Facebook Twitter YouTube

Feasibility and Potential Impact of Using CT Volume As a Predictor of Specimen Weight In a Subgroup of Patients with Low Risk Wilms Tumors Registered On COG Study AREN03B2

Sunday, October 21, 2012: 8:00 AM
Grand Ballroom B (Hilton Riverside)
Fernando A. Ferrer Jr., MD, FAAP1, Katherine W. Herbst, M.Sc.1, Conrad Fernandez, M.D.2, Geetika Khanna, M.D.3, Elizabeth Mullen, M.D.4, Jeffrey S. Dome, MD, PhD5, Robert C. Shamberger, MD6, Michael L. Ritchey, MD7 and Peter F. Ehrlich, MD, FACS8, (1)Division of Urology, Connecticut Children's Medical Center, Hartford, CT, (2)Department of Pediatrics, IWK Health Center, Halifax, NS, Canada, (3)Mallinckrodt Institute of Radiology, Washington University School of Medicine, (4)Dana-Farber/Children's Hospital Cancer Care, Boston, MA, (5)Center for Cancer and Blood Disorders, Children's National Medical Center, Washington, DC, (6)department of Surgery, Children's Hospital Boston, Harvard Medical School, Boston, MA, (7)Department of Urology, Phoenix Children's Hospital, Phoenix, AZ, (8)Division of Pediatric Surgery, University of Michigan, Ann Arbor, MI



Patients with favorable histology Wilms tumor (stage I, age ≤2 years, tumor < 550 grams) may not require therapy beyond primary nephrectomy.  Accurate estimation of specimen weight prior to nephrectomy allows for preoperative planning with the family regarding staging and informs the decision of whether or not to place a central venous access device for chemotherapy. The study's aims were to determine if a linear relationship existed between tumor weight and CT estimated tumor volume and, if so, to describe the accuracy of the slope intercept equation in estimating tumor specimen weight in a sample population.



On-study age, gender, tumor weight, and port placement were abstracted from 105 age ≤2 patients enrolled in COG study AREN03B2. One radiologist estimated tumor size using each subject's CT scan. Volume (length x width x height) was calculated for tumor mass, linear regression performed, and a slope-intercept equation calculated.  Equation estimated tumor weight was determined from the slope-intercept equation. Then using sample population weight, positive predictive value (PPV), or test precision, was calculated for both the equation and the actual outcome (line placement yes or no).



Gender was evenly distributed (50% male, 50% female). Median on-study age was 14 months (range >1-24 months). Median volume was 653 cc (range 4-2,252 cc), and median tumor weight was 409 grams (range 37-1,366 grams). Fifty-five ports were actually placed, twenty-nine potentially unnecessarily (tumor weight <550 grams), and six were not placed in patients requiring them (tumor weight >550).

Linear regression demonstrated a strong relationship between tumor volume and tumor weight, and a statistically significant slope (p<.001) (Fig. 1).  A slope-intercept equation for weight (W) based on volume (V) was calculated: W = 0.54(V) + 58.75 (C.I.  0.50-0.58(V) + 21.0296.47). PPV for the equation was 0.8437 vs 0.4727 actually seen in the sample population.



A strong relationship exists between tumor volume and weight, allowing for a viable slope-intercept equation. Precision of the equation was 84% compared to a precision of 47% in our sample population, granted not all factors influencing port placement could be abstracted. Twenty-nine ports (28%) potentially could have been avoided (Type I error), and six ports (6%) were not placed that may have been (Type II error). If applied to the study population, the equation's Type I and Type II errors would have been five cases (5%). If the slope-intercept equation had been used to determine need for port placement, twenty-four fewer ports may have been placed, while one child would not have been exposed to additional surgery for port placement. As the slope-intercept equation's accuracy is limited by the accuracy of radiological measurements, precision may vary due to inter-rater reliability.