The influence of body-mass index on survival of advanced melanoma patients

Fontes, Pedro and Megiddo, Itamar and Mueller, Tanja and Clarke, Julie and Barry, Sarah and Kleczkowski, Adam (2022) The influence of body-mass index on survival of advanced melanoma patients. In: British Oncology Pharmacy Association (BOPA) 2022 – 25th Annual Symposium, 2022-10-07 - 2022-10-09, Liverpool.

[thumbnail of Fontes-etal-BOPA-2022-Poster-The-influence-of-Body-Mass-Index-on-survival]
Preview
 Text. Filename: Fontes-etal-BOPA-2022-Poster-The-influence-of-Body-Mass-Index-on-survival.pdf
Accepted Author Manuscript
License: Strathprints license 1.0
Download (790kB)| Preview

Abstract

Objectives: Melanoma is the deadliest form of skin cancer, accounting for 90% of all skin cancer-related deaths globally. Although treatment of advanced melanoma has improved over the last decade, we have limited knowledge about the most effective treatment pathways.1 During the course of treatment patients may experience changes in therapy or physiology that can influence survival and should be considered in survival models. Incorporating this variability may help better assess the treatment effectiveness in real-world clinical practice for informing treatment decisions. Survival probability is usually calculated according to information at baseline. The aim of this research is to understand the impact of body-mass index (BMI) changes over time on survival of melanoma. BMI was recorded at most appointments along with other patient information, demographics and laboratory results. BMI was categorised into groups according to the parameters defined by World Health Organization.2 Thus, survival analysis of these patients according to BMI and BMI changes was studied. Methods: The study was conducted on a cohort of 350 patients, who had 2784 appointments between 14 March 2008 and 30 March 2018 from NHS Greater Glasgow and Clyde diagnosed with advanced melanoma. The methods used were Cox proportional hazards models and log-rank tests. The event investigated was survival and the main outcome of interest was the hazard ratio. Multivariable standard Cox regression model and time-dependency adjusted Cox models were fitted to the data to assess survival probability. A log-rank test was performed in these Cox models to assess covariates’ statistical significance towards overall survival. A time-dependency-adjusted (TDA) Cox regression model using BMI as a time-dependent covariate was fitted to the data. Discussion: The TDA Cox model for BMI revealed patients with higher BMI have an increased probability of survival. It found that increases in BMI over time may enhance/improve patient survival. Patients with an increase of 1 unit in BMI (e.g. 29–30 kg/m2) are 5% more likely to survive in comparison with the previous BMI measured. Recent studies on the impact of BMI at baseline and obesity on survival suggest that in patients with advanced melanoma, obesity is associated with improved outcomes such as progression free-survival and overall survival.3,4 The results also showed that ECOG PS Performance Score and LDH levels have a statistically significant impact on survival. Figure 1. Forest plot of survival model with BMI as a time-dependent covariate. Note: BMI is included as a time-dependent covariate in a multivariable Cox regression model; Other covariates are time-fixed (baseline); BMI – Body-Mass Index; SIMD – Scottish Index of Multiple Deprivation; PS – ECOG Performance Score; LDH – Lactate dehydrogenase. Chemotherapy includes dacarbazine, temozolomide and paclitaxel + carboplatin; Immunotherapy includes ipilimumab, nivolumab, pembrolizumab and ipilimumab + nivolumab; Targeted therapy includes dabrafenib, dabrafenib + trametinib and vemurafenib. Conclusion: A cancer patient’s journey for survival should not be solely determined by baseline information. Along their treatment course, changes in therapeutic or regimen medication dose, variations in laboratory results and/or appearance of adverse events, for example, can influence overall survival. Time-dependency-adjusted Cox models could be more suited to predict survival as these can incorporate changes over time. Thus, further investigation into these models is needed to understand their applicability of these models in a real-world population.

CORE (COnnecting REpositories)