ASCO GU 2017: Comparative effectiveness of tumor response assessment methods: Standard-of-care versus computer-assisted response evaluation.

Orlando, Florida USA (UroToday.com) Dr. Allen presented work from a large collaborative effort to produce a computerized tool that the aids in the assessment of radiographic response to targeted therapy in metastatic renal cell carcinoma (RCC). The current standard of care to evaluate metastatic tumor response to antiangiogenic agents is by using the RECIST 1.1 criteria. Unfortunately, this criteria underestimates tumor response and can miss progression in some cases. The Choi criteria were also developed that uses percent change in tumor length and attenuation to evaluate objective response. The MASS (Morphology, Attenuation, Size, and Structure) criteria can also be used to assess for objective response to treatment.

In this multi-institutional study, 20 patients were chosen from the sunitinib arm from a large randomized Phase III trial of metastatic RCC patients treated with sunitinib vs. interferon. After one cycle of sunitinib, CT scans were analyzed by 11 readers from 10 institutions along the RECIST, Choi, and MASS criteria. The authors developed a way to automate the input of measurement criteria in an effort to standardize and more effectively objectify radiographic changes after treatment. CARE (computer assisted response evaluation) is an automated system that combines some manual input of tumor data with a host of automatically acquired radiographic data via programmed data acquisition.

In this study, CARE was compared to standard of care, which was obtained by having the readers manually input all necessary measurements into a database. The authors looked evaluate the difference in measurement errors by the two techniques. They did this by identifying 15 common errors in measuring tumor response and refining the CARE program to eliminate those errors. They used a complex cross-over study design to ensure washout of bias related to reader and recall.

CARE resulted in fewer errors in the measurement of metastatic lymph node measurements, as well as improved measurements in attenuation and size change compared to manual measurement. Quite remarkably, CARE resulted in no errors with respect to RECIST, Choi, and MASS criteria measurements, whereas manual entry resulted in 10-25% error rates for assessment of these criteria. CARE was also twice as fast as manual measurement.

An ideal imaging marker should be able to be used as an early indicator for PFS and tumor response. It should be straightforward, applicable regardless of contrast-enhanced status, easily learned, reproducible, and usable across multiple institutions. CARE appears to fit most of these criteria.

Even though it is unclear how clinically meaningful the errors in measurements made by the manual entry arm actually are, objectifying measurements and reducing errors can only be a positive step. The more clearly we understand the response to treatment in metastatic RCC, the better we will be able to study and treat these patients in an individualized fashion. This is an important step forward to marry advanced automation with clinical diagnostics.

First author: Brian Allen, Duke University

Written By: Shreyas Joshi, MD, Fox Chase Cancer Center

at the 2017 Genitourinary Cancers Symposium - February 16 - 18, 2017 – Orlando, Florida USA

Login