This month's news from across the specialty

September 1, 2021

In this monthly column, members of the DermWorld Editorial Advisory Workgroup identify exciting news from across the specialty.  

The parable of the blind men and the elephant speaks to the limitations of an individual’s perception in understanding the full context of the world. This parable is well-aligned with the challenges of determining patient behaviors with limited sensory data in that a full understanding of patients’ behavior is limited by isolated information and incomplete datasets. Pediatric atopic dermatitis presents a clinical need for the improvement in monitoring patients’ behavior. As clinicians, we understand that quantifying pruritus by the patients’ reported history is limited by the subjective reporting of the nocturnal perceptions of children. Collecting objective scratching data as a surrogate for pruritus can be performed with the labor-intensive method of video monitoring in the lab. Previous work to assess scratching behavior with motion data has been limited by false signals from activities with similar movements such as typing. A recent study by Chun et al in Science Advances (2021; 7; 18: eabf9405. 30 April 2021) applies data from motion and sound sensors to objectively quantify scratching in pediatric patients with atopic dermatitis. A single wearable adhesive monitor with a seven-day battery was applied to the dorsum of the patient’s dominant hand. Machine learning was applied to improve the pattern recognition of scratching behavior. Like a pedometer, the device has been trained to ignore patterns of sound and motion that mimic elements of scratching. Texting on a phone and clicking a mouse are distinct patterns that can be automatically ignored by the system. The algorithm performed on par with the gold standard of annotated video. Future work is needed to further validate this approach in real-world use before broader applications. Additionally, the patient-reported outcome measures of pruritus and quality of life will continue to hold the primary importance for our patients in assessing treatment responses.

The future role of augmented intelligence and machine learning in health care has doctors abuzz across the house of medicine. Naturally, the visual nature of our specialty has sparked ongoing dialogue and debate around the role that augmented intelligence may play in our practice in the coming years. These debates have largely been fueled by the now multiple studies that have demonstrated the ability of neural networks to aid in the diagnosis and treatment of both malignant and non-malignant skin disease. Earlier this month, JAAD released an important piece on this topic titled "Bias In, Bias Out: Underreporting and Underrepresentation of Diverse Skin Types in Machine Learning Research for Skin Cancer Detection – A Scoping Review". This study found that of 136 published studies that used machine learning to classify malignant or premalignant skin lesions, skin type of patients was disclosed in only six studies (4.41%) and race/ethnicity was disclosed in only 12 studies (8.82%). Importantly, of those studies that disclosed race/ethnicity only one study included Black or African American individuals, one study included American Indian or Alaska Native individuals, and two studies included Hispanic individuals. While skin cancer is less common in individuals with skin of color, it has been shown to lead to greater rates of morbidity and mortality, often due to delays in diagnosis. Almost without a doubt, augmented intelligence is here to stay. The only question that remains is how fast and in what way these technologies will find their way into our practice. It is crucial that image sets used to train neural networks and machine learning algorithms utilize images across skin tones to ensure these technologies help to mitigate rather than widen existing disparities in diagnoses and treatment in dermatology. The time to shape these technologies to become equitable tools to augment the care of patients across racial and ethnic backgrounds is now, while the technologies are still new and fluid. Once the concrete sets, we may be too late.

Hepatitis B reactivation (HBVr) in the setting of immunosuppression is one risk that we encounter routinely in practice. Chronic, resolved, and occult HBV infection each confer varying levels of risk for HBVr. Reactivation first manifests with an increase in serum HBV DNA, followed by an increase in hepatitis B surface antigen about one month later. Thereafter, hepatitis with transaminitis occurs.

I read with great interest about a nine-year (2009-2018) multicenter cohort study based out of Taiwan that looked at predictors of hepatitis B and C virus reactivation in patients with psoriasis treated with biologic agents. Of note, the Bureau of National Health Insurance is a mandatory single-payer insurance plan established in 1995 in Taiwan. This system only reimburses the prophylactic use of antiviral therapy in those undergoing chemotherapy, or those with organ transplants receiving immunosuppression. Thus, patients on biologic drugs for psoriasis who declined to pay for prophylaxis themselves formed an at-risk cohort.

Biologics used included etanercept, adalimumab, golimumab, ustekinumab, and secukinumab. Out of 2,060 patients with psoriasis, 359 patients had HBV, and there were 88 HBVr episodes during this time. HBVr was significantly higher for chronic infection (34.3%) vs. occult (3.2%) or resolved (5.0%) infection. On multivariate analysis, hepatitis B surface antigen seropositivity, hepatitis B e-antigen seropositivity, and TNF-alpha-inhibitor therapy were risk factors for HBVr. Reactivation was much more common with TNF-alpha-inhibitor than IL-17-inhibitor therapy as well (adjusted hazard ratio, 2.67). Antiviral prophylaxis, on the other hand, significantly reduced the risk of HBVr.

The COVID pandemic has been part of our lives for almost 18 months. With the FDA emergency use authorization for Pfizer and Moderna and the subsequent introduction of COVID vaccines in December 2020, we began to see cutaneous reactions to the vaccines. As the vaccine rollout has picked up, increasing reports of these cutaneous reactions have been reported (J Amer Acad Dermatol. 2021. 85(1):46-55). This article looked at reactions reported after administration of the Moderna and Pfizer vaccines. The authors created a registry with data on 414 individual cases. The most common reactions were delayed large local reactions as well as local injection site reactions and urticarial and morbilliform eruptions. These reactions were more common with Moderna vaccines, 83%, and only 17% in Pfizer vaccines. The authors report other less common reactions of pernio/chilblains, cosmetic filler reactions, zoster and herpes flares. The delayed local reactions occurred about eight days after the first dose with fewer (less than 50%) reporting reactions after the second dose, and no severe reactions after the second dose in patients who reported reactions with the first dose.

COVID remains a hot topic of interest impacting all of us and with the increase in cases due to the Delta variant, this will persist. The questions around the vaccine will continue as rollout expands to younger patients and many of those holding out on vaccination continue to consider the risks and benefits. Articles like these help us to better understand the cutaneous reactions and their prevalence. It allows us to have more information to better counsel patients regarding vaccine reactions. As the variant cases continue to rise, and boosters remain a possibility, our ability to present factual information that supports the self-limited nature of the cutaneous reactions, should help calm concerns for those considering vaccination.

Additional DermWorld Resources