Comparative assessment of different radiological criteria to identify paradoxical hyperprogression (HPD) to IO drugs

• Published in: Journal of clinical oncology Vol. 38; no. 15_suppl; p. e15229
• Main Authors: Gomes de Morais, Ana Luiza, Cardenas, Jose M, de Miguel Luken, Maria Jose, Boni, Valentina, Moreno, Irene, Ao, Gerileto, Liu, Run Han, de Hoyos, Fernando Bergaz, Cubillo, Antonio, Calvo, Emiliano
• Format: Journal Article
• Language: English
• Published: 20-05-2020
• ISSN: 0732-183X; 1527-7755
• DOI: 10.1200/JCO.2020.38.15_suppl.e15229

Abstract only e15229 Background: HPD is a new pattern of response consisting of accelerated tumor growth due to immunotherapy (5%-20% of pts on IO drugs). The tumor growth rate (TGR, Champiat et al.) estimates the differential increase in tumor volume over time, pre- and on IO drugs, to assess for HPD, but it is not easily calculated in practice. A comparison of the different methods to identify HPD is missing. Methods: Retrospective study of 182 consecutive pts treated in Phase 1 studies of IO drugs at START Madrid-CIOCC in 2017-8, comparing 3 HPD measurement criteria (Champiat, Matos -which does not use pre-baseline CT scans-, and Saâda-Bouzid). Cohen’s Kappa index was calculated as a measure of the agreement between the 3 methods, being those values closer to 1 the more concordant ones. Overall survival (OS) rates were calculated by Kaplan-Meier (p < 0.05 to be significant). Results: 99 (54%) of 182 pts had progressive disease at cycle 1 of treatment and in 62 (34%) pre-baseline CT scans were available to calculate TGR. The Champiat method identified 18 (9.9%) pts with HPD. Of 61 cases validated to comparison, the Matos criteria labeled 27 pts (14.8% of 182) with HPD, of whom only 10 pts coincided with those identified by Champiat with a low agreement (Kappa: 0.140, p:0.251). No significant differences in OS between pts with non-HPD vs HPD by Matos criteria were seen (p = 0.16). With Saâda-Bouzid method, of 59 cases validated to comparison, 17 pts (9.3% of 182) were labeled with HPD and 13 of them coincided with those identified by Champiat method with a high agreement (Kappa:0.706, p:0.000). Differences in OS between non-HPD vs HPD pts were statistically significant (p = 0.038). Conclusions: HPD might have a detrimental effect to 10% of pts on IO drugs, also in our series. Every effort should be done to obtain pre-baseline CT scans to identify HPD response to IO drugs, based on differential TGR. The use of Saâda-Bouzid method is preferred due to its practical application and its correlation with survival. [Table: see text]