A recent Australian melanoma conference put the spotlight on early detection, dermoscopy and artificial intelligence, write dermatologists Dr Helena Collgros and A/Prof Tony Caccetta.
The Australasian Melanoma Conference which took place in Sydney recently brought together specialists in the field of melanoma, including dermatologists, oncologists, surgeons, radiation oncologists, pathologists, skin cancer GPs and researchers.
Speakers outlined a road map toward achieving ‘zero deaths from melanoma,’ focusing on topics such as prevention, early diagnosis, early melanoma, high-risk patients and advanced melanoma.
For early detection, several risk prediction tools are available that consider well-known classical risk factors such as fair skin type, multiple naevi, and a personal or family history of melanoma.
However, these tools do not differentiate between in situ and invasive melanoma development. Moreover, the absolute number of thicker tumours is still increasing, so we are still missing patients with no apparent risk factors who will present with thick melanoma.
It is also essential to redirect awareness and early detection messages to individuals with darker skin tones, encouraging them to check acral surfaces such as hands, feet and nails, where melanoma is more common in these skin types.
A significant focus of the conference was on artificial intelligence (AI), or augmented intelligence, for early melanoma diagnosis.
Most AI applications currently target single-lesion diagnosis using dermoscopy images. However, there is greater potential for AI in real-world applications such as enhancing screening through total body photography or as a second opinion tool for patients already under surveillance.
This real-world contextualised approach is currently being realised with newer datasets containing macroscopic total body image tiles, rarer diagnoses (e.g. amelanotic melanoma, mucosal and nail melanoma) and darker skin types. AI is also showing early promise in predicting the risk of melanoma metastasis and assisting in adjuvant therapy selection for patients.
The issue of overdiagnosis was discussed including screening practices that lead to detection of a silent disease reservoir – melanomas in situ and thin ‘minimally invasive’ T1 melanomas that may never metastasise.
Other factors contributing to overdiagnosis are over-biopsy by clinicians and overcalling by pathologists, who may mistakenly classify benign lesions as malignant. Research indicates poor intra- and inter-observer agreement among pathologists regarding melanocytic lesions, combined with a downward trend in diagnostic thresholds leading to more malignant diagnoses.
Overdiagnosis can result in overtreatment, which may cause harm to patients and increase healthcare costs. The challenge remains to determine which early melanomas will progress to life-threatening tumours, as the tools for this differentiation are still lacking. Future research hopes to address this concern.
After the diagnosis of early melanoma, several risk progression prediction tools are available to help predict which melanomas will advance.
Some of these tools include panels of gene expression profiles (GEPs) and combinations of clinical and pathological variables with GEPs. Recent publications have proposed using ‘deep dermoscopy’ alongside histopathological indicators (such as Breslow thickness and ulceration) to predict melanoma metastasis and early-stage melanoma recurrence based on clinical and histopathological data extracted from electronic medical records.
For high-risk patients, guidelines recommend utilising total body photography (TBP) and sequential digital dermoscopy (SDD) to assist in early melanoma diagnosis while reducing the biopsy rate of benign naevi.
TBP serves as a comparative reference for future examinations since primary melanomas typically arise de novo (from new) or in association with pre-existing melanocytic naevi.
SDD is useful for flat melanocytic lesions that are not suggestive enough to warrant excision but not completely inconspicuous, as well as for mildly atypical lesions that show change over time. Evidence suggests that SDD identifies featureless melanomas and reduces the ratio of benign to malignant excisions and the number of benign lesions removed.
In the near future, AI may be employed to classify lesions, providing more accurate and reliable diagnoses. Although AI is already integrated into certain photography machines, it is not approved by the Therapeutic Goods Administration and is currently limited to research (in silico and real world) applications.
The main takeaway in the field of advanced melanoma is the concept of ‘de-escalation of surgery’. Neoadjuvant immunotherapy has now become the standard of care for stage III melanoma.
Recent research indicates that patients respond better when immunotherapy is administered before lymph node dissection rather than undergoing surgery first and receiving immunotherapy afterwards.
There is still a need to develop new treatment options for patients resistant to both immunotherapy and targeted therapy, such as personalised mRNA vaccines and new drugs currently in clinical trials. We are moving toward a new era of precision oncology that combines clinical data with pathology, genomics, and other ‘omics’ to deliver personalised and effective treatments to patients.
The conference also featured a presentation of the Conquering Skin Cancer movie that launched in cinemas this November. This powerful movie is a must-see for all as it will undoubtedly raise awareness about skin cancer and melanoma.
It inspires us all to take action and make a difference in the fight against this disease, contributing to the goal of zero deaths from melanoma.
ED: Dr Helena Collgros and A/Prof Tony Caccetta are Fellows of the Australasian College of Dermatologists (FACD).