A simple MRI-derived correction substantially reduces bias in caliper-based tumor volume estimation in subcutaneous tumor models

Abstract

Accurate longitudinal quantification of tumor burden is critical in preclinical oncology studies. Calipers remain the standard tool, being fast, inexpensive and anesthesia-free, yet the conventional caliper equation systematically overestimates subcutaneous tumor volume, because caliper jaws enclose overlying skin and fatty tissue in external measurements. We quantified and corrected this bias in two breast cancer xenograft subtypes (HER2-positive and triple-negative), untreated and under neoadjuvant chemotherapy, using MRI-derived 3D volumetry as reference. In a derivation subset of 12 segmentations from 6 animals, spanning both subtypes and all imaging timepoints, principal component analysis of tumor-only and tumor-plus-skin masks yielded a constant tissue offset of 1.04 mm. This offset represents the summed non-tumor tissue on both tumor sides and is therefore subtracted once per caliper dimension ( \(\delta \) ~ 1 mm). The corrected equation was evaluated on the remaining 106 paired measurements (47 animals, 5 to 1700 mm 3 ). The conventional equation overestimated volume by 158% (< 100 mm 3 ) to 54% (> 400 mm 3 ) with significant proportional bias ( p < 0.001). The MRI-corrected equation reduced mean relative overestimation from 50 to 21% and eliminated proportional bias (p = 0.99). Variance ratios of 1.47–1.73 indicated improved statistical power and reduced animal numbers, directly supporting the 3Rs principle and doubling-time (DT) estimates improved by up to 24%. The MRI-corrected equation improves tumor volume accuracy while preserving the simplicity, speed, and low cost of caliper measurements.