The predictive value of treatment response using FDG PET performed on day 21 of chemoradiotherapy in patients with oesophageal squamous cell carcinoma. A prospective, multicentre study (RTEP3)

Abstract

Purpose

FDG PET has been suggested to have predictive value in the prognosis of oesophageal carcinoma. However, the retrospective studies reported in the literature have shown discordant results. Additionally, only four studies have evaluated FDG PET during chemoradiotherapy (CRT) in patients with different histological lesions. The purpose of this study was to investigate the predictive value of FDG PET performed early during CRT (on day 21) in a population of patients with oesophageal squamous cell carcinoma.

Methods

Included in this prospective study were 57 patients with a histological diagnosis of squamous cell carcinoma of the oesophagus. Of these 57 patients, 48 (84 %) were evaluated (aged 63 ± 11 years; 44 men, 4 women). Each patient underwent FDG PET (4.5 MBq/kg) before CRT, according to the Herskovic protocol (t0; PET₁) and on day 21 ± 3 from the start of CRT (d21; PET₂). The response assessment included a clinical examination, CT scan or FDG PET and histological analysis 3 months and 1 year after PET₁. The patients were classified as showing a complete response (CR) or a noncomplete response. A quantitative analysis was carried out for PET₁ and PET₂ using the following parameters: SUVmax, SUVmean (with SUVmean₄₀ as the 3-D volume at an SUVmax threshold of 40 % and SUVmean_p as that defined by a physician), tumour volume (TV, with TV₄₀ defined as the TV at 40 % of SUVmax, and TV_p as that defined by a physician); and the total lesion glycolysis (TLG, SUVmean × TV, with TLG₄₀ defined as the TLG at 40 % of SUVmax, and TLG_p as that defined by a physician). The differences in responses at 3 months and 1 year between PET₁ (t0) and PET₂ (d21) were assessed in terms of variations in SUV, TV and TLG using a repeated measures of variance (ANOVA).

Results

SUVmax, SUVmean and TLG decreased significantly between PET₁ (t0) and PET₂ (d21; p < 0.0001). The TV significantly decreased only when assessed as TV_p (p = 0.02); TV₄₀ did not decrease significantly. With respect to the predictive value of PET₁, only TV_{40_1} and TV_{p_1} values, and therefore TLG_{40_1} and TLG_{p_1}, but not the SUV values, were significantly lower in patients with CR at 3 months. SUVmax₁, TV_{p_1} and TLG_{p_1} were significantly lower in patients with CR at 1 year. With respect to the predictive value of PET₂, only TV_{40_2} and TV_{p_2} values, and therefore TLG_{40_2} and TLG_{p_2}, but not the SUV values, were significantly lower in patients with CR at 3 months. None of the PET₂ parameters had significant value in predicting patient outcome at 1 year. The changes in SUVmax, TV₄₀, TV_p, TLG₄₀ and TLG_p between PET₁ and PET₂ had no relationship to patient outcome at 3 months or 1 year.

Conclusion

This prospective, multicentre study performed in a selected population of patients with oesophageal squamous cell cancer demonstrates that the parameters derived from baseline PET₁ are good predictors of response to CRT. Specifically, a high TV and TLG are associated with a poor response to CRT at 3 months and 1 year, and a high SUVmax is associated with a poor response to CRT at 1 year. FDG PET performed during CRT on day 21 appears to have less clinical relevance. However, patients with a large functional TV on day 21 of CRT have a poor clinical outcome (ClinicalTrials.gov NCT 00934505).