Diagnostic performance of 68Gallium-PSMA-11 PET/CT to detect significant prostate cancer and comparison with 18FEC PET/CT

Background Radiolabeled prostate-specific membrane antigen (PSMA) has proven to be a highly accurate method to detect recurrence and metastases of prostate cancer, but only sparse data is available about its performance in the diagnosis of clinically significant primary prostate cancer. Methods We compared 68Ga-PSMA-11 PET/CT in 25 patients with 18FEC PET/CT in 40 patients with suspected prostate carcinoma based on an increased PSA level. The PET/CT results were compared with the histopathologic Gleason Score (GS) of biopsies. Results The 68Ga-PSMA-11 PET/CT revealed highly suspect prostatic lesions (maximum standardized uptake value/SUVmax >2.5) in 21/25 patients (84%), associated with GS≥6 (low-grade/high-grade carcinoma). Two histopathologic non-malignancy-relevant cases (GS<6) had PSMA-SUVmax ≤2.5; all histopathologic high-grade cases (GS≥7b) showed PSMA-SUVmax >12.0 which further increased with rising GS. There were 2 false positives and no false negative findings for high-grade prostate cancer using a cut off-level for SUVmax of 2.5. In contrast, the 18FEC PET/CT showed suspected malignant lesions in 38/40 patients (95%), which included 3 lesions with GS<6. The mean SUVmax values did not differ with different GS. There were 11 false positives and 1 false negative for detection of high-grade prostate cancer (cut off 2.5). By means of ROC analysis a SUVmax of 5.4 was found to be an optimal cut off-level to distinguish between low- and high-grade carcinoma in 68Ga-PSMA-11 PET/CT (AUC=0.9692; 95% CI 0.9086;1.0000;SD(AUC)=0.0309)). Choosing a cut off-level of SUVmax5.4, 68Ga-PSMA-11 PET/CT was able to distinguish between GS ≤7a/≥7b with a sensitivity of 84%, a specificity of 100%, a negative predictive value (NPV) of 67%, and an efficiency of 88% (p<0.001). The ROC analysis revealed a SUVmax 6.5 as an optimal cut off-level to distinguish between low- and high-grade carcinoma in 18FEC PET/CT (AUC=0.7470; 95% CI 0.5919;0.9020;SD(AUC)=0.0791) with a sensitivity of 61% and a specificity of 92%; but the efficiency was only 70% and the NPV 50% (p=0.01). Conclusion 68Ga-PSMA-11 PET/CT guided biopsy of the prostate increases diagnostic precision and is likely to help to reduce overtreatment of low-grade malignant disease as well as detect the foci of the highest Gleason pattern. Both methods (68Ga-PSMA-11,18FEC) were suitable to detect primary prostate cancer, but the excellent image quality, the higher specificity and the good correlation of positive scans with GS are advantages of 68Ga-PSMA-11.


INTRODUCTION
Exact diagnosis and staging of primary prostate cancer is crucial for optimal treatment decisions. The differentiation between an indolent low-grade tumor (GS 6, 7a) which may allow conservative management such as active surveillance and an aggressive high-grade tumor (GS≥7b) [1][2][3] leading to prostatectomy, radio-/ chemotherapy or androgen-deprivation is essential for the patient's quality of life and prognosis. Epstein and Montironi showed that GS 7 is misleading, as GS7b tumors have a much worse prognosis than GS7a tumors and should be considered differently for treatment and prognostic purposes [2].
Morphological imaging techniques, e.g. transrectal ultrasound, computed tomography and magnetic resonance imaging, have limited accuracy in the diagnosis of primary prostate cancer [4,5]. Imaging is the preferred base for sampling biopsies to minimize undetected lesions leading to under treatment [6][7][8][9], and over diagnosis resulting in overtreatment of low risk lesions [10].
Molecular imaging with specific tracers should improve diagnostic accuracy. Some studies report a high sensitivity of PET/CT with radioactive labeled choline derivatives for the detection of primary prostate cancer lesions by focally increased choline uptake [11][12][13][14]. But although choline PET/CT is widely used, other studies report low sensitivity and specificity, particularly at low PSA levels and high GS [15][16][17][18][19]. The major factors which compromise the diagnostic accuracy of choline PET/CT for primary prostate cancer are the tumor configuration, especially the detection of small carcinoma, and the differentiation of prostatitis, benign prostatic hyperplasia or high-grade intraepithelial neoplasia [4,5,13,20,21].
Prostate-specific membrane antigen (PSMA) is a transmembrane protein which is significantly overexpressed in prostate carcinoma cells, and its expression increases with tumor aggressiveness [4,5,13,20,21], metastatic disease and recurrence while normal prostatic tissue expresses PSMA sparsely [22][23][24]. PSMA PET/CT and multiparametric magnetic resonance imaging (mpMRI) correlated well with regard to tumor allocation in patients with a high pretest probability for large tumors [25]. Preliminary results of 37 patients with recurrent prostate cancer and rising PSA levels (background ratio between 18.8 and 28.3) indicated that the labeling of PSMA ligand with 68 Ga could detect relapses and metastases of prostate carcinoma with high contrast compared to normal tissue [23].
Afshar-Oromieh et al. (2014) used an 68 Galabeled HBED-CC conjugate of the PSMA-specific pharmacophore Glu-NH-CO-NH-Lys (" 68 Ga-PSMA-11"), and detected at least one lesion characteristic for prostate cancer in 86.5% of patients, but only in 26 of 37 (70.3%) patients with 18 FEC PET/CT. In patients with PSA values ≤2.82 ng/ml at least one lesion characteristic of prostate cancer was identified in 68.8% of patients with 68 Ga-PSMA-11 PET/CT, but only in 43.8% of patients with 18 FEC PET/CT. All lesions detected by 18 FEC PET/CT were also seen in 68 Ga-PSMA-11 PET/CT [19].
Despite positive results for recurrent prostate cancer the availability of data for the primary diagnosis is poor. Therefore, the aim of this study was the evaluation of the diagnostic performance of 68 Ga-PSMA-11 PET/CT for detection of primary prostate cancer in patients with increased PSA levels and comparison of the results to 18 FEC PET/CT. The intent of this study is to improve the diagnostic precision of prostate cancer detection in patients with increased PSA by use of 68 Ga-PSMA-11 PET/CT with subsequent imaging guided biopsy. Furthermore, we have demonstrated that this approach may be able to predict the histological aggressiveness of the underlying tumor.

RESULTS
68 Ga-PSMA-11 PET/CT was conducted in 25 patients with a mean age of 67.0 ± 8.1 years and a basic PSA of 20.4 ± 33.50 ng/ml. Another 40 patients, who underwent 18 FEC PET/CT, were aged 69.4 ± 7.7 years and had a mean basic PSA of 55.0 ± 56.9 ng/ml ( Table 1).
The 68 Ga-PSMA PET/CT scans showed prostatic lesions with accumulation of the radiotracer above SUV max 2.5 in 21/25 patients (84%), which related to lesions with GS≥6 (low-grade and high-grade carcinoma). The mean prostatic SUV max was 12.1 ± 13.9 (1.5-56.0) in the whole group and 12.1 ± 14.7 (1.5-38.7) in 18 patients without metastases. In the group of 18 FEC PET/CT scans such lesions (SUV max >2.5, GS≥6) were seen in 38/40 patients (95%) and the mean prostatic SUV max accounted for 6.3 ± 6.4 in the whole group and 5.8 ± 2.8 (2.4-14.5) in 25 patients without metastases. In 68 Ga-PSMA PET/CT, but not in 18 FEC PET/CT, there was a tendency towards increasing SUV max with rising PSA as shown in Figure 1   After histopathologic examination of biopsies and application of the GS, 8% of patients in the 68 Ga-PSMA PET/CT group had non-malignancy-relevant (GS<6), 16% low-grade (GS 6, 7a) and 76% high-grade (GS≥7b) lesions. The corresponding distribution in the 18 FEC PET/ CT-group was 10%, 20%, and 70%, respectively ( Table  2). Rising SUV max values were associated with rising GS categories in both 68 Ga-PSMA PET/CT (R 0.642, p = 0.005) and 18 FEC PET/CT (R 0.404, p = 0.009).
As shown in Figure 2 the two histopathologic nonmalignancy-relevant lesions (GS<6) in the 68 Ga-PSMA PET/CT group had SUV max ≤2.5 and the SUV max means of all histopathologic high-grade categories (GS≥7b) continuously increased starting from SUV max values above 12.0, whereas in the 18 FEC PET/CT group the SUV max means remained stable across all GS categories.
Using an SUV max of 2.5 as the cut off-level between not-malignancy-relevant and malignant lesions (GS<6 vs. GS≥6) 68 Ga-PSMA PET/CT revealed no false malignant and 2/23 false non-malignancy-relevant (8.7%) results indicating a sensitivity of 91%, a specificity of 100% and a NPV of 50%. With regard to the distinction between low-grade (GS≤7a) and high-grade carcinoma (GS≥7b) the sensitivity was 100%, the specificity 67% and the NPV 100%. Using ROC analysis SUV max 5.4 was found to be an optimal cut off-level to distinguish between low-and high-grade carcinoma by means of 68 Ga-PSMA PET/CT (AUC = 0.9692; 95% CI 0.9086; 1.0000; SD(AUC) = 0.0309) ( Figure 3). Applying this split-point analysis the specificity increased to 100% while the sensitivity was 84%, which resulted in an efficiency (= sum of all correct diagnoses) of 88% and a NPV of 67% (p<0.001) ( Table 3).
In 7 patients 68 Ga-PSMA PET/CT showed metastases in lymph nodes (n = 4), bones (n = 4) and/or lung (n = 1). Also biopsy results demonstrated perineural invasion in 8 of 25 patients, which was associated with PET-positive nodal metastases in 2 patients, but with no metastases in 6 patients.
The 18 FEC scans showed metastases in 15 patients, mainly in the lymph nodes (n = 13) and bones (n = 8). A lung metastasis was found in one patient. In 10 patients the biopsy results revealed perineural invasion, which was associated with nodal metastases in five cases. Another 5 patients had a perineural invasion without metastases.

DISCUSSION
In this study, we present the results from two groups of patients with increased PSA plasma levels undergoing 68 Ga-PSMA-11 PET/CT (n = 25) or 18 FEC PET/CT (n = 40) and imaging guided biopsy to detect significant cancer of the prostate. 18 FEC PET/CT has been in widespread use for the diagnosis of prostate carcinoma [19], but in many malignant lesions the choline metabolism is not increased while most prostate carcinomas overexpress PSMA [19]. By labeling of PSMA ligands with 68 Ga, relapses and metastases of prostate carcinoma with high contrast compared to normal tissues can be detected [23]. Therefore, 68 Ga-PSMA PET/CT might also be superior to 18 FEC PET/CT in the detection of primary prostate cancer. In two groups of patients with elevated PSA levels, scheduled for biopsy, we analyzed 68 Ga-PSMA-11 PET/ CT-scans of 25 patients with an increased PSA and compared the results to 18  Choosing an SUV max of 2.5 as cut off-level between histologically confirmed non-malignancy-relevant and malignant lesions (GS<6/≥6), 68 Ga-PSMA-11 PET/CT detected 2/2 cases (100%) as correct benign, but 18 FEC PET/CT only 1/4 (25%). False negative 68 Ga-PSMA-11 PET/CT scans (not used in our study) may occur if a prostate tumor is poorly differentiated and displays neuroendocrine aberrations [23,26]. The specificity of 100% for 68 Ga-PSMA-11 PET/CT and of 25% for 18 FEC PET/CT for differentiating between non-malignancy-relevant and malignancy-relevant lesions (GS<6/≥6) observed here, has to be interpreted with caution because of the low rate of benign results in both groups. Our study was conducted in patients with a high probability of malignancy due to an increased PSA which is regarded as a marker for progressive disease because there is a strong association between PSA-level and positive 68 Ga-PSMA-11 PET/CT scans [27] but not with 18 FEC PET/CT [12]. Studies including a higher proportion of patients with benign prostate alterations are needed to verify the results.
In our study we distinguished between nonmalignancy-relevant (GS<6) and malignant (GS≥6) and between low-grade carcinoma (GS≤7a) and high-grade carcinoma (GS≥7b). The latter for the reason, that several studies showed significantly worse prognosis with regard to prostate cancer with GS≥7b [27]. in spite of such correlation having been described in the literature [28][29][30][31]. They explained this by their low number of patients with a GS 5-6 which might have caused substantial variability in the statistical analysis. In our study we observed a possible correlation between GS and the SUV max values measured by 68 Ga-PSMA PET/CT: The two histopathologic nonmalignancy-relevant lesions (GS<6) had PSMA-SUV max ≤2.5 and all histopathologic high-grade lesions (GS≥7b) showed PSMA-SUV max >2.5. Starting from GS 7b the mean SUV max increased with rising GS category, while the SUV max means of 18 FEC PET/CT scans did not vary between different GS categories. Statistical analysis revealed a correlation between SUV max and GS for both methods ( 68 Ga-PSMA PET/CT R 0.642, p = 0.005; 18 FEC PET/CT R 0.404, p = 0.009). However the sensitivity for distinction between GS≤7a/≥7b by means of 68 Ga-PSMA-11 PET/CT vs. 18 FEC PET/CT at a cut off-level of SUV max 2.5 was 100% vs. 96%, the specificity 67% vs. 8%, NPV 100% vs. 50% and PPV 90% vs. 71%. This means that only 68 Ga-PSMA-11 PET/CT was able to correctly predict an aggressive prostate cancer.
Assuming that a greater amount of prostate tissue is altered in higher GS categories this difference might be due to a higher specificity of the radiolabeled 68 Ga-PSMA for cancer cells than 18 FEC. Because of the high affinity of PSMA to prostate cancer cells every lesion with an accumulation of PSMA should be regarded as prostate cancer or prostate cancer metastasis until proven otherwise [27]. This means that a significant focal increase of PSMA metabolism might be predictive for a high GS, i.e. an aggressive cancer.
Our data showed that ROC analysis indicated an optimal cut off-level of SUV max 5.4 for 68 Ga-PSMA-11 PET/CT, which increased the specificity and PPV for separating GS≤7a/≥7b up to 100% but slightly decreased sensitivity to 84% and NPV to 67%. Using the calculated optimal cut off-level of SUV max 6.5 for 18 FEC PET/CT all prognostic parameters were less favorable than for 68 Ga-PSMA-11 PET/CT. Fendler et al. (2016) evaluated the accuracy of 68 Ga-PSMA-11 PET/CT to localize cancer in the prostate and surrounding tissue. They found a statistically higher SUV max in histopathologically positive segments (11.8 ± 7.6) compared to negative segments (4.9 ± 2.9; p<0.001). ROC revealed an optimal SUV max cut off-value of 6.5 for discrimination between positive/negative segments [32], i.e. the results were concordant with our data. Fendler et al. concluded that 68 Ga-PSMA-11 PET/CT accurately detects location and extent of primary prostate cancer and might be a promising tool for noninvasive tumor characterization and biopsy guidance [32]. Koerber et al. observed a significantly higher mean SUV max in tumors with higher d`Amico risk classification and GS from biopsy (p<0.001 for grouped analyses) [33]. Giesel et al. published that based on the eight-segment resolution of biopsy, mpMRI and PSMA PET/CT presented identical tumor allocations [25]. MpMRI combined with ultrasound fusion guided biopsy of the prostate is a widely accepted method for T-Staging. However, no statistical significance could be shown regarding the differentiation between high-and low-grade carcinomas in a recently published study by our group [34].
High PSA values, T2b-T3 stage, poor tumor differentiation and perineural invasion are associated with high risk of nodal metastases [35,36]. Measurement of PSA alone is not helpful in predicting lymph node metastases [37]. With regard to the therapeutic strategy, T-staging determines local surgery and radiotherapy while the pretreatment nodal status defines the extent of pelvic lymph node dissection or radiotherapy [38,39]. PSMA PET/CT may provide valuable information in planning for focal radiation to the dominant lesions [40], because of its ability to detect even small lymph metastases, primarily due to a high radiotracer uptake [19].
In our study 68 Ga-PSMA-11 PET/CT revealed metastases in 7/25 patients (28%) and 18   bone metastases and in 1 patient a lung metastasis were demonstrated by 68 Ga-PSMA-11 PET/CT. Whereas by 18 FEC PET/CT, 13 patients showed positive lymph nodes, 8 patients bone metastases and 1 patient a lung metastasis. A perineural invasion was seen in 8 patients (biopsy results according to 68 Ga-PSMA-11 PET/CT) and in 10 patients (biopsy results according to 18 FEC PET/CT), respectively. But a perineural invasion was associated with nodal metastases in only 2 of 8 ( 68 Ga-PSMA-11 PET/CT) and 5 of 10 patients ( 18 FEC PET/CT).
It is assumed that 68 Ga-PSMA-11 PET/CT reveals the highest contrast in lymph node metastases, followed by bone metastases, local relapses and soft tissue metastases [27]. 68 Ga-PSMA-11 PET/CT allows the detection of bone and organ metastases -due to low background signal [19] -which may lead to systemic therapy, but if excluded may lead to curative therapy [41]. 18 F-PSMA agents are an attractive alternative to 68 Ga-PSMA compounds. 18 F-PSMA can be produced in larger amounts per batch in PET radiopharmaceuticals with an on-site cyclotron. Additionally, the average lower positron range of 18 F reduces blurring effects leading to a higher spatial resolution and the longer half-life of 18 F (110 min.) in comparison to 68 Ga (67 min.) optimizes the production and distribution of 18 F [42].

Patients
In this retrospective study we included 61 consecutive patients from the Central Military Hospital Koblenz and 4 consecutive patients from the Practice of Radiology and Nuclear Medicine Cologne Triangle with suspected prostate carcinoma due to an elevated PSA (≥4.0, depending on age). Patients with a history of specific cancer pretreatment, surgical intervention, or inconspicuous PSA plasma levels were excluded. In 25 patients we conducted a 68 Ga-PSMA-11 PET/CT (7/2015 -08/2016) and in another 40 patients an 18 FEC PET/ CT (2/2010 -7/2015). In all patients, prostate cancer was verified histologically with transrectal ultrasound (TRUS)-guided biopsy, and the GS results of TRUS biopsy served as reference for the PET findings. Detailed information on patient characteristics is shown in Table  1. Biopsy specimens were histopathologically evaluated based on the Gleason System on ISUP criteria 2014 and stratified by categorization into low-grade (GS≤3+4 = 7a) and high-grade malignancies (GS≥4+3 = 7b), respectively [43].
Our study was in accordance with the Helsinki Declaration and with our national legislation (German Medicinal Products Act, AMG § 13 Abs. 2b), and all patients gave their written informed consent. The retrospective study was approved by the ethics committee (Landesärztekammer Rheinland-Pfalz). A contrast-enhanced diagnostic CT scan (140 keV, 100-400 mAs, dose modulation) or a low-dose CT scan (120 keV, 55 mAs) was performed for attenuation correction at the time of the PET scan. Contrast media could not be used in the following cases: renal insufficiency, contrast media allergy, hyperthyreosis, and oral administration of the pharmaceutical Metformin. PET was acquired in 3D (matrix: 168x168/Koblenz;144x144/ Cologne). Each bed position (axial field of view of 21.8 cm/Koblenz;19 cm/Cologne) was acquired for 3 min. Random, scatter and decay correction were applied to the emission data. An ordered-subsets expectation maximization (OSEM) algorithm was used for reconstruction (two iterations, fourteen subsets, Gaussian filtering, 4.2 mm/Koblenz;5.3 mm Cologne transaxial resolution, full-width at half-maximum). CT data were obtained for attenuation correction.

PET/CT imaging protocols
The uptake of 68 Ga-PSMA-11 and 18 FEC, i.e. the tracer concentration of the hypermetabolic cancer region detected in the image, was quantified in terms of SUV max .
SUV max values above 2.5 were related to clear and reproducible visual detection of PET positivity and therefore a cut off of SUV max 2.5 was used to discriminate PET positivity from PET negativity for both tracers [27].

Statistical analysis
The PET/CT results were related to the GS obtained by histopathologic analysis of biopsies. Additionally, we analyzed correlations between PSA, SUV max , GS and detection of metastases.
The data analysis was performed using descriptive statistics (relative and absolute frequencies, arithmetic means, standard deviation).
Spearman rank correlation coefficients were calculated to specify the relationship between two variables. The SUV max values were related to the GS obtained by histopathologic analysis of biopsies. Additionally, we analyzed correlations between PSA, SUV max , GS and detection of metastases. www.impactjournals.com/oncotarget By means of ROC analysis we tested the ability of the method to distinguish between high-and lowgrade cancer (significant/not significant) by plotting the true positive cases (sensitivity) against the false positive cases (1-specificity) for various SUV max cut offlevels. Area under the ROC curve (AUC) together with 95%-confidence interval (CI) and standard deviation (SD) were calculated to characterize the quality of the discrimination between the two groups.
All tests were carried out using the software BiAS; p-values of less than 0.05 were stated as significant.

CONCLUSION
Our results show that 68 Ga-PSMA-11 PET/CT and 18 FEC PET/CT are both suitable for the detection of primary prostate cancer. An advantage for 68 Ga-PSMA-11 PET/CT is its excellent imaging quality, its high specificity and a correlation of positive scans with GS which may allow a differentiation between low-and high-grade carcinoma. Our results support the view that 68 Ga-PSMA-11 PET/CT promotes higher detection rates of significant malignancies requiring intervention as does the diagnostic procedure using 18 FEC PET/CT.
Therefore 68 Ga-PSMA-11 PET/CT is intended to increase diagnostic precision (avoiding false-negative results and understaging) to guide prostate biopsy and might help to reduce overtreatment of low-grade malignant disease as well as detect the foci of the highest Gleason pattern.
68 Ga-PSMA-11 PET/CT is already clinically accepted in detecting metastases in patients with biochemical recurrence [31], but it may play an important role also in initial tumor staging similar to and in conjunction with mpMRI-supported biopsy [33,42].