OBJECTIVES : We performed a multidimensional NMA of trials comparing long active release (LAR) somatostatin analogs (SSA) against placebo for the treatment of advanced neuroendocrine tumors (NET).

METHODS : A targeted search of published literature performed using Embase^® and MEDLINE^® from database inception to February 2020 identified two randomized controlled trials assessing octreotide LAR (PROMID) and lanreotide LAR (CLARINET) against placebo in advanced NET patients. The Kaplan Meier curves for progression-free survival (PFS) and overall survival (OS) outcomes were “digitized” using DigitizeIt and regenerated individual patient level data using Guyot algorithm. The Bayesian NMA was performed using a series of parametric (Weibull, Gompertz, log-logistic and log-normal) and fractional polynomial (FP) models using survivalNMA package in R (v3.6.2) and parameters were estimated accordingly. A deviance information criterion (DIC) was used to compare the goodness-of-fit.

RESULTS : Among the parametric and FP models, log-logistic showed the best fit (DIC=431.65; 248.18), followed by FP1 and Weibull for both PFS and OS outcomes. The predicted survival of lanreotide LAR, octreotide LAR, and placebo was 44%, 40% and 6% in terms of PFS at 5 years, and 54%, 57% and 54% in terms of OS at 10 years. Lanreotide LAR showed highest PFS, followed by octreotide LAR and Placebo. Regarding OS, octreotide LAR showed the highest survival, while lanreotide LAR and placebo showed near-identical survival prospects. The findings were not sensitive to different distributions, although the survival probabilities varied substantially. The ordering of treatments was also consistent across distributions.

CONCLUSIONS : Octreotide LAR and lanreotide LAR significantly delayed the time to progression compared to placebo. The results of our multi-dimensional evidence synthesis approach are consistent with previously published HR based meta-analyses. The use of this approach may be more informative for any subsequent cost-effectiveness modelling over constant HRs.