Sunitinib for the treatment of gastrointestinal stromal tumours: a critique of the submission from Pfizer

A proposed approach for analyzing post-study therapy effect in survival analysis

Clinical trials that explore long-term endpoints may confound the analysis when post-study therapy effects are considered. This article introduces a procedure to mediate the effects of confounding and allow inferences of first-line experimental treatments in the presence of post-study therapy. The procedure is evaluated by intensive simulation analyses and applied to an analysis of a clinical cancer trial.

Health utility of patients with advanced gastrointestinal stromal tumors (GIST) after failure of imatinib and sunitinib: findings from GRID, a randomized, double-blind, placebo-controlled phase III study of regorafenib versus placebo

BACKGROUND

In this analysis we report patients with advanced gastrointestinal stromal tumors (GIST) refractory to imatinib and sunitinib therapy as derived from the EuroQol-5D (EQ-5D) for progression-free (PF) and progressive disease health status.

METHODS

Data were analyzed from a phase III trial conducted at 57 hospitals in 17 countries (trial registration number, NCT01271712). Patients with advanced GIST were randomized (2:1) to receive blinded treatment using oral regorafenib 160 mg daily or placebo, plus best supportive care (BSC) in both groups, for the first 3 weeks of each 4-week cycle. EQ-5D-3L was administered on day 1 of each cycle before contact with their physician and before any study-related procedures. The effect of disease progression on the utility of EQ-5D was tested with paired-samples comparison and general linear mixed modeling (GLMM).

RESULTS

One hundred and eighty five patients [93 % of the intention-to-treat (ITT) population] completed 803 EQ-5D questionnaires: 77.7 % in progression-free (PF) state, 6.5 % at progression, 13.9 % following first progression, and 1.9 % after second progression. Mean baseline utility was 0.767 (SD 0.221) with no significant between-group differences for active treatment and BSC. The first post-progression health state was 0.647 (SD 0.343), suggesting significantly impaired health-related quality of life after confirmed disease progression showed a decrease of -0.120 (paired samples t test, p = 0.001). GLMM showed no effect of study treatment or cycle number on utility.

CONCLUSIONS

We demonstrate a significant and clinically meaningful difference in health state utility values between PF and progression. Utility values remained stable over successive regorafenib cycles after controlling for disease status and treatment type.

Prevalence of gastrointestinal stromal tumour (GIST) in the United Kingdom at different therapeutic lines: an epidemiologic model

BACKGROUND

The prevalence of patients with gastrointestinal stromal tumourgst (GIST) who fail currently available treatments imatinib and sunitinib (third-line treatment-eligible GIST) is unknown, but is expected to be below an ultra-orphan disease threshold of 2/100,000 population used in England and Wales. Our study was designed to estimate the prevalence and absolute number of UK patients with unresectable/metastatic GIST at first-, second- and eventually third-line treatment.

METHODS

Our open population model estimates the probability that the prevalence of UK third-line treatment-eligible GIST patients will remain under the ultra-orphan disease threshold. Model parameters for incidence, proportion of unresectable/metastatic disease and survival estimates for GIST patients were obtained from a targeted literature review and a UK cancer register. The robustness of the results was checked through differing scenarios taking extreme values of the input parameters.

RESULTS

The base-case scenario estimated a prevalence of third-line treatment-eligible GIST of 1/100,000 and a prevalence count of 598 with a 99.9% likelihood of being below the ultra-orphan disease threshold. The extreme scenarios, one-way and probabilistic sensitivity analyses and threshold analysis confirmed the robustness of these results.

CONCLUSIONS

The prevalence of third-line treatment-eligible GIST is very low and highly likely below the ultra-orphan disease threshold.