br Fig KaplaneMeier curve of progression free survival and
Fig 2. KaplaneMeier curve of progression-free survival and liver-specific progression-free survival after selective internal Bafilomycin-A1 therapy in patients with metastatic colorectal cancer. Ninety-five per cent confidence intervals are shaded; numbers at risk at 3-month intervals are displayed.
studies of unresectable, chemotherapy-refractory patients with CRC liver metastases treated with SIRT, which identified 23 studies (one RCT, two retrospective comparative studies and several single-arm observational studies) reporting
overall survival (Supplementary Table S1). For 2517 patients in these studies, the pooled weighted overall survival esti-mate was 9.6 months (95% confidence interval 8.9e10.4; range 6.0e12.7 months). The patients included in our study had similar performance status and a similar rate of extra-hepatic disease (around 40% of patients). Ongoing clinical studies are listed in Supplementary Table S3.
Published evidence on the efficacy of SIRT in the salvage setting is of limited quality and at risk of bias. Statistically significant improvements in overall survival in patients treated with SIRT were observed in two retro-spective studies: patients receiving BSC survived for a median of 6.6 months compared with 11.9 months in pa-tients who received SIRT  or 8.3 months versus 3.5 months . In a small (n ¼ 44 total) RCT comparing fluorouracil chemotherapy alone to SIRT plus chemo-therapy, PFS and LPFS were improved in the SIRT arm (PFS 2.1 versus 4.5 months; hazard ratio 0.51; P ¼ 0.03; LPFS
2.1 versus 5.5 months; hazard ratio 0.38; P ¼ 0.003), showing prolonged control of liver tumour growth . In this trial, patients were permitted to cross-over after progression. The overall survival estimate reported in our dataset of 7.6 months aligns with the SIRT arm of the retrospective comparative study by Seidensticker et al. . It also consistent with the lower end of the range of
Total number of patients with severe day-of-treatment complications and total number of all-cause adverse events and abnormal laboratory value events
Severe day-of-treatment complications Number of patients
Adverse event category Number of adverse events Number of grade 3 adverse events Fatigue 89 8 Abdominal pain 58 3 Nausea 22 Vomiting 14 Fever 10 1 Gastritis 5 Gastrointestinal ulcer 1 REILD 1 Radiation pneumonitis Radiation cholecystitis Radiation pancreatitis Other 53 7 Total adverse events 253 19
Abnormal laboratory result event category Number of events Number of grade 3 events AST increased 79 ALT increased 73 1 Hypoalbuminaemia 67 4 Hyperbilirubinaemia 44 8 INR increased 1 Neutrophil count decreased 10 3 Platelet count decreased 28 Other 51 2 Total abnormal laboratory result events 353 18
ALT, alanine aminotransferase; AST, aspartate aminotransferase; REILD, radio-embolisation-induced liver disease; INR, International Normalised Ratio.
previously published data. It is possible that this may be due to the patient selection that occurs during a clinical trial, leading to inclusion of patients who may be in worse health in this registry-based study.
The study reported here is the largest, prospective, registry-based study to examine the survival of patients with unresectable, chemotherapy-refractory mCRC treated with SIRT. Unfortunately, the ‘real-world’ setting of the treatment and data collection in NHS centres resulted in missing data and in variability in the assessment criteria, which does add some uncertainty to the conclusions. However, the real-life setting may have led to the inclusion of a patient population more representative of the patients to be treated within the NHS.
The absence of a contemporaneous comparator group limits our interpretation of the clinical data reported. As a registry study, no minimum follow-up was specified. How-ever, long-term data were available for most of the patients included. Data collection for health-related quality of life questionnaires varied significantly between centres. High levels of missing data meant that reliable conclusions about the impact of SIRT on patient quality of life could not be drawn from this study. Despite these reservations, the clin-ical data presented here will aid treatment decisions reached between clinicians and patients in day-to-day practice.
PFS and LPFS in this cohort were 3.0 months and 3.7 months, respectively. Both values are at the lower end of the range from published studies of 2.8e9.2 months (nine studies; 437 patients) for PFS and 2.0e9.0 months (eight studies, 376 patients) for LPFS (Supplementary Table S1). PFS estimates should be interpreted with caution given the inherent risk of bias in this measure ; PFS relies on interval-censored data, which will probably inflate the sur-vival estimate. We report a 0.7 month higher LPFS compared with PFS, which mirrors results from other studies .