Welcome back to the journal review. This is the September twenty twenty-six issue of the Journal of the American Academy of Dermatology, and we've got four brief reports to walk through today — one on electronic brachytherapy for basal cell carcinoma, and three that all circle around the same theme from different angles: racial and ethnic disparities in melanoma outcomes, one in vulvar melanoma specifically, one in veterans, and one a large SEER trends analysis. Let's get into it.

First up, a prospective multicenter study on long-term local control and cosmesis of electronic skin surface brachytherapy, or ESSB, for basal cell carcinoma, out of Memorial Sloan Kettering. This is a straightforward original prospective study, small but with unusually long follow-up for a radiotherapy device paper.

The background here is one you already know clinically: radiotherapy is a reasonable option for select keratinocytic carcinoma patients, especially those who are poor surgical candidates, and electronic skin surface brachytherapy is the newer entrant — a handheld surface applicator that mimics the rapid dose falloff of traditional brachytherapy without needing an afterloader or radioactive source. The gap the authors identify is simple: short-term data on ESSB already look good, but nobody had published durable, multi-year outcomes with truly prospective, independent assessment.

Methodologically, this was a single-arm prospective cohort — thirty-four patients, mostly elderly, mostly facial basal cell carcinoma, treated with a standardized hypofractionated regimen, forty-two gray in six fractions, and followed for a median of five years. There's no comparator arm and no real statistical modeling here, so there's no formal statistics lesson to teach on this one — it's essentially descriptive, and that's the right call for a small single-arm device follow-up study where the goal is simply to characterize the durability of an outcome, not to test a hypothesis against a control. What is worth noting methodologically is the multimodal outcome assessment: clinical exam corroborated by reflectance confocal microscopy and high-frequency ultrasound, plus patient- and physician-reported cosmesis on visual analog scales, plus formal quality-of-life instruments — the Skindex-16 and the Skin Cancer Index. That layering is what makes the long-term data credible rather than just eyeballed.

The results: only one recurrence in the entire cohort over five years of follow-up — a low, clinically meaningful local control rate that compares favorably to the roughly eighty-six percent five-year local control seen in a large radiotherapy meta-analysis of nearly ten thousand patients. Cosmesis was excellent and durable — about nine in ten patients rated "good" cosmesis at twelve weeks, and roughly eight in ten still rated it "good" at five years, comparing favorably to a benchmark of sixty-nine percent good cosmesis from a randomized trial of other radiotherapy modalities for basal cell carcinoma. Acute toxicity was substantial but self-limited — Grade 3 radiation dermatitis hit about four in ten patients at two weeks, but resolved to grade one or less by twelve weeks. The late toxicity signal is the one to flag for patient counseling: hypopigmentation and telangiectasia emerged after twelve months and persisted in a majority of patients — around six in ten with hypopigmentation and a third with telangiectasia — through the full five years. Quality of life dipped predictably at two weeks post-treatment and then recovered progressively.

Limitations the authors are upfront about: this is overwhelmingly a basal cell carcinoma cohort — only two of thirty-four were non-BCC — so you can't generalize to squamous cell carcinoma with any confidence. And local control was assessed clinically and photographically rather than with routine post-treatment biopsy, so there's an inherent risk of missing subclinical persistence; they note a companion trial is underway to correlate clinical assessment with histopathology and imaging.

Practical takeaway: this is genuinely useful confirmatory data, not practice-changing in the sense of shifting your treatment algorithm. For your appropriately selected elderly or surgery-averse basal cell carcinoma patients, this reinforces that ESSB can deliver durable control and cosmesis that holds up over five years, but you should counsel patients specifically about the likelihood of persistent hypopigmentation and telangiectasia — this isn't a free lunch cosmetically in the very long run, even though the global "good" cosmesis rating stays high. File this under solid reassurance for a modality you may already be using or referring for, not a reason to change your patient selection criteria.

Now to the disparities cluster, starting with the vulvar melanoma paper — a retrospective cohort study using the National Cancer Database.

Background: vulvar melanoma is rare, aggressive, and carries substantially worse survival than cutaneous melanoma, and while racial disparities in cutaneous melanoma are well described, nobody had really nailed down whether the same pattern holds in this specific mucosal subtype.

Methodologically, this is a registry study — over four thousand five hundred patients from 2004 through 2022, stratified into non-Hispanic White, non-Hispanic Black, Asian/Pacific Islander/American Indian/Alaska Native combined, and Hispanic groups. The core analytic tool is a multivariable Cox proportional hazards model. This is the right tool here, and it's worth spending a moment on why. A Cox model estimates the hazard, essentially the instantaneous risk of dying at any given moment, and compares that hazard between groups while allowing for the fact that patients entered the registry at different times and were followed for different durations — some died early, some were censored at last follow-up, some are still alive. Rather than needing everyone followed for the same fixed period, as a simple proportion-surviving calculation would require, the Cox model uses all the follow-up time each patient contributes. The "proportional hazards" assumption is that the ratio of risk between groups stays roughly constant over time — it doesn't say Black patients have double the absolute risk of dying at every moment in the same way, just that the relative risk ratio is stable. And because it's multivariable, it lets the authors report an adjusted hazard ratio that controls simultaneously for stage, ulceration, treatment, and other variables, isolating race as an independent variable rather than something that's really just a proxy for showing up with more advanced disease.

The results here are stark. Black and Hispanic patients presented with a heavier burden of advanced-stage disease and ulceration, and had lower rates of surgery and higher use of radiation, consistent with more advanced presentation. Immunotherapy use was actually higher across all non-White groups compared with White patients. Five-year overall survival was lowest for non-Hispanic Black patients at under half, roughly forty-three percent, compared with about fifty-five percent for White patients, fifty-six percent for Hispanic patients, and sixty percent for the combined Asian/Pacific Islander/American Indian/Alaska Native group. The critical finding is that after full multivariable adjustment for stage, treatment, and other clinicopathologic factors, Black race remained independently associated with roughly a thirty percent lower risk of death compared to White, Hispanic, and Asian/Pacific Islander/American Indian/Alaska Native patients relative to Black patients — meaning the survival gap for Black patients didn't wash out even once you account for the fact that they present with more advanced tumors. That's the clinically meaningful point: measured tumor burden and treatment differences don't fully explain the gap.

Limitations are the classic registry constraints: no melanoma-specific survival or recurrence data in the National Cancer Database, no HIV or immunosuppression status captured, and the inherent risk of residual confounding no multivariable model can fully erase. The authors are appropriately cautious about invoking biology, noting that genomic data on vulvar melanoma across racial groups are simply too sparse to support a claim of intrinsic biological difference.

Practical takeaway: this is not practice-changing in terms of a specific new intervention, but it is an important awareness point for anyone involved in the multidisciplinary care of vulvar melanoma — Black patients are both presenting later and dying faster even independent of stage, which should raise your index of suspicion for expediting workup and referral in symptomatic pigmented vulvar lesions in Black patients, and reinforces the need for care pathways that don't rely on patients self-advocating their way to earlier biopsy.

Next, the veterans’ sentinel lymph node biopsy paper — a comparative registry analysis between the VA Cancer Registry and SEER.

The background: prior work already established that veterans present with thicker, higher-stage melanoma than the general SEER population, but nobody had looked at whether sentinel node biopsy utilization or positivity rates differed in this population, which matters directly for staging and adjuvant therapy decisions in your Mohs and oncology practices that see VA-referred patients.

Methodologically, this pulled nearly three thousand VA patients and about thirty-five thousand SEER patients with T2 through T4 melanoma from 2010 through 2019. They used chi-square tests for the basic utilization and positivity comparisons, and multivariable logistic regression to identify factors independently associated with receiving a sentinel node biopsy and with a positive result. Logistic regression is the natural choice here because the outcome is binary — biopsy performed or not, positive or not — rather than a time-to-event outcome, so instead of modeling hazard over time like the vulvar melanoma study, this model estimates the odds of the binary outcome as a function of multiple predictors simultaneously, again letting you isolate, say, the independent effect of rurality or marital status after controlling for age and tumor thickness. They also ran competing risk regression, specifically Fine-Gray models, as a secondary analysis for melanoma-specific mortality — worth a quick note on why: with an older, comorbid population like veterans, a lot of patients die of things other than melanoma, and a standard Kaplan-Meier approach would inappropriately treat those non-melanoma deaths as if the patient were simply censored, potentially distorting the estimated melanoma-specific risk. Fine-Gray competing risk models handle this by explicitly accounting for the competing event, giving a more accurate picture of melanoma-specific mortality risk in a population where dying of something else is a real and common alternative.

Results: veterans were substantially less likely to receive sentinel node biopsy than SEER patients across every age and T-stage stratum — for instance, in patients under sixty-five, about seventy-three percent of veterans got the biopsy versus eighty-five percent in SEER, and in patients sixty-five and older, about fifty-three percent versus sixty-four percent, both statistically significant gaps. But — and this is the clinically interesting twist — when veterans did get biopsied, positivity rates were higher, not lower, particularly in older patients with T3 and T4 tumors, where positivity ran twenty-five versus eighteen percent for T3 and forty versus twenty-nine percent for T4. Younger age, Midwest residence, and extremity tumor location predicted higher biopsy uptake in both cohorts; in SEER specifically, being married predicted higher uptake while rural residence and lentigo maligna histology predicted lower uptake.

The authors' interpretation, and this is their stated reasoning, is that the lower utilization in the VA likely reflects a mix of selection — reserving biopsy for patients where prognostic information is felt to matter most — plus real access barriers and higher competing comorbidity burden that make watchful management more attractive in some veterans. The higher positivity rate despite lower overall utilization suggests that when VA clinicians do pursue the biopsy, they're selecting appropriately for higher pretest probability, though it also raises the concern that lower-risk veterans who might still benefit from staging information are being systematically missed.

Limitations: this is registry data without granular comorbidity capture, no ability to quantify specific access barriers, and it excludes veterans who get some or all of their melanoma care outside the VA system, which could bias the denominator in either direction.

Practical takeaway: if you're in a VA-affiliated practice or see veteran referrals, this is a call to actively examine your own threshold for offering sentinel node biopsy in the sixty-five-and-older veteran population — the data suggest real underutilization relative to guideline-eligible T2 through T4 disease, and given that positivity runs higher in this population, there may be actionable missed staging information with downstream implications for adjuvant immunotherapy eligibility.

Now the last of the disparities trio, and the largest in scope: a SEER database analysis of temporal trends and mediators of racial and ethnic disparities in cutaneous melanoma survival, spanning two thousand one through twenty twenty, out of Xiangya Hospital in China.

The background is stated plainly: everyone knows disparities in melanoma survival exist, but the long-term trend — are they narrowing or widening — and what specifically is driving them, remained poorly characterized.

This is the most methodologically rich paper of the four, and it's worth slowing down here. They pulled over three hundred sixty thousand patients from SEER. Rather than simple Kaplan-Meier comparisons, their primary technique was an accelerated failure time model, generating what's called a survival time ratio between racial groups. This is worth teaching because it's a nice contrast with the Cox model we just covered in the vulvar melanoma paper. Where a Cox model asks "how does the instantaneous risk of death compare between groups at every moment," an accelerated failure time model asks a subtly different question: "how does belonging to one group stretch or compress a patient's overall survival timeline compared to the reference group." A time ratio below one means that group's effective survival time is compressed relative to White patients — so a time ratio of zero-point-six-eight, which is roughly what they found for Black versus White patients overall, means Black patients' survival trajectory is compressed to about two-thirds the pace of White patients', accounting for age. The authors also ran a sensitivity check using a generalized gamma distribution instead of the standard parametric form, and got consistent results, which is a reasonable robustness check for whether the shape assumption of the model was driving the conclusion.

On top of that, they layered a mediation analysis, which is the more advanced statistical concept worth flagging here. Mediation analysis tries to partition a total observed effect — in this case, the survival gap between racial groups — into how much of that gap is statistically "explained" by an intermediate variable, or mediator, that sits causally between race and the outcome. For example, if Black patients present with more advanced stage, and advanced stage independently worsens survival, then stage is a mediator, and the analysis quantifies what proportion of the total racial survival gap is attributable to that pathway rather than to race having some direct, unexplained effect. This matters immensely for interpretation, because it moves the conversation from "there is a disparity" to "here is specifically what is and isn't accounting for it."

The results are sobering. Melanoma-specific mortality declined about one and a half percent overall across the twenty-year study period — but increased about two and a half percent among Black patients, the only group to move in the wrong direction. Relative disparities widened significantly for Black, Hispanic, and Asian/Pacific Islander patients relative to White patients across the two decades. For Black patients specifically, the overall survival time ratio dropped from zero-point-seven-seven in the earliest period to zero-point-five-eight in the most recent period, and the absolute gap in five-year melanoma-specific mortality rose from about thirteen percent to nearly eighteen percent. The mediation analysis found that stage at diagnosis and histologic subtype — for instance, higher rates of acral lentiginous melanoma in minority patients — each accounted for up to roughly thirteen percent of the disparity, and marital status accounted for up to about ten percent. Income had a genuinely interesting divergent effect: it mediated a small slice, three to four percent, of the Black-White gap in the expected direction, but for Asian/Pacific Islander patients, income actually acted as a suppressor, meaning higher income in that group was masking an even larger disparity that would otherwise be apparent — a subtlety worth sitting with rather than glossing over.

The authors' own discussion is direct: advances in melanoma detection and treatment over the past twenty years have not been equitably distributed, and the widening gap for Black patients in particular reflects persistent barriers to early diagnosis rather than treatment access per se, since SEER doesn't even capture detailed treatment data like immunotherapy use — which the authors flag as a real limitation. They also note reliance on area-level rather than individual socioeconomic measures, and the general residual confounding inherent to registry data.

Practical takeaway: this one is more sobering than actionable in the sense of a single procedural change, but it is a strong argument for practice-level intervention on the pathway you actually control — earlier biopsy and diagnosis. Given that stage at diagnosis is one of the largest identified mediators, and that acral lentiginous subtype and delayed presentation are disproportionately represented in Black patients, this reinforces the importance of lower biopsy thresholds for atypical acral and mucosal lesions in patients of color, and supports continued investment in community-level early detection education, since the data suggest the survival gap is actively widening rather than closing on its own.

That wraps our four reports for this issue. To bring it together: solid, reassuring long-term device data for electronic brachytherapy in basal cell carcinoma; and a consistent, uncomfortable throughline across three separate registry analyses — vulvar melanoma, veterans, and SEER national trends — that racial disparities in melanoma outcomes are real, only partially explained by stage and treatment, and in the case of Black patients specifically, getting worse rather than better over time. Thanks for listening, and I'll see you next issue.