Purpose: The clinical investigator must understand that errors in measuring tumors can greatly affect such clinical-trial end points as tumor response. We performed a prospective, controlled study of tumor measurements that compared circumferential measurements made with a loop planimeter with linear measurements made with a standard caliper.
Methods: Using a cross-over design, 76 clinical oncology fellows/junior oncology faculty members attending a Methods in Clinical Cancer Research Workshop sponsored by the American Association for Cancer Research and the American Society of Clinical Oncology measured five pulmonary nodule phantoms that ranged in size from 1.76 to 13.21 cm(2) and five surface nodule phantoms with sizes ranging from 2.3 to 12.9 cm(2). To perform these measurements, they used both a loop planimeter and a caliper. Forty-two and 40 participants repeated measurements 3 days later on pulmonary and surface nodules. Accuracy, reproducibility, and time efficiency were evaluated.
Results: The linear caliper measurements overestimated pulmonary nodule and surface nodule size by a median of 37% and 23%, respectively. Circumferential loop planimeter measurements overestimated pulmonary nodule size and surface nodule size by a median of 8% and 17%, respectively. Interobserver reproducibility for the planimeter was greater than that for the caliper, as evidenced by thinner measurement interquartile ranges. Furthermore, intraobserver reproducibility was higher for the planimeter, with its variability being only 31.4% and 25.5% as large as that of the caliper when measuring the pulmonary and surface nodules, respectively.
Conclusion: Circumferential measurements provide better accuracy, reproducibility, and speed in measuring both pulmonary and surface nodules than do perpendicular diameters.