Ali Guermazi MD, Presenter: Stockholder, Synarc Inc

Souhila Ounadjela MD, Abstract Co-Author: Nothing to Disclose

Nadia Bouzegaou MD, Abstract Co-Author: Nothing to Disclose

Agnes Coulon MD, Abstract Co-Author: Nothing to Disclose

Radhika Sivaramakrishna PhD, Abstract Co-Author: Nothing to Disclose

Souhil Zaim MD, Abstract Co-Author: Nothing to Disclose

To report the error rate when a paper CRF is used in the tumor response assessment of patients treated for cancer using RECIST.

165 patients with metastatic renal cell carcinoma were included in a prospective multicenter clinical trial. Standardized CT was performed every 8 weeks. An independent single evaluation using RECIST was performed by 2 trained onco-radiologists. For each time point, the longest diameter of up to 10 target lesions were performed manually and their sum was calculated. Similarly, non-target lesions and any new lesion were identified and recorded for each time point. The percentage of change of the sum of the longest diameters and the changes in the appearance of non-target lesions as well as the appearance of new lesions were integrated to determine an overall response for each time point. The calculations and determination of overall response were recorded manually by the radiologist using a paper CRF. Subsequently, an independent radiologist reviewed all CRFs twice for the accuracy of calculations and the correct use of the response algorithm as defined by RECIST.

There were 47 errors identified in 34 (20.6%) of the 165 patients: 10 were calculation errors, 21 were data capture typing errors, and 16 were errors in response determination. Of the 47 errors, 26 resulted in an inaccurate determination of patient response according to RECIST in 16 (9.7%) patients: 11 errors were in the overall response, 12 errors in the best response, 2 errors in the date of best response, and 1 error in reporting the date of progressive disease. There were double errors in 10 patients and single errors in the remaining 6 patients. Of the 47 errors, 21 had no consequence on patient response determination.

Data capture and manual calculations using paper CRF for the assessment of tumor response are error prone and may compromise patient status in reporting endpoint. The use of integrated software algorithms combined with electronic data capture overcomes this risk.

Manual calculation and reporting of tumor response assessment using measurement-based criteria is error prone and calls for the use of computer-assisted integrated response assessment algorithms.

Guermazi, A, Ounadjela, S, Bouzegaou, N, Coulon, A, Sivaramakrishna, R, Zaim, S, Paper Case Report Form (CRF) Assessment Using Response Evaluation Criteria in Solid Tumor (RECIST) in Patients Treated for Cancer.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5014090.html