For one thing, melanin’s extinction overlaps with common laser lines, which affects the safety and efficacy of laser treatments in dermatology, but also in imaging and wearable devices that use LEDs in the visible range. “Pheomelanin and eumelanin are chemically very similar and both have this property of having very high extinction coefficients in the visible range, meaning that melanins both absorb and scatter light which we commonly use for laser treatments and for wearable medical devices,” Dr. Marks, a research scientist in dermatology at Massachusetts General Hospital, Boston, said during a virtual course on laser and aesthetic skin therapy. “Melanins also shadow a number of other biological signals that we look for in the skin, such as those from hemoglobin.”
A number of different scales can be used to estimate the amount of eumelanin, or darkly pigmented melanin, in the skin, but the most famous is Fitzpatrick skin typing, the classification system that ranges from I to VI originally intended to quantify the skin’s response to UV light. “It’s so famous that it’s used in the emoji modifier of the Unicode Consortium lookup table,” said Dr. Marks, who spoke on behalf of the Wellman Anti-Racism Effort (WARE), a grassroots working group within the Wellman Center for Photomedicine at Massachusetts General Hospital. (The mission of the group is to eradicate racism in STEM, medicine, and academia starting with its own research and Center.)
Dr. Marks referred to a Northwestern University study published in 2013, which found that both patients and dermatologists failed to accurately determine Fitzpatrick skin type (FST) when compared with reflectance spectrophotometry used to measure melanin index objectively. “There is a need to classify skin type with reliable questions with responses suitable for all skin types,” the authors concluded.
Plenty more can go wrong when clinicians ignore or misunderstand the role of melanin as a background contrast agent, Dr. Marks continued. She cited the common misconception that melanomas do not occur in darker pigmented skin, a topic discussed in an article published online in January 2020 in Cancer Cytopathology.
“While they do occur at a lower rate, this misconception leads to an alarmingly low survival rate for black melanoma patients,” Dr. Marks said. “Acral lentiginous melanoma is one example of this. It is not related to sun exposure, yet it occurs in 30% to 70% of melanomas in black patients. This also exposes a mortality rate of 1 in 3 for Black melanoma patients, compared with 1 in 11 for White patients. In fact, Black patients face a lower survival for most cancers, often attributed to social and economic disparities rather than biological differences.”
Another significant contributing factor may be the lack of data and awareness of clinical research related to patients with skin of color. The Skin of Color Society’s “Find a Doctor” database is attempting to address this by improving patients’ access to board-certified dermatologists who specialize in skin of color. “Some of the discrepancies in dermatology education, screening, and treatment for Black, indigenous, and people of color is likely attributed to the fact that only 4.5% of images in general medical textbooks show darker skin, as they are only 5% of clinical trial participants despite making up 17% of the U.S. population,” Dr. Marks said at the meeting, which was sponsored by Harvard Medical School, Massachusetts General Hospital, and the Wellman Center for Photomedicine. Mind The Gap, a handbook of clinical signs and symptoms in black and brown skin, was published in 2020 by students and staff at St. George’s University of London. It can be downloaded for free.
Some 40 years after Kodak was criticized for not acknowledging inherent biases in their film stocks based on its “Kodak Shirley” color correction card, Dr. Marks said that camera makers are still ignoring racial bias in their technologies. “This is likely a ‘garbage in, garbage out’ phenomenon,” she said. “Due to the lack of diverse images, these biases get ingrained into machine learning models themselves, either because patients were not served in the first place, resulting in missing data, or because of mislabeling due to a lack of knowledge in properly classifying these images. So, while machine learning has the potential to step in where dermatologists fall short, we must be very diligent about recognizing any bias we are ingraining into these algorithms,” she said.
“That said, no technology is ‘born racist,’ of course; it is up to us to prevent history from repeating itself and prevent these biases from being ingrained in our work,” she added. “We can start by holding ourselves and others accountable when designing studies that have exclusion criteria, by challenging our sponsors on the exclusion of Fitzpatrick V and VI if you feel it is not scientifically sound, and by ensuring inclusive algorithm development. If these things are not possible, please use a disclaimer to make these limitations clear.”
According to Dr. Marks and WARE, clinicians can increase diversity in clinical trials by widening eligibility criteria, tapping into community-based medical centers, connecting with patient advocacy groups, using point-of-care and telemedicine technologies, supporting diversity-focused public policy on a larger scale, and making diversity an internal mandate, “within your institution, and within yourselves.”
Some community efforts stemming from Wellman inventions so far include the Texas-based Removery INK-nitiative program, which removes racist and hateful tattoos for free via laser tattoo removal technology that was invented at Wellman. Dr. Marks and her WARE colleagues also work with the Dream Beam Foundation, which is a global initiative bringing laser-based technologies to children in Vietnam, Armenia, Israel, Brazil, and Lebanon.
Dr. Marks reported having no financial disclosures.