Filtering for skin cancer diagnosis

Abstract

The study by Nervil et al. provides a comprehensive epidemiological analysis of skin biopsy practice in Denmark over a 15-year period.¹ With over 4.4 million biopsies analysed, the work provides not only a unique insight into the reality of care, but also a differentiated view of diagnostic efficiency in different sectors of the healthcare system. Of particular note are the Number Needed to Biopsy for Melanoma (NNB) and the Benign to Malignant Ratio (BMR).

A key finding of the study is the clear discrepancy in diagnostic accuracy between specialty groups and between cancer and routine pathways. While the NNB for hospital-based plastic surgeons was a remarkably low 2.8 in the cancer pathway, it was a surprisingly high 127.2 in the routine primary care pathway. These figures do not primarily reflect differences in diagnostic expertise, but rather the structural implications of a stratified healthcare system with effective filtering triage.

It is particularly important that the authors include a health economic perspective. The estimated direct costs for biopsies of benign lesions in the cancer pathway alone were more than €6.6 million in 2021 – a significant proportion for a publicly funded system. This highlights the importance of structured decision making and the potential for optimization in the management of patient pathways.

The low 1.5:1 NNB in the Danish cancer pathway illustrates the potential of stepped care systems with clearly defined filter functions. At the same time, however, it also shows that this diagnostic gain comes at a considerable cost – especially as most benign biopsies (69%) are performed in high-cost hospital structures. This raises the question of whether this allocation of resources is sustainable in the long term, especially in view of demographic trends and the increasing incidence of skin tumours. Indeed, the authors' modelling showed that the greatest cost savings could be achieved by treating suspicious lesions directly in primary care.

The remarkably low benign to malignant ratio (BMR), which appears to be exceptionally favourable in the cancer pathway, deserves particular attention. Critically, this ratio is not limited to melanoma but includes all malignant lesions, including all non-melanoma skin cancers. This relativizes the significance of the BMR in relation to melanoma-specific diagnostics. In addition, the analysis shows that, overall, significantly more biopsies were performed in the private sector and that most malignant tumours were diagnosed outside the cancer pathway. This raises further questions about the efficiency of current triage structures. The strict separation between the cancer pathway and the routine pathway also makes a direct international comparison difficult, as the Danish system is largely unique in its structure.

A pioneering aspect could be the use of AI-supported systems in monitoring skin lesions.² In initial studies, modern deep learning algorithms show at least a comparable sensitivity to experienced dermatologists in the detection of malignant lesions.³ Their use in the context of teledermatology services or as a decision-making aid in primary care could help to reduce unnecessary biopsies that could be further reduced by implementing advanced imaging techniques like confocal microscopy or optical coherence tomography.⁴ This opens up new opportunities to ensure the quality of care and at the same time conserve economic resources. However, the successful integration of such systems requires robust validation studies, legal clarity, and targeted education and training strategies for medical staff. It is important to understand the efficiency and burden of biopsy-based diagnostics in skin cancer management.

For dermatology professionals, the study provides a valuable basis for critically reviewing existing processes and initiating evidence-based reforms – for example, through the use of modern technology and structured training programs.

None.