Strategies for improving melanoma survival : from early detection to therapies for advanced disease

CM is one of the deadliest skin cancers, with increasing incidence in a majority of countries with predominantly fair-skinned populations. Despite the introduction of systemic therapies for metastasized CM showing promising effects on survival, CM is still responsible for considerable morbidity and mortality.

The aim of this thesis is therefore to investigate the potential role of AI as a tool to improve clinicians' performance in CM diagnosis, to study the effect of some commonly used drugs with a putative immune modulatory effect on CM survival, and to investigate the outcome of systemic therapies and clinical markers in patients with metastatic CM in a real-world setting.

In study I, we conducted a systematic review and meta-analysis of use of AI- assistance for skin cancer diagnosis including CM. In our search, we identified 2'983 studies, of these, ten could be included for meta-analysis. The results indicated that clinicians within all levels of experience benefitted from AI-assistance and those with least experience in dermatology showed the most significant improvement. However, only few studies were conducted in real-life settings.

In study II, we performed a population-based cohort study of patients diagnosed with a primary invasive CM in Sweden between 2007-2014, using the SweMR linked to national population-based registries with high coverage. We identified 23'507 patients with primary invasive CM and included 1'162 patients with a diagnosis of type 2 diabetes mellitus in the study. The use of metformin pre-, peri-, and postdiagnostically was assessed among the patients. We showed that metformin use was consistently associated with significantly improved overall survival (OS) regardless of timing of treatment, although with no significant association on the melanoma-specific survival (MSS).

In study III a cohort of 24'562 patients with a primary invasive CM diagnosed between 2006-2014 was identified, of whom 1'253 patients were using at least one of the six most common groups of H1-antihistamines. Drug use was assessed peri- and postdiagnostically. Desloratadine and loratadine use were associated with significantly improved MSS, in the analysis of peri- and postdiagnostic use, respectively, but no association on the MSS was seen for the other groups of antihistamines.

In study IV, 252 patients with a diagnosis of metastatic (stage IV) CM and first-line treatment with immune checkpoint inhibitors (ICIs) or targeted therapy (TT) were identified from the records at the Department of Oncology/Skin Cancer Center, the Karolinska University Hospital, Sweden, between 1 January 2010 and 31 December 2017, and were followed until 31 March 2019. Patients receiving ICIs experienced significantly longer survival in terms of both overall- and progression-free survival (PFS) compared to patients receiving TTs, although not reaching statistical significance in the multivariable analysis for the latter. Several clinical markers were associated with therapy response and survival. Specifically, male sex and more advanced M-stage were associated with significantly lower response to both therapies and ICI, respectively. Furthermore, increasing lactase dehydrogenase levels were associated with shorter PFS and OS among both patient groups, and more advanced M-stage with lower OS and PFS among ICI-treated patients. Interestingly, decreasing albumin levels were associated with shorter PFS in patients treated with both therapies, lower OS for ICI-treated patients, and with a lower response rate to TT.

In conclusion, we report that AI may be a useful tool for skin cancer diagnosis including CM, though further evaluation in prospective clinical trials is needed. Metformin, desloratadine, and loratadine are commonly used, cheap, and well- tolerated drugs that may have secondary antineoplastic effects, potentially contributing to improved survival among CM patients. Moreover, ICI and TT in first- line treatment of metastatic disease showed a survival benefit in a real-world setting. Finally, albumin may be a prognostic marker that needs further validation.

List of scientific papers

I. Isabelle Krakowski*, Jiyeong Kim*, Zhuo Ran Cai, Roxana Daneshjou, Jan Lapins, Hanna Eriksson, Anastasia Lykou, Eleni Linos. Human-AI interaction in skin cancer diagnosis: a systematic review and meta-analysis. NPJ Digit Med 2024; 7(1): 78. https://doi.org/10.1038/s41746-024-01031-w

II. Isabelle Krakowski, Henrike Habel, Kari Nielsen, Christian Ingvar, Therese M L Andersson, Ada Girnita, Karin E Smedby, Hanna Eriksson. Association of metformin use and survival in patients with cutaneous melanoma and diabetes, Br J Dermatol 2023; 188(1): 32-40 https://doi.org/10.1093/bjd/ljac003

III. Ildikó Fritz, Philippe Wagner, Matteo Bottai, Hanna Eriksson, Christian Ingvar, Isabelle Krakowski, Kari Nielsen, Håkan Olsson. Desloratadine and loratadine use associated with improved melanoma survival, Allergy 2020; 75(8): 2096- 9. https://doi.org/10.1111/all.14273

IV. Isabelle Krakowski, Matteo Bottai, Henrike Häbel, Giuseppe Masucci, Ada Girnita, Karin E Smedby, Hanna Eriksson. Impact of modern systemic therapies and clinical markers on treatment outcome for metastatic melanoma in a real-world setting, J Eur Acad Dermatol Venereol 2021; 35(1): 105-15. https://doi.org/10.1111/jdv.16678