Endoscopic management of upper tract urothelial carcinoma: Improved prediction of invasive cancer using a ureteroscopic scoring model

A Novel Preoperative Prediction Model Based on Deep Learning to Predict Neoplasm T Staging and Grading in Patients with Upper Tract Urothelial Carcinoma

Objectives: To create a novel preoperative prediction model based on a deep learning algorithm to predict neoplasm T staging and grading in patients with upper tract urothelial carcinoma (UTUC). Methods: We performed a retrospective cohort study of patients diagnosed with UTUC between 2001 and 2012 at our institution. Five deep learning algorithms (CGRU, BiGRU, CNN-BiGRU, CBiLSTM, and CNN-BiLSTM) were used to develop a preoperative prediction model for neoplasm T staging and grading. The Matthews correlation coefficient (MMC) and the receiver-operating characteristic curve with the area under the curve (AUC) were used to evaluate the performance of each prediction model. Results: The clinical data of a total of 884 patients with pathologically confirmed UTUC were collected. The T-staging prediction model based on CNN-BiGRU achieved the best performance, and the MMC and AUC were 0.598 (0.592–0.604) and 0.760 (0.755–0.765), respectively. The grading prediction model [1973 World Health Organization (WHO) grading system] based on CNN-BiGRU achieved the best performance, and the MMC and AUC were 0.612 (0.609–0.615) and 0.804 (0.801–0.807), respectively. The grading prediction model [2004 WHO grading system] based on BiGRU achieved the best performance, and the MMC and AUC were 0.621 (0.616–0.626) and 0.824 (0.819–0.829), respectively. Conclusions: We developed an accurate UTUC preoperative prediction model to predict neoplasm T staging and grading based on deep learning algorithms, which will help urologists to make appropriate treatment decisions in the early stage.

Phase II multi-center trial of optical coherence tomography as an adjunct to white light cystoscopy for intravesical real time imaging and staging of bladder cancer

BACKGROUND

Optical coherence tomography (OCT) is a novel imaging modality that provides microstructural information of different tissue layers using near-infrared light. This prospective, multicenter phase II trial aimed to assess the accuracy of OCT-assisted cystoscopy for bladder tumor staging.

METHODS

Patients with primary or recurrent bladder tumors (Ta,T1) identified by outpatient cystoscopy were included. The primary objective was to assess the accuracy and positive predictive value of for determining tumor stage ≥T1 correlated by histopathology. 72 suspicious lesions from 63 patients were eligible to analyze in the study. All suspected lesions were evaluated with conventional cystoscopy, interpreted in real-time using OCT, and then resected. All results were compared to pathology. A total of 363 OCT images of tumor and normal mucosa in 25 patients were obtained to evaluate diagnostic efficacy of the computer-aided texture analysis algorithm.

RESULTS

Sensitivity and specificity for predicting invasive tumors (≥ T1, n = 17) were 58.8% and 92.7% for cystoscopy, 64.7% and 100% for OCT-assisted cystoscopy, respectively. Accuracy of cystoscopy and OCT-assisted cystoscopy for predicting invasive tumor was 84.7% and 91.7% (P = 0.063), respectively. Cystoscopy and OCT-assisted cystoscopy correctly predicted T stage in 52/72 and 59/72 cases, respectively (P = 0.016). Cystoscopy missed 2 more invasive tumors than OCT-assisted cystoscopy. Cystoscopy (14.3%, 1/7) and OCT-assisted cystoscopy (28.6%, 2/7) showed relatively low sensitivity in detecting muscle invasion. Computer aided texture analysis demonstrated 75.1% sensitivity, 64.0% specificity, and 74.4% accuracy for differentiating tumor and normal urothelium.

CONCLUSION

OCT-assisted cystoscopy is a real time noninvasive and simple procedure that enhanced the accuracy of staging bladder tumors and prediction of any tumor invasion. Though the study did not meet the prespecified primary endpoint, OCT imaging is a promising adjunct to cystoscopy that may supplement intraoperative decision-making during transurethral resection of bladder tumors and additional prospective studies are warranted.

Novel Classification for Upper Tract Urothelial Carcinoma to Better Risk-stratify Patients Eligible for Kidney-sparing Strategies: An International Collaborative Study

BACKGROUND

The European Association of Urology risk stratification dichotomizes patients with upper tract urothelial carcinoma (UTUC) into two risk categories.

OBJECTIVE

To evaluate the predictive value of a new classification to better risk stratify patients eligible for kidney-sparing surgery (KSS).

DESIGN, SETTING, AND PARTICIPANTS

This was a retrospective study including 1214 patients from 21 centers who underwent ureterorenoscopy (URS) with biopsy followed by radical nephroureterectomy (RNU) for nonmetastatic UTUC between 2000 and 2017.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

A multivariate logistic regression analysis identified predictors of muscle invasion (≥pT2) at RNU. The Youden index was used to identify cutoff points.

RESULTS AND LIMITATIONS

A total of 811 patients (67%) were male and the median age was 71 yr (interquartile range 63-77). The presence of non-organ-confined disease on preoperative imaging (p < 0.0001), sessile tumor (p < 0.0001), hydronephrosis (p = 0.0003), high-grade cytology (p = 0.0043), or biopsy (p = 0.0174) and higher age at diagnosis (p = 0.029) were independently associated with ≥pT2 at RNU. Tumor size was significantly associated with ≥pT2 disease only in univariate analysis with a cutoff of 2 cm. Tumor size and all significant categorical variables defined the high-risk category. Tumor multifocality and a history of radical cystectomy help to dichotomize between low-risk and intermediate-risk categories. The odds ratio for muscle invasion were 5.5 (95% confidence interval [CI] 1.3-24.0; p = 0.023) for intermediate risk versus low risk, and 12.7 (95% CI 3.0-54.5; p = 0.0006) for high risk versus low risk. Limitations include the retrospective design and selection bias (all patients underwent RNU).

CONCLUSIONS

Patients with low-risk UTUC represent ideal candidates for KSS, while some patients with intermediate-risk UTUC may also be considered. This classification needs further prospective validation and may help stratification in clinical trial design.

PATIENT SUMMARY

We investigated factors predicting stage 2 or greater cancer of the upper urinary tract at the time of surgery for ureter and kidney removal and designed a new risk stratification. Patients with low or intermediate risk may be eligible for kidney-sparing surgery with close follow-up. Our classification scheme needs further validation based on cancer outcomes.

Cryobiopsy in the Upper Urinary Tract: Preclinical Evaluation of a Novel Device

OBJECTIVE

To develop a novel device for cryobiopsy of the upper urinary tract (UUT) and to evaluate its feasibility in a standardized preclinical setting.

MATERIALS AND METHODS

Flexible cryoprobes (diameter 0.9 mm; cooling agent CO2) were developed and used to extract biopsies in porcine UUTs. Cryosamples obtained by ureterorenoscopy were systematically compared with biopsy specimens obtained with standard of care devices in terms of physical characteristics (deflection angle and irrigation flow rates) and histologic criteria (assessability).

RESULTS

Irrigation flow rates were significantly higher with introduced BIGopsy (2.8 ± 0.1) compared with standard forceps (0.94 ± 0.06; P < .001) and cryoprobe (1.1 ± 0.1; P < .001). Angular deflection was significantly reduced by the inserted cryoprobe (130.7° ± 1.2° vs 166.9° ± 1.1° [BIGopsy] or 161.4° ± 1.9° [standard forceps]; both P < .001). Significantly larger UUT tissue samples were obtained by the cryoprobe (mean specimen area 7.5 ± 2.5 vs 4.6 ± 2.5 mm² [BIGopsy] or 1.4 ± 1.4 mm² [standard forceps]; both P < .001). No crush artifacts were observed in cryosamples. Superior histologic assessability scores were achieved in samples obtained by the cryoprobe (mean 2.8 ± 0.8) and BIGopsy (2.3 ± 1.9) when compared with standard forceps (0.4 ± 0.9; P < .001).

CONCLUSION

Cryobiopsy in the UUT is feasible and represents a viable new option to improve the diagnostic accuracy of histopathologic evaluation. Larger and more representative tissue samples can be obtained using a cryoprobe and artifacts may be avoided. Further optimization of the probe will reduce possible restrictions of ureterorenoscopy handling when the device is inserted.

Multi-institutional Evaluation of Upper Urinary Tract Biopsy Using Backloaded Cup Biopsy Forceps, a Nitinol Basket, and Standard Cup Biopsy Forceps

OBJECTIVE

To compare the performance of 3 contemporary ureteroscopic biopsy devices for the histopathologic diagnosis of upper tract urothelial carcinoma (UTUC).

METHODS

We retrospectively reviewed 145 patients who underwent 182 urothelial biopsies using 2.4F backloaded cup biopsy forceps, a nitinol basket, or 3F standard cup biopsy forceps at 3 tertiary academic centers between 2011 and 2016. Experienced genitourinary pathologists provided an assessment of each specimen without knowledge of the device used for biopsy. For patients who underwent nephroureterectomy without neoadjuvant chemotherapy within 3 months of biopsy-proven UTUC diagnosis, the biopsy grade was compared with both the grade and stage of the surgical specimen.

RESULTS

Biopsy utilization varied among the 3 institutions (P <.0001). Significant variabilities in specimen size (P = .001), the presence of intact urothelium (P = .008), and crush artifact (P = .028) were found among the biopsy devices. The quality of specimens from backloaded cup forceps was rated similarly to the nitinol basket (P >.05) and was favored over standard cup forceps specimens. Grade concordance was not affected by specimen size (P >.05), morphology (P >.1), or location (P >.5). No difference existed among the devices in the rate of acquiring a grade concordant biopsy; however, the backloaded cup forceps provided concordant biopsies that could be distinguished as low- and high-grade (P = .02).

CONCLUSION

The backloaded cup forceps and nitinol basket obtained a higher quality urothelial specimen compared with standard cup forceps. Ureteroscopic biopsy device selection did not significantly impact the accuracy of the histologic diagnosis of UTUC.