Author's Reply: Unrecognized Biases and Validation Gaps in TraceOrg for Automated Autosomal Dominant Polycystic Kidney Disease Volumetry

We thank Sheu and Tsai for their Letter on our JASN article1 and for recognizing that “TraceOrg represents a major advance in automated [autosomal dominant polycystic kidney disease] volumetry.”2 Yes, there is sequence bias,3 and a major strength of TraceOrg is that it works on most magnetic resonance imaging (MRI) sequences, which makes averaging possible for mitigating this bias.4 Measuring total kidney volume (TKV) on multiple MRI sequences also makes exclusion or correction of outlier values possible, which can further improve measurement reproducibility. In our practice, autosomal dominant polycystic kidney disease abdominal MRI examinations always acquire a spectrum of pulse sequences, including T1, T2, and steady-state free precession in multiple planes, allowing us to estimate TKV as the average of five or six sequences. We agree that a patient on the borderline between two Mayo Imaging Classifications could switch classifications with a minor TKV error. For this reason, TraceOrg reports show the TKV on the Mayo Imaging Classification plot, making it is easy to identify these borderline cases that can be affected by a minor error. Plans are underway for additional validations as well as strategies for accurately segmenting smaller cysts. We are also working toward deployment of longitudinal TKV assessment to show TKV growth rates on TraceOrg for examining the effects of therapeutic interventions, such as tolvaptan treatment.5