Risk of Severe Radiation Morbidity in Radiotherapy Trials: A Critical Evaluation of the TAME Risk Classification System

Abstract

Reporting of treatment-related morbidity and complications in clinical trials are important. Several scoring systems have been suggested, and some of these have been validated. However, it has generally been difficult to summarize and compare the level of adverse effects between clinical trials or treatment strategies. To this end, the T-A-M End-results system has been suggested as a new and concise index for short-term toxicity (T), adverse effects (A), and complication mortality (M) (Pajak et al., ASCO 2005; Trotti et al., ASTRO 2005). The aim of the current study was to validate the TAME adverse event summary system in large clinical morbidity data sets from DAHANCA and IAEA. Individual morbidity data from a total of 2854 patients included in the following randomized trials were used: DAHANCA 6 and 7 (5 vs. 6 fractions, 1476 patients), IAEA-ACC (5 vs. 6 fractions, 900 patients), IAEA-MMC (RT ± MMC, 478 patients). Estimation of T and A scores were obtained by dividing the total number of Grade 3–4 events by the number of patients in the cohort. Four common predefined scores were used in all data sets; mucositis, skin reaction, fibrosis, and late edema. In addition, acute edema + atrophy were included for DAHANCA data, and pain + xerostomia for the IAEA data. For the three identical standard treatment arms (conventional radiotherapy only), estimated T + A scores varied from 0.55 + 0.27 (DAHANCA) to 0.67 + 0.02 (MMC) and 0.86 + 0.08 (ACC). For the accelerated treatments, estimated T + A scores varied from 0.77 + 0.27 (DAHANCA) to 1.04 + 0.05 (ACC). The T + A scores for RT + MMC were similar to the radiotherapy alone in the same trial (0.70 + 0.01). Mortality risk estimates varied from 0 in the MMC trial to 2.8% in DAHANCA. For comparison, the T-A-M for RTOG radiotherapy alone was 0.77 + 0.59 + 3%. The T score was reasonably stable for acute morbidity in identical treatments (0.55 to 0.86), and the T score was also able to pick up the intensification seen in accelerated radiotherapy. In contrast, the significant variation in A scores (from 0.02 to 0.27) for long-term morbidity—even between identical treatments—was of concern. Since the TAME system is nonactuarial, analysis of treatments with poor outcome or short follow-up (low number of events, patients dying or lost for follow-up) will result in a low A score, as seen in e.g. the IAEA studies. The present results indicate that the TAME system is useful only when comparing a defined set of endpoints from treatments in confined settings, and not across borders and cooperative groups. The TAME system is not an appropriate alternative to actuarial analysis of late radiotherapy morbidity.