Head-to-Head Comparison between 18F-FES PET/CT and 18F-FDG PET/CT in Oestrogen Receptor-Positive Breast Cancer: A Systematic Review and Meta-Analysis

18F-FDG PET/CT is a powerful diagnostic tool in breast cancer (BC). However, it might have a reduced sensitivity in differentiated, oestrogen receptor-positive (ER+) BC. In this setting, specific molecular imaging with fluorine-oestradiol (18F-FES) PET/CT could help in overcoming these limitations; however, the literature on the diagnostic accuracy of this method is limited. We therefore planned this systematic review and meta-analysis to compare 18F-FDG and 18F-FES PET/CT in ER+ BC patients. We performed a literature search to identify all studies performing a head-to-head comparison between the two methods; we excluded review articles, preclinical studies, case reports and small case series. Finally, seven studies were identified (overall: 171 patients; range: 7–49 patients). A patients-based analysis (PBA) showed that 18F-FDG and 18F-FES PET/CT had a similar high pooled sensitivity (97% and 94%, respectively) at the lesion-based analysis (LBA), 18F-FES performed slightly better than 18F-FDG (pooled sensitivity: 95% vs. 85%, respectively). Moreover, when we considered only the studies dealing with the restaging setting (n = 3), this difference in sensitivity was even more marked (98% vs. 81%, respectively). In conclusion, both tracers feature an excellent sensitivity in ER+ BC; however, 18F-FES PET/CT could be preferred in the restaging setting.


Introduction
F-FDG PET/CT is a very useful imaging tool in staging breast cancer (BC) patients with locally advanced disease [1][2][3] or in restaging patients with biochemical relapse without evidence of structural recurrence on conventional imaging procedures [4,5]. However, the 18 F-FDG avidity of BC is associated with histopathological features and can thus show relevant variations [6,7]. In particular, patients with less differentiated, triple-negative BC often present with 18 F-FDG avid metastases and, in this setting, 18 F-FDG PET/CT can provide relevant additional information when compared with morphological imaging modalities or with surgical staging [8]. Conversely, well-differentiated and oestrogen receptor-positive (ER+) BCs often present metastases with relatively low glucose consumption and thus inconstant positivity on 18 F-FDG PET/CT. Therefore, in this particular field of patients, more specific PET/CT agents, such as receptor tracers, still represent an unmet diagnostic need. 2 of 12 Fluorine-oestradiol 18 F-FES PET/CT has been introduced as an effective imaging procedure in patients with metastatic BC and previous histological confirmation of ER+ status on the primary tumour, to assess whether ERs are also expressed in loco-regional or distant metastases [9]. Indeed, 18 F-FES PET/CT provides excellent whole-body information about heterogeneity in ER expression in BC metastases, thus allowing to predict the response to endocrine therapy [10]. However, little can be said about the diagnostic role of this receptor-specific tracer and, although some studies proved its low sensitivity in detecting BC liver metastases [11,12], others showed its relevant diagnostic impact on BC bone metastases [11,13]. Overall, it is not clear yet whether 18 F-FES PET/CT could be proposed as a core imaging procedure in ER+ BC patients with suspected distant metastases and whether it could be preferred to 18 F-FDG PET/CT, at least in some specific instances.
To clarify this diagnostic issue, we performed a systematic search of the literature to identify original studies reporting a diagnostic head-to-head comparison of 18 F-FES PET/CT and 18 F-FDG PET/CT in detecting ER+ BC metastases. We also carried out a meta-analysis of the available data using sensitivity as a diagnostic outcome measure.

Materials and Methods
The systematic review was conducted according with the PRISMA statement [14].

Search Strategy
Two authors (A.P. and G.B.) searched the available literature independently. The search and selection process consisted of four separate steps.
In the first step, so-called "sentinel" studies were identified in PubMed by entering various combinations of the following keywords: 18 F-FES, 18 F-FDG, PET/CT, breast cancer and oestrogen receptor. In the second step, the results were used to identify specific MeSH terms in PubMed. In the third step, PubMed, CENTRAL, Scopus, Web of Science and the web were searched using the selected MeSH terms. In the final step, we evaluated the studies that compared the sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT in identifying disease localization in patients affected by ER+ BC. All initially eligible articles were screened, and those reporting a head-to-head comparison of 18 F-FES PET/CT and 18 F-FDG PET/CT in BC patients were included. Review articles, studies based on preclinical data, phantom studies, case reports, small case series and studies with overlapping data were excluded.
The references of the included studies were searched to identify further potential matches. The search process was concluded on 21 December 2021.

Data Extraction
The two authors (A.P. and G.B.) extracted independently:

1.
General characteristics of the studies (authors, year of publication, country, study design, population).

2.
Technical parameters (mode of acquisition, fasting before 18 F-FDG injection and premedication, mean injected activity, uptake time, interval elapsed between the two imaging procedures, PET/CT scan field of view, PET/CT image analysis and use of reference standard).

3.
Sensitivity of the two imaging procedures: this parameter was computed as a patientbased analysis (PBA) and a lesion-based analysis (LBA).
In the evaluation phase, full-text articles and their supplementary materials were included; in case of missing data, the responsible corresponding authors were contacted via e-mail. The extracted data were cross-checked and any discrepancy was discussed through a consensus meeting.

Study Quality Assessment
The risk of bias of included studies was assessed by one author (F.F.) using the QUADAS-2 method [15]. Briefly, for each study, an evaluation of the seven QUADAS-2 items was performed and each point was scored as having high, low or unclear risk of bias. High and unclear risk of bias were assigned 1 and 0.5 points, respectively; studies that totalled four or more points were excluded from the meta-analysis.

Statistical Analysis
A proportion meta-analysis was carried out using a random-effects statistical model. Pooled data are presented with their 95% confidence interval (95% CI) values. Heterogeneity across studies was assessed with the I-square statistic (I 2 ): a value of 50% or higher was considered as a high heterogeneity. Publication bias was evaluated with Egger's test [16].
The multidisciplinary follow-up was used as SOR to be able to compute the sensitivity of both image modalities (PBA/LBA). The OpenMeta[Analyst] statistical software (CEBM, Providence, RI, USA) was used for the statistical analyses.

Literature Search Outcome
A total of 55 records were initially identified after duplicate removal, and their titles and abstracts were assessed; 12 articles had to be excluded since they reported single cases or small cases series. Out of the remaining 43 records, 36 were excluded because they did not meet the set inclusion criteria. Therefore, seven articles were finally selected [12,[17][18][19][20][21][22], including 171 ER+ BC patients ( Figure 1). via e-mail. The extracted data were cross-checked and any discrepancy was discussed through a consensus meeting.

Study Quality Assessment
The risk of bias of included studies was assessed by one author (F.F.) using the QUADAS-2 method [15]. Briefly, for each study, an evaluation of the seven QUADAS-2 items was performed and each point was scored as having high, low or unclear risk of bias. High and unclear risk of bias were assigned 1 and 0.5 points, respectively; studies that totalled four or more points were excluded from the meta-analysis.

Statistical Analysis
A proportion meta-analysis was carried out using a random-effects statistical model. Pooled data are presented with their 95% confidence interval (95% CI) values. Heterogeneity across studies was assessed with the I-square statistic (I 2 ): a value of 50% or higher was considered as a high heterogeneity. Publication bias was evaluated with Egger's test [16].
The multidisciplinary follow-up was used as SOR to be able to compute the sensitivity of both image modalities (PBA/LBA). The OpenMeta[Analyst] statistical software (CEBM, Providence, RI, USA) was used for the statistical analyses.

Literature Search Outcome
A total of 55 records were initially identified after duplicate removal, and their titles and abstracts were assessed; 12 articles had to be excluded since they reported single cases or small cases series. Out of the remaining 43 records, 36 were excluded because they did not meet the set inclusion criteria. Therefore, seven articles were finally selected [12,[17][18][19][20][21][22], including 171 ER+ BC patients ( Figure 1).

Qualitative Analysis (Systematic Review)
The seven articles included in the systematic review had been published between 2013 and 2021. All of them had a retrospective design. Three studies were carried out in China, while India, Republic of Korea, USA and Italy, contributed one study each. The characteristics of the studies and their patients' populations are summarized in Table 1.

Technical Aspects
The imaging modality consisted of PET/CT with low-dose computed tomography settings in all cases. Information on fasting before 18 F-FDG (4-6 h) injection were available in all articles. On the other hand, fasting was not required before 18 F-FES injection.
The injected radiotracer activity ranged from 185 to 370 MBq for 18 F-FDG PET/CT and from 111 to 222 for 18 F-FES PET/CT. The time interval between radiotracer injection and PET/CT image acquisition was similar across the studies, being 60 min for both tracers in the majority of cases but one [19] in which 18 F-FES PET/CT acquisition started 80-100 min after the tracer injection.
In all studies, PET image analysis was performed by a combination of qualitative (visual) and semi-quantitative analysis through the calculation of the maximum standardized uptake values (SUV max ).
On visual analysis, 18 F-FDG focal uptake greater than the surrounding normal tissue that could not be explained by physiological activity was considered as positive in five studies [12,18,20,22,23]. In the remaining two studies, the criteria to classify 18 F-FDG PET findings as positive were not clearly specified [17,21]. When 18 F-FES PET/CT was considered, in three studies a SUV max cut-off was introduced to interpret as positive each focal tracer uptake [18,20,22]. On the other hand, a visual interpretation (i.e., uptake higher than surrounding background) was used in another study [12], and in a further three analyses, the criteria were not reported [17,19,21]. All technical aspects are summarized in Table 2.

Diagnostic Performance
The seven articles selected for the systematic review were published between 2013 and 2021 and included populations consisting of 7 to 49 patients affected by ER+ BC (Table 1). Table 3 details the rate of positive cases at the PBA and LBA.
Overall, 18 F-FDG PET/CT and 18 F-FES PET/CT showed very high sensitivity in detecting sites of disease in ER+ BC patients. Indeed, in two studies, 18 F-FDG PET/CT identified more patients with BC lesions/metastases than 18 F-FES PET/CT [19,22]. Conversely, in one study, 18 F-FES PET/CT showed more BC lesions than 18 F-FDG PET/CT [21]. In the other four studies [12,17,18,20], both diagnostic procedures identified at least one BC lesion in the 75 patients analysed. Overall, no significant differences between the two methods were reported.
When the ability for detecting each single lesion/metastasis was considered (i.e., LBA), 18 F-FES PET/CT identified more BC lesions in three studies [18,21,22] and 18 F-FDG PET/CT identified more BC lesions in another three studies [12,19,20]. In one study, including only patients with primary BC, both modalities identified the same number of lesions [17]. As the main finding, we can point out that the studies reporting a higher number of 18 F-FDG-positive lesions included patients often affected by ductal carcinoma (i.e., 96%) [19] among which were also included those with ER+ HER2 + BC (11%) [19] and those with liver metastases (>10%) [12]. In addition, these patients were studied at a time of suspicious relapse when heterogeneity, due to the comparison of metastatic ERclones, is more frequent [12,20]. On the other hand, studies showing a higher number of 18 F-FES-positive lesions often analysed patients affected by lobular BC [21] or patients with a high prevalence of bone metastases [22]. Moreover, patients were predominantly affected by ER+ HER-BC [18,22] and were often studied at the time of their first staging [18].
When the impact on clinical decision making was considered, only one study [18] reported that 18 F-FES PET/CT was able to change therapeutic strategies in 5 out of the 19 (26.5%) patients analysed at the time of first staging. Indeed, this diagnostic procedure was able to properly downstage two patients and upstage three when compared with 18 F-FDG PET/CT.

Quality Assessment of the Studies
The risk of bias was assessed according to seven items, which are listed in Table 4. The overall bias score ranged from none to 2.5; therefore, no study had to be excluded because of high bias risk. The most frequent sources of possible bias were the "study test" and "reference standard categories" since, in some studies, it was unclear whether a blinded evaluation of the two methods had been performed. In particular, in the study by Ulaner et al. [21], the same reader assessed both examinations. In opposition to this, risks regarding the feasibility were almost never detected.

Quantitative Analysis (Meta-Analysis)
The pooled sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT in terms of PBA and LBA was computed (Table 5 and Figures 2-5) based on the available data (see Table 3). Regarding PBA, the pooled sensitivity of 18 F-FDG PET/CT and 18 F-FES PET/CT was 97% and 94%, respectively, with overlapping 95% confidence intervals. In the LBA, however, we observed a wider difference between the pooled sensitivity of the two methods (95% for 18 F-FES vs. 85% for 18 F-FDG). Although the 95% confidence intervals of the two methods were overlapping, the values of 18 F-FES PET/CT were consistently at the higher end of the spectrum (93-97%), while those of 18 F-FDG PET/CT showed a much larger spread (68-100%). In all the analyses, high heterogeneity was found (Table 5).
On the basis of the above results, the heterogeneity of PBA and LBA was explored by using the following covariates: timing of the studies (i.e., PET/CT assessment on staging and on time of relapse), their sample sizes, the prevalence of ductal or lobular BC, and the prevalence of bone and liver metastases. longer present in the PBA when we considered these two groups separately. In fact, in the staging scenario, the sensitivity in the PBA of both the 18 F-FES PET/CT and 18 F-FDG PET/CT was 97% (Figures 2 and 3). Conversely, at the time of restaging, the patient-based sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT was 90% and 95% respectively ( Figures  2 and 3).  When LBA was conducted, the sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT at the time of first staging was 88% and 89% respectively, while at the time of restaging it was 81% and 98%, respectively (Figures 4 and 5). Interestingly, the sensitivity of 18 F-FES PET/CT at the time of restaging was significantly higher than that of the same procedure at the time of staging (Figure 4).  .97 (0.91, 1.02 longer present in the PBA when we considered these two groups separately. In fact, in the staging scenario, the sensitivity in the PBA of both the 18 F-FES PET/CT and 18 F-FDG PET/CT was 97% (Figures 2 and 3). Conversely, at the time of restaging, the patient-based sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT was 90% and 95% respectively ( Figures  2 and 3).  When LBA was conducted, the sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT at the time of first staging was 88% and 89% respectively, while at the time of restaging it was 81% and 98%, respectively (Figures 4 and 5). Interestingly, the sensitivity of 18 F-FES PET/CT at the time of restaging was significantly higher than that of the same procedure at the time of staging (Figure 4).

Discussion
In this systematic review and meta-analysis, we aimed to clarify the diagnostic role of 18 F-FES PET/CT in BC patients compared to 18 F-FDG PET/CT. In fact, 18 F-FES PET/CT has gained significant attraction as a non-invasive means to predict the effectiveness of hormone blockade. However, its potential in the mere diagnostic evaluation is still debated [1,8,21]. In this study, we gathered all the available evidence of its sensitivity in ER+ BC patients when compared with the more commonly used 18 F-FDG PET/CT.
Our qualitative and quantitative assessment in this particular BC subpopulation showed that there are no significant differences in terms of sensitivity between the two PET tracers at the PBA. Indeed, both molecular imaging modalities proved able to detect patients affected by ER+ BC with similarly high sensitivity. The lack of significant difference between these two modalities can, however, be explained by the selection of patients included in the analysis. These patients were evaluated at the time of first staging for an already known primary tumour or who underwent PET at the time of relapse to evaluate the extension or the ER expression of the metastatic disease. This population is indeed characterized by a very high prevalence of a true positive BC lesion; the probability However, these last variables could not be tested, given incomplete and inconsistent data, and only the variable "timing of the studies" was explored. As illustrated in Figures 2 and 3, the heterogeneity of the 18 F-FES PET/CT and 18 F-FDG PET/CT results was no longer present in the PBA when we considered these two groups separately. In fact, in the staging scenario, the sensitivity in the PBA of both the 18 F-FES PET/CT and 18 F-FDG PET/CT was 97% (Figures 2 and 3). Conversely, at the time of restaging, the patient-based sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT was 90% and 95% respectively (Figures 2 and 3).
When LBA was conducted, the sensitivity of 18 F-FES PET/CT and 18 F-FDG PET/CT at the time of first staging was 88% and 89% respectively, while at the time of restaging it was 81% and 98%, respectively (Figures 4 and 5). Interestingly, the sensitivity of 18 F-FES PET/CT at the time of restaging was significantly higher than that of the same procedure at the time of staging (Figure 4).

Discussion
In this systematic review and meta-analysis, we aimed to clarify the diagnostic role of 18 F-FES PET/CT in BC patients compared to 18 F-FDG PET/CT. In fact, 18 F-FES PET/CT has gained significant attraction as a non-invasive means to predict the effectiveness of hormone blockade. However, its potential in the mere diagnostic evaluation is still debated [1,8,21]. In this study, we gathered all the available evidence of its sensitivity in ER+ BC patients when compared with the more commonly used 18 F-FDG PET/CT.
Our qualitative and quantitative assessment in this particular BC subpopulation showed that there are no significant differences in terms of sensitivity between the two PET tracers at the PBA. Indeed, both molecular imaging modalities proved able to detect patients affected by ER+ BC with similarly high sensitivity. The lack of significant difference between these two modalities can, however, be explained by the selection of patients included in the analysis. These patients were evaluated at the time of first staging for an already known primary tumour or who underwent PET at the time of relapse to evaluate the extension or the ER expression of the metastatic disease. This population is indeed characterized by a very high prevalence of a true positive BC lesion; the probability that at least one these lesions was detected by one of the two molecular imaging procedures was very high. The small diagnostic advantage of the 18 F-FDG PET/CT over 18 F-FES PET/CT (97% vs. 94%) reported in our analysis seems related to the variability in terms of the clinical characteristics of the patients. Indeed, the high prevalence of metastatic heterogeneity, often present at the time of restaging, can be associated to false negative 18 F-FES PET/CT results. When the diagnostic performances of these two imaging procedures at the time of recurrence was explored by means of a PBA, the sensitivity of 18 F-FDG PET/CT and that of 18 F-FES PET/CT was 95% and 90%, respectively.
Conversely, when we investigated the diagnostic sensitivity of these diagnostic tools by means of an LBA, we found that the sensitivity of 18 F-FDG PET/CT and that of 18 F-FES PET/CT was 85% and 95%, respectively. Indeed, when the analysis was focused on the time of disease relapse, 18 F-FES PET/CT was more sensitive than 18 F-FDG PET/CT (98% vs. 81%) with a trend towards significance (95% CI: 97-100% and 56-100%, respectively).
Overall, the use 18 F-FDG PET/CT as a first-line examination appears to be the best strategy to stage and restage the ER+ BC subjects, being able to identify the highest number of true positive patients. However, if this approach is applied to a population of ER+ BC with low metastatic heterogeneity (i.e., lobular breast cancer, or ductal breast cancer with a high percentage of ER positive and HER2 negative cells) it could underestimate the actual disease burden since such a clone could show low glucose avidity. Indeed, although 18 F-FES PET/CT has low sensitivity in detecting liver metastases (given the intense background tracer uptake in this organ), it has a very high sensitivity in disclosing peripheral lesions in other organs or tissues, such as bone lesions, which represent the most frequent sites of disease in ER+ BC [22]. In addition, as reported by Gupta et al. [12], one of the diagnostic advantages of this receptor PET tracer is its higher specificity in characterizing cervical, axillary and mediastinal lymph nodes, which may often present unspecific uptake at 18 F-FDG PET/CT and thus be misinterpreted on 18 F-FDG PET/CT. Although our systematic review and meta-analysis reported interesting and useful results to understand the weaknesses and strengths of the two imaging procedures, some limitations should also be borne in mind. First, only seven studies, investigating relatively small patient populations, were fit for inclusions in this meta-analysis. In addition, all these seven studies showed a retrospective design, and three out of seven included a low number of ER+ BC patients. However, the selection criteria limited us to studies that tested both 18 F-FDG PET/CT and 18 F-FES PET/CT within a restricted time frame. Second, a computation of specificity was not possible since none of the studies reported the true negative rate. Third, the absence of histological confirmation of suspected distant metastases detected by both modalities is an important bias that could have affected some of the studies included in this analysis [12,[20][21][22], and we cannot exclude the possibility that some of the metastases detected by PET tracers may have been false-positive findings. However, given the elevated prevalence of patients with disseminated disease, the likelihood of any given finding to be a false negative was relatively low. Moreover, ethical and practical reasons prevented the execution of a biopsy evaluation of each single lesion. In addition, in at least 3 out of 7 studies, a multidisciplinary follow-up (consisting of clinical and imaging evaluation) was available. Fourth, only two out of the seven studies reported information regarding the menopausal status of the patients. Indeed, these data could be of value to correctly estimate the real sensitivity of 18 F-FES PET/CT that could theoretically be affected by competitive binding of high oestradiol concentration. However, the exact endogenous oestrogen concentration that can have a measurable effect on tumour 18 F-FES uptake remains hitherto unexplored [9]. Statistically significant heterogeneity of the 18 F-FES PET/CT pooled heterogeneity was found across studies. Regrettably, the sparse data available in the studies did not allow exploration of the heterogeneity with the exception of the variable "timing of the studies".
Finally, at this time, data about the cost/effectiveness of 18 F-FES PET/CT in comparison with that of 18 F-FDG PET/CT do not exist. However, it is conceivable that, when the overall expense is considered, the additional information provided by FES might help in optimizing the cost by guiding the choice of the most appropriate therapeutic protocol.

Conclusions
The present data show that both 18 F-FDG PET/CT and 18 F-FES PET/CT represent accurate imaging procedures in patients with ER+ BC, providing comparable results in terms of sensitivity. However, in the field of lesion detection, we observed a non-negligible difference in favour of 18 F-FES PET/CT. Thus, the use of 18 F-FES PET/CT as a first-line molecular imaging procedure might be considered in lobular breast cancer or ductal breast cancer with a high percentage of ER and HER2 negative. However, larger, multicentre and prospective studies are required to confirm these preliminary indications.
Author Contributions: Conceiving the study and writing the manuscript draft, A.P., G.B., P.T.; performing the literature search and selecting the studies, F.F., G.B.; performing the meta-analysis, P.T.; reviewing the manuscript draft, F.F., G.T. All authors have read and agreed to the published version of the manuscript.