Annals of Oncology Advance Access published online on May 2, 2008
Annals of Oncology, doi:10.1093/annonc/mdn170
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[18F]-FDG–PET in clinical stage I/II non-seminomatous germ cell tumours: results of the German multicentre trial
1 Internal Medicine, Hematology and Oncology, Vivantes Klinikum Neukölln, Berlin
2 Department of Nuclear Medicine, University Clinic Eppendorf, Eppendorf
3 Department of Urology, Military Hospital Hamburg, Hamburg
4 Department of Nuclear Medicine, University of Dresden, Dresden
5 Department of Nuclear Medicine, Saarland University Medical Center, Homburg
6 Department of Urology, Saarland University, Saarland
7 Department of Nuclear Medicine
8 Department of Urology, University Hospital Münster, Münster
9 Department of Medicine
10 Department of Nuclear Medicine, Ludwig-Maximilian University, München
11 Department of Urology, Clinic Uckermark, Schwedt/Oder
12 Department of Nuclear Medicine, Clinic for Nuclear Medicine, Campus Mitte, Charité, Berlin
13 Department of Radiology, Clinic for Radiology, Campus Virchow, Charité, Berlin
14 Department of Nuclear Medicine, Diakonie Hospital Rotenburg (W), Rotenburg
15 Department of Nuclear Medicine, University Schleswig-Holstein, Kiel
16 Department of Nuclear Medicine, Klinikum Ingolstadt, Ingolstadt
17 Department of Nuclear Medicine, University Hospital Tübingen, Tübingen, Germany
* Correspondence to: Dr M. de Wit, Department of Internal Medicine–Hematology and Oncology, Vivantes Klinikum Neukölln, Rudower Strasse 48, 12351 Berlin, Germany. Tel: +49-30-130 14 2250; Fax: + 49-30-130 14 2494; E-mail: maike.dewit{at}vivantes.de
 |
Abstract |
|---|
 Top Abstract introductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
Purpose: The aim of this study was to determine the predictive values of 2-[fluorine-18]fluoro-2-deoxy-D-glucose–positron emission tomography (FDG–PET) in primary staging in patients with newly diagnosed non-seminomatous germ cell tumour (NSGCT) clinical stage I/II.
Patients and methods: The hypothesis was that FDG–PET would improve the negative predictive value (NPV) from 70% to 90%, thus requiring a total of 169 patients. All scans underwent visual analysis by a reference team of nuclear medicine physicians. Results were validated by histology following retroperitoneal lymph node dissection.
Results: Only 72 of the planned 169 patients were included, due to poor accrual. The prevalence of nodal involvement was 26%. Correct nodal staging by FDG–PET was achieved in 83% compared with correct computed tomography (CT) staging in 71%. CT had a sensitivity and specificity of 41% and 95%, respectively. Positive predictive value (PPV) and NPV were 87% and 67%, respectively. FDG–PET had a sensitivity and specificity of 66% and 98%, respectively. PPV was 95%. The primary end point was not reached, with an NPV of 78%.
Conclusion: FDG–PET as a primary staging tool for NSGCT yielded only slightly better results than CT. Both methods had a high specificity while false-negative findings were more frequent with CT. FDG–PET is mostly useful as a diagnostic tool in case of questionable CT scan.
FDG–PET, germ cell tumour, NSGCT, staging
|
introduction |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
Conventional radiological staging for retroperitoneal lymph nodes in patients with germ cell tumour results in 20%–30% false upstaging and about the same percentage of false downstaging. Using spiral computed tomography (CT) and lowering the threshold for lymph node size to <1 cm as indicative for metastases improves sensitivity to >90% but decreases specificity to 50% [1].
2-[Fluorine-18]fluoro-2-deoxy-D-glucose–positron emission tomography (FDG–PET) as a functional rather than a morphological tool is applied in order to improve staging by exploiting the metabolic difference between normal cells and cancer cells [2]. Compared with other non-invasive diagnostic methods, improved sensitivity and specificity were reported for FDG–PET in the detection of primary or recurrent tumours [3–9].
Very few data are available for FDG–PET in early-stage germ cell tumours. The aim of our study was to test if FDG–PET increases staging accuracy in early-stage testicular germ cell tumours compared with spiral CT.
|
patients and methods |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
In 1998, a German multicentre trial was initiated for FDG–PET staging of germ cell tumour patients. Patients were stratified according to stage. The study recruited patients with non-seminomatous germ cell tumours (NSGCTs) at an early stage (I and II) undergoing primary retroperitoneal lymph node dissection (RPLND).
As required by German law, the protocol was fully approved by the ethical committees of each participating hospital and the national radiation protection authorities and passed a review for quality control by the German Cancer Society. The study was carried out in accordance with the Declaration of Helsinki. Every patient signed a written fully informed consent form before PET scanning.
inclusion criteria
Patients were included if they were diagnosed with NSGCT at the clinical stage I or IIA/B according to tumour–node–metastasis classification. FDG–PET and CT were carried out following orchiectomy. Evaluation of PET scans took place in the local PET centre with knowledge of the clinical data. Following RPLND, lymph nodes were histologically evaluated as gold standard.
exclusion criteria
Patients with seminoma or patients not undergoing primary RPLND as well as germ cell tumour patients receiving primary chemotherapy after orchiectomy followed by surgery were excluded.
statistics
The hypotheses were that the negative predictive value (NPV) would improve from 70% to 90% and the positive predictive value (PPV) from 75% to 90% by using FDG–PET in comparison to CT at early-stage NSGCT. The trial was aimed at detecting a difference in PPV between CT and FDG–PET at a 5% significance level with a power of 80% under the premises that PPV of CT is 75% and PPV of FDG–PET is 90%. Considering that 40% of the patients present with stage II and 60% present with stage I, 169 patients would have to be included with at least 37 patients at stage II.
Primary analysis variable was staging by FDG–PET scan and CT scan with lymph node histology as gold standard.
Due to changes in the treatment approach to stage I patients, the number of patients undergoing primary RPLND was constantly decreasing. As it appeared impossible to reach the required number of 169 patients within a reasonable time frame, the study was closed in 2003.
comparison between FDG–PET and CT
Confidence intervals (CIs) for sensitivity, specificity, PPV and NPV were calculated according to the method of Clopper and Pearson. Sensitivity and specificity as well as PPV, NPV and accuracy of both imaging methods were compared using McNemar's exact (i.e. binomial) test.
computed tomography
Standardised CT scans of the retroperitoneum and abdomen were carried out after orchiectomy in all but six patients following oral contrast media 1 h before CT. The liver was examined without i.v. contrast media followed by an i.v. bolus of 120-ml contrast media for spiral CT or 150 ml for conventional CT. Maximum slice thickness allowed was 10 mm. Contrast media for thorax CT was 100 ml for spiral techniques and 120 ml for conventional CT with slices of maximum 10 mm. Different windows for lung and soft tissue were examined. If only one CT was carried out, 120–150 ml contrast media was injected i.v. immediately before examination and the craniocaudal CT was carried out reaching as far caudal as possible.
Lymph nodes with a maximum diameter of >1 cm in CT were classified as lymph node metastases. To compare the different methods, classification of all lymph nodes in the scanned field was required.
CT and FDG–PET results were reported locally and validated by a panel of three independent nuclear physicians and radiologists, blinded to local and histological results.
positron emission tomography
Before the study, all PET scanners were calibrated by one physicist using a standardised procedure [10]. The accuracy of the dose calibrators was checked and a phantom of a known volume was scanned for 60 min. Nine of 19 scanners displayed an error of <5% in three-dimensional acquisition modes. Four scanners displayed an error of 10% in the three-dimensional mode. By thorough application of straightforward standard procedures, an accuracy of at least 5%–10% could be achieved for almost all dedicated PET scanners tested and maintained by permanent monitoring.
FDG–PET was carried out following orchiectomy (median time 15 days, range 2–68 days) but before RPLND in all patients. The median time between FDG–PET and CT was 9 days (range –5 to 54 days). Twenty-nine patients waited for >2 weeks. For three patients, FDG–PET was carried out before CT. The median time between FDG–PET and surgery was 5 days (range 0–61 days) and 15 patients waited for >2 weeks without concurrent therapeutic interventions. Patients fasted for at least 6 h before FDG–PET scanning in order to minimise glucose utilisation of normal tissue [11]. FDG–PET emission scanning commenced at a median of 60 min (range 45–120 min) following i.v. FDG injection of median 388 MBq FDG (range 185–730 MBq) using dedicated PET scanners. A transmission scan of the retroperitoneum was required for attenuation correction. After iterative reconstruction (n = 33) or filtered back-projection (n = 39), visual analysis was carried out.
|
results |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
Due to changes in the treatment approach to stage I patients, the number of patients undergoing primary RPLND was constantly decreasing and the study was closed early after enrolment of 87 patients.
patients
Four patients included in this study had suffered from malignancies before non-seminoma, three patients from seminoma and one patient from thyroid cancer 6 years before non-seminoma. Of 87 patients with early-stage germ cell tumour enrolled after inguinal orchiectomy, 15 had to be excluded. In one of the latter, RPLND was carried out before FDG–PET scanning, and 14 patients were observed without undergoing RPLND.
Thus, the assessable patient group included 72 patients with a median age of 28 years ranging from 18 to 57 years. The blood glucose levels reached a median of 87 mg/dl (range 57–103 mg/dl) before FDG–PET; however, these were not available in four patients.
histology
The primary testicular tumours were mostly classified as mixed NSGCT (44) according to World Health Organisation classification [12], and additionally, 12 showed embryonal carcinoma, eight embryonal carcinoma with teratoma, three mature teratomas, three teratomas with areas of malignant transformation and two choriocarcinomas.
primary stage
Pathological stage of the primary tumour was pT1 in 45 patients (68%), pT2 in 26 patients and pT3 in one patient.
The prevalence of nodal involvement was 26% (n = 19). Of 57 patients classified by morphological analysis with CT as cN0 [clinical stage (CS) I], 19 were at pathological stage II. In the pathologist's report, the size of the pathological lymph nodes involved with tumour was not described for four patients, six patients showed tumour involvement of <1 cm, in six patients lymph nodes measured between 1 and 2 cm, while three showed lymph nodes of >2 cm.
scan results
On the basis of a per-patient analysis, 60 FDG–PET scans and 51 CT scans predicted lymph node involvement correctly while 21 CT and 12 FDG–PET results were incorrect (Table 1). Combining PET and CT results improved either sensitivity or specificity. Selecting those patients who had negative results in PET and CT decreased the sensitivity to 34% but yielded no false-positive results. Selecting those patients who had positive results in PET and CT optimised the sensitivity (72%), reducing specificity to 93%.
|
Detailed analysis of FDG–PET and CT results of 40 pathological stage I patients showed that 37 patients were correctly classified by CT and FDG–PET, two patients were falsely classified as CS II by CT and one patient was falsely judged as CS II by FDG–PET (Table 2). There was ‘no’ patient for whom both methods jointly produced false-positive results.
|
Of 32 pathological stage II patients, 11 were correctly classified by CT and FDG–PET while 19 (by CT) and 11 (by FDG–PET) were falsely staged as CS I. Nine patients were incorrectly staged by CT and FDG–PET (Table 2). PET changed the stage for 15 patients compared with CT. This was correct for 12 of 15 patients and incorrect for 3 of 15 patients with respect to pathological staging. For 11 patients, FDG–PET resulted in an upstaging. For 10 of these 11 patients, this was correct, while for one patient the change from CS I to CS II was incorrect.
This patient-based evaluation showed a superior sensitivity of 66% for FDG–PET compared with 41% for CT and a similar specificity for both methods of 98% for FDG–PET and 95% for CT. The comparison of CT and FDG–PET is compromised by the lower than planned patient numbers; nevertheless, the results show that FDG–PET had a significantly superior sensitivity (66% versus 41%, P = 0.038) and NPV (78% versus 67%, P = 0.05). Specificity, PPV and accuracy, however, were not significantly different (Table 3).
|
False-negative FDG–PET results were not only a function of size but also observed in metastatic lymph nodes of 2 cm. The smallest tumour detected by FDG–PET was a lymph node measuring only 0.2 cm.
The accuracy of CT and FDG–PET was independent of the histology of the primary tumour. For patients with histologically proven teratomatous foci in the tumour (n = 14), lymph node metastases of one patient were confirmed by FDG–PET. The other patients showed no FDG uptake but did not show pathological lymph node metastases either. Thus, the inclusion of teratoma did not result in lower predictive values than in the other subgroups of NSGCT.
|
discussion |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
The aim of this study was to determine whether the predictive values of FDG–PET in primary staging of retroperitoneal lymph node metastases in patients with newly diagnosed early-stage NSGCT could reach both an NPV and a PPV of
90%. In the evaluation on a per-patient basis, the PPV of FDG–PET was better than that of CT (95% versus 87%). The NPV of FDG–PET was slightly higher than that of CT (78% versus 67%) but it was still not as high as desired due to the low sensitivity of both methods.
The accuracies of PET and CT were not significantly different with overlapping CIs, although there was a trend towards a higher accuracy with FDG–PET. The results were reported locally but a central review was added (not reported). This improved the PPV and lowered the NPV of CT and FDG–PET but did not improve the sensitivity of either method.
The study failed the goal to improve the NPV from 70% to 90% and the PPV from 75% to 90% by the use of FDG–PET in early-stage NSGCT. The drawbacks of this study are that the recruitment of patients could not be completed and the statistical power to determine significant differences between the two methods is limited.
Some patients waited for >2 weeks between PET and CT or between staging and surgery. For the two groups, the percentage of false-negative results did not differ according to a waiting time of either more or less than 2 weeks, neither for CT nor for PET. Thus, the study was not compromised by these delays.
Inflammation is a typical pitfall for FDG–PET causing false-positive results. As only few lymph nodes showed histological signs of inflammation, FDG–PET as well as CT presented a high specificity. Lymph node histology showed inflammation in only four patients, and none of them had false-positive PET results. FDG–PET detected several small lymph nodes associated with cancer, but as expected, it was not able to detect them all. In general, lymph nodes smaller than the resolution of the reconstructed images can be detected only in the case of a very high tracer uptake. Depending on the PET scanner and the imaging mode, the reconstructed resolution usually ranges from 5 to 10 mm. Albers et al. [13] reported undetectability by PET for lymph nodes <0.5 cm. In our study, we observed two patients with detectable lymph node involvement by FDG–PET in nodes with a diameter of 0.2 and 0.5 cm. Lymph node metastases of such a small size, however, are not routinely detectable by FDG–PET and account for the low sensitivity of this method in comparison to histopathology. The same is obviously true for lymph node micrometastases that are not detectable by any current imaging modality.
Very few data are available for the use of FDG–PET in early-stage germ cell tumours. Assuming all patients developing metastases during surveillance will be cured, it seems not necessary to detect lymph node involvement at primary diagnosis. Risk-adapted therapy depending on FDG–PET results, however, may be preferable to current practice.
Albers et al. [13] published a report on 37 patients with stage I and II NSGCT. FDG–PET showed no false-positive results. The same findings were reported by Spermon et al. [14] (n = 12) as well as in the analysis of 46 NSGCT patients at CS I published by the Danish Group [15]. In our patient group, one false-positive FDG result was observed. In all reports, false-positive results are rarely seen in FDG–PET at stage I NSGCT in the hands of experienced nuclear medicine physicians. False-negative results, however, remain a serious problem even for the combined use of PET and CT. According to our results and published reports of others, FDG–PET results are as good as CT and may be particularly helpful for patients with lymph nodes seen in CT (CS II) who do not want to receive primary chemotherapy. Since patients with FDG accumulation and morphological signs of lymph node metastases by CT always suffer from histologically proven lymph node involvement, a watch-and-wait policy seems reasonable only in FDG-negative patients.
Patients who are considered to be at high risk for metastatic lymph node disease receive adjuvant chemotherapy instead of RPLND. Upstaging by means of PET imaging will have a therapeutic impact in the number of treatment cycles. FDG–PET resulted in an upstaging in our patient group. In terms of patient management, this means that these patients will need three cycles of cisplatinum, etoposide and bleomycine chemotherapy instead of two cycles of adjuvant chemotherapy.
Furthermore, positive FDG uptake may alter the adjuvant treatment decision in patients who were clinically judged as being low risk for lymph node involvement.
In contrast to the Danish trial [15] with its NPV of 92%, the NPV of this trial was only 78% and does not allow excluding lymph node involvement. Our findings are concordant with the results of the Medical Research Council (UK) (MRC) trial TE22 [16].
The MRC trial examined the question whether FDG–PET could identify patients without occult metastatic disease and whether patients with negative PET scans could be observed closely without risk for the patients [16]. A far too high 1-year relapse-free survival of 56% was, however, observed in FDG–PET-negative patients at the high-risk CS I NSGCT. Thus, the study recruitment was suspended. The study group reported to be unable to reproduce the sensitivity of >70% which had been reported previously. The same is true for this study with a sensitivity of 66% (95% CI 47% to 81%).
The MRC trial stated that FDG–PET identified a proportion of high-risk patients (20%) with a presumed disease not detected by CT scan. Additionally, in this study we are able to report the PPV of FDG–PET with 95% in an unselected group of NSGCT patients. Both studies were prospective trials and both conclude that we cannot recommend a negative FDG–PET scan to replace current strategies for managing stage I NSGCT patients.
|
conclusion |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
FDG–PET as a primary staging tool in patients with CS I and II non-seminomatous germ cell cancer yielded more accurate results than CT. Both methods had a high specificity, but false-negative findings frequently resulted in a relative low sensitivity for both CT and FDG–PET. Sensitivity and PPV were significantly better for FDG–PET. FDG–PET is mostly useful as a diagnostic tool in case of questionable CT scan.
|
funding |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
Deutsche Krebshilfe (wi702329).
|
Acknowledgements |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
The remarkable contribution of all participating physicians and technical assistants is gratefully acknowledged. Statistical analyses were carried out by Dr J. Hüsing, KKS Heidelberg. Results were partly presented at American Society of Clinical Oncology 2005 as oral presentation. The following were the members of the study group: W. Abenhardt, München; P. Albers, Kassel; H. Amthauer, Berlin; R. Bares, Tübingen; H. Bender, Bonn; H. J. Biersack, Bonn; H. Bihl, Stuttgart; U. Blum, Aachen; K. Bohuslavizki, Hamburg; C. Bokemeyer, Hamburg; A. Bockisch, Essen; W. Brenner, Hamburg; U. Büll, Aachen; D. Bumann, Hamburg; M. Clausen, Hamburg; U. Cremerius, Ingolstadt; H. Dittmann, Tübingen; B. M. Dohmen; Tübingen/Rotenburg (W); R. Felix, Berlin; C. Franzius, Münster; A. Gerl, München; L. Geworski, Berlin; F. K. Gratz, Hannover; U. Haberkorn, Mannheim; M. Hartmann, Hamburg; R. Hautmann, Ulm; S. Heicappell, Schwedt; D. Hellwig, Homburg; J. Hüsing, Heidelberg; H. Huland, Hamburg; V. Ivancevic, Berlin; G. Jakse, Aachen; W. Jenicke, Hamburg; S. Kliesch, Münster; C.-M. Kirsch, Homburg; W. H. Knapp, Hannover; H. H. Knispel, Berlin; J. Kotzerke, Dresden; S. Krege, Essen; J. Kropp, Dresden; J. Lehmann, Homburg; S. Mihm, Mannheim; K. Miller, Berlin; St Müller, Essen; D. L. Munz, Berlin; R. Naumann, Dresden; K. Oechsle, Tübingen; W. Pust, Ulm; S. Reske, Ulm; J. Rübben, Essen; H. Schirrmeister, Heide; M. Schlemmer, München; K. Tatsch, München; St Ventz, Berlin; L. Weißbach, Berlin; M. Wirth, Dresden; and M. Ziegler, Homburg.
Received for publication September 28, 2007. Revision received February 17, 2008. Accepted for publication March 20, 2008.
|
References |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionconclusionfundingAcknowledgementsReferences |
|---|
1. Leibovitch L, Foster RS, Kopecky KK, et al. Improved accuracy of computerized tomography based clinical staging in low stage nonseminomatous germ cell tumor using size criteria of retroperitoneal lymph nodes. J Urol (1995) 154:1759–1763.[CrossRef][Web of Science][Medline]
2. Smith TAD. FDG uptake, tumor characteristics and response to therapy: a review. Nucl Med Commun (1998) 1:97–105.
3. Higashi T, Sakahara H, Torizuka T, et al. Evaluation of intraoperative radiation therapy for unresectable pancreatic cancer with FDG PET. J Nucl Med (1999) 40:1424–1433.
4. Kinkel K, Lu Y, Both M, et al. Detection of hepatic metastases from cancers of the gastrointestinal tract by using noninvasive imaging methods (US, CT, MR imaging, PET): a meta-analysis. Radiology (2002) 224:748–756.
5. Valk PE, Abella-Columna E, Haseman MK, et al. Whole-body PET imaging with [18F]fluorodeoxyglucose in management of recurrent colorectal cancer. Arch Surg (1999) 134:503–511.
6. Kern KA, Brunetti A, Norton JA, et al. Metabolic imaging of human extremity musculoskeletal tumors by PET. J Nucl Med (1988) 29:181–186.
7. Newman JS, Francis IR, Kaminski MS. Imaging of lymphoma with PET with 2-(F-18)-fluoro-2-deoxy-D-glucose: correlation with CT. Radiology (1994) 190:111–116.
8. Dewan NA, Gupta NC, Redepenning LS, et al. Diagnostic efficacy of PET-FDG imaging in solitary pulmonary nodules: potential role in evaluation and management. Chest (1993) 104:997–1002.[CrossRef][Web of Science][Medline]
9. de Wit M, Bumann D, Beyer W, et al. Whole-body positron-emission tomography (PET) for diagnosis of residual mass in patients with lymphoma. Ann Oncol (1997) 8:57–60.
10. Geworski L, Knoop BO, de Wit M, et al. Multicenter comparative evaluation of calibration and cross calibration of positron emission tomographs. J Nucl Med (2002) 43:635–639.
11. Minn H, Leskinen-Kallio S, Lindholm P, et al. (18-F)Fluorodeoxyglucose uptake in tumors: kinetic vs. steady-state methods with reference to plasma insulin. J Comput Assist Tomogr (1993) 17:115–123.[Web of Science][Medline]
12. Mostofi FK, Sesterhenn IK. Revised international classification of testicular tumors. In: Germ Cell Tumors III. Advances in the Biosciences—Jones WG, Harnden P, Appleyard I, eds. (1994) Volume 9. Oxford, UK: Pergamon Press. 153–158.
13. Albers P, Bender H, Yilmaz H, et al. Positron emission tomography in the clinical staging of patients with stage I and II testicular germ cell tumors. Urology (1999) 53:808–811.[CrossRef][Web of Science][Medline]
14. Spermon JR, De Geus-Oei LF, Kiemeney LA, et al. The role of (18)-fluoro-2-deoxyglucose positron emission tomography in initial staging and re-staging after chemotherapy for testicular germ cell tumors. BJU Int (2002) 89:549–556.[CrossRef][Web of Science][Medline]
15. Lassen U, Daugaard G, Eigtved A, et al. Whole-body FDG-PET in patients with stage I non-seminomatous germ cell tumors. Eur J Nucl Med (2003) 30:396–402.[Web of Science]
16. Huddart RA, O'Doherty MJ, Padhani A, et al. 18Fluorodeoxyglucose positron emission tomography in the prediction of relapse in patients with high risk, clinical stage I nonseminomatous germ cell tumors: preliminary report of MRC Trial TE22—the NCRI Testis Tumour Clinical Study Group. J Clin Oncol (2007) 25:3090–3095.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||