High Resolution Imaging for Early and Better Detection of Bladder Cancer
Agency: Cancer Prevention Research Institute of Texas
Collaborators: Rebecca Richards-Kortum (Rice), Richard Schwarz (Rice), Nadeem Dhanani (UT Health Science Center-Houston)
An estimated 563,640 people are living with bladder cancer in the US, where bladder cancer accounts for 4.4% of all new cancer cases and 2.6% of all cancer deaths. The five-year survival rate for patients diagnosed with stage 1 bladder cancer is 88%, but declines to 15% for patients diagnosed with stage 4 cancer. Most patients diagnosed with early bladder cancer can be treated with organ-sparing therapy. However, bladder cancer has the highest rate of recurrence of any cancer, and as a result, is the most costly cancer to treat.
Surveillance for bladder cancer includes endoscopic inspection of the bladder via the urethra using a cystoscope. White light cystoscopy has a number of limitations; many bladder cancers present as small, subtle, flat lesions that are difficult to differentiate from benign changes due to infection or inflammation. Patients who have been successfully treated for bladder cancer are recommended to undergo surveillance endoscopy every 3-6 months because of the high risk of recurrence. Unfortunately, recurrent lesions are more likely to be flat and difficult to detect. When a suspicious lesion is visualized at endoscopy, a biopsy is performed to establish a diagnosis of cancer. Recent reports suggest that more than 1/3 of biopsies performed during endoscopy show only benign changes and are unnecessary. These unnecessary biopsies add increased risks and anxiety for the patient and additional costs to the healthcare system.
Thus, there is an important need to improve the ability of white light cystoscopy to: (1) identify bladder lesions with high sensitivity, and (2) characterize them as benign or malignant with high specificity. Improved endoscopy techniques such as narrow band imaging and fluorescence cystoscopy have shown promise to improve the ability to identify malignant bladder lesions with high sensitivity; unfortunately, they have even lower specificity than white light cystoscopy. Here, we propose to develop and evaluate a new high-resolution micro-endoscope (HRME) that can be used during white light, narrow band or fluorescence endoscopy to improve the ability to characterize bladder lesions in real time as benign or malignant with higher specificity. The HRME provides images with sub-cellular resolution, revealing morphologic detail that is normally only available from a biopsy and histology. We hypothesize that high resolution imaging of visually suspicious lesions in the bladder may improve the ability to discriminate flat, early neoplastic lesions from benign lesions, reducing the number of unnecessary biopsies performed. The goal of this proposal is to develop, optimize and validate the HRME to improve early detection of bladder cancer in high risk patients. In Aim 1, we will collect HRME images of suspicious bladder lesions from 50 patients at high risk for bladder cancer who are scheduled to undergo standard of care cystoscopy. Using histologic diagnosis as the gold standard, we will develop a classification algorithm to distinguish malignant from benign lesions. In Aim 2, we will develop a second generation HRME system that incorporates real time image analysis and diagnosis based on the algorithm developed in Aim 1. Finally, in Aim 3 we will prospectively evaluate the performance of the HRME in a follow up study of 50 patients at high risk for bladder cancer who are scheduled to undergo standard of care cystoscopy, using histology as the gold standard.
When used in conjunction with new endoscopic imaging modalities, HRME imaging has the potential to dramatically improve early detection for bladder cancer. Patients diagnosed at an earlier stage have better outcomes with less expensive and less morbid treatment. Moreover, improved tools for continued surveillance are essential if we are to prevent recurrence of bladder cancer in patients who are at high risk. The low cost ($1500) of the HRME and the ability to automate image analysis support the potential for rapid translation of the technology once accuracy is established.