Advertisement

Current applications and challenges of radiomics in urothelial cancer

  • Gumuyang Zhang
  • Lili Xu
  • Hao SunEmail author
  • Zhengyu JinEmail author
Review
  • 63 Downloads

Abstract

New discoveries and technologies have begun to change paradigms of urothelial cancer therapy in recent years. One of the novel techniques which emerged in the imaging community is radiomics, which refers to the high-throughput extraction of quantitative image features from medical images. Radiomics, being noninvasive and easy to perform, has shown great potential in oncology by providing valuable information about tumor type, aggressiveness, progression, response to treatment and prognosis and enabling us to gain insights into the true utility of personalized medicine in the management of cancer in the near future. With rapid development in this area, radiomics has already been applied in urothelial cancer to predict pathological grade, clinical stage, lymph node metastasis and treatment response demonstrating promising results. In this review, we highlight advances in clinical applications of radiomics in urothelial cancer, discuss about the challenges and implications of radiomics for radiologists and suggest the future directions that we could move toward in order to fully realize the potentials of radiomics to improve personalized management of patients with urothelial cancer.

Keywords

Urothelial cancer Radiomics Imaging biomarker Precision medicine 

Introduction

Urothelial cancer is a common malignancy worldwide. More than 90% of the cases are bladder cancer while upper tract urothelial cancer (UTUC) is a rare subset [1]. Bladder cancer is the sixth most common cancer in the USA with an estimation of 81,000 new cases and 17,000 deaths each year, and the standardized mortality rate varies from 2 to 10/100,000 per year in men and 0.5 to 4/100,000 per year in women [2, 3]. According to the National Central Cancer Registry of China, the new cases and deaths for bladder cancer are about 80,500 cases and 32,900 cases, respectively, with an upward trend in recent years [4]. Over the past three decades, few signs of progress have been made in the treatment for urothelial cancer. For patients with nonmuscle-invasive bladder cancer, around one-third of patients experience recurrences or progression despite receiving standard treatment. The 5-year survival of patients with muscle-invasive bladder cancer is less than 50%, and the median overall survival of patients with advanced or metastatic bladder cancer is about 15 months [5, 6]. The deadlock of urothelial cancer treatment has been broken with significant advances in our understanding of underlying tumor biology and immunology in recent years. The advent of immune checkpoint inhibitors (ICIs) including anti-programmed cell death 1 (PD-1) and anti-programmed cell-death ligand 1 (PD-L1) antibodies has revolutionized the treatment for many advanced solid tumors including urothelial cancer [7]. In spite of dramatic improvements in clinical outcomes in certain patients, there are still quite a few important unmet clinical needs in urothelial cancer management, for example, how to identify patients most likely to benefit from ICIs and how to predict treatment response for individuals [8]. It is necessary to develop new tools that have the potential to tackle those difficulties in tailoring treatment for each patient with urothelial cancer, especially in the era of precision medicine.

A wide range of “-omic” technologies, such as genomics and proteomics, have been investigated in the field of oncology to improve current biomarkers for the diagnosis and therapy of tumors including urothelial cancer. The term “radiomics” has been introduced several years ago, and it has become a novel research field with rapid development. Radiomics refers to analysis and translation of medical images into mineable and measurable high-dimensional data producing quantitative features in relation to prediction targets such as gene expression and clinical outcomes [9, 10]. Radiomics generally involves five major steps: data selection, medical imaging, feature extraction, exploratory analysis and modeling. Figure 1 generally describes the process. The whole process mainly relies on computer algorithms rather than human visual assessment, and this advantage of quantitative analysis allows radiomics to reveal information related to cellular and molecular properties of the tissue that may not be perceived by human naked eyes [11]. As tumors are extremely heterogeneous, a biopsy of a limited tissue sample is unlikely to represent the entire tumor. Unlike biopsy, radiomics can examine a tumor as a whole and evaluate intratumoral heterogeneity at the same time [12]. These unique abilities of radiomics allow it to be applied as a promising biomarker to noninvasively indicate biological processes, pathological changes or responses to therapeutic intervention.
Fig. 1

Radiomics in urothelial cancer. Representative images of a bladder cancer, with region of interest (ROI) segmentation shown in red followed by feature extraction, feature selection and model construction

Up to now, many published papers have demonstrated the huge potential of radiomics to improve clinical diagnosis, disease monitoring and outcome predictions in various solid tumors including urothelial cancer. In this review, we briefly summarize the current clinical applications of radiomics in urothelial cancer along with the latest development in the field. Then, we discuss the challenges we met in this field and the implications for radiologists. At last, we offer our perspectives of future research directions of radiomics in urothelial cancer.

Radiomics applications in urothelial cancer

The past few years have witnessed considerable scientific advances in applications of artificial intelligence in human malignant neoplasms. Different technical methods including texture analysis, machine learning, deep learning and radiomics have been employed in studies regarding urothelial cancer. We would like to focus on radiomics studies in this review but we will also mention related studies using other technical methods as well, so as to give a comprehensive review of achievements in this filed in the last several years. In the following paragraphs, we are going to group these studies by the targeted clinical question and analyze in detail the data related to each topic.

Evaluation of pathological grade

Pathological grade of urothelial cancer has important implications for prognosis and treatment selection. Low-grade urothelial cancer has a lower rate of recurrence and stage progression, and it could be treated with less invasive techniques [13]. A few studies have explored the feasibility of texture analysis, machine learning and radiomics to distinguish between low- and high-grade urothelial cancer. A study by our group in 105 patients with urothelial carcinoma found that low-grade tumors demonstrated lower texture features of mean, entropy and mean of positive pixels (MPP) quantified from CT images and MPP could differentiate low- from high-grade tumors with an area under the curve (AUC) of 0.779 [14]. Mammen et al. performed a similar texture analysis of CT scans of 48 patients with UTUC and found entropy was also greater in high-grade tumors with an AUC of 0.83 [15]. In a study by Zhang et al. in 61 patients with bladder cancer, 102 texture features were extracted from diffusion-weighted images (DWI) and apparent diffusion coefficient (ADC) maps, 47 of which were found to be significantly different between low- and high-grade tumors [16]. By the method of support vector machine with recursive feature elimination (SVM-RFE), 22 features were selected to build the classifier and reached an AUC of 0.861. Wang et al. conducted a similar but more comprehensive radiomics analysis of MRI images for preoperative evaluation of pathological grade [17]. They examined T2-weighted (T2W), DWI and ADC maps of 70 patients and validated them in a cohort of 30 patients with bladder cancer. Multimodal features were combined to construct radiomics models which achieved an AUC around 0.92 in both training and validation cohorts. The above studies revealed that texture features extracted from CT or MRI images could reflect the difference between low- and high-grade urothelial carcinoma. A radiomics approach has the potential to act as a noninvasive tool to assist in preoperative grading of urothelial cancer. But the sample size of these studies is small, and as UTUC is quite rare compared to bladder cancer, only our group and Mammen et al. have included UTUC [14, 15]. More data should be gathered to further evaluate the ability of radiomics for predicting pathological grade, especially in UTUC.

Evaluation of clinical T stage

Accurate local staging of bladder cancer is key to realize optimal management of an individual patient. Patients with nonmuscle-invasive bladder cancer (NMIBC, stage ≤ T1) are mostly treated with bladder-sparing methods such as transurethral resection of a bladder tumor (TURBT) and intravesical therapies, whereas patients with muscle-invasive bladder cancer (MIBC, stage ≥ T2) are treated with cystectomy, radiation therapy or chemotherapy and usually have a poor prognosis [13, 18]. Thus, discrimination between NMIBC and MIBC has great value in guiding therapeutic choices. In a study by Garapati et al., morphological and texture features were extracted from CT images of 84 bladder cancer lesions and a linear discriminant analysis, a neural network, a SVM and a random forest classifier were used to combine the features to stratify the stage of bladder cancer into two groups: ≥ T2 and < T2 groups [19]. The classification accuracies of the four classifiers were similar with AUCs around 0.9. Several MRI-based radiomics studies have also reported promising results of accurate stratification for stages of bladder cancer. Xu et al. extracted a total of 1104 radiomic features from T2W and DW images of 44 patients with bladder cancer and selected 19 features to build an optimal discriminative model by the method of SVM-RFE and synthetic minority oversampling technique (SMOTE) to differentiate between NMIBC and MIBC or stage ≥ T2 and < T2 [20]. An AUC of 0.9857 was reached by the SVM-RFE + SMOTE classifier, which outperformed the diagnostic accuracy by experts. In a study by Tong et al., T2W images of 65 patients with bladder cancer were used to quantify intensity and texture parameters to classify patients into ≥ T2 or < T2 [21]. Nine optimal features were selected from a total of 15,834 features and demonstrated an AUC of 0.813 for the differentiation purpose. Lim et al. performed a slightly different study from the previous ones [22]. They evaluated whether T2W and ADC texture features of bladder cancer and extravesical fat as well in 36 patients could be used to predict MIBC (≥ T2) and extravesical disease (≥ T3) after TURBT. Results show that greater entropy of bladder cancers and extravesical fat was found in category ≥ T3 than in category ≤ T2 and in category ≥ T2 than in category T1 tumors with AUCs in the range of 0.74–0.85. These studies indicated that radiomics could help with local staging of bladder cancer and has the potential to improve current patient management. The main limitation of this study is the very small sample size, and none of the above studies has a patient population over 100. It is no doubt that these preliminary studies proved the concept of applying radiomics to evaluate the local stage of bladder cancer, but where the study results are still valid or radiomics could really promote precise local staging of bladder cancer needs further investigation and validation in larger cohorts.

Prediction of lymph node (LN) metastasis

LN metastasis in patients with bladder cancer indicates a poorer prognosis; thus, the accurate prediction of LN metastasis in patients with bladder cancer assists in treatment decision making. Routine CT and MRI identify positive LN metastasis according to the size of LN, but the efficacy is quite low with a sensitivity of 31–45%, which indicates a certain proportion of patients being understaged [23, 24, 25]. A study by Wu et al. in 118 patients with bladder cancer found 150 radiomic features in each patient’s arterial-phase CT images, among which nine LN status-related features were used to build the radiomic signature for LN metastasis and achieved favorable prediction efficacy [26]. The radiomics nomogram incorporating the radiomics signature and CT-reported LN status also demonstrated good calibration and discrimination in the training set (AUC 0.9262) and the validation set (AUC 0.8986). The same research group conducted another study in 103 patients with bladder cancer in the purpose of developing and validating an MRI-based radiomics signature for the individual preoperative prediction of LN metastasis [25]. A total of 718 radiomic features were extracted from T2W images, and nine features were selected to construct the radiomic signature which showed a favorable outcome in the training set with an AUC of 0.9005 and in the validation set with an AUC of 0.8447. The radiomics signature and the MRI-reported LN status constituted the nomogram, and it demonstrated good calibration and discrimination in the training (AUC 0.9118) and validation (AUC 0.8902) sets. These two studies proved the promising value of radiomics in the prediction of LN metastasis in bladder cancer. Despite the relatively satisfactory performance of CT- and MRI-based radiomics, a shared limitation of the two studies is that they lack external validation. Multicenter validation with larger cohorts is required to confirm the ability of radiomics to accurately predict LN metastasis. Genetic markers have been shown to be predictive of LN metastasis, the addition of genetic markers to the nomogram might further improve the accuracy of radiomics to predict LN metastasis and further studies may work on this issue.

Prediction of recurrence risk

A prominent characteristic of bladder cancer is its high recurrent rate, which could reach up to 61% for patients with nonmuscle-invasive bladder cancer in the first 2 years (TFTY) after TURBT [27]. Preoperative prediction of recurrence risk is critical for prognostication and individualized follow-up regimens for patients. Xu et al. developed and validated a nomogram combining MRI-based radiomics and clinical predictors for predicting the TFTY recurrence risk [28]. Of the 1872 features extracted from T2W, DW, ADC and dynamic contrast-enhanced images, the 32 features with the highest AUC were selected for calculating Rad-Score. The nomogram developed by Rad-Score and clinical predictor of muscle-invasive status produced a good performance in the training (accuracy 88%, AUC 0.915) and validation cohorts (accuracy 80.95%, AUC 0.838). This preliminary study demonstrates the ability of radiomics together with clinical factors to address the important clinical issue of recurrence risk prediction for bladder cancer. So far, we have not found any study investigating the potential of CT-based radiomics for prediction of recurrence risk for bladder cancer. As CT plays an important role in preoperative evaluation and postoperative follow-up in patients with bladder cancer, it is worth exploring the value of CT-based radiomics in this clinical issue as well.

Treatment response assessment

Neoadjuvant chemotherapy before cystectomy has been shown to improve survival but only 30% of the patients have complete treatment response; a reliable prediction of the efficacy of neoadjuvant chemotherapy is beneficial for patients with bladder cancer [29]. In a study by Cha et al., they explored the feasibility of three CT-based radiomics models employing different design principles to distinguish between patients with and without complete chemotherapy responses [30]. The three models included a model using deep-learning convolution neural network (DL-CNN), a model using radiomic features extracted from segmented lesions and a model using radiomic features extracted from pre- and post-treatment paired regions of interest. All the three models produced comparable AUCs compared to two expert radiologists ranging from 0.69 to 0.77. It is obvious that the accuracy is not satisfactory in terms of AUCs but this study is the first to indicate the potential of using DL-CNN and radiomics methods to assess treatment response of chemotherapy for patients with bladder cancer. The small sample size of this study (82 patients in the training set and 42 patients in the test set) could be a major factor that impacted the performance of prediction models. More radiomics studies with larger cohorts targeting the prediction of treatment responses for urothelial cancer should be conducted in the future.

Challenges and implications for radiologists

The above studies provide improved insight into the utility of radiomics in the management of urothelial cancer. These studies demonstrate the capability of radiomics to assist more precise characterization and stratification of patients with urothelial cancer. As it is impossible to biopsy each and every lesion, radiomics offers a noninvasive and economic approach to reveal the tumor heterogeneity in different individuals, different lesions and even within the same lesion. By using radiomics as biomarkers, we may begin to appreciate the complexity of tumor biology and tailor treatment for each patient with urothelial cancer.

There is no doubt that radiomics could facilitate the process of clinical decision making, but up to now, radiomics for urothelial cancer remains in research and not in clinical use. There are quite a lot of challenges ahead of us for applying radiomics in daily practice to improve patient care. The workflow of radiomics includes data selection, medical imaging, feature extraction, exploratory analysis and modeling and implementation of radiomics is rather a complicated process. One of the major challenges lies in the optimal collection and integration of multiple data sources that can produce accurate and robust predictions. Currently, the field of radiomics lacks standardized evaluation criteria and reporting guidelines; the clinical utility of those published prediction models still needs to be further evaluated for their performance [31]. To promote the development and acceptance of radiomics, Lambin et al. have proposed the radiomics quality score (RQS) to evaluate the quality of radiomic studies [11]. The RQS evaluates each necessary step in a radiomic analysis, both rewards and penalizes the methodology and analyses of a study. Investigators should be encouraged to follow the rigorous evaluation criteria and reporting guidelines to avoid overly optimistic claims about robustness and generalizability.

High reproducibility and replicability are essential for the widespread acceptance of radiomics-based models or decision support systems in clinical practice. As radiomics studies comprise multiple steps, and each could be affected by a wide range of factors, details of these subprocesses should be disclosed by researchers; otherwise, reproducibility and replicability in radiomics would not be possible. Large-scale data sharing is imperative for the validation and generalization of radiomics; thus, disclosure of imaging protocols, analyzed scans, segmentations, details of feature extraction and modeling methodology should be provided as supplementary material in future publications.

Future directions

Medical imaging is evolving from being a diagnostic tool to becoming a vital part in the era of personalized medicine. It is of great importance to promoting precision medicine, especially in countries like China which has a large population but limited government investment in health care and low average expense per patient need. Application and generalization of novel techniques could help to provide the optimal treatment for patients while avoiding unnecessary cost, relieving the heavy economic burden of diseases for both individuals and society. Radiomics, with the advantages of being noninvasive and economical, is worthy of further investigation and application.

With the advances in radiomics, it has made it possible to correlate clinically feasible quantitative imaging with tissue pathophysiology. Radiogenomics highlighting the link between radiomic features and gene expression patterns allows the acceleration of their incorporation into personalized medicine approaches [32]. Over the years, there have been many studies investigating the application of gene expression signatures for prediction of tumor characteristics and outcomes of urothelial cancer, including stage, risk of recurrences, the progression of nonmuscle-invasive and muscle-invasive bladder cancer and survival [9, 31, 33, 34, 35, 36, 37]. But up to now, studies focusing on identifying the association between specific imaging traits and gene profile of urothelial cancer have not been reported yet. The research in radiogenomics of urothelial cancer is still at the initial stage and remains to be further explored. Standardized gene assay and radiomics workflow would enable radiogenomics biomarkers to meaningfully improve diagnosis, prognosis and prediction of response to treatment of urothelial cancer.

The past few years have been an exciting time for the field of urothelial cancer. With the introduction of ICIs such as atezolizumab and pembrolizumab, significant advances have been made in the treatment for urothelial cancer. ICIs have been proved to be effective with safe and tolerable side effects in a subset of patients with urothelial cancer, but the majority have primary disease progression [38, 39]. It is crucial to identify patients who are most likely to benefit from ICIs. Certain pathological markers assessed by immunohistochemistry have shown the potential to predict the treatment response of ICIs [40]. Noninvasive imaging biomarkers for optimal patient selection are still under investigation. Promising results have been reported recently that the radiomic signature of tumor-infiltrating CD8 cells could be useful in inferring clinical outcomes for patients with cancer treated with immunotherapy. But only eight patients with urothelial cancer were included in the study [41]. The full potential of radiomics as a biomarker for immunotherapy in patients with urothelial cancer should be investigated and validated in a larger cohort and ideally in prospective randomized trials.

Conclusions

Urothelial cancer is among the most prevalent cancers worldwide and, as one of the most heterogeneous cancers known, needs a personalized approach for diagnosis and treatment. The development of radiomics to obtain quantitative features from imaging traits has shown the potential to aid diagnosis, guide therapy and monitor treatment response of urothelial cancer. To an extent, the applicability of radiomics in clinical practice depends on standardized data collection, evaluation criteria and reporting guidelines, and large-scale data sharing is fundamental for the full potential that radiomics represents. The research in radiogenomics of urothelial cancer and radiomics as a biomarker for immunotherapy has just started and needs further investigation. It is promising that radiomics-based decision support system for precision diagnosis and treatment for urothelial cancer will improve the quality of patient care in the near future.

Notes

Acknowledgements

This study was funded by the Fundamental Research Funds for the Central Universities (Grant No. 3332018022); Beijing Municipal Natural Science Foundation (Grant No. 7192176); Basic Scientific Research Program of Chinese Academy of Medical Sciences (Grant Nos. 2019PT320008 and 2018PT32003); and National Natural Science Foundation of China (Grant No. 81901742).

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

References

  1. 1.
    Miyazaki J, Nishiyama H. Epidemiology of urothelial carcinoma. Int J Urol. 2017;24(10):730–4.  https://doi.org/10.1111/iju.13376.CrossRefPubMedGoogle Scholar
  2. 2.
    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin. 2018;68(1):7–30.  https://doi.org/10.3322/caac.21442.CrossRefGoogle Scholar
  3. 3.
    Mahdavifar N, Ghoncheh M, Pakzad R, Momenimovahed Z, Salehiniya H. Epidemiology, incidence and mortality of bladder cancer and their relationship with the development index in the world. Asian Pac J Cancer Prev. 2016;17(1):381–6.CrossRefGoogle Scholar
  4. 4.
    Chen W, Zheng R, Baade PD, Zhang S, Zeng H, Bray F, et al. Cancer statistics in China, 2015. CA Cancer J Clin. 2016;66(2):115–32.  https://doi.org/10.3322/caac.21338.CrossRefPubMedGoogle Scholar
  5. 5.
    Advanced Bladder Cancer Meta-analysis C. Neoadjuvant chemotherapy in invasive bladder cancer: update of a systematic review and meta-analysis of individual patient data advanced bladder cancer (ABC) meta-analysis collaboration. Eur Urol. 2005;48(2):202–5.  https://doi.org/10.1016/j.eururo.2005.04.006.CrossRefGoogle Scholar
  6. 6.
    Koshkin VS, Grivas P. Emerging role of immunotherapy in advanced urothelial carcinoma. Curr Oncol Rep. 2018;20(6):48.  https://doi.org/10.1007/s11912-018-0693-y.CrossRefPubMedGoogle Scholar
  7. 7.
    Felsenstein KM, Theodorescu D. Precision medicine for urothelial bladder cancer: update on tumour genomics and immunotherapy. Nat Rev Urol. 2018;15(2):92–111.  https://doi.org/10.1038/nrurol.2017.179.CrossRefPubMedGoogle Scholar
  8. 8.
    Theodora Katsila ML, Patrinos GP, Aristotelis B, Dimitrios K. The new age of -omics in urothelial cancer—re-wording its diagnosis and treatment. EBioMedicine. 2018;28:43–50.  https://doi.org/10.1016/j.ebiom.2018.01.044.CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Aerts HJ, Velazquez ER, Leijenaar RT, Parmar C, Grossmann P, Carvalho S, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006.  https://doi.org/10.1038/ncomms5006.CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441–6.  https://doi.org/10.1016/j.ejca.2011.11.036.CrossRefPubMedPubMedCentralGoogle Scholar
  11. 11.
    Lambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14(12):749–62.  https://doi.org/10.1038/nrclinonc.2017.141.CrossRefPubMedGoogle Scholar
  12. 12.
    Neri E, Del Re M, Paiar F, Erba P, Cocuzza P, Regge D, et al. Radiomics and liquid biopsy in oncology: the holons of systems medicine. Insights Imaging. 2018;9(6):915–24.  https://doi.org/10.1007/s13244-018-0657-7.CrossRefPubMedPubMedCentralGoogle Scholar
  13. 13.
    Spiess PE, Agarwal N, Bangs R, Boorjian SA, Buyyounouski MK, Clark PE, et al. Bladder cancer, version 5.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Cancer Netw. 2017;15(10):1240–67.  https://doi.org/10.6004/jnccn.2017.0156.CrossRefGoogle Scholar
  14. 14.
    Zhang G-M-Y, Sun H, Shi B, Jin Z-Y, Xue H-D. Quantitative CT texture analysis for evaluating histologic grade of urothelial carcinoma. Abdom Radiol. 2016;42(2):561–8.  https://doi.org/10.1007/s00261-016-0897-2.CrossRefGoogle Scholar
  15. 15.
    Mammen S, Krishna S, Quon M, Shabana WM, Hakim SW, Flood TA, et al. Diagnostic accuracy of qualitative and quantitative computed tomography analysis for diagnosis of pathological grade and stage in upper tract urothelial cell carcinoma. J Comput Assist Tomogr. 2018;42(2):204–10.  https://doi.org/10.1097/RCT.0000000000000664.CrossRefPubMedGoogle Scholar
  16. 16.
    Zhang X, Xu X, Tian Q, Li B, Wu Y, Yang Z, et al. Radiomics assessment of bladder cancer grade using texture features from diffusion-weighted imaging. J Magn Reson Imaging. 2017;46(5):1281–8.  https://doi.org/10.1002/jmri.25669.CrossRefPubMedPubMedCentralGoogle Scholar
  17. 17.
    Wang H, Hu D, Yao H, Chen M, Li S, Chen H, et al. Radiomics analysis of multiparametric MRI for the preoperative evaluation of pathological grade in bladder cancer tumors. Eur Radiol. 2019.  https://doi.org/10.1007/s00330-019-06222-8.CrossRefPubMedPubMedCentralGoogle Scholar
  18. 18.
    Nishiyama H. Asia consensus statement on NCCN clinical practice guideline for bladder cancer. Jpn J Clin Oncol. 2018;48(1):3–6.  https://doi.org/10.1093/jjco/hyx130.CrossRefPubMedGoogle Scholar
  19. 19.
    Garapati SS, Hadjiiski L, Cha KH, Chan H-P, Caoili EM, Cohan RH, et al. Urinary bladder cancer staging in CT urography using machine learning. Med Phys. 2017;44(11):5814–23.  https://doi.org/10.1002/mp.12510.CrossRefPubMedPubMedCentralGoogle Scholar
  20. 20.
    Xu X, Zhang X, Tian Q, Wang H, Cui LB, Li S, et al. Quantitative identification of nonmuscle-invasive and muscle-invasive bladder carcinomas: a multiparametric MRI radiomics analysis. J Magn Reson Imaging. 2019;49(5):1489–98.  https://doi.org/10.1002/jmri.26327.CrossRefPubMedGoogle Scholar
  21. 21.
    Tong Y, Udupa JK, Wang C, Chen J, Venigalla S, Guzzo TJ, et al. Radiomics-guided therapy for bladder cancer: using an optimal biomarker approach to determine extent of bladder cancer invasion from t2-weighted magnetic resonance images. Adv Radiat Oncol. 2018;3(3):331–8.  https://doi.org/10.1016/j.adro.2018.04.011.CrossRefPubMedPubMedCentralGoogle Scholar
  22. 22.
    Lim CS, Tirumani S, van der Pol CB, Alessandrino F, Sonpavde GP, Silverman SG, et al. Use of quantitative T2-weighted and apparent diffusion coefficient texture features of bladder cancer and extravesical fat for local tumor staging after transurethral resection. Am J Roentgenol. 2019;212(5):1060–9.  https://doi.org/10.2214/ajr.18.20718.CrossRefGoogle Scholar
  23. 23.
    Abol-Enein H, Tilki D, Mosbah A, El-Baz M, Shokeir A, Nabeeh A, et al. Does the extent of lymphadenectomy in radical cystectomy for bladder cancer influence disease-free survival? A prospective single-center study. Eur Urol. 2011;60(3):572–7.  https://doi.org/10.1016/j.eururo.2011.05.062.CrossRefPubMedGoogle Scholar
  24. 24.
    Zehnder P, Studer UE, Skinner EC, Dorin RP, Cai J, Roth B, et al. Super extended versus extended pelvic lymph node dissection in patients undergoing radical cystectomy for bladder cancer: a comparative study. J Urol. 2011;186(4):1261–8.  https://doi.org/10.1016/j.juro.2011.06.004.CrossRefPubMedGoogle Scholar
  25. 25.
    Wu S, Zheng J, Li Y, Wu Z, Shi S, Huang M, et al. Development and validation of an MRI-based radiomics signature for the preoperative prediction of lymph node metastasis in bladder cancer. EBioMedicine. 2018;34:76–84.  https://doi.org/10.1016/j.ebiom.2018.07.029.CrossRefPubMedPubMedCentralGoogle Scholar
  26. 26.
    Wu S, Zheng J, Li Y, Yu H, Shi S, Xie W, et al. A Radiomics nomogram for the preoperative prediction of lymph node metastasis in bladder cancer. Clin Cancer Res. 2017;23(22):6904–11.  https://doi.org/10.1158/1078-0432.Ccr-17-1510.CrossRefPubMedGoogle Scholar
  27. 27.
    Soukup V, Capoun O, Cohen D, Hernandez V, Burger M, Comperat E, et al. Risk stratification tools and prognostic models in non-muscle-invasive bladder cancer: a Critical Assessment from the European Association of Urology Non-muscle-invasive Bladder Cancer Guidelines Panel. Eur Urol Focus. 2018.  https://doi.org/10.1016/j.euf.2018.11.005.CrossRefPubMedGoogle Scholar
  28. 28.
    Xu X, Wang H, Du P, Zhang F, Li S, Zhang Z, et al. A predictive nomogram for individualized recurrence stratification of bladder cancer using multiparametric MRI and clinical risk factors. J Magn Reson Imaging. 2019.  https://doi.org/10.1002/jmri.26749.CrossRefPubMedGoogle Scholar
  29. 29.
    Meeks JJ, Bellmunt J, Bochner BH, Clarke NW, Daneshmand S, Galsky MD, et al. A systematic review of neoadjuvant and adjuvant chemotherapy for muscle-invasive bladder cancer. Eur Urol. 2012;62(3):523–33.  https://doi.org/10.1016/j.eururo.2012.05.048.CrossRefPubMedGoogle Scholar
  30. 30.
    Cha KH, Hadjiiski L, Chan H-P, Weizer AZ, Alva A, Cohan RH, et al. Bladder cancer treatment response assessment in CT using radiomics with deep-learning. Sci Rep. 2017.  https://doi.org/10.1038/s41598-017-09315-w.CrossRefPubMedPubMedCentralGoogle Scholar
  31. 31.
    Vickers AJ. Prediction models: revolutionary in principle, but do they do more good than harm? J Clin Oncol. 2011;29(22):2951–2.  https://doi.org/10.1200/JCO.2011.36.1329.CrossRefPubMedGoogle Scholar
  32. 32.
    Rutman AM, Kuo MD. Radiogenomics: creating a link between molecular diagnostics and diagnostic imaging. Eur J Radiol. 2009;70(2):232–41.  https://doi.org/10.1016/j.ejrad.2009.01.050.CrossRefPubMedGoogle Scholar
  33. 33.
    Blaveri E, Brewer JL, Roydasgupta R, Fridlyand J, DeVries S, Koppie T, et al. Bladder cancer stage and outcome by array-based comparative genomic hybridization. Clin Cancer Res. 2005;11(19 Pt 1):7012–22.  https://doi.org/10.1158/1078-0432.CCR-05-0177.CrossRefPubMedGoogle Scholar
  34. 34.
    Blaveri E, Simko JP, Korkola JE, Brewer JL, Baehner F, Mehta K, et al. Bladder cancer outcome and subtype classification by gene expression. Clin Cancer Res. 2005;11(11):4044–55.  https://doi.org/10.1158/1078-0432.CCR-04-2409.CrossRefPubMedGoogle Scholar
  35. 35.
    Birkhahn M, Mitra AP, Williams AJ, Lam G, Ye W, Datar RH, et al. Predicting recurrence and progression of noninvasive papillary bladder cancer at initial presentation based on quantitative gene expression profiles. Eur Urol. 2010;57(1):12–20.  https://doi.org/10.1016/j.eururo.2009.09.013.CrossRefPubMedGoogle Scholar
  36. 36.
    Catto JWF, Abbod MF, Wild PJ, Linkens DA, Pilarsky C, Rehman I, et al. The application of artificial intelligence to microarray data: identification of a novel gene signature to identify bladder cancer progression. Eur Urol. 2010;57(3):398–406.  https://doi.org/10.1016/j.eururo.2009.10.029.CrossRefPubMedGoogle Scholar
  37. 37.
    Smith SC, Baras AS, Dancik G, Ru YB, Ding KF, Moskaluk CA, et al. A 20-gene model for molecular nodal staging of bladder cancer: development and prospective assessment. Lancet Oncol. 2011;12(2):137–43.  https://doi.org/10.1016/S1470-2045(10)70296-5.CrossRefPubMedPubMedCentralGoogle Scholar
  38. 38.
    Powles T, Smith K, Stenzl A, Bedke J. Immune checkpoint inhibition in metastatic urothelial cancer. Eur Urol. 2017;72(4):477–81.  https://doi.org/10.1016/j.eururo.2017.03.047.CrossRefPubMedGoogle Scholar
  39. 39.
    Rijnders M, de Wit R, Boormans JL, Lolkema MPJ, van der Veldt AAM. Systematic review of immune checkpoint inhibition in urological cancers. Eur Urol. 2017;72(3):411–23.  https://doi.org/10.1016/j.eururo.2017.06.012.CrossRefPubMedGoogle Scholar
  40. 40.
    Buder-Bakhaya K, Hassel JC. Biomarkers for Clinical Benefit of Immune Checkpoint Inhibitor Treatment-A Review From the Melanoma Perspective and Beyond. Front Immunol. 2018;9:1474.  https://doi.org/10.3389/fimmu.2018.01474.CrossRefPubMedPubMedCentralGoogle Scholar
  41. 41.
    Sun R, Limkin EJ, Vakalopoulou M, Dercle L, Champiat S, Han SR, et al. A radiomics approach to assess tumour-infiltrating CD8 cells and response to anti-PD-1 or anti-PD-L1 immunotherapy: an imaging biomarker, retrospective multicohort study. Lancet Oncol. 2018;19(9):1180–91.  https://doi.org/10.1016/s1470-2045(18)30413-3.CrossRefPubMedGoogle Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  1. 1.Department of Radiology, Peking Union Medical College HospitalPeking Union Medical College and Chinese Academy of Medical SciencesBeijingPeople’s Republic of China

Personalised recommendations