Real world outcomes with systemic anti-cancer treatments for advanced melanoma : experience from the West of Scotland 2010-2018

Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2021
Thesis identifier
  • T16521
Person Identifier (Local)
  • 201774208
Qualification Level
Qualification Name
Department, School or Faculty
  • Background: Ipilimumab was the first systemic anti-cancer therapy (SACT) to show survival benefit in clinical trials for advanced (unresectable or metastatic) melanoma. (Hodi et al. 2010) Vemurafenib, dabrafenib with trametinib, pembrolizumab and nivolumab followed but it is recognised that clinical trial findings are not always representative of real world practice (Donia et al. 2017). It was proposed that electronic record linkage (ERL) of routinely captured healthcare data with SACT prescriptions would generate real world data (RWD) to supplement clinical trial results for treatment decision making. Methods: Patients starting SACT for advanced melanoma between 1.11.10 and 31.12.17 in the West of Scotland were identified and followed up until 31.3.18. Prescribing records from the Chemotherapy Electronic Prescribing and Administration System (CEPAS) were linked with routinely captured healthcare data including the Scottish Cancer Registry, anonymised then accessed in the NHSGGC Safe Haven. Multivariable Cox regression models estimated impact of patient baseline characteristics and SACT on OS. ERL methodology was validated using results from individual patient case notes (IPLR). Results: Median OS varied from 18.5 months (95%CI 14.4-not estimable) for ipilimumab with nivolumab to 5.6 months (4.5-7.3) with dabrafenib. Dabrafenib with trametinib (HR 0.42, p=0.0014) and ipilimumab with nivolumab (HR 0.50, p=0.0352) had a positive impact on OS compared to ipilimumab monotherapy. Baseline characteristics including LDH levels above the upper limit of normal; ECOG performance status ≥2 and mucosal melanomas had a negative impact on OS. Data availability differences between ERL and IPLR (some laboratory results and disease characteristics) did not have a statistically significant impact on the hazard ratios for each SACT, except dabrafenib, in the multivariable Cox model. Conclusion: Our RWD showed reduced median OS compared to the pivotal clinical trials but it included patients often excluded from clinical trials. ERL is a valid method for obtaining real world outcomes.
Advisor / supervisor
  • Bennie, Marion
Resource Type
Date Created
  • 2020