Supervised machine learning enables non-invasive lesion characterization in primary prostate cancer with [68Ga]Ga-PSMA-11 PET/MRI

L Papp, C P Spielvogel, B Grubmüller, M Grahovac, D Krajnc, B Ecsedi, R A M Sareshgi, D Mohamad, M Hamboeck, I Rausch, M Mitterhauser, W Wadsak, A R Haug, L Kenner, P Mazal, M Susani, S Hartenbach, P Baltzer, T H Helbich, G Kramer, S F Shariat, T Beyer, M Hartenbach, M Hacker, L Papp, C P Spielvogel, B Grubmüller, M Grahovac, D Krajnc, B Ecsedi, R A M Sareshgi, D Mohamad, M Hamboeck, I Rausch, M Mitterhauser, W Wadsak, A R Haug, L Kenner, P Mazal, M Susani, S Hartenbach, P Baltzer, T H Helbich, G Kramer, S F Shariat, T Beyer, M Hartenbach, M Hacker

Abstract

Purpose: Risk classification of primary prostate cancer in clinical routine is mainly based on prostate-specific antigen (PSA) levels, Gleason scores from biopsy samples, and tumor-nodes-metastasis (TNM) staging. This study aimed to investigate the diagnostic performance of positron emission tomography/magnetic resonance imaging (PET/MRI) in vivo models for predicting low-vs-high lesion risk (LH) as well as biochemical recurrence (BCR) and overall patient risk (OPR) with machine learning.

Methods: Fifty-two patients who underwent multi-parametric dual-tracer [18F]FMC and [68Ga]Ga-PSMA-11 PET/MRI as well as radical prostatectomy between 2014 and 2015 were included as part of a single-center pilot to a randomized prospective trial (NCT02659527). Radiomics in combination with ensemble machine learning was applied including the [68Ga]Ga-PSMA-11 PET, the apparent diffusion coefficient, and the transverse relaxation time-weighted MRI scans of each patient to establish a low-vs-high risk lesion prediction model (MLH). Furthermore, MBCR and MOPR predictive model schemes were built by combining MLH, PSA, and clinical stage values of patients. Performance evaluation of the established models was performed with 1000-fold Monte Carlo (MC) cross-validation. Results were additionally compared to conventional [68Ga]Ga-PSMA-11 standardized uptake value (SUV) analyses.

Results: The area under the receiver operator characteristic curve (AUC) of the MLH model (0.86) was higher than the AUC of the [68Ga]Ga-PSMA-11 SUVmax analysis (0.80). MC cross-validation revealed 89% and 91% accuracies with 0.90 and 0.94 AUCs for the MBCR and MOPR models respectively, while standard routine analysis based on PSA, biopsy Gleason score, and TNM staging resulted in 69% and 70% accuracies to predict BCR and OPR respectively.

Conclusion: Our results demonstrate the potential to enhance risk classification in primary prostate cancer patients built on PET/MRI radiomics and machine learning without biopsy sampling.

Keywords: Biochemical recurrence prediction; Lesion risk prediction; Machine learning; Overall patient risk prediction; PET/MRI; Prostate cancer; Radiomics.

Conflict of interest statement

The authors declare that they have no conflict of interest.

Figures

Fig. 1
Fig. 1
The analysis workflow of the collected dataset. The pre-study of the prospective randomized trial NCT02659527 provided data records of 122 patients between 2014 and 2015. Patients having a dual-tracer positron emission tomography/magnetic resonance imaging (PET/MRI), prostate-specific antigen (PSA) screening, and whole-mount histopathology through undergone surgery were included in the analysis (n = 52). Only [68Ga]Ga-PSMA-11 PET, apparent diffusion coefficient (ADC), and transverse relaxation time-weighted (T2w) MRI images were selected for radiomic analysis. Overall 121 PET/MRI-positive lesions were delineated from the 52 patients followed by radiomics feature extraction. The 121 lesions underwent prostate specific membrane antigen (PSMA) standardized uptake value (SUV) and volume area under the receiver operator characteristics curve (AUC) analysis. Monte Carlo (MC) cross-validation scheme was utilized to generate patient training and validation sets 1000-times. This MC scheme was utilized to build lesion low-vs-high (LH) prediction models via machine learning (MLH). Biochemical recurrence (BCR, n = 36) and overall patient risk (OPR, n = 50) patient prediction models were built across the same MC folds (MBCR and MOPR respectively). All machine learning models underwent confusion matrix analytics, sham data analysis, and AUC analysis across MC folds. BCR and OPR were also predicted by standard D’Amico score
Fig. 2
Fig. 2
(A) Positron emission tomography/magnetic resonance imaging (PET/MRI) views of a prostate cancer patient with volumes of interests (VOIs) drawn over lesions with Gleason 4 (red) and high-grade pin (blue) patterns. Standard iso-count 3D VOIs were drawn over the [68Ga]Ga-PSMA-11 PET in the Hermes Hybrid 3D software. First row: [68Ga]Ga-PSMA-11 PET; second row: apparent diffusion coefficient (ADC) MRI; third row: fused [68Ga]Ga-PSMA-11 PET and transverse relaxation time-weighted (T2w) MRI images. Note that each image is represented in its own frame of reference, while the fused PET/MRI view is aligned to the frame of reference of the T2-weighted MRI. Hence, the cross-sections of the drawn VOIs look different on each view. (B) An example histopathological slice with the same color codes as in case of the PET/MRI views (red: Gleason 4, blue: high-grade pin)
Fig. 3
Fig. 3
Area under the receiver operator characteristics curves (AUC) of conventional standardized uptake values (SUV) as well as lesion volume together with the machine learning low-vs-high lesion risk scores. Note that the MLH AUC performance is a conservative estimate, as it is a Monte Carlo cross-validation AUC, while the SUV and volume curves were measured directly from the whole dataset
Fig. 4
Fig. 4
Occurrence of the highest ranked features across the 1000-fold Monte Carlo cross-validation scheme. PSMA—[68Ga]Ga-PSMA-11 positron emission tomography (PET); stat.cov: coefficient of variation; cm.info.corr.1—gray level co-occurrence matrix information correlation type 1; ADC—apparent diffusion coefficient; stat.iqr—interquartile range; cm.joint.entr—gray level co-occurrence matrix joint entropy; dzm.hgze—gray level distance zone matrix high gray zone emphasis
Fig. 5
Fig. 5
Left: validation performance estimations of predicting biochemical recurrence (BCR) by MBCR and clinical standard models. Right: validation performance estimations of predicting overall patient risk (OPR) MOPR and the clinical standard models. SENS—sensitivity; SPEC—specificity; ACC—accuracy; PPV—positive predictive value; NPV—negative predictive value. Confusion matrix values are in percentages. Note that standard risk estimator had a confusion analytics performance estimation in the whole dataset, as it is an established model, while the performance of MBCR and MOPR models was calculated through Monte Carlo cross-validation

References

    1. Ferlay J, Colombet M, Soerjomataram I, Mathers C, Parkin DM, Piñeros M, et al. Estimating the global cancer incidence and mortality in 2018: GLOBOCAN sources and methods. Int J Cancer [Internet]. 2018;144(8):1941–53. 10.1002/ijc.31937.
    1. WCRF . Protate cancer statistics. 2018.
    1. Mottet N, Bellmunt J, Bolla M, Briers E, Cumberbatch MG, De Santis M, et al. EAU-ESTRO-SIOG guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent. Eur Urol [Internet] 2017;71(4):618–629. doi: 10.1016/j.eururo.2016.08.003.
    1. Hernandez DJ, Nielsen ME, Han M, Partin AW. Contemporary evaluation of the D’Amico risk classification of prostate cancer. Urol Int. 2007;70(5):931–935.
    1. Fenton JJ, Weyrich MS, Durbin S, Liu Y, Bang H, Melnikow J. Prostate-specific antigen-based screening for prostate cancer: a systematic evidence review for the U.S. Rockville (MD): Preventive Services Task Force; 2018.
    1. Hatt M, Tixier F, Pierce L, Kinahan PE, Le Rest CC, Visvikis D. Characterization of PET/CT images using texture analysis: the past, the present… any future? Eur J Nucl Med Mol Imaging [Internet]. 2017;44(1):151–65. 10.1007/s00259-016-3427-0.
    1. Preisser F, Bandini M, Marchioni M, Nazzani S, Tian Z, Pompe RS, et al. Extent of lymph node dissection improves survival in prostate cancer patients treated with radical prostatectomy without lymph node invasion. Prostate [Internet]. 2018;78(6):469–75. 10.1002/pros.23491.
    1. Salmasi A, Faiena I, Wu J, Sisk AE, Sachveda A, Vandel JJ, et al. Radical prostatectomy then and now: surgical overtreatment of prostate cancer is declining from 2009 to 2016 at a tertiary referral center. Urol Oncol Semin Orig Investig [Internet] 2018;36(9):401.e19–401.e25.
    1. Chow K, Herrera P, Stuchbery R, Peters JS, Costello AJ, Hovens CM, et al. Late biochemical recurrence after radical prostatectomy is associated with a slower rate of progression. BJU Int [Internet]. 2019;123(6):976–84. 10.1111/bju.14556.
    1. Galgano SJ, Valentin R, McConathy J. Role of PET imaging for biochemical recurrence following primary treatment for prostate cancer. Transl Androl Urol [Internet] 2018;7(S4):S462–S476. doi: 10.21037/tau.2018.06.09.
    1. Herlemann A, Wenter V, Kretschmer A, Thierfelder KM, Bartenstein P, Faber C, et al. 68Ga-PSMA positron emission tomography/computed tomography provides accurate staging of lymph node regions prior to lymph node dissection in patients with prostate cancer. Eur Urol [Internet] 2016;70(4):553–557. doi: 10.1016/j.eururo.2015.12.051.
    1. Polanec SH, Andrzejewski P, Baltzer PAT, Helbich TH, Stiglbauer A, Georg D, et al. Multiparametric [11C]acetate positron emission tomography-magnetic resonance imaging in the assessment and staging of prostate cancer. Gelovani JG editor. PLoS One [Internet]. 2017;12(7):e0180790. 10.1371/journal.pone.0180790.
    1. Bouchelouche K, Choyke PL. Advances in prostate-specific membrane antigen PET of prostate cancer. Curr Opin Oncol [Internet] 2018;30:189–196. doi: 10.1097/CCO.0000000000000439.
    1. Afshar-Oromieh A, Avtzi E, Giesel FL, Holland-Letz T, Linhart HG, Eder M, et al. The diagnostic value of PET/CT imaging with the 68Ga-labelled PSMA ligand HBED-CC in the diagnosis of recurrent prostate cancer. Eur J Nucl Med Mol Imaging [Internet]. 2015;42(2):197–209. 10.1007/s00259-014-2949-6.
    1. Grubmüller B, Baltzer P, Hartenbach S, D’Andrea D, Helbich TH, Haug AR, et al. PSMA ligand PET/MRI for primary prostate cancer: staging performance and clinical impact. Clin Cancer Res [Internet]. 2018;24(24):6300–7. 10.1158/1078-0432.CCR-18-0768.
    1. Stone L. Predicting 68Ga-PSMA-PET–CT positivity for recurrent disease. Nat Rev Urol [Internet]. 2018;15:137. 10.1038/nrurol.2018.15.
    1. Nair R, Porpiglia F, Zargar H. Re: MRI-targeted or standard biopsy for prostate-cancer diagnosis. Eur Urol [Internet]. 2018;74(4):524–5. 10.1056/NEJMoa1801993.
    1. Hartenbach M, Hartenbach S, Bechtloff W, Danz B, Kraft K, Klemenz B, et al. Combined PET/MRI improves diagnostic accuracy in patients with prostate cancer: a prospective diagnostic trial. Clin Cancer Res [Internet]. 2014;20(12):3244–53. 10.1158/1078-0432.CCR-13-2653.
    1. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data. Radiology [Internet]. 2016;278(2):563–77. 10.1148/radiol.2015151169.
    1. Manetta R, Palumbo P, Gianneramo C, Bruno F, Arrigoni F, Natella R, et al. Correlation between ADC values and Gleason score in evaluation of prostate cancer: multicentre experience and review of the literature. Gland Surg [Internet] 2019;8(S3):S216–S222. doi: 10.21037/gs.2019.05.02.
    1. Le MH, Chen J, Wang L, Wang Z, Liu W, Cheng K-T(T), et al. Automated diagnosis of prostate cancer in multi-parametric MRI based on multimodal convolutional neural networks. Phys Med Biol [Internet] 2017;62(16):6497–6514. doi: 10.1088/1361-6560/aa7731.
    1. Liu C, Liu T, Zhang N, Liu Y, Li N, Du P, et al. 68Ga-PSMA-617 PET/CT: a promising new technique for predicting risk stratification and metastatic risk of prostate cancer patients. Eur J Nucl Med Mol Imaging [Internet]. 2018;45(11):1852–61. 10.1007/s00259-018-4037-9.
    1. Zamboglou C, Carles M, Fechter T, Kiefer S, Reichel K, Fassbender TF, et al. Radiomic features from PSMA PET for non-invasive intraprostatic tumor discrimination and characterization in patients with intermediate- and high-risk prostate cancer - a comparison study with histology reference. Theranostics [Internet] 2019;9(9):2595–2605. doi: 10.7150/thno.32376.
    1. Cysouw MCF, Jansen BHE, van de Brug T, Oprea-Lager DE, Pfaehler E, de Vries BM, et al. Machine learning-based analysis of [18F]DCFPyL PET radiomics for risk stratification in primary prostate cancer. Eur J Nucl Med Mol Imaging [Internet]. 2020. 10.1007/s00259-020-04971-z.
    1. Epstein JI, Allsbrook WC, Amin MB, Egevad LL. The 2005 International Society of Urological Pathology (ISUP) consensus conference on Gleason grading of prostatic carcinoma. Am J Surg Pathol [Internet] 2005;29(9):1228–1242. doi: 10.1097/01.pas.0000173646.99337.b1.
    1. Papp L, Pötsch N, Grahovac M, Schmidbauer V, Woehrer A, Preusser M, et al. Glioma survival prediction with combined analysis of in vivo 11C-MET PET features, ex vivo features, and patient features by supervised machine learning. J Nucl Med. 2018;59(6):892–899. doi: 10.2967/jnumed.117.202267.
    1. Phillips DL, Marks DG. Spatial uncertainty analysis: propagation of interpolation errors in spatially distributed models. Ecol Model. 1996;91(1–3):213–229. doi: 10.1016/0304-3800(95)00191-3.
    1. Stytz MR, Parrott RW. Using kriging for 3d medical imaging. Comput Med Imaging Graph. 1993;17(6):421–442. doi: 10.1016/0895-6111(93)90059-V.
    1. Zwanenburg A, Leger S, Vallières M, Löck S, Initiative for the IBS. Image biomarker standardisation initiative. arXiv [Internet]. 2016;(November).
    1. Sagi O, Rokach L. Ensemble learning: a survey. Wiley Interdiscip Rev Data Min Knowl Discov [Internet]. 2018;8(4):1249. 10.1002/widm.1249.
    1. Parmar C, Grossmann P, Bussink J, Lambin P, Aerts HJWL. Machine learning methods for quantitative radiomic biomarkers. Sci Rep. 2015;5(1):13087. doi: 10.1038/srep13087.
    1. Feurer M, Hutter F. Hyperparameter optimization; 2019. p. 3–33. 10.1007/978-3-030-05318-5_1.
    1. van Timmeren JE, Leijenaar RTH, van Elmpt W, Reymen B, Oberije C, Monshouwer R, et al. Survival prediction of non-small cell lung cancer patients using radiomics analyses of cone-beam CT images. Radiother Oncol. 2017;123(3):363–369. doi: 10.1016/j.radonc.2017.04.016.
    1. Lacroix M, Frouin F, Dirand A-S, Nioche C, Orlhac F, Bernaudin J-F, et al. Correction for magnetic field inhomogeneities and normalization of voxel values are needed to better reveal the potential of MR radiomic features in lung cancer. Front Oncol [Internet]. 2020;10:43. 10.3389/fonc.2020.00043.
    1. Amin A, Anwar S, Adnan A, Nawaz M, Howard N, Qadir J, et al. Comparing oversampling techniques to handle the class imbalance problem: a customer churn prediction case study. IEEE Access. 2016;4(October):7940–7957. doi: 10.1109/ACCESS.2016.2619719.
    1. Papp L, Rausch I, Grahovac M, Hacker M, Beyer T. Optimized feature extraction for radiomics analysis of 18F-FDG PET imaging. J Nucl Med. 2019;60(6):864–872. doi: 10.2967/jnumed.118.217612.
    1. Woythal N, Arsenic R, Kempkensteffen C, Miller K, Janssen J-C, Huang K, et al. Immunohistochemical validation of PSMA expression measured by 68 Ga-PSMA PET/CT in primary prostate cancer. J Nucl Med [Internet]. 2018;59(2):238–43. 10.2967/jnumed.117.195172.
    1. Bravaccini S, Puccetti M, Bocchini M, Ravaioli S, Celli M, Scarpi E, et al. PSMA expression: a potential ally for the pathologist in prostate cancer diagnosis. Sci Rep. 2018;8(1):4254. doi: 10.1038/s41598-018-22594-1.
    1. Fehr D, Veeraraghavan H, Wibmer A, Gondo T, Matsumoto K, Vargas HA, et al. Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images. Proc Natl Acad Sci [Internet]. 2015;112(46):E6265–73. 10.1073/pnas.1505935112.
    1. Traverso A, Kazmierski M, Zhovannik I, Welch M, Wee L, Jaffray D, et al. Machine learning helps identifying volume-confounding effects in radiomics. Phys Medica [Internet] 2020;71:24–30. doi: 10.1016/j.ejmp.2020.02.010.
    1. D’Amico A. D’Amico Risk classification for Prostate Cancer [Internet].
    1. Cook GJR, Siddique M, Taylor BP, Yip C, Chicklore S, Goh V. Radiomics in PET: principles and applications. Clin Transl Imaging. 2014;2(3):269–276. doi: 10.1007/s40336-014-0064-0.
    1. Verwer EE, Oprea-Lager DE, van den Eertwegh AJM, van Moorselaar RJA, Windhorst AD, Schwarte LA, et al. Quantification of 18F-fluorocholine kinetics in patients with prostate cancer. J Nucl Med [Internet]. 2015;56(3):365–71. 10.2967/jnumed.114.148007.

Source: PubMed

スポンサーと協力者

医学的状態

薬物療法

3
購読する