224O - Survival differences between EU and Asia for head and neck cancer: Results of the RARECAREnet-Asia collaboration

Background

The epidemiology of head and neck cancer (HNC) differs between East Asia and Europe: some HNCs are common and have better survival in East Asia than in Europe. We analyzed the impact of sex, age, HNC subsite and histologies on survival differences between Europe and Asia leveraging the RARECAREnet-Asia collaboration between European and Asian population-based cancer registries (PBCR).

Methods

We analysed all HNCs: oral cavity, hypopharynx, larynx, nasal cavity, salivary gland, nasopharynx, oropharynx. Oropharynx subsites were grouped in tonsil and non-tonsil related. Nasopharyngeal tumours were differentiated in keratinising and non-keratinising. European data were provided by 94 PBCR (diagnosis period 2000-2007; follow-up 2008). Asian data were provided by the Taiwan and Japan PBCR (diagnosis period 2009-2011, follow-up 2016). Relative survival was modelled with a generalised linear model providing relative excess risks of death after adjusting for sex, age, subsite and histological grouping. Analyses were performed using federated learning which enables analyses without sharing sensitive data.

Results

HNC subsites and histologies distribution and survival differed between the 3 countries. Compared to Europe, after adjusting for covariates, we confirmed better survival in Taiwan and in Japan for each HNC.

Conclusions

The lower survival in Europe was attributed to different period of diagnosis (in Europe intensity modulated radiotherapy was not yet standard), more advanced stage at diagnosis and worse treatment outcome. Our data confirm the need of international collaboration to improve early diagnosis and treatment appropriateness for HNC. RARECAREnet-Asia was established in the framework of Rare Cancers Asia to build a bridge between Asia and Europe to learn from each other in addressing the challenges in the respective continents faced by rare cancers considering the different cancer epidemiology. Our analyses have confirmed federated learning as a feasible approach for data analyses that addresses the challenges of data sharing between organisations and geographies.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.