br Procedures br Each site followed their
Each site followed their local process for undertaking SIRT procedures. All patients received a hepatic arteriogram and a liver-to-lung breakthrough nuclear medicine scan to ensure suitability and to plan the delivery of the Y-90 microspheres. Selective coil embolisation of 70323-44-3 to the stomach, duo-denum or other visceral structures was carried our as required to prevent non-target Y-90 delivery. SIRT was car-ried out using an established method. One of two brands of Conformite Europeene-marked active implantable medical devices was used to carry out the SIRT procedure: (i) SIR-Spheres (Sirtex Medical Ltd, Australia) resin microspheres;
(ii) TheraSphere (Biocompatibles UK Ltd, UK) glass micro-spheres. Dosing of SIR-Spheres and TheraSpheres was car-ried out as per manufacturer instructions. It should be noted that the dosing method is different for the two products, so the activity administered in GBq cannot be directly compared . Administration of concomitant chemotherapy and post-SIRT chemotherapy was at the discretion of the treating clinician. Sites were expected to follow up patients every 2e3 months after their SIRT procedure until liver progression was confirmed on scan. Adverse events were assessed and recorded throughout the follow-up period.
Data Collection and Outcomes
Data were collected by clinical teams and entered into an anonymised online registry. The final dataset was extracted
in March 2017 and sent to Cedar for analysis. The full evaluation report from the SIRT CtE project has recently been published online by NHS England . Patients with a missing diagnosis or missing SIRT administration date were excluded from the analysis. Data were only collected on patients who received SIRT.
Overall survival was defined as the duration from the first SIRT procedure until death from any cause. Patients with no date of death recorded were right censored at the date at which they were lost to follow-up. Survival pro-portions at 3, 6, 12, 24 and 36 months were reported for patients for whom these data were available. Hepatic and extrahepatic tumour response assessments were carried out locally by a radiologist and recorded in the SIRT registry. Typically, the Response Evaluation Criteria for Solid Tu-mours (RECIST) were used . PFS was defined as the duration from the first SIRT administration to the earliest date of detection of progressive disease (either hepatic or extrahepatic) by computed tomography, magnetic reso-nance imaging or positron emission tomography scan, or to the date of death from any cause if progression was not recorded. Patients with no progressive disease recorded were censored at the most recent date of non-progression (complete response, partial response or stable disease). LPFS was defined as the duration from the first SIRT administration to the date of progression in the liver or death from any cause. Patients with no progressive disease in the liver were censored at the most recent date of non-progression in the liver. Adverse events were recorded us-ing Common Terminology Criteria for Adverse Events version 4.0. Causality was determined by the treating physician on site.
The sample size of the study was estimated by NHS En-gland based on the number of patients who could be treated at 10 specialist centres over a 3-year period . All statis-tical analyses were conducted in IBM SPSS Statistics version 126.96.36.199 (IBM Corp. Armonk, NY, USA) or R (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org/).
Descriptive statistics for continuous variables were re-ported as appropriate. For each statistical comparison, P-values and confidence intervals were reported. P < 0.05 was considered statistically significant and all tests were two-sided. Median overall survival, PFS and LPFS were esti-mated using the KaplaneMeier analysis . Survival curves were presented with 95% confidence intervals and numbers at risk displayed. Potentially important baseline covariates were agreed in advance and tested to identify statistically significant prognostic factors associated with survival in the CRC cohort using the pairwise Log-rank test. Hazard ratios for baseline covariates were estimated for overall survival by univariate Cox proportional hazards models. The following covariates were selected: number of previous lines of chemotherapy (categories: 0, 1, 2, 3, 4), Eastern Cooperative Oncology Group (ECOG) performance status (categories: 0, 1, 2), age (either as continuous or categories:
<65 years, 65 years), sex, primary tumour in situ or not, prior biological therapy (including bevacizumab, cetux-imab, aflibercept, panitumumab), presence of extrahepatic metastases (categories: yes, no), extent of liver involvement (continuous or categories: <25%, 25e50%, >50%), prior liver surgery (categories: yes, no), number of liver tumours (categories: 1e5, 6e10, >10). The reverse KaplaneMeier method was used to calculate the median follow-up time.