Tumor Volume Growth as Surrogate Endpoint in IDH-mt Glioma—Response

Historically the choice of a modeling framework for the growth of brain tumor volume has been disputed, lacking a clear consensus (1–4). In our recent manuscript in Clinical Cancer Research (1), we explored both linear and exponential volume growth models and observed an excellent fit for each, with a marginal preference for the exponential model (1). Dr. Mandonnet's suggestion to consider linear modeling of tumor diameter over time is well-taken from his response to our manuscript as well as from his prior studies on glioma growth (3–5).In addressing Dr. Mandonnet's inquiry, we characterized the growth of tumor diameter over time by performing linear mixed modeling. We incorporated an interaction term for 1p19q status to assess the differences in diameter growth between patients with intact versus codeleted tumors. Analyses were performed in SAS v9.4 (Cary, NC).Across our entire cohort, we observed an estimated diameter growth rate of 2.2 mm per year [95% confidence interval (CI), 1.9–2.5 mm]. Consistent with our original findings, there was no discernible difference in diameter growth rates by 1p19q status. Patients with intact tumors had an estimated diameter growth rate of 2.4 mm per year (95% CI, 1.9–3.0 mm), whereas patients with codeleted tumors had an estimated diameter growth rate of 2.1 mm per year (95% CI, 1.8–2.4 mm; P value = 0.25).Dr. Mandonnet and we have recognized the advantages and shortcomings of modeling diameter growth (1, 4). Although the approach is straightforward, it tends to overestimate volume, especially in cases with tumor contour irregularity. The cubic and exponential models share the characteristics that they are convex and upward sloping functions. Therefore, in the presence of noise, distinguishing the two may be challenging or impossible. Regardless of the modeling framework, our findings did not demonstrate a statistically significant difference based on 1p19q status in postoperative glioma growth rates; rather only a trend towards faster growth in 1p19q non-codeleted gliomas. We thank Dr. Mandonnet for his interest in our work and his modeling suggestion, and we wholeheartedly agree on the importance of integrating tumor kinetics into the management of patients with IDH-mutated glioma. We believe our work advances this mission by characterizing postoperative growth of these tumors, and, importantly, incorporating molecular genetics.See the original Letter to the Editor, p. 638A. Bhatia reports serving on the advisory board for Servier outside the submitted work. A.S. Reiner reports grants from NIH during the conduct of the study. I.K. Mellinghoff reports serving as a consultant for 501c3 Global Coalition for Adaptive Research; honoraria from the Doris Duke Charitable Foundation; serving on advisory board for Black Diamond Therapeutics, Roche Therapeutics, Prelude Therapeutics Incorporated, Voyager Therapeutics; research support from Erasca Therapeutics, Servier Pharmaceuticals LLC, Kazia Therapeutics, Vigeo Therapeutics, and Samus Therapeutics, Inc. R.J. Young reports personal fees from ICON plc, NordicNeuroImaging; and personal fees from Olea Sphere outside the submitted work. No disclosures were reported by the other authors.