Melanoma spread pattern model

For her PhD thesis Dr Hayley Reynolds, from the Auckland Bioengineering Institute, developed a computerised 3D model of the body to help doctors predict where a patient’s cancerous melanoma cells are more likely to spread. Melanoma develops in skin cells but can metastasise (spread) from the skin to other sites in the body. Treatments for advanced melanoma that has spread are limited, and certain therapies are not publicly funded in New Zealand.

It spreads very quickly and not always in predictable ways. The earliest sign that melanoma has spread is usually found in the lymph nodes, but it is often difficult for doctors to know which lymph nodes to check and to be sure they have checked all the likely lymph nodes – there are over 600 lymph nodes in the body with 43 lymph node fields where melanoma is likely to spread.

Learn more about lymphatic system.

The aim of Hayley’s research was to develop an efficient way of displaying which lymph nodes melanoma is most likely to have spread to, from any skin site in the body.

Building the computer model

To build her 3D anatomical computer model, Hayley used photographic images from the Visible Human Project, which has over 18,000 cross-section photos of the human body, obtained from slicing a male cadaver into 1 mm slices. Hayley ‘stacked’ these 2D photos in 3D computer space and used this to create a 3D computer skin mesh. She then located and mapped lymph nodes onto her 3D model, also using the Visible Human. Hayley’s 3D man is an accurate model of the skin and lymph nodes.

Using lymphoscintigraphy data

As well as building the computer model of skin and lymph nodes, Hayley needed to record data about melanoma sites and dissemination on the model. She did this using lymphoscintigraphy data from melanoma patients.

In some hospitals, all patients with early melanoma undergo a hospital test called lymphoscintigraphy (LSG) – a radioactive tracer and blue tracer dye are injected into the skin around the melanoma, and over 2–3 hours, these travel to the lymph nodes. The first lymph node(s) that drains the melanoma site is called the sentinel lymph node(s). Obviously, this is very useful to know before surgery.

Hayley used lymphoscintigraphy data from the Sydney Melanoma Unit (SMU) to develop her computer model. The SMU has the world’s largest LSG database of over 5,000 melanoma patients. They have recorded in 2D the location of each patient’s melanoma site and which lymph node(s) it has spread to. Hayley has used this 2D data to map onto her 3D computerised model to reveal patterns of melanoma spread. This allows doctors to predict where tumours may develop if the melanoma has spread so they can focus on these high risk areas.

Using the computer model

Instead of looking at data from over 5,000 patients on 2D sheets of paper, doctors can click on an area of skin on the 3D body displayed on the screen and can see all the sentinel lymph nodes sites previously found by the SMU. They can instantly see a display of the lymph nodes where cancer cells have spread to from that site in other patients. This means they can decide where to check for signs of cancer spread.

Hayley has also developed heat maps that show the likelihood of spread to a particular lymph node site. A skin selection tool allows doctors to select an area of skin to see potential lymph node sites and how many patients have been sampled with melanoma in the particular skin site.

Why this kind of research in particular?

New Zealand has one of the highest melanoma rates in the world, with around 300 deaths per year. According to New Zealand skin cancer statistics, skin cancer is the most common cancer here, with melanoma in the three most diagnosed cancers for both men and women. Hayley, whose 17-year-old brother Wayne lost a long battle with leukaemia, wanted to pursue research that would have a useful practical outcome, something that would enable doctors to diagnose and treat their patients more quickly and effectively.

Hayley’s research was supervised by Maurice Wilkins Centre investigators Professor Rod Dunbar (School of Biological Sciences) and Dr Nicolas Smith (Bioengineering Institute) at the University of Oxford.

Key Findings

Hayley completed her thesis in 2008. The melanoma spread pattern model she created to visualise existing melanoma drainage (i.e. spread) data created new knowledge. Her work showed that: 

• Different people with melanoma in the same general skin area may drain to different lymph node areas. That is, drainage is not perfectly predictable by simple anatomical zones. 

• Sappey’s lines do not accurately predict melanoma drainage. Sappey’s lines are old anatomical ‘boundary lines’ that doctors were using to predict how lymph fluid drains from different areas of the skin. 

• Melanoma draining to dual lymph node areas is common. For example, a melanoma near the waist region had a non-trivial probability of draining to both the left underarm and left groin areas. 

• Symmetry mostly holds, with exceptions. For many skin regions, drainage is symmetric (left versus right side).  

• Head and neck melanomas have complex drainage patterns, but there were discernible patterns. 

• Melanomas on the torso have ambiguous drainage patterns – some areas may drain differently, not fitting neatly into a single area. This is important for planning because it warns against assuming a single drainage site.