The added value of Tumor Growth Rate (TGR) at RECIST evaluation in the era of targeted agents
The question of how to measure antitumor treatment activity in the era of targeted or novel immunologic agents is a true challenge. Because these agents exert more cytostatic than cytotoxic activity or, at least, induce limited tumor shrinkage and sometimes only tumor stabilization, questions have been raised about the usefulness of commonly accepted RECIST method for patients receiving such treatments. Namely, the granularity of RECIST 1.1 is insufficient to refine the response to treatments and to identify early the patients in whom treatments may have a beneficial effect, for two main reasons : (i) it does not take into account the tumor growth characteristics and notably its pre-treatment component ; (ii) it reduces the information with a qualitative assignment of the treatment effect into only four categories, thus classifying the majority of patients as stable disease, between arbitrary cutoffs on the continuous response scale: -30% for partial response (PR), +20% or occurrence of new lesions for progressive disease. As a result, the current version of RECIST may actually be hampering drug development and miss early non-responding patients, thus possibly putting these ones at toxicity risk without any clear benefit. In the recent years, different research groups and especially the Innovative Therapeutics and Early Drug Development Depatment at the Institut Gustave Roussy, a leading oncology research and treatment center in Europe, have reported on the potential value of tumor kinetics in phase I and phase III trials to better evaluate tumor response.
Based on the original work of Gomez-Roca et al ( Eur J Cancer, 2011 ; 2512-2516), Ferté and colleagues have shown that TGR in phase I patients is simple and provides clinically relevant information: (i) an early and subtle assessment of signs of antitumor activity; (ii) independent association with PFS; and (iii) it reveals drug-specific profiles, suggesting potential utility for guiding the further development of the investigational drugs (Clin Cancer Res; 20(1); 246–52). Responding to a comment of Dienstmann and Tabernero, the authors observed that the management of the new lesions, a major issue in the evaluation of drug efficacy, could not be taken into account by the TGR assessment, mainly because dimension measurements of new lesions are mostly lacking from RECIST. They acknowledged that the mandatory retrieval and assessment of pretreatment radiologic data is a challenge for the TGR mathod.
The potential inadequacies of the RECIST criteria appear also critical in phase III trials in some metastatic cancers for which registered molecular targeted agents are widely used. Ferté et al have have provided evidence that, in metastatic Renal Cell Cancer patients, TGR is independently associated with prognosis (PFS, OS), allows for a subtle and quantitative characterization of drug activity at the first evaluation, and reveals clear drug-specific profiles at progression (Eur Uro, 2014; 65 : 713-20). Notably, the authors demonstrated persistent activity of sorafenib at progression and an apparent flare-up effect after drug discontinuation, posing the crucial, and yet non solved, ethical and economical questions of how and when continuing treatment beyond (limited) progression in such settings.
Last but not least, S. Litière, on the behalf of the RECIST Committee, explored whether a more refined categorisation of tumor response and/or these components of progression, varying over time, can improve prediction of overall survival (OS) in the RECIST database (Eur J Cancer 2014;50 :1847–53). In short, hhe authors concluded that « Modelling target lesion tumour growth did not show a marked improvement in OS prediction over and above the other components : the appearance of new lesions and the progression of non-target disease » . However, Ferté et al have pointed out three majors limitations of the study (i) the definition of tumor growth used in this article as an increase in mm per week has no justification and does not correspond to the TGR commonly used to measure tumor growth (ii) they wonder whether conclusions on tumor growth can be extrapolated to all-comer metastatic cancer patients since these analyses were performed only a fraction (about 30%) of the entire set of patients, (iii) the original aim of RECIST criteria was not to predict prognosis but to provide a better assessment of the treatment effect. They believe that further evolution of the RECIST criteria should focus on improving the evaluation of antitumor effects of treatments (Clin Cancer Res. 2014; 20: 2497).
In conclusion, TGR provides useful clinical information for patients. It allows for an early and precise assessment of the tumor response by taking in account the pre-treatment tumor kinetics, is independently associated with outcome both in early clinical trials and in phase III setting, and the variation of TGR reveals drug-specific profiles, suggesting its potential use for the early assessment of drug activity. In practice, it only needs RECIST sums and CT-Scans, each patient considered as his own control, it is simple to compute and must be considered as a potential tool to help personalized medicine trials. In addition, taking into account that pre-treatment imaging is needed, a web tool is available to educate clinicians/ clinical researchers for its routine use in clinical practice.