- Email: [email protected]
CT for diagnosis of primary prostate cancer compared to histopathology
Accepted Manuscript RM2-PET/CT of prostate cancer compared to histopathology
Thomas F. Fassbender, Florian Schiller, Michael Mix, Helmut R. Maecke, Selina Kiefer, Vanessa Drendel, Philipp T. Meyer, Cordula A. Jilg PII: DOI: Reference:
S0969-8051(18)30356-1 https://doi.org/10.1016/j.nucmedbio.2019.01.009 NMB 8056
To appear in:
Nuclear Medicine and Biology
Received date: Revised date: Accepted date:
9 December 2018 17 January 2019 21 January 2019
Please cite this article as: T.F. Fassbender, F. Schiller, M. Mix, et al., RM2-PET/CT of prostate cancer compared to histopathology, Nuclear Medicine and Biology, https://doi.org/10.1016/j.nucmedbio.2019.01.009
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Accuracy of [68Ga]Ga-RM2-PET/CT for Diagnosis of Primary Prostate Cancer Compared to Histopathology Short title: RM2-PET/CT of prostate cancer compared to histopathology
Authors:
PT
Thomas F. Fassbender1, Florian Schiller1, Michael Mix1, Helmut R. Maecke1,2, Selina Kiefer3, Vanessa Drendel3, Philipp T. Meyer1,2, Cordula A. Jilg4 1
German Cancer Consortium (DKTK), Partner Site Freiburg, Germany
SC
2
RI
Department of Nuclear Medicine, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
3
Department of Pathology, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany 4
MA
NU
Department of Urology, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
Thomas F. Fassbender, MSc, MD; email: [email protected] Florian Schiller, MSc, PhD; email: [email protected]
D
Michael Mix, MSc, PhD; email: [email protected]
PT E
Helmut R. Maecke, MSc, PhD; email: [email protected] Selina Kiefer; email: [email protected]
CE
Vanessa Drendel, MD; email: [email protected] Philipp T. Meyer, MD, PhD; email: [email protected]
AC
Cordula A. Jilg, MD; email: [email protected]
Corresponding Author: Thomas F. Fassbender Department of Nuclear Medicine, Medical Center – University of Freiburg, Germany Hugstetter Str. 55 D-79106 Freiburg Telephone: +4976127039970 Telefax: +4976127039300 [email protected]
Keywords: prostate cancer, GRPR, bombesin, [68Ga]-RM2, PET/CT, histopathology
1
ACCEPTED MANUSCRIPT Abstract: Introduction: Prostate cancer (PCa) often shows an overexpression of the gastrin-releasing peptide receptor (GRPr). Therefore, GRPr is a possible theragnostic target. An interesting antagonist
PT
GRPr-ligand is RM2 or BAY86-7548. This study examines the accuracy of positron emission
RI
tomography (PET) with [68Ga]Ga-RM2 for diagnostic imaging of primary PCa (pPCa)
SC
compared to histopathology in patients undergoing radical prostatectomy (RP).
Methods:
NU
[68Ga]Ga-RM2-PET examinations were performed in 15 patients before RP. All prostate
MA
specimens were histopathologically examined based on predefined spatial octants. Each prostate volume on PET was subdivided into octants, which were correlated to histopathology and evaluated according to presence of tumor by two experienced examiners.
D
Additionally, PET data was evaluated by volume of interest (VOI) analyses in terms of
PT E
maximum standardized uptake value (SUVmax) and normalized SUVmax relative to background activity (rSUVmax). Receiver operating characteristic (ROC) curves for SUVmax
AC
Results:
CE
and rSUVmax were calculated.
At least one focus of increased [68Ga]Ga-RM2 uptake corresponding to a tumor manifestation on histology was found in 14 of 15 patients (93%). Spatial concordance of visual PET readings with histopathology was very variable. Intraindividual agreement reached from ≤ 2 octants in three, 3-5 octants in six to ≥ 6 octants in six patients, resulting in a relatively low correlation of visual PET readings with histopathology (accuracy=0.63; p=0.0018). Lesion-based analysis found a sensitivity of 69% and a positive predictive value
2
ACCEPTED MANUSCRIPT of 73%. Concordantly, the octant-based ROC curves for SUVmax and rSUVmax indicated a relatively low diagnostic performance (area under the curve of 0.59 and 0.61, respectively).
Conclusions: [68Ga]Ga-RM2-PET shows only a relatively low diagnostic accuracy for pPCa compared to
PT
histopathology on an octant basis, which may be explained to some extent by methodological weaknesses. Further studies need to explore, whether the observed high
RI
interindividual variability of agreement between [68Ga]Ga-RM2-PET and histopathology can
AC
CE
PT E
D
MA
NU
SC
be explained by different tumor biologies or other coincident prostatic pathologies.
3
ACCEPTED MANUSCRIPT Background: For optimal treatment of primary prostate cancer (pPCa) by radical prostatectomy (RP) or radiotherapy, imaging is an essential method for confirmation of suspected diagnosis, guiding prostate biopsy and staging the disease. Different modalities are commonly
PT
employed, e.g. ultrasound, magnetic resonance imaging (MRI) and positron emission tomography (PET). Recently, several PET tracers with different target structures on prostate
RI
cancer (PCa) have been evaluated for diagnosing and staging of PCa: e.g. 11C- and 18F-
SC
labelled choline as markers of membrane metabolism [1–3] or several tracers targeting the prostate-specific membrane antigen (PSMA) [4, 5]. Another interesting target on PCa is the
NU
gastrin-releasing peptide receptor (GRPr), also known as bombesin receptor subtype 2. GRPr is a G-protein coupled receptor being physiologically expressed in several human
MA
tissues, that is involved in various physiologic functions such as thermoregulation, control of circadian rhythm or gastric functions [6]. It is also involved in cellular growth signal pathways
D
of normal and cancerous human tissue [6]. Overexpression of GRPr has been shown to be
PT E
present in a variety of human cancer types [7, 8], most notably including prostate cancers. GRPr overexpression gradually increases in prostatic carcinogenesis reaching from lowgrade prostatic intraepithelial neoplasia (PIN) over high-grade PIN to PCa, whereas GRPr
CE
shows only little expression in normal prostate tissue and in benign prostate hyperplasia (BPH) [9, 10]. Markwalder et al. show in their study that “all of the 30 cases with primary
AC
invasive prostatic carcinomas were found to express GRP receptors. In the majority (83%) of the cases, a high to very high density of receptors (>1000 dpm/mg tissue) was found” [9]. Several GRPr-specific radiotracers have been developed [7, 11–14], some being receptor agonists and, more recently, others antagonists. Antagonists have the advantage of avoiding side effects caused by triggering cellular signal pathways. Furthermore, the density of binding sites for antagonists is higher than for agonists leading to better tumor-to-normal tissue ratios [15]. There have also been preliminary studies to use the GRPr as a therapeutic target [16, 17]. 4
ACCEPTED MANUSCRIPT A particularly interesting GRPr antagonist used for PET is [68Ga]Ga-DOTA-4-amino-1carboxymethyl-piperidine-D-Phe- Gln-Trp-Ala-Val-Gly-His-Sta-Leu-NH2, also named BAY867548, or, and further mentioned here as, [68Ga]Ga-RM2 [18]. It has been used in several preliminary studies examining its diagnostic potential for prostate cancers: Kähkönen et al. have shown its diagnostic potential for the diagnosis of pPCa in 11 men prior to undergoing
PT
RP with a sensitivity, specificity, and accuracy of 88 %, 81 % and 83 %, respectively [13]. For the detection of lymph node metastases they found a sensitivity of 70 %. Recently, our group
RI
showed that PET/CT undergone with [68Ga]Ga-RM2 ([68Ga]Ga-RM2-PET) was able to
SC
localize PCa recurrence in 10 of 16 men (62.5 %) with biochemical PCa relapse but negative (n = 14) or inconclusive (n =2) preceding 18F-labeled choline PET/CT [19]. In another study,
NU
Minamimoto et al. compared [68Ga]Ga-RM2-PET with [68Ga]Ga-PSMA-11-PET in 7 men with suspected biochemical recurrence of PCa [20]. They mention small, but interesting
MA
differences outside the expected physiologic biodistributions, e.g., lymph node metastases that were either negative on [68Ga]Ga-RM2-PET, but positive on [68Ga]Ga-PSMA-11-PET or
D
positive on both examinations more easily visualized by [68Ga]Ga-RM2-PET due to lower
PT E
intestinal uptake. These results emphasize the interest in [68Ga]Ga-RM2-PET as a diagnostic tool for PCa.
CE
Against this background, the aim of the present study is to investigate the diagnostic accuracy of [68Ga]Ga-RM2-PET compared to histopathology in pPCa in a larger sample
AC
group.
Materials and Methods: In this single-institution analysis 15 patients with biopsy-proven pPCa underwent [68Ga]GaRM2-PET for staging purposes on a compassionate-use basis at the Department of Nuclear Medicine prior to RP with pelvic lymphadenectomy at the Department of Urology. Patients who met these criteria were chosen retrospectively for this study. All patients gave written informed consent. This study was approved by the local ethics committee (number 562/15).
5
ACCEPTED MANUSCRIPT PET/CT Imaging The RM2 precursor was provided by Piramal Imaging (Berlin, Germany) and the synthesis was done under GMP-conditions as previously described [8]. All patients received an intravenous injection of 176 ± 42 MBq (4,8 ± 1,1 mCi) of [68Ga]Ga-RM2. Whole-body PET with additional computed tomography (CT) scans were performed 65 ± 10 minutes after
PT
injection from the proximal femur to the base of the skull with 2 min per bed position for PET
RI
imaging (6 patients on a GEMINI TF 64-slice PET/CT and 9 patients on a GEMINI TF Big Bore PET/CT, Philips Healthcare, Cleveland, USA; both scanners had the same PET-
SC
detector system). 11 patients received a contrast-enhanced diagnostic CT; 3 patients
NU
received a diagnostic CT with only oral contrast agent application; 1 patient underwent only a low-dose non-enhanced CT.
MA
PET data were fully corrected for attenuation, scatter, decay and randoms and expressed as standardized uptake value (SUV; i.e., local radioactivity concentration normalized to injected
PT E
Histopathology
D
dose per body weight).
CE
For all prostate specimens, a standard workup protocol based on current guidelines (German guidelines of oncology, German Cancer Society [21]) was performed by an experienced
AC
pathologist as follows:
1. Resection margins of apical urethra, basal area towards bladder neck and deferent duct were prepared. Then the prostate was sectioned in apicobasal direction. Organ slices (3-4 mm thickness) were embedded in paraffin. Orientation was ensured by specific designation of each slide in apicobasal direction on the tissue cassette and in the macroscopic report. Additionally, the capsule was inked in 4 different colors (left/right lobe, ventral/dorsal portion) for histology. At last, sampling of seminal vesicles was also performed.
6
ACCEPTED MANUSCRIPT 2. All tumors were routinely classified according to the present WHO TNM classification [22] and grading was performed according to the current ISUP/WHO modified Gleason system [23] [24]. 3. For each of the 8 different octants (see PET analysis), the number, diameter (in mm) and Gleason pattern of invasive tumor areas were documented. Different tumor areas were
PT
measured separately if spatially separated by at least 5 mm (i.e., approximate spatial resolution of PET/CT). Tumors extending over more than one octant were labeled to be
SC
RI
identified as such.
PET Analysis
NU
A CT-based volume of interest (VOI) analysis was performed using the commercial software
MA
package PMOD (version 3.6, PMOD Technologies, Zürich, Switzerland). This entailed the following steps: An individual organ VOI was defined on CT including the entire prostate volume of each patient. Whenever available, we also used T2-weighted MRI data of the
D
prostate to guide VOI definition. The prostate VOI was intersected in a left and right part. The
PT E
bilateral VOI were then intersected into an anterior and posterior part, which were chosen perpendicular to the orientation of the histology slices as indicated by the pathologist. Finally,
CE
a height intersection was done perpendicular to aforementioned sections in parallel to histologic slice orientation, separating the prostate into an apical and a basal half, each
AC
containing four sectors, hence eight octants. Each octant is defined by three relative coordinates for unique identification (e.g., “apical anterior right”). Of note, since the axis of pathological organ sectioning was frequently tilted slightly to the axial axis of the PET/CT scan (i.e., pathological organ sectioned were tilted by -10° to 45° to axial imaging slices), aforementioned axial segment nomenclatures do not necessarily strictly correspond to axial segments on routine imaging. These VOI sets were then transferred to the PET datasets of each patient. See Figure 1 for example images of this process.
7
ACCEPTED MANUSCRIPT Due to the inherent time delay between CT and PET scanning of several minutes, continuous bladder filling, bowel and possible patient’s movements may lead to a slight misalignment of the prostate on PET and CT images (typical dorsocaudal movement of the prostate gland by a few millimeters). Thus, accurate positioning of the CT-based VOI set on the corresponding PET image was carefully checked and adjusted manually, if felt
PT
necessary. Voxels containing obvious spill-over of bladder activity were manually removed from the VOI.
RI
Under guidance of the octant VOI set, all PET images were visually analyzed for the
SC
presence of pathologically increased focal tracer uptake by two experienced nuclear medicine physicians in consensus. PET positivity was defined as focal tracer uptake beyond
NU
local background using a standardized inverted grey-scale image display (display range: 0 to
MA
5 SUV). In analogy to the pathologic evaluation, all PET/CT analyses and readings were performed in a blinded manner regarding the histopathologic results. PET positive foci were
D
labeled by letters A-D to be identified.
PT E
Furthermore, the mean standardized uptake value (SUVmean) of a non-prostate reference region was assessed in each patient. For this purpose, we employed the mean value of two VOI spheres (25 mm in diameter) positioned over the left M. gluteus maximus and M. gluteus
CE
medius of each patient.
AC
In a preliminary analysis, lymph nodes (LN) with increased tracer uptake were identified and compared to pathological analysis after surgery.
Statistics
Data of visual readings were evaluated octant-based using contingency patterns with corresponding accuracies for each individual patient and for the entire sampling group. A χ2-Test was performed on the overall distribution to investigate the significance of the association between visual PET reads and histopathology.
8
ACCEPTED MANUSCRIPT Data of visual readings were also evaluated lesion-based, meaning each single tumor lesion on histopathology (possibly extending across multiple octants) was counted as true positive, if at least one affected octant showed pathological [68Ga]Ga-RM2 uptake on PET (false negative otherwise). Lesions showing pathological [68Ga]Ga-RM2 uptake (possibly extending across multiple octants) were taken as false positive if the affected octants showed no
PT
histopathologic tumor lesions. In analogy to Kähkönen et al. [13] the lesion-based analysis was repeated after excluding all octants with PCa lesions ≤ 0.5 cm. Sensitivity and positive
RI
predictive value (PPV) were calculated.
SC
Maximum SUV (SUVmax) were read out octant-based. Additionally, for whole prostate volumes SUVmax and SUVmean were assessed. Normalized SUVmax (rSUVmax) were
NU
calculated by dividing SUVmax by SUVmean of each individual’s reference region.
MA
Using SUVmax of all octants, we calculated a receiver operating characteristic (ROC) curve with the binary histopathologic result (0 = no tumor, 1 = tumor) as reference. In a second
D
step we repeated the ROC analysis after excluding 9 octants with PCa lesions ≤ 0.5 cm (see
PT E
above). Both steps were also repeated for rSUVmax. ROC analyses were statistically compared according to DeLong et al. [25].
CE
Results:
AC
Patient Characteristics
Of all tested patients (n = 15) 40 % were categorized as high-risk and 60 % as intermediaterisk (D'Amico et al., 1998 [26]). Patients’ characteristics are summarized in Table 1.
Results of visual readings At least one focus of increased [68Ga]Ga-RM2 uptake corresponding to a tumor manifestation on histology was found in 14 of 15 patients (93%).
9
ACCEPTED MANUSCRIPT Intraindividually, the accuracy of visual reads compared to histopathology reached from ≤ 2 octants in three, 3 to 5 octants in six to ≥ 6 octants in six patients. This variability is illustrated in Figure 2 where example images of 5 cases are presented. Shown are one axial mid-range PET slice of basal octants (red) and one axial mid-range slice of apical octants (green). Note that due to varying cutting angles of histopathologic cuts compared to PET axial plane, parts
PT
of other octants may appear in some PET slices. Each octant was rated binary whether it contained pathological focal [68Ga]Ga-RM2 uptake on PET and PCa on histopathology or
RI
not.
SC
There are cases where PET readings correspond quite well to histopathology with high contrast between PCa and normal tissue, e.g. patients 1 and 11, showing an intermediate
NU
accuracy of 0.75, or patient 2 with a large tumor covering all octants on PET and
MA
histopathology with an accuracy of 1.0. But there are also cases showing a different behavior, resulting in a large overall variability of accuracies calculated for each individual case (e.g. patient 3, also with a large tumor mass, being false negative on PET in 7 of 8
D
octants, or patient 9, showing false positive [68Ga]Ga-RM2 uptake in multiple octants
PT E
resulting in an accuracy of 0.13). The detailed results of the octant-based analyses for each patient are summarized in Table 2. Of note, coefficients of variation of SUVmax (cv = 0.41)
CE
and SUVmean (cv = 0.38) of the whole prostate volumes were considerably larger than that of the SUVmean of the muscular reference region (cv= 0.16).
AC
The overall accuracy across all patients was 62 % and 63 % for analyses with and without inclusion of octants with tumors ≤ 0.5 cm in diameter, respectively. A low, but significant correlation of the visual PET reads with histopathology was found (χ²-test: p=0.0018). Table 3 compares visual PET reads to predominant ISUP scores of respective octants. There was no significant correlation (including all lesions: χ2-test: p = 0.59; only lesions > 5 mm: p = 0.47).
10
ACCEPTED MANUSCRIPT Comparing whole prostate SUVmax values to postoperative T-category or postoperative ISUP score did not reveal any significant relationship. [68Ga]Ga-RM2 uptake did not show a significant correlation with d’Amico risk categorization, either.
Lesion-based analysis
PT
Lesion-based analysis across all patients revealed 23 true positive tumor lesions, 8 false positive lesions and 16 false negative lesions resulting in a sensitivity of 59 % and a PPV of
RI
74%. Leaving out octants with a single lesion with maximum tumor diameter ≤ 0.5 cm results
SC
in 22 true positive tumor lesions, 8 false positive and 10 false negative lesions leading to a sensitivity of 69 % and a PPV of 73 %. This is summarized in Table 4 comparing lesion-
NU
based SUVmax values to lesion-based predominant ISUP score.
MA
ROC analysis of octant based SUV values
ROC analyses employing the SUVmax value of each octant and the binary histopathologic
D
result as reference provided an area under the curve (AUC) of 0.57, p = 0.277 (see Figure
PT E
3a), with no significant difference from chance (AUC = 0.5). Leaving out octants with maximum tumor diameter ≤ 0.5 cm results in an AUC = 0.59 (Figure 3b, p = 0.155).
CE
ROC analyses for rSUVmax provided a similar AUC = 0.58 (see Figure 3c, p = 0.164), leaving out octants with maximum tumor diameter ≤ 0.5 cm results in an AUC = 0.61 (Figure
AC
3d, p = 0.069).
Lymph node findings PET/CT revealed only 1 LN region with increased [68Ga]Ga-RM2 uptake. This finding was confirmed by histopathology, showing that the entire LN region (left obturatoria) contained 1 LN metastasis in total (size of tumor deposit within lymph node: width 0,7 cm). Seven additional LN regions (in 4 patients) were found to contain 1 or 2 LN metastases, which were
11
ACCEPTED MANUSCRIPT all rated as negative by PET readings (size of tumor deposit within lymph node: width of one: 0,4 cm, all others < 0,3 cm).
Discussion: The aim of the present study was to investigate the diagnostic accuracy of [68Ga]-RM2-
PT
PET/CT compared to histopathology in pPCa in 15 patients.
RI
On the patient level, 14 of the 15 patients (93 %) with biopsy-proven PCa showed at least one focus of pathological [68Ga]Ga-RM2 uptake. However, using the octant-based visual
SC
evaluation the overall accuracy across all patients was only 63 %, whereas intraindividual
NU
accuracies proofed to be very variable, ranging from 13 % to 100 %. Octant-based ROC curves across all patients for SUVmax and rSUVmax yielded an AUC of only 0.59 and 0.61
MA
(not significantly different from chance), respectively, when octants with a maximum tumor diameter of less than 0.5 cm are not taken into account.
D
In a comparable study to ours, Kähkönen et al. analyzed 11 patients who underwent
PT E
[68Ga]Ga-RM2-PET prior to RP [13]. They also compared visual [68Ga]Ga-RM2-PET readings to histopathologic results. Their model consisted of 12 regions instead of 8, like ours, by adding a second height intersection, dividing the prostate volumes in apical, middle and
CE
basal parts. This resulted in an overall accuracy across all patients of 83 %. This is a
AC
remarkable difference compared to our results. Furthermore, they found a sensitivity of 79 % and a PPV of 71 % in a lesion-based analysis, whilst we found a sensitivity of 69 % and a comparable PPV of 73 % in our study. Whereas the patient populations of both studies were comparable in terms of age, size of dominant lesion, PSA value at time of PET and postoperative Gleason score (see supplemental Figure 1a), the patient population of their study showed significantly higher T-categories (see supplemental Figure 1b; Mann-Whitney U test: p = 0.038). Thus, the higher sensitivity in the study of Kähkönen et al. might be explained by the more advanced tumor stage in that study. The same explanation may also account, at least partially, for the differences in octant-based analysis. 12
ACCEPTED MANUSCRIPT The difference between octant-based overall accuracy (63 %) and patient-based accuracy (93 %) in our study might also be a result of oversampling, meaning that the presented method does not sufficiently separate the octants on PET images according to octant separation on histopathology. Several limitations of this method of comparison need to be noted: CT-based volume contouring might be suboptimal. For other methods relying on
PT
prostate anatomy, such as fusion biopsy, it is recommended to use anatomic data from MRI [27]. Furthermore, the manually adjusted position of prostate volume to account for patient
RI
movement or bladder filling between CT and PET acquisition and the manually indicated
SC
slice angle of histopathologic slice position may lead to inaccuracies. Although done in all conscience these adjustments may lead to wrong positioning of image foci compared to the
NU
histopathologic result.
MA
Probably more important, large differences in intraindividual accuracy in our patient group (some being nearly completely true positive, false negative or false positive) indicate substantial differences of GRPr expression in normal prostate and/or cancer tissue across
D
patients. Further immunohistopathologic studies need to explore, whether the strong
PT E
interindividual variability of [68Ga]Ga-RM2 performance can be explained by differences in tumor biology, premalignant lesions and coinciding benign pathologies and their GRPr
CE
expression [9, 10]. This will also be essential for selecting the optimal imaging strategy (e.g., GRPr or other targets like PSMA) in patients with primary or recurrent PCa.
AC
Conclusions:
In the present study, the comparison of [68Ga]Ga-RM2-PET to histopathology shows a good accuracy for pPCa patient-based. However, a relatively low diagnostic accuracy was found octant-based, which may be explained to some extent by methodological weaknesses. Additionally, further studies need to explore, whether the observed high interindividual variability on agreement between [68Ga]Ga-RM2-PET and histopathology can be explained by different tumor biologies or coincident prostatic pathologies. 13
ACCEPTED MANUSCRIPT References: [1]
Hara T, Kosaka N, Kishi H. PET imaging of prostate cancer using carbon-11-choline. J Nucl Med. 1998;39:990-995.
[2]
Reske SN, Blumstein NM, Neumaier B, et al. Imaging Prostate Cancer with 11 C-
Degrado TR, Coleman RE, Wang S, et al. Synthesis and evaluation of 18F-labeled
RI
[3]
PT
Choline PET/CT. J Nucl Med. 2006;47:1249-1254.
SC
choline as an oncologic tracer for positron emission tomography: Initial findings in prostate cancer. Cancer Res. 2001;61:110-117.
Eder M, Neels O, Müller M, et al. Novel preclinical and radiopharmaceutical aspects of
NU
[4]
[68Ga]Ga-PSMA-HBED-CC: A new PET tracer for imaging of prostate cancer.
Zamboglou C, Schiller F, Fechter T, et al. Ga-68 HBED-CC-PSMA PET/CT versus
D
[5]
MA
Pharmaceuticals. 2014;7:779-796.
PT E
histopathology in primary localized prostate cancer: A voxel-wise comparison. Theranostics. 2016;6:1619-1628. Nakagawa T, Hocart SJ, Schumann M, et al. Identification of key amino acids in the
CE
[6]
gastrin-releasing peptide receptor (GRPR) responsible for high affinity binding of
[7]
AC
gastrin-releasing peptide (GRP). Biochem Pharmacol. 2005;69:579-593. Reubi JC, Wenger S, Schmuckli-Maurer J, Schaer J-C, Gugger M. Bombesin Receptor Subtypes in Human Cancers: Detection with the Universal Radioligand 125I[D-Tyr6,b-ALA11,PHE13,NLE14]Bombesin(6-14). Clin Cancer Res. 2002;8:11391146. [8]
Stoykow C, Erbes T, Maecke HR, et al. Gastrin-releasing Peptide Receptor Imaging in Breast Cancer Using the Receptor Antagonist (68)Ga-RM2 And PET. Theranostics. 2016;6:1641-1650. 14
ACCEPTED MANUSCRIPT [9]
Markwalder R, Reubi JC. Gastrin-releasing peptide receptors in the human prostate: relation to neoplastic transformation. Cancer Res. 1999;59:1152-1159.
[10]
Körner M, Waser B, Rehmann R, Reubi JC. Early over-expression of GRP receptors in prostatic carcinogenesis. Prostate. 2014;74:217-224. Van de Wiele C, Dumont F, Vanden Broecke R, et al. Technetium-99m RP527, a GRP
PT
[11]
analogue for visualisation of GRP receptor-expressing malignancies: A feasibility
Sah BR, Burger IA, Schibli R, et al. Dosimetry and First Clinical Evaluation of the New
SC
[12]
RI
study. Eur J Nucl Med. 2000;27:1694-1699.
18F-Radiolabeled Bombesin Analogue BAY 864367 in Patients with Prostate Cancer.
Kähkönen E, Jambor I, Kemppainen J, et al. In vivo imaging of prostate cancer using
MA
[13]
NU
J Nucl Med. 2015;56:372-378.
[68Ga]-labeled bombesin analog BAY86-7548. Clin Cancer Res. 2013;19:5434-5443. Nock BA, Kaloudi A, Lymperis E, et al. Theranostic Perspectives in Prostate Cancer
D
[14]
PT E
with the Gastrin-Releasing Peptide Receptor Antagonist NeoBOMB1: Preclinical and First Clinical Results. J Nucl Med. 2017;58:75-80. Mansi R, Wang X, Forrer F, et al. Evaluation of a 1,4,7,10-tetraazacyclododecane-
CE
[15]
1,4,7,10-tetraacetic acid-conjugated bombesin-based radioantagonist for the labeling
AC
with single-photon emission computed tomography, positron emission tomography, and therapeutic radionuclides. Clin Cancer Res. 2009;15:5240-5249. [16]
Kim K, Zhang H, Rosa S La, et al. Bombesin antagonist-based radiotherapy of prostate cancer combined with WST-11 vascular targeted photodynamic therapy. Clin Cancer Res. 2017;23:3343-3351.
[17]
Dumont R a, Tamma M, Braun F, et al. Targeted radiotherapy of prostate cancer with a gastrin-releasing peptide receptor antagonist is effective as monotherapy and in 15
ACCEPTED MANUSCRIPT combination with rapamycin. J Nucl Med. 2013;54:762-769. [18]
Mansi R, Wang X, Forrer F, et al. Development of a potent DOTA-conjugated bombesin antagonist for targeting GRPr-positive tumours. Eur J Nucl Med Mol Imaging. 2011;38:97-107. Wieser G, Popp I, Christian Rischke H, et al. Diagnosis of recurrent prostate cancer
PT
[19]
with PET/CT imaging using the gastrin-releasing peptide receptor antagonist Ga-68
RI
RM2: Preliminary results in patients with negative or inconclusive
[20]
SC
[18F]Fluoroethylcholine-PET/CT. Eur J Nucl Med Mol Imaging. 2017;44:1463-1472. Minamimoto R, Hancock S, Schneider B, et al. Pilot Comparison of Ga-68 RM2 PET
NU
and Ga-68 PSMA-11 PET in Patients with Biochemically Recurrent Prostate Cancer. J
[21]
MA
Nucl Med. 2016;57:557-562.
Wirth M, Weißbach L, Ackermann R, et al. Interdisziplinäre Leitlinie der Qualität S3 zur
D
Früherkennung, Diagnose und Therapie der verschiedenen Stadien des
[22]
PT E
Prostatakarzinoms. Therapie. 2009;1-620. Sobin L, Gospodarowicz M WC. TNM Classification of Malignant Tumours, 7th Edition.
[23]
CE
Wiley.
Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A Contemporary Prostate Cancer Grading
[24]
AC
System: A Validated Alternative to the Gleason Score. Eur Urol. 2016;69:428-435. Epstein JI, Egevad L, Amin MB, Delahunt B, Srigley JR, Humphrey PA. The 2014 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma. Am J Surg Pathol. 2015:1. [25]
DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach. Biometrics. 1988;44:837. 16
ACCEPTED MANUSCRIPT [26]
D’Amico A V, Whittington R, Malkowicz SB, et al. Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA. 1998;280:969-974.
[27]
Leitlinienprogramm Onkologie (Deutsche Krebsgesellschaft, Deutsche Krebshilfe, AWMF): Interdisziplinäre Leitlinie der Qualität S3 zur Früherkennung, Diagnose und
PT
Therapie der verschiedenen Stadien des Prostatakarzinoms, Leitlinienreport, Version
RI
4.0. AWMF Registernummer: 043/022OL. doi:AWMF Registernummer: 043/022OL.
SC
Acknowledgements
NU
Special thanks to all people who helped preparing and conducting the scans and the preparation of prostate specimens, particularly Helmut R. Maecke, Constantinos Zamboglou, Selina Kiefer, Christine Linstedt and Daniela Lickert.
MA
Conflict of Interests
AC
CE
PT E
D
Author Michael Mix received research grants from Philips Medical Systems. Author Philipp T. Meyer received a research grant from Piramal Imaging, served as consultant for IBA Pharma and received honorary for a lecture from Desitin Arzneimittel GmbH. Author Thomas F. Fassbender declares that he has no conflict of interest. Author Florian Schiller declares that he has no conflict of interest. Author Helmut Mäcke declares that he has no conflict of interest. Author Selina Kiefer declares that she has no conflict of interest. Author Vanessa Drendel declares that she has no conflict of interest. Author Cordula A. Jilg declares that she has no conflict of interest.
17
ACCEPTED MANUSCRIPT Figures: Figure 1: Correlating Histopathologic Cuts to PET Slices. Example images showing the process of correlating histopathologic cuts with PET images. a) Representative histopathologic cut of the prostate. b) VOI of prostate volume drawn on CT image. c) The same VOI was transferred to the corresponding PET series. d) PET VOI were then divided into octants. Note that the green contour marks parts of the slice that belong to other octants due to tilting of histopathologic cuts compared to strictly axial orientation of voxel slices in PET (standardized image thresholding, i.e. inverted grey scale, 0 - 5 SUV).
SC
RI
PT
Figure 2: Examples of Octant Intersections of PET Slices. Examples of 5 patients with different outcomes regarding the diagnostic accuracy of [68Ga]-RM2 for PCa, showing one axial mid-range basal (red) and one apical (green) PET slice for each patient. Note: Due to varying cutting angles of histopathologic cuts compared to PET axial planes, parts of other truncated octants appear in some slices (standardized image thresholding, i.e. inverted grey scale, 0 - 5 SUV).
AC
CE
PT E
D
MA
NU
Figure 3: Receiver Operating Characteristic Curves. a) Receiver Operating Characteristic (ROC) curve of SUVmax of all octants compared to the binary histopathologic result. b) The same ROC curve as in a) without including octants with small tumors (max. diameter ≤ 0.5 cm). c) ROC curve of rSUVmax compared to the binary histopathologic result. d) The same ROC curve as in c) without including octants with small tumors (max. diameter ≤ 0.5 cm).
18
ACCEPTED MANUSCRIPT Tables:
D’Amico risk level
Time to surgery [days]
Postoperative TNM stage
Postoperative Gleason score
1
25
high
54
pT2c pN0
3+4
3+4
2
52.4
high
117
pT3b pN1
4+4
60
3+4
2
42
high
47
pT2c pN1
3+4
4
68
4+3
3
8.5
intermediate
41
pT3a pN0
3+4
5
49
3+3
1
5.3
intermediate
14
pT2c pN0
3+3
6
52
3+4
2
39.4
high
50
pT3b pN1
5+4
7
61
3+4
2
10.6
intermediate
118
pT2c pN0
3+4
8
66
3+4
2
5.3
intermediate
8
pT3b pN0
3+4
9
74
3+3
1
13.8
intermediate
35
pT2c pN0
3+4
10
62
3+4
2
8.9
intermediate
46
pT2c pN0
4+3
11
45
4+3
3
7.6
intermediate
5
pT2b pN0
3+4
12
65
4+3
3
9.4
intermediate
1
pT2c pN0
4+3
13
72
3+4
2
25
high
5
pT3a pN0
4+3
14
61
4+3
3
23.5
high
4
pT3a pN1
4+3
15
65
3+4
2
5.0
intermediate
4
pT2c pN0
4+3
3+3
2
62
3
RI
73
SC
1
AC
CE
PT E
D
Age at PET [years]
MA
Biopsy ISUP/ Gleason WHO score
Patient-Nr.
PT
PSA at imaging [ng/mL]
NU
Table 1: Patients’ characteristics
19
ACCEPTED MANUSCRIPT Table 2: Individual results Octant-based Dominant Lesion Diameter range [mm] (total lesions)
1
2
2
8
3
8
4
4
5
Whole Prostate SUVmax [g/ml]
Whole Prostate SUVmean [g/ml]
Average SUVmean of muscles (mSUVmean) [g/ml]
Octants PETpositive [n]
TP [n]
FN [n]
FP [n]
TN [n]
Accuracy
9-14 (n=4)
8.6
2.8
0.26
4
2
0
2
4
0.75
36 (n=1)
10.1
6.9
0.30
8
8
0
0
0
1.00
35-40 (n=29)
6.0
2.3
0.30
1
1
7
0
0
0.13
5-12 (n=17)
16.0
3.0
0.24
0
0
4
0
2
0.33
4
1-14 (n=11)
6.8
3.3
0.26
6
4
0
2
1
0.71
6
8
36-44 (n=1)
4.4
2.6
0.17
6
6
2
0
0
0.75
7
6
10-22 (n=11)
5.0
2.5
0.30
3
2
4
1
1
0.38
8
8
13-35 (n=31)
14.9
4.7
0.29
7
7
1
0
0
0.88
9
1
15 (n=5)
11.0
3.2
0.30
PT
Patient
Octants with tumor on histopath. [n]
8
1
0
7
0
0.13
10
4
3-23 (n=11)
8.1
3.1
0.33
6
4
0
2
1
0.71
11
3
12-19 (n=2)
13.3
5.4
0.30
5
3
0
2
3
0.75
12
6
1-20 (n=17)
8.7
2.3
0.23
4
4
2
0
1
0.71
13
6
3-32 (n=28)
12.7
4.1
0.36
7
6
0
1
0
0.86
14
5
3-20 (n=20)
4.0
2.4
0.33
3
3
2
0
1
0.67
15
2
2-19 (n=3)
6.6
2.5
0.27
4
2
0
2
3
0.71
Σ 75
1-44
9.1 ± 3.7
3.4 ± 1.3
0.28 ± 0.05
Σ 72
Σ 53
Σ 22
Σ 19
Σ 17
0.63
RI
SC
NU
e
a
b
c
d
AC
CE
PT E
D
MA
Octants positive on histopathology compared to PET and intraindividual contingency patterns with accuracy (leaving out octants with tumor diameter ≤ 0.5 cm). a TP = true positive, b FN = false negative, c FP = false positive, d TN = true negative, e the diameter range from mean lower to mean higher end, f the overall accuracy
20
f
ACCEPTED MANUSCRIPT Table 3: Categorization of octants compared to ISUP score ISUP Value
True Positive
False Negative
AC
CE
PT E
D
MA
NU
SC
RI
PT
1 1 (0) 0 (0) 2 8 (7) 6 (4) 3 29 (27) 10 (8) 4 4 (4) 1 (1) 5 14 (13) 9 (9) Values in parenthesis denote lesions with size > 5 mm.
21
ACCEPTED MANUSCRIPT Table 4: SUVmax and predominant ISUP score of PET-positive lesions Patient
Lesion a count
Lesion A b SUVmax
Lesion A c ISUP
Lesion B SUVmax
Lesion B ISUP
Lesion C SUVmax
Lesion C ISUP
Lesion D SUVmax
Lesion D ISUP
AC
CE
PT E
D
MA
NU
SC
RI
PT
1 2 8.62 -- d 4.81 3 2 2 9.79 5 5.76 5 3 1 6.26 5 4 3 6.56 4 23.88 3 3.92 -- d 5 2 6.46 2 4.46 3 6 3 4.1 5 3.99 5 3.92 5 7 1 4.95 3 8 2 14.94 3 9.74 3 9 2 11.03 3 5.56 -- d 10 4 7.07 3 8.06 2 4.66 3 5.16 5 d 11 2 9.85 2 7.47 -12 1 8.67 3 13 2 12.71 3 6.1 2 14 2 3.93 3 3.74 3 15 1 6.63 3 a Number of separate lesions found in whole prostate, bSUVmax in lesion-based VOI analysis, c predominant ISUP score found in the respective octant; dfalse-positive PET foci; note that PETpositive lesions may extend over multiple octants
22
Figure 1
Figure 2
Figure 3
Figure 4
Thomas F. Fassbender, Florian Schiller, Michael Mix, Helmut R. Maecke, Selina Kiefer, Vanessa Drendel, Philipp T. Meyer, Cordula A. Jilg PII: DOI: Reference:
S0969-8051(18)30356-1 https://doi.org/10.1016/j.nucmedbio.2019.01.009 NMB 8056
To appear in:
Nuclear Medicine and Biology
Received date: Revised date: Accepted date:
9 December 2018 17 January 2019 21 January 2019
Please cite this article as: T.F. Fassbender, F. Schiller, M. Mix, et al., RM2-PET/CT of prostate cancer compared to histopathology, Nuclear Medicine and Biology, https://doi.org/10.1016/j.nucmedbio.2019.01.009
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Accuracy of [68Ga]Ga-RM2-PET/CT for Diagnosis of Primary Prostate Cancer Compared to Histopathology Short title: RM2-PET/CT of prostate cancer compared to histopathology
Authors:
PT
Thomas F. Fassbender1, Florian Schiller1, Michael Mix1, Helmut R. Maecke1,2, Selina Kiefer3, Vanessa Drendel3, Philipp T. Meyer1,2, Cordula A. Jilg4 1
German Cancer Consortium (DKTK), Partner Site Freiburg, Germany
SC
2
RI
Department of Nuclear Medicine, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
3
Department of Pathology, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany 4
MA
NU
Department of Urology, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
Thomas F. Fassbender, MSc, MD; email: [email protected] Florian Schiller, MSc, PhD; email: [email protected]
D
Michael Mix, MSc, PhD; email: [email protected]
PT E
Helmut R. Maecke, MSc, PhD; email: [email protected] Selina Kiefer; email: [email protected]
CE
Vanessa Drendel, MD; email: [email protected] Philipp T. Meyer, MD, PhD; email: [email protected]
AC
Cordula A. Jilg, MD; email: [email protected]
Corresponding Author: Thomas F. Fassbender Department of Nuclear Medicine, Medical Center – University of Freiburg, Germany Hugstetter Str. 55 D-79106 Freiburg Telephone: +4976127039970 Telefax: +4976127039300 [email protected]
Keywords: prostate cancer, GRPR, bombesin, [68Ga]-RM2, PET/CT, histopathology
1
ACCEPTED MANUSCRIPT Abstract: Introduction: Prostate cancer (PCa) often shows an overexpression of the gastrin-releasing peptide receptor (GRPr). Therefore, GRPr is a possible theragnostic target. An interesting antagonist
PT
GRPr-ligand is RM2 or BAY86-7548. This study examines the accuracy of positron emission
RI
tomography (PET) with [68Ga]Ga-RM2 for diagnostic imaging of primary PCa (pPCa)
SC
compared to histopathology in patients undergoing radical prostatectomy (RP).
Methods:
NU
[68Ga]Ga-RM2-PET examinations were performed in 15 patients before RP. All prostate
MA
specimens were histopathologically examined based on predefined spatial octants. Each prostate volume on PET was subdivided into octants, which were correlated to histopathology and evaluated according to presence of tumor by two experienced examiners.
D
Additionally, PET data was evaluated by volume of interest (VOI) analyses in terms of
PT E
maximum standardized uptake value (SUVmax) and normalized SUVmax relative to background activity (rSUVmax). Receiver operating characteristic (ROC) curves for SUVmax
AC
Results:
CE
and rSUVmax were calculated.
At least one focus of increased [68Ga]Ga-RM2 uptake corresponding to a tumor manifestation on histology was found in 14 of 15 patients (93%). Spatial concordance of visual PET readings with histopathology was very variable. Intraindividual agreement reached from ≤ 2 octants in three, 3-5 octants in six to ≥ 6 octants in six patients, resulting in a relatively low correlation of visual PET readings with histopathology (accuracy=0.63; p=0.0018). Lesion-based analysis found a sensitivity of 69% and a positive predictive value
2
ACCEPTED MANUSCRIPT of 73%. Concordantly, the octant-based ROC curves for SUVmax and rSUVmax indicated a relatively low diagnostic performance (area under the curve of 0.59 and 0.61, respectively).
Conclusions: [68Ga]Ga-RM2-PET shows only a relatively low diagnostic accuracy for pPCa compared to
PT
histopathology on an octant basis, which may be explained to some extent by methodological weaknesses. Further studies need to explore, whether the observed high
RI
interindividual variability of agreement between [68Ga]Ga-RM2-PET and histopathology can
AC
CE
PT E
D
MA
NU
SC
be explained by different tumor biologies or other coincident prostatic pathologies.
3
ACCEPTED MANUSCRIPT Background: For optimal treatment of primary prostate cancer (pPCa) by radical prostatectomy (RP) or radiotherapy, imaging is an essential method for confirmation of suspected diagnosis, guiding prostate biopsy and staging the disease. Different modalities are commonly
PT
employed, e.g. ultrasound, magnetic resonance imaging (MRI) and positron emission tomography (PET). Recently, several PET tracers with different target structures on prostate
RI
cancer (PCa) have been evaluated for diagnosing and staging of PCa: e.g. 11C- and 18F-
SC
labelled choline as markers of membrane metabolism [1–3] or several tracers targeting the prostate-specific membrane antigen (PSMA) [4, 5]. Another interesting target on PCa is the
NU
gastrin-releasing peptide receptor (GRPr), also known as bombesin receptor subtype 2. GRPr is a G-protein coupled receptor being physiologically expressed in several human
MA
tissues, that is involved in various physiologic functions such as thermoregulation, control of circadian rhythm or gastric functions [6]. It is also involved in cellular growth signal pathways
D
of normal and cancerous human tissue [6]. Overexpression of GRPr has been shown to be
PT E
present in a variety of human cancer types [7, 8], most notably including prostate cancers. GRPr overexpression gradually increases in prostatic carcinogenesis reaching from lowgrade prostatic intraepithelial neoplasia (PIN) over high-grade PIN to PCa, whereas GRPr
CE
shows only little expression in normal prostate tissue and in benign prostate hyperplasia (BPH) [9, 10]. Markwalder et al. show in their study that “all of the 30 cases with primary
AC
invasive prostatic carcinomas were found to express GRP receptors. In the majority (83%) of the cases, a high to very high density of receptors (>1000 dpm/mg tissue) was found” [9]. Several GRPr-specific radiotracers have been developed [7, 11–14], some being receptor agonists and, more recently, others antagonists. Antagonists have the advantage of avoiding side effects caused by triggering cellular signal pathways. Furthermore, the density of binding sites for antagonists is higher than for agonists leading to better tumor-to-normal tissue ratios [15]. There have also been preliminary studies to use the GRPr as a therapeutic target [16, 17]. 4
ACCEPTED MANUSCRIPT A particularly interesting GRPr antagonist used for PET is [68Ga]Ga-DOTA-4-amino-1carboxymethyl-piperidine-D-Phe- Gln-Trp-Ala-Val-Gly-His-Sta-Leu-NH2, also named BAY867548, or, and further mentioned here as, [68Ga]Ga-RM2 [18]. It has been used in several preliminary studies examining its diagnostic potential for prostate cancers: Kähkönen et al. have shown its diagnostic potential for the diagnosis of pPCa in 11 men prior to undergoing
PT
RP with a sensitivity, specificity, and accuracy of 88 %, 81 % and 83 %, respectively [13]. For the detection of lymph node metastases they found a sensitivity of 70 %. Recently, our group
RI
showed that PET/CT undergone with [68Ga]Ga-RM2 ([68Ga]Ga-RM2-PET) was able to
SC
localize PCa recurrence in 10 of 16 men (62.5 %) with biochemical PCa relapse but negative (n = 14) or inconclusive (n =2) preceding 18F-labeled choline PET/CT [19]. In another study,
NU
Minamimoto et al. compared [68Ga]Ga-RM2-PET with [68Ga]Ga-PSMA-11-PET in 7 men with suspected biochemical recurrence of PCa [20]. They mention small, but interesting
MA
differences outside the expected physiologic biodistributions, e.g., lymph node metastases that were either negative on [68Ga]Ga-RM2-PET, but positive on [68Ga]Ga-PSMA-11-PET or
D
positive on both examinations more easily visualized by [68Ga]Ga-RM2-PET due to lower
PT E
intestinal uptake. These results emphasize the interest in [68Ga]Ga-RM2-PET as a diagnostic tool for PCa.
CE
Against this background, the aim of the present study is to investigate the diagnostic accuracy of [68Ga]Ga-RM2-PET compared to histopathology in pPCa in a larger sample
AC
group.
Materials and Methods: In this single-institution analysis 15 patients with biopsy-proven pPCa underwent [68Ga]GaRM2-PET for staging purposes on a compassionate-use basis at the Department of Nuclear Medicine prior to RP with pelvic lymphadenectomy at the Department of Urology. Patients who met these criteria were chosen retrospectively for this study. All patients gave written informed consent. This study was approved by the local ethics committee (number 562/15).
5
ACCEPTED MANUSCRIPT PET/CT Imaging The RM2 precursor was provided by Piramal Imaging (Berlin, Germany) and the synthesis was done under GMP-conditions as previously described [8]. All patients received an intravenous injection of 176 ± 42 MBq (4,8 ± 1,1 mCi) of [68Ga]Ga-RM2. Whole-body PET with additional computed tomography (CT) scans were performed 65 ± 10 minutes after
PT
injection from the proximal femur to the base of the skull with 2 min per bed position for PET
RI
imaging (6 patients on a GEMINI TF 64-slice PET/CT and 9 patients on a GEMINI TF Big Bore PET/CT, Philips Healthcare, Cleveland, USA; both scanners had the same PET-
SC
detector system). 11 patients received a contrast-enhanced diagnostic CT; 3 patients
NU
received a diagnostic CT with only oral contrast agent application; 1 patient underwent only a low-dose non-enhanced CT.
MA
PET data were fully corrected for attenuation, scatter, decay and randoms and expressed as standardized uptake value (SUV; i.e., local radioactivity concentration normalized to injected
PT E
Histopathology
D
dose per body weight).
CE
For all prostate specimens, a standard workup protocol based on current guidelines (German guidelines of oncology, German Cancer Society [21]) was performed by an experienced
AC
pathologist as follows:
1. Resection margins of apical urethra, basal area towards bladder neck and deferent duct were prepared. Then the prostate was sectioned in apicobasal direction. Organ slices (3-4 mm thickness) were embedded in paraffin. Orientation was ensured by specific designation of each slide in apicobasal direction on the tissue cassette and in the macroscopic report. Additionally, the capsule was inked in 4 different colors (left/right lobe, ventral/dorsal portion) for histology. At last, sampling of seminal vesicles was also performed.
6
ACCEPTED MANUSCRIPT 2. All tumors were routinely classified according to the present WHO TNM classification [22] and grading was performed according to the current ISUP/WHO modified Gleason system [23] [24]. 3. For each of the 8 different octants (see PET analysis), the number, diameter (in mm) and Gleason pattern of invasive tumor areas were documented. Different tumor areas were
PT
measured separately if spatially separated by at least 5 mm (i.e., approximate spatial resolution of PET/CT). Tumors extending over more than one octant were labeled to be
SC
RI
identified as such.
PET Analysis
NU
A CT-based volume of interest (VOI) analysis was performed using the commercial software
MA
package PMOD (version 3.6, PMOD Technologies, Zürich, Switzerland). This entailed the following steps: An individual organ VOI was defined on CT including the entire prostate volume of each patient. Whenever available, we also used T2-weighted MRI data of the
D
prostate to guide VOI definition. The prostate VOI was intersected in a left and right part. The
PT E
bilateral VOI were then intersected into an anterior and posterior part, which were chosen perpendicular to the orientation of the histology slices as indicated by the pathologist. Finally,
CE
a height intersection was done perpendicular to aforementioned sections in parallel to histologic slice orientation, separating the prostate into an apical and a basal half, each
AC
containing four sectors, hence eight octants. Each octant is defined by three relative coordinates for unique identification (e.g., “apical anterior right”). Of note, since the axis of pathological organ sectioning was frequently tilted slightly to the axial axis of the PET/CT scan (i.e., pathological organ sectioned were tilted by -10° to 45° to axial imaging slices), aforementioned axial segment nomenclatures do not necessarily strictly correspond to axial segments on routine imaging. These VOI sets were then transferred to the PET datasets of each patient. See Figure 1 for example images of this process.
7
ACCEPTED MANUSCRIPT Due to the inherent time delay between CT and PET scanning of several minutes, continuous bladder filling, bowel and possible patient’s movements may lead to a slight misalignment of the prostate on PET and CT images (typical dorsocaudal movement of the prostate gland by a few millimeters). Thus, accurate positioning of the CT-based VOI set on the corresponding PET image was carefully checked and adjusted manually, if felt
PT
necessary. Voxels containing obvious spill-over of bladder activity were manually removed from the VOI.
RI
Under guidance of the octant VOI set, all PET images were visually analyzed for the
SC
presence of pathologically increased focal tracer uptake by two experienced nuclear medicine physicians in consensus. PET positivity was defined as focal tracer uptake beyond
NU
local background using a standardized inverted grey-scale image display (display range: 0 to
MA
5 SUV). In analogy to the pathologic evaluation, all PET/CT analyses and readings were performed in a blinded manner regarding the histopathologic results. PET positive foci were
D
labeled by letters A-D to be identified.
PT E
Furthermore, the mean standardized uptake value (SUVmean) of a non-prostate reference region was assessed in each patient. For this purpose, we employed the mean value of two VOI spheres (25 mm in diameter) positioned over the left M. gluteus maximus and M. gluteus
CE
medius of each patient.
AC
In a preliminary analysis, lymph nodes (LN) with increased tracer uptake were identified and compared to pathological analysis after surgery.
Statistics
Data of visual readings were evaluated octant-based using contingency patterns with corresponding accuracies for each individual patient and for the entire sampling group. A χ2-Test was performed on the overall distribution to investigate the significance of the association between visual PET reads and histopathology.
8
ACCEPTED MANUSCRIPT Data of visual readings were also evaluated lesion-based, meaning each single tumor lesion on histopathology (possibly extending across multiple octants) was counted as true positive, if at least one affected octant showed pathological [68Ga]Ga-RM2 uptake on PET (false negative otherwise). Lesions showing pathological [68Ga]Ga-RM2 uptake (possibly extending across multiple octants) were taken as false positive if the affected octants showed no
PT
histopathologic tumor lesions. In analogy to Kähkönen et al. [13] the lesion-based analysis was repeated after excluding all octants with PCa lesions ≤ 0.5 cm. Sensitivity and positive
RI
predictive value (PPV) were calculated.
SC
Maximum SUV (SUVmax) were read out octant-based. Additionally, for whole prostate volumes SUVmax and SUVmean were assessed. Normalized SUVmax (rSUVmax) were
NU
calculated by dividing SUVmax by SUVmean of each individual’s reference region.
MA
Using SUVmax of all octants, we calculated a receiver operating characteristic (ROC) curve with the binary histopathologic result (0 = no tumor, 1 = tumor) as reference. In a second
D
step we repeated the ROC analysis after excluding 9 octants with PCa lesions ≤ 0.5 cm (see
PT E
above). Both steps were also repeated for rSUVmax. ROC analyses were statistically compared according to DeLong et al. [25].
CE
Results:
AC
Patient Characteristics
Of all tested patients (n = 15) 40 % were categorized as high-risk and 60 % as intermediaterisk (D'Amico et al., 1998 [26]). Patients’ characteristics are summarized in Table 1.
Results of visual readings At least one focus of increased [68Ga]Ga-RM2 uptake corresponding to a tumor manifestation on histology was found in 14 of 15 patients (93%).
9
ACCEPTED MANUSCRIPT Intraindividually, the accuracy of visual reads compared to histopathology reached from ≤ 2 octants in three, 3 to 5 octants in six to ≥ 6 octants in six patients. This variability is illustrated in Figure 2 where example images of 5 cases are presented. Shown are one axial mid-range PET slice of basal octants (red) and one axial mid-range slice of apical octants (green). Note that due to varying cutting angles of histopathologic cuts compared to PET axial plane, parts
PT
of other octants may appear in some PET slices. Each octant was rated binary whether it contained pathological focal [68Ga]Ga-RM2 uptake on PET and PCa on histopathology or
RI
not.
SC
There are cases where PET readings correspond quite well to histopathology with high contrast between PCa and normal tissue, e.g. patients 1 and 11, showing an intermediate
NU
accuracy of 0.75, or patient 2 with a large tumor covering all octants on PET and
MA
histopathology with an accuracy of 1.0. But there are also cases showing a different behavior, resulting in a large overall variability of accuracies calculated for each individual case (e.g. patient 3, also with a large tumor mass, being false negative on PET in 7 of 8
D
octants, or patient 9, showing false positive [68Ga]Ga-RM2 uptake in multiple octants
PT E
resulting in an accuracy of 0.13). The detailed results of the octant-based analyses for each patient are summarized in Table 2. Of note, coefficients of variation of SUVmax (cv = 0.41)
CE
and SUVmean (cv = 0.38) of the whole prostate volumes were considerably larger than that of the SUVmean of the muscular reference region (cv= 0.16).
AC
The overall accuracy across all patients was 62 % and 63 % for analyses with and without inclusion of octants with tumors ≤ 0.5 cm in diameter, respectively. A low, but significant correlation of the visual PET reads with histopathology was found (χ²-test: p=0.0018). Table 3 compares visual PET reads to predominant ISUP scores of respective octants. There was no significant correlation (including all lesions: χ2-test: p = 0.59; only lesions > 5 mm: p = 0.47).
10
ACCEPTED MANUSCRIPT Comparing whole prostate SUVmax values to postoperative T-category or postoperative ISUP score did not reveal any significant relationship. [68Ga]Ga-RM2 uptake did not show a significant correlation with d’Amico risk categorization, either.
Lesion-based analysis
PT
Lesion-based analysis across all patients revealed 23 true positive tumor lesions, 8 false positive lesions and 16 false negative lesions resulting in a sensitivity of 59 % and a PPV of
RI
74%. Leaving out octants with a single lesion with maximum tumor diameter ≤ 0.5 cm results
SC
in 22 true positive tumor lesions, 8 false positive and 10 false negative lesions leading to a sensitivity of 69 % and a PPV of 73 %. This is summarized in Table 4 comparing lesion-
NU
based SUVmax values to lesion-based predominant ISUP score.
MA
ROC analysis of octant based SUV values
ROC analyses employing the SUVmax value of each octant and the binary histopathologic
D
result as reference provided an area under the curve (AUC) of 0.57, p = 0.277 (see Figure
PT E
3a), with no significant difference from chance (AUC = 0.5). Leaving out octants with maximum tumor diameter ≤ 0.5 cm results in an AUC = 0.59 (Figure 3b, p = 0.155).
CE
ROC analyses for rSUVmax provided a similar AUC = 0.58 (see Figure 3c, p = 0.164), leaving out octants with maximum tumor diameter ≤ 0.5 cm results in an AUC = 0.61 (Figure
AC
3d, p = 0.069).
Lymph node findings PET/CT revealed only 1 LN region with increased [68Ga]Ga-RM2 uptake. This finding was confirmed by histopathology, showing that the entire LN region (left obturatoria) contained 1 LN metastasis in total (size of tumor deposit within lymph node: width 0,7 cm). Seven additional LN regions (in 4 patients) were found to contain 1 or 2 LN metastases, which were
11
ACCEPTED MANUSCRIPT all rated as negative by PET readings (size of tumor deposit within lymph node: width of one: 0,4 cm, all others < 0,3 cm).
Discussion: The aim of the present study was to investigate the diagnostic accuracy of [68Ga]-RM2-
PT
PET/CT compared to histopathology in pPCa in 15 patients.
RI
On the patient level, 14 of the 15 patients (93 %) with biopsy-proven PCa showed at least one focus of pathological [68Ga]Ga-RM2 uptake. However, using the octant-based visual
SC
evaluation the overall accuracy across all patients was only 63 %, whereas intraindividual
NU
accuracies proofed to be very variable, ranging from 13 % to 100 %. Octant-based ROC curves across all patients for SUVmax and rSUVmax yielded an AUC of only 0.59 and 0.61
MA
(not significantly different from chance), respectively, when octants with a maximum tumor diameter of less than 0.5 cm are not taken into account.
D
In a comparable study to ours, Kähkönen et al. analyzed 11 patients who underwent
PT E
[68Ga]Ga-RM2-PET prior to RP [13]. They also compared visual [68Ga]Ga-RM2-PET readings to histopathologic results. Their model consisted of 12 regions instead of 8, like ours, by adding a second height intersection, dividing the prostate volumes in apical, middle and
CE
basal parts. This resulted in an overall accuracy across all patients of 83 %. This is a
AC
remarkable difference compared to our results. Furthermore, they found a sensitivity of 79 % and a PPV of 71 % in a lesion-based analysis, whilst we found a sensitivity of 69 % and a comparable PPV of 73 % in our study. Whereas the patient populations of both studies were comparable in terms of age, size of dominant lesion, PSA value at time of PET and postoperative Gleason score (see supplemental Figure 1a), the patient population of their study showed significantly higher T-categories (see supplemental Figure 1b; Mann-Whitney U test: p = 0.038). Thus, the higher sensitivity in the study of Kähkönen et al. might be explained by the more advanced tumor stage in that study. The same explanation may also account, at least partially, for the differences in octant-based analysis. 12
ACCEPTED MANUSCRIPT The difference between octant-based overall accuracy (63 %) and patient-based accuracy (93 %) in our study might also be a result of oversampling, meaning that the presented method does not sufficiently separate the octants on PET images according to octant separation on histopathology. Several limitations of this method of comparison need to be noted: CT-based volume contouring might be suboptimal. For other methods relying on
PT
prostate anatomy, such as fusion biopsy, it is recommended to use anatomic data from MRI [27]. Furthermore, the manually adjusted position of prostate volume to account for patient
RI
movement or bladder filling between CT and PET acquisition and the manually indicated
SC
slice angle of histopathologic slice position may lead to inaccuracies. Although done in all conscience these adjustments may lead to wrong positioning of image foci compared to the
NU
histopathologic result.
MA
Probably more important, large differences in intraindividual accuracy in our patient group (some being nearly completely true positive, false negative or false positive) indicate substantial differences of GRPr expression in normal prostate and/or cancer tissue across
D
patients. Further immunohistopathologic studies need to explore, whether the strong
PT E
interindividual variability of [68Ga]Ga-RM2 performance can be explained by differences in tumor biology, premalignant lesions and coinciding benign pathologies and their GRPr
CE
expression [9, 10]. This will also be essential for selecting the optimal imaging strategy (e.g., GRPr or other targets like PSMA) in patients with primary or recurrent PCa.
AC
Conclusions:
In the present study, the comparison of [68Ga]Ga-RM2-PET to histopathology shows a good accuracy for pPCa patient-based. However, a relatively low diagnostic accuracy was found octant-based, which may be explained to some extent by methodological weaknesses. Additionally, further studies need to explore, whether the observed high interindividual variability on agreement between [68Ga]Ga-RM2-PET and histopathology can be explained by different tumor biologies or coincident prostatic pathologies. 13
ACCEPTED MANUSCRIPT References: [1]
Hara T, Kosaka N, Kishi H. PET imaging of prostate cancer using carbon-11-choline. J Nucl Med. 1998;39:990-995.
[2]
Reske SN, Blumstein NM, Neumaier B, et al. Imaging Prostate Cancer with 11 C-
Degrado TR, Coleman RE, Wang S, et al. Synthesis and evaluation of 18F-labeled
RI
[3]
PT
Choline PET/CT. J Nucl Med. 2006;47:1249-1254.
SC
choline as an oncologic tracer for positron emission tomography: Initial findings in prostate cancer. Cancer Res. 2001;61:110-117.
Eder M, Neels O, Müller M, et al. Novel preclinical and radiopharmaceutical aspects of
NU
[4]
[68Ga]Ga-PSMA-HBED-CC: A new PET tracer for imaging of prostate cancer.
Zamboglou C, Schiller F, Fechter T, et al. Ga-68 HBED-CC-PSMA PET/CT versus
D
[5]
MA
Pharmaceuticals. 2014;7:779-796.
PT E
histopathology in primary localized prostate cancer: A voxel-wise comparison. Theranostics. 2016;6:1619-1628. Nakagawa T, Hocart SJ, Schumann M, et al. Identification of key amino acids in the
CE
[6]
gastrin-releasing peptide receptor (GRPR) responsible for high affinity binding of
[7]
AC
gastrin-releasing peptide (GRP). Biochem Pharmacol. 2005;69:579-593. Reubi JC, Wenger S, Schmuckli-Maurer J, Schaer J-C, Gugger M. Bombesin Receptor Subtypes in Human Cancers: Detection with the Universal Radioligand 125I[D-Tyr6,b-ALA11,PHE13,NLE14]Bombesin(6-14). Clin Cancer Res. 2002;8:11391146. [8]
Stoykow C, Erbes T, Maecke HR, et al. Gastrin-releasing Peptide Receptor Imaging in Breast Cancer Using the Receptor Antagonist (68)Ga-RM2 And PET. Theranostics. 2016;6:1641-1650. 14
ACCEPTED MANUSCRIPT [9]
Markwalder R, Reubi JC. Gastrin-releasing peptide receptors in the human prostate: relation to neoplastic transformation. Cancer Res. 1999;59:1152-1159.
[10]
Körner M, Waser B, Rehmann R, Reubi JC. Early over-expression of GRP receptors in prostatic carcinogenesis. Prostate. 2014;74:217-224. Van de Wiele C, Dumont F, Vanden Broecke R, et al. Technetium-99m RP527, a GRP
PT
[11]
analogue for visualisation of GRP receptor-expressing malignancies: A feasibility
Sah BR, Burger IA, Schibli R, et al. Dosimetry and First Clinical Evaluation of the New
SC
[12]
RI
study. Eur J Nucl Med. 2000;27:1694-1699.
18F-Radiolabeled Bombesin Analogue BAY 864367 in Patients with Prostate Cancer.
Kähkönen E, Jambor I, Kemppainen J, et al. In vivo imaging of prostate cancer using
MA
[13]
NU
J Nucl Med. 2015;56:372-378.
[68Ga]-labeled bombesin analog BAY86-7548. Clin Cancer Res. 2013;19:5434-5443. Nock BA, Kaloudi A, Lymperis E, et al. Theranostic Perspectives in Prostate Cancer
D
[14]
PT E
with the Gastrin-Releasing Peptide Receptor Antagonist NeoBOMB1: Preclinical and First Clinical Results. J Nucl Med. 2017;58:75-80. Mansi R, Wang X, Forrer F, et al. Evaluation of a 1,4,7,10-tetraazacyclododecane-
CE
[15]
1,4,7,10-tetraacetic acid-conjugated bombesin-based radioantagonist for the labeling
AC
with single-photon emission computed tomography, positron emission tomography, and therapeutic radionuclides. Clin Cancer Res. 2009;15:5240-5249. [16]
Kim K, Zhang H, Rosa S La, et al. Bombesin antagonist-based radiotherapy of prostate cancer combined with WST-11 vascular targeted photodynamic therapy. Clin Cancer Res. 2017;23:3343-3351.
[17]
Dumont R a, Tamma M, Braun F, et al. Targeted radiotherapy of prostate cancer with a gastrin-releasing peptide receptor antagonist is effective as monotherapy and in 15
ACCEPTED MANUSCRIPT combination with rapamycin. J Nucl Med. 2013;54:762-769. [18]
Mansi R, Wang X, Forrer F, et al. Development of a potent DOTA-conjugated bombesin antagonist for targeting GRPr-positive tumours. Eur J Nucl Med Mol Imaging. 2011;38:97-107. Wieser G, Popp I, Christian Rischke H, et al. Diagnosis of recurrent prostate cancer
PT
[19]
with PET/CT imaging using the gastrin-releasing peptide receptor antagonist Ga-68
RI
RM2: Preliminary results in patients with negative or inconclusive
[20]
SC
[18F]Fluoroethylcholine-PET/CT. Eur J Nucl Med Mol Imaging. 2017;44:1463-1472. Minamimoto R, Hancock S, Schneider B, et al. Pilot Comparison of Ga-68 RM2 PET
NU
and Ga-68 PSMA-11 PET in Patients with Biochemically Recurrent Prostate Cancer. J
[21]
MA
Nucl Med. 2016;57:557-562.
Wirth M, Weißbach L, Ackermann R, et al. Interdisziplinäre Leitlinie der Qualität S3 zur
D
Früherkennung, Diagnose und Therapie der verschiedenen Stadien des
[22]
PT E
Prostatakarzinoms. Therapie. 2009;1-620. Sobin L, Gospodarowicz M WC. TNM Classification of Malignant Tumours, 7th Edition.
[23]
CE
Wiley.
Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A Contemporary Prostate Cancer Grading
[24]
AC
System: A Validated Alternative to the Gleason Score. Eur Urol. 2016;69:428-435. Epstein JI, Egevad L, Amin MB, Delahunt B, Srigley JR, Humphrey PA. The 2014 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma. Am J Surg Pathol. 2015:1. [25]
DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach. Biometrics. 1988;44:837. 16
ACCEPTED MANUSCRIPT [26]
D’Amico A V, Whittington R, Malkowicz SB, et al. Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA. 1998;280:969-974.
[27]
Leitlinienprogramm Onkologie (Deutsche Krebsgesellschaft, Deutsche Krebshilfe, AWMF): Interdisziplinäre Leitlinie der Qualität S3 zur Früherkennung, Diagnose und
PT
Therapie der verschiedenen Stadien des Prostatakarzinoms, Leitlinienreport, Version
RI
4.0. AWMF Registernummer: 043/022OL. doi:AWMF Registernummer: 043/022OL.
SC
Acknowledgements
NU
Special thanks to all people who helped preparing and conducting the scans and the preparation of prostate specimens, particularly Helmut R. Maecke, Constantinos Zamboglou, Selina Kiefer, Christine Linstedt and Daniela Lickert.
MA
Conflict of Interests
AC
CE
PT E
D
Author Michael Mix received research grants from Philips Medical Systems. Author Philipp T. Meyer received a research grant from Piramal Imaging, served as consultant for IBA Pharma and received honorary for a lecture from Desitin Arzneimittel GmbH. Author Thomas F. Fassbender declares that he has no conflict of interest. Author Florian Schiller declares that he has no conflict of interest. Author Helmut Mäcke declares that he has no conflict of interest. Author Selina Kiefer declares that she has no conflict of interest. Author Vanessa Drendel declares that she has no conflict of interest. Author Cordula A. Jilg declares that she has no conflict of interest.
17
ACCEPTED MANUSCRIPT Figures: Figure 1: Correlating Histopathologic Cuts to PET Slices. Example images showing the process of correlating histopathologic cuts with PET images. a) Representative histopathologic cut of the prostate. b) VOI of prostate volume drawn on CT image. c) The same VOI was transferred to the corresponding PET series. d) PET VOI were then divided into octants. Note that the green contour marks parts of the slice that belong to other octants due to tilting of histopathologic cuts compared to strictly axial orientation of voxel slices in PET (standardized image thresholding, i.e. inverted grey scale, 0 - 5 SUV).
SC
RI
PT
Figure 2: Examples of Octant Intersections of PET Slices. Examples of 5 patients with different outcomes regarding the diagnostic accuracy of [68Ga]-RM2 for PCa, showing one axial mid-range basal (red) and one apical (green) PET slice for each patient. Note: Due to varying cutting angles of histopathologic cuts compared to PET axial planes, parts of other truncated octants appear in some slices (standardized image thresholding, i.e. inverted grey scale, 0 - 5 SUV).
AC
CE
PT E
D
MA
NU
Figure 3: Receiver Operating Characteristic Curves. a) Receiver Operating Characteristic (ROC) curve of SUVmax of all octants compared to the binary histopathologic result. b) The same ROC curve as in a) without including octants with small tumors (max. diameter ≤ 0.5 cm). c) ROC curve of rSUVmax compared to the binary histopathologic result. d) The same ROC curve as in c) without including octants with small tumors (max. diameter ≤ 0.5 cm).
18
ACCEPTED MANUSCRIPT Tables:
D’Amico risk level
Time to surgery [days]
Postoperative TNM stage
Postoperative Gleason score
1
25
high
54
pT2c pN0
3+4
3+4
2
52.4
high
117
pT3b pN1
4+4
60
3+4
2
42
high
47
pT2c pN1
3+4
4
68
4+3
3
8.5
intermediate
41
pT3a pN0
3+4
5
49
3+3
1
5.3
intermediate
14
pT2c pN0
3+3
6
52
3+4
2
39.4
high
50
pT3b pN1
5+4
7
61
3+4
2
10.6
intermediate
118
pT2c pN0
3+4
8
66
3+4
2
5.3
intermediate
8
pT3b pN0
3+4
9
74
3+3
1
13.8
intermediate
35
pT2c pN0
3+4
10
62
3+4
2
8.9
intermediate
46
pT2c pN0
4+3
11
45
4+3
3
7.6
intermediate
5
pT2b pN0
3+4
12
65
4+3
3
9.4
intermediate
1
pT2c pN0
4+3
13
72
3+4
2
25
high
5
pT3a pN0
4+3
14
61
4+3
3
23.5
high
4
pT3a pN1
4+3
15
65
3+4
2
5.0
intermediate
4
pT2c pN0
4+3
3+3
2
62
3
RI
73
SC
1
AC
CE
PT E
D
Age at PET [years]
MA
Biopsy ISUP/ Gleason WHO score
Patient-Nr.
PT
PSA at imaging [ng/mL]
NU
Table 1: Patients’ characteristics
19
ACCEPTED MANUSCRIPT Table 2: Individual results Octant-based Dominant Lesion Diameter range [mm] (total lesions)
1
2
2
8
3
8
4
4
5
Whole Prostate SUVmax [g/ml]
Whole Prostate SUVmean [g/ml]
Average SUVmean of muscles (mSUVmean) [g/ml]
Octants PETpositive [n]
TP [n]
FN [n]
FP [n]
TN [n]
Accuracy
9-14 (n=4)
8.6
2.8
0.26
4
2
0
2
4
0.75
36 (n=1)
10.1
6.9
0.30
8
8
0
0
0
1.00
35-40 (n=29)
6.0
2.3
0.30
1
1
7
0
0
0.13
5-12 (n=17)
16.0
3.0
0.24
0
0
4
0
2
0.33
4
1-14 (n=11)
6.8
3.3
0.26
6
4
0
2
1
0.71
6
8
36-44 (n=1)
4.4
2.6
0.17
6
6
2
0
0
0.75
7
6
10-22 (n=11)
5.0
2.5
0.30
3
2
4
1
1
0.38
8
8
13-35 (n=31)
14.9
4.7
0.29
7
7
1
0
0
0.88
9
1
15 (n=5)
11.0
3.2
0.30
PT
Patient
Octants with tumor on histopath. [n]
8
1
0
7
0
0.13
10
4
3-23 (n=11)
8.1
3.1
0.33
6
4
0
2
1
0.71
11
3
12-19 (n=2)
13.3
5.4
0.30
5
3
0
2
3
0.75
12
6
1-20 (n=17)
8.7
2.3
0.23
4
4
2
0
1
0.71
13
6
3-32 (n=28)
12.7
4.1
0.36
7
6
0
1
0
0.86
14
5
3-20 (n=20)
4.0
2.4
0.33
3
3
2
0
1
0.67
15
2
2-19 (n=3)
6.6
2.5
0.27
4
2
0
2
3
0.71
Σ 75
1-44
9.1 ± 3.7
3.4 ± 1.3
0.28 ± 0.05
Σ 72
Σ 53
Σ 22
Σ 19
Σ 17
0.63
RI
SC
NU
e
a
b
c
d
AC
CE
PT E
D
MA
Octants positive on histopathology compared to PET and intraindividual contingency patterns with accuracy (leaving out octants with tumor diameter ≤ 0.5 cm). a TP = true positive, b FN = false negative, c FP = false positive, d TN = true negative, e the diameter range from mean lower to mean higher end, f the overall accuracy
20
f
ACCEPTED MANUSCRIPT Table 3: Categorization of octants compared to ISUP score ISUP Value
True Positive
False Negative
AC
CE
PT E
D
MA
NU
SC
RI
PT
1 1 (0) 0 (0) 2 8 (7) 6 (4) 3 29 (27) 10 (8) 4 4 (4) 1 (1) 5 14 (13) 9 (9) Values in parenthesis denote lesions with size > 5 mm.
21
ACCEPTED MANUSCRIPT Table 4: SUVmax and predominant ISUP score of PET-positive lesions Patient
Lesion a count
Lesion A b SUVmax
Lesion A c ISUP
Lesion B SUVmax
Lesion B ISUP
Lesion C SUVmax
Lesion C ISUP
Lesion D SUVmax
Lesion D ISUP
AC
CE
PT E
D
MA
NU
SC
RI
PT
1 2 8.62 -- d 4.81 3 2 2 9.79 5 5.76 5 3 1 6.26 5 4 3 6.56 4 23.88 3 3.92 -- d 5 2 6.46 2 4.46 3 6 3 4.1 5 3.99 5 3.92 5 7 1 4.95 3 8 2 14.94 3 9.74 3 9 2 11.03 3 5.56 -- d 10 4 7.07 3 8.06 2 4.66 3 5.16 5 d 11 2 9.85 2 7.47 -12 1 8.67 3 13 2 12.71 3 6.1 2 14 2 3.93 3 3.74 3 15 1 6.63 3 a Number of separate lesions found in whole prostate, bSUVmax in lesion-based VOI analysis, c predominant ISUP score found in the respective octant; dfalse-positive PET foci; note that PETpositive lesions may extend over multiple octants
22
Figure 1
Figure 2
Figure 3
Figure 4