**9. Recurrence & survival**

With bi-modality regimes the local recurrence rates were reported to be above 70% [10, 13]. Despite the advent in treatment regimes, local recurrence still occurs in about 40% of the patients [43]; it is expected that local recurrence rate is higher in patients with T4 disease because complete resection can be achieved in less than half of the patients with c-T4 disease [17]. More specifically [44] complete resection rate was achieved in only 64% of tumour stage T3 and nodal stage N 0 and 39% of T4N0 tumours. It is apparent however, that loco-regional relapse is predominant in R1-2 resections, whereas distant recurrence is frequent in R0 resections.

One would expect that a shift in the trend of clinical recurrences towards distant metastasis is to be currently expected because of the fact that tri-modality treatment facilitates better R0 resection. As per Pourel et al [15] the most frequent site of relapse was distant metastasis in 66% of the patients, (mainly brain) with the loco-regional recurrence rate of 18%. Likewise Kwong et al [18] reported brain metastasis in 25% and local recurrence rate in 19% of the cases. A small series that had bi-modality treatment however had an incidence of loco-regional recurrence of 17.2% [14].

Survival has been extensively reviewed by Attar et al [45]. Overall survival at 5 years after surgery was 46% for T3N0, 13% for T4N0, and 0% for lesions with N2 disease [44]. Particularly noteworthy [17]was the reproducibility of the favourable survival data, with a 5-year overall survival rate of 44% in the United States trial (SWOG) and 56% in JCOG trial, which were clearly superior to the historical value of 30%.

There is wide variability in overall 5-year survival rates reported in larger series,[19, 24, 26, 39, 41, 44, 46] with figures ranging 10 to 35% probably because of the heterogeneity in studied populations, operative techniques, and preoperative and postoperative treatments. Such heterogeneity is probably responsible for the difference in the percentage of T3 and T4 tumours as well as in the rates of complete resection. Comparison of long-term results of different studies is difficult also for the frequent lack of information about survival according to the pathologic stage.
