga-labeled PSMA are still lacking. Therefore, the aims of this meta-analysis were to compare the DRs of radiotracers, including choline, fluciclovine, and PSMA PET/CT, for biochemical recurrence with PSA levels less than 2 ng/mL and to perform subgroup analyses based on 18 F-labeled and 68 Ga-labeled PSMA.

Search strategy
This meta-analysis was performed in accordance with the PRISMA-DTA statement [14]. Two reviewers searched the library databases of PubMed and Embase that involved the DR of PET/CT using choline, fluciclovine, and PSMA agents between 2012 and July 2021. The search terms included the following: prostatic neoplasms, prostate cancer, recurrence, biochemical recurrence, 1 8 F-choline, 11

Study selection
Both retrospective and prospective studies involving males with PCa with BCR who underwent PET/CT using choline, fluciclovine, and PSMA agents between 2012 and July 2021 were included. In addition, single-arm trials, comparative, single-center, multicenter, and clinical trials were also included. Studies were excluded as follows: abstracts, comments, letters, conference records, case reports, reviews, and meta-analyses, non-English articles, studies for staging purposes, and studies assessing specific types of metastatic disease, such as that of bones or lymph nodes. If the studies included patients from the same group, the largest sample was reviewed.

Data extraction and quality assessment
Two reviewers independently extracted and confirmed the data. Information was recorded from each study, including year of publication, radiotracer, imaging protocols, country of origin, study design (prospective or retrospective, multicenter or single center), patient age, sample size, treatment, PSA stratified into tiers (PSA level less than 0.5 ng/mL ng/mL, 0.5-0.99 ng/mL, and 1-1.99 ng/mL), and DR. We used the revised Quality Assessment of Studies of Diagnostic Accuracy, which was included in the QUADAS-2 tool for quality assessment [15]. Each item was judged as "yes," "no," or "unclear." Any disagreements were resolved by consensus.

Statistical analysis
The pooled estimates with 95% CIs were the DRs of PSAstratified patients with biochemical recurrence after treatment. The pooled estimates for the DRs of different radiotracers were compared using a chi-square test. Forest plots with 95% confidence intervals were used to visually assess the results. The inconsistency index (I 2 ) was used to assess statistical heterogeneity of the included studies. The Cochrane Q with p < 0.05, and I 2 > 50% indicated significant heterogeneity. A randomeffects model was applied, and marked heterogeneity was observed. Statistical significance was set at a p-value less than 0.05. Egger's test and funnel plot tests were conducted to assess the publication bias. The open-source statistical software R was used to conduct all statistical analyses (version 3.6.3; www.rproject.org/). The QUADAS quality evaluation was conducted using RevMan (version 5.3). Figure 1 presents an overview of the inclusion process. Initially, 1759 total articles were identified through PubMed and Embase databases using the search terms and keywords (1396 in PubMed, and 363 in Embase). In total, 324 duplicate records were removed. After screening titles and abstracts, 1345 records were excluded; 900 because they were not relevant to the study; 90 as they were abstracts or conference records; 180 as they were letters, reviews, or meta-analyses; and 175 as they were case reports and comments.

Literature search
Of the remaining studies, 90 full-texts were reviewed, and 64 studies were included in this meta-analysis. Of these, 48 studies focused on the performance of PSMA PET/CT for patients with BCR. Seven and 12 studies were included for fluciclovine and choline PET/CT, respectively. Of these, two studies evaluated both PSMA and choline [16,17], and one study evaluated both chorine and fluciclovine [18]. The characteristics of the included articles are presented in Tables 1, 2 and 3. Figure 2 shows the proportion of different tracers in the included studies under PSA stratification.

Publication bias and heterogeneity and quality assessment
A symmetrical funnel-shaped distribution of PSA < 0.5 ng/mL (p = 0.96) and PSA levels of 0.5-1.0 ng/mL (p = 0.12) indicated that there was no significant publication bias. However, publication bias was found in the cohorts with PSA levels of 1-1.99 ng/mL (p < 0.001). The Egger test was used to quantify significant asymmetry.

Discussion
Whereas other meta-analyses have investigated the diagnostic roles of tracers applied in PCa with BCR [2,6,7,78,79], to our knowledge, this is the first comparative meta-analysis that focuses on all three relevant tracers for the early detection of this disease. This meta-analysis showed a higher pooled DR for PCa with BCR using PSMA compared to that with fluciclovine and choline PET/CT for three PSA levels, and we observed a significant difference. These results are in accordance with a previous meta-analysis that included only 18 F-labeled choline, fluciclovine, and PSMA, which reported that PSMA was better than choline and fluciclovine [6]. A meta-analysis recently showed that there is a trend but no significant difference when the PSA level is < 0.5 ng/mL and 0.5-0.9 ng/mL. However, our study found an absolute statistical difference when comparing the DRs of PSMA and fluciclovine. A possible reason for this is the limited number of studies. In general, PET/CT imaging is more likely to be negative with low PSA values. Radiation therapy remains the gold standard for intermediate-and high-risk, localized prostate cancer. While these are effective forms of management, approximately 30-40% of cancers still recur following treatment, manifesting as a  Fig. 2 The proportion of three tracers in different PSA stratification rising prostate-specific antigen (PSA). The key issue for patients with BCR is the early and correct identification of recurrent or metastatic disease. Conventional imaging modalities consisting of CT, bone scan, and MRI have been used for patients with PCa, but their diagnostic performance in detecting minimal or occult lesions is limited. At this stage of the disease, it is important to determine the location and extent of metastases to determine the next course of management. PET is an established, non-invasive, molecular imaging modality that uses different radiolabeled tracers, a combination of a radionuclide and a biologically active molecule, targeted to specific receptors to localize disease. PET/CT has a higher detection rate of intra-prostatic tumors that might have clinical implications regarding focal therapy such as radiotherapy and surgical planning. The current study has demonstrated that PSMA-based tracer PET/CT imaging seems to be a promising tool and shows clear superiority in the detection of PCa with BCR and PSA when compared to choline-and fluciclovinebased tracers. In clinical practice, choline PET/CT is the most commonly used radioactive tracer [17]. Significantly, choline PET/CT exhibits a higher DR only at high PSA levels [80]. A previous study has shown that the DR of PCa with BCR and PSA < 1.5 ng/mL was only below 30% when using cholinebased tracers PET/CT [81,82], which is in accordance with our findings. In our study, 18 F-fluciclovine also showed a higher DR than choline (37% vs 24% for 0.2-0.5 ng/mL, 44% vs 36% for 0.5-1.0 ng/mL). Similarly, in a prospective study, the authors showed that the overall performance with [ 18 F]FACBC, a relatively new radiotracer, was higher than that of 11 C-choline, and they found that this difference in DR was particularly significant for PCa with BCR and a PSA level < 1 ng/mL 18 . Furthermore, current EAU guidelines recommend that choline PET/CT only be used for non-early PCa recurrence with serum PSA levels > 1 ng/mL [83]. Afshar-Oromieh et al calculated the detection rate of [ 68 Ga]Ga-PSMA-11 to be 46% (32/ 69) for PSA < 0.2 ng/mL, 46% (50/108) for 0.2-0.5 ng/mL, and 73% (87/119) for 0.5-1.0 ng/mL [43]. In addition, a higher detection rate of PCa with BCR with low PSA levels was suggested by other research [59]. Considering these Fig. 3 Depiction of the study quality assessment using the QUADAS2 tool However, PSMA PET/CT is an increasingly used tracer for patients with BCR and achieves a high DR for early PCa recurrence (PSA ≤ 2.0 ng/mL) [84]. Furthermore, the strength of evidence was limited by publication bias, multiple reference standards, and a lack of consistent clinical follow-up times. In particular, our meta-analyses also performed subgroup analyses based on 18 F-labeled and 68 Ga-labeled PSMA. We observed statistically significant differences when comparing 18 F-labeled and 68 Ga-labeled PSMA (PSA < 0.5 ng/mL: 58% vs 44%, p < 0.01; PSA level 0.5-1.0 ng/mL: 72% vs 56%, p < 0.01; PSA 1.0-1.99 ng/mL: 88% vs 78%, p < 0.01). To date, to our knowledge, this is the first evidencebased study to evaluate the DRs of 18 F-labeled and 68 Ga-labeled PSMA. Recently, there have been multiple metaanalyses showing that the summary DR of 18 F-labeled PSMA in patients with BCR was approximately 49% for PSA < 0.5 ng/mL [1,78,85], which is slightly better than the 44.9% detection rate of 68 Ga-PSMA in a recent prospective study [73]. Compared to 18 F-PSMA, 68 Ga-PSMA ligands have a short half-life (68 min), and thus are inconvenient for transport [86]. Moreover, they are characterized by a lower signal-to-noise ratio for images [87], limiting its clinical  18 F-PSMA analogs seemed to be more favorable due to their longer half-life and a higher physical spatial resolution, and [ 18 F]PSMA-1007, as a secondgeneration 18 F-labeled PSMA tracer, demonstrated high labeling yields, better tumor uptake, and hepatobiliary excretion, making it an ideal PSMA-target tracer for diagnostic imaging in patients with BCR. Accordingly, this might explain why some authors considered [ 18 F]DCFPyL to be a good replacement for recurrent PCa.
Our meta-analysis has several limitations. First, significant heterogeneity was observed in all cohorts. Second, because the sample size was limited, retrospective, singleinstitutional studies accounted for a large amount, which might be one of the reasons for the selection bias. Additionally, the different PET/CT scanners, radiotracers, patient populations, and various treatment modes increased the risk of bias and significant heterogeneity.
In conclusion, our meta-analysis revealed that PSMAradiotracers demonstrate a potentially promising DR with low PSA levels in biochemically recurrent PCa. PSMA has a relatively higher DR than fluciclovine and choline in PCa patients with BCR and with PSA < 2.0 ng/mL. Additionally, 18Flabeled PSMA achieved a higher DR than 68 Ga-labeled PSMA.
Funding Key Laboratory of Functional Molecular Imaging of Tumor and Interventional Diagnosis and Treatment of Shaoxing City.

Declarations
Guarantor The scientific guarantor of this publication is Jiwei Mao.

Conflict of interest The authors declare no competing interests.
Statistics and biometry One of the authors performed statistical analysis for this paper.
Informed consent Written informed consent was not required for this study because this study was designed as a systematic review with metaanalysis.
Ethical approval Institutional Review Board approval was not required because this study was designed as a systematic review with metaanalysis.

Methodology
• systematic review • meta-analysis Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .