Patients and treatment for LRRC
A total of 26 males and 14 females were enrolled, with a median age of 68.5 years (36-81). All patients received radical resection with curative intent after pre-operative CRT. Among the 40 patients, 22 underwent total pelvic exenteration, while 10 patients underwent abdominoperineal resection, and 6 underwent low anterior resection. Sacral bone resection was concomitantly performed in 21 patients (52.5%) to secure a negative surgical margin. Resection was considered to be curative (R0 resection) in 36 patients and microscopically incomplete (R1) in 4 patients. According to Mandard’s criteria 11, 17 of the 40 patients were classified as responders (TRG1-TRG2) and 23 patients were classified as nonresponders (TRG3-TRG5). Clinical characteristics of the patients are described in Table 1-a.
Response assessment by PET-CT
Pre-SUVs ranged from 1.0 to 26.3 (8.2±6.1, median 5.9). Post-SUVs ranged from 0.0 to 18.7 (3.81±4.0, median 3.0), and Post-SUVs were found to be significantly lower than the pre-SUVs (P < 0.0001). The mean ⊿SUV was 4.4±4.8 (range, -1.3～22.5, median 2.8). The mean DR was 48.1±30.3% (range, -30.2～100, median 44.7%) (Table 1-b).
Twelve patients underwent PET-CT after the initial 2 weeks of CRT. In this pilot phase study, Post-SUVs were significantly lower than Pre-SUVs (P = 0.0442). However, Mid-SUVs were not significantly different from Pre-SUVs and Post-SUVs (P = 0.215 and 0.4068, respectively) (Supplementary Figure 1). Subsequent to the pilot phase, Mid-SUVs were no longer examined because this time point appeared to be too soon to assess the effect of CRT.
Post-SUVs in the responder (TRG1-2, n = 17) group were significantly lower than those in the nonresponder (TGR3-4, n = 23) group (2.0±1.7 vs. 5.1±3.9, P = 0.0038). DR (%) was significantly higher in the responder group than in the nonresponders (65.3±32.3 vs. 35.4±21.7, P = 0.0012). Pre-SUVs and ⊿SUV were not significantly different between the responder and the nonresponder groups (P = 0.5103 and 0.2502, respectively) (Table 2).
Response assessment by CT and MRI
Table 3 shows the relationship between the CT/MRI response evaluation and pathological response grade. Most of the patients were classified in the nonresponder group by CT or MRI evaluation (26/40: 65.0% and 28/40: 70.0%, respectively), though almost half of these nonresponders were classified as responders using the pathological criteria (11/26: 42.3% and 11/28: 39.2%, respectively). There was no significant correlation between histological response classification and CT/MRI response classification (P > 0.9999 and P > 0.9999, respectively). One other observation of note is that CT or MRI could not evaluate LRRC lesion in some cases (9/40: 22.5%, 3/40: 7.5%).
Local re-recurrence free survival
Patients’ age, gender, primary lesion-related factors, and locally recurrent lesion-related factors were analyzed by univariate analysis. The extent of resection and post-SUV were significant prognostic factors for local re-recurrence-free survival (P = 0.0299 and 0.0102, respectively) (Figure 1-A). When analyzed with these statistically significant parameters by univariate analysis, multivariate Cox regression analysis revealed that post-SUV was a significant prognostic factor (P = 0.0383) (Table 4).
In univariate analysis, extent of resection, pathologic tumor response, and post-SUV were significantly associated with prognostic factors for overall survival (P = 0.0035, 0.0411, and 0.0009, respectively) (Figure 1-B). Multivariate Cox regression analysis demonstrated that Post-SUV was a significant prognostic factor for overall survival (P = 0.0195) (Table 5).