Welcome back to the journal review. This month we're working through the August 2026 issue of the Journal of the American Academy of Dermatology, and we've got four pieces on deck — an ethics column on post-transplant skin surveillance, a full clinical review of dermatofibrosarcoma protuberans, an original ecological study on geographic disparities in melanoma mortality, and a sharp little letter to the editor about a wound care mixing practice that turns out to have a chemistry problem. Let's get into it.

First up is an Ethics Journal Club piece — not a study, more of an advice-column format, structured as a letter to a "Dr. Dermatoethicist." The question comes from a dermatologist doing annual total body skin exams on solid organ transplant recipients and wondering whether blanket, indefinite, high-frequency screening for every transplant patient — including those many years out with no prior skin cancer and no major risk factors — is really the best use of limited dermatologic capacity, especially when it might be crowding out higher-need patients.

The response walks through the tension pretty directly. Solid organ transplant recipients as a group do carry markedly elevated skin cancer risk from chronic immunosuppression, with tumors that behave more aggressively and recur more often than in the general population, and that's exactly why American and European guidelines recommend total body skin exams every six to twelve months, particularly for kidney and thoracic organ recipients. But the author points out that risk is not homogeneous across this population — it varies by organ type, age at transplant, skin type, cumulative sun exposure, and the specific immunosuppressive regimen. Liver and pancreas recipients, for instance, generally carry lower cumulative risk than kidney, heart, or lung recipients. Treating everyone identically, regardless of that variation, isn't necessarily feasible or equitable, especially when low-yield visits displace patients who need the appointment more urgently.

The proposed solution is risk stratification, and the specific tool mentioned is the SUNTRAC score — a composite of demographic and clinical features that sorts patients into risk categories with suggested surveillance intervals ranging anywhere from every six months to every two years. The author is careful to flag that this isn't yet fully validated for universal adoption, but it's offered as a structured framework for individualizing surveillance rather than defaulting to one-size-fits-all intervals. The piece also emphasizes shared decision-making — informing patients of their individualized risk and letting some opt for more frequent reassurance while others, especially lower-risk patients, choose wider intervals. And there's a systems-level point too: since primary care physicians perform a large share of skin exams in the U.S. but have limited lesion-detection sensitivity without dedicated training, expanding structured dermatology education for primary care could help absorb some of this surveillance burden in a risk-stratified model.

There's no results or limitations section here in the traditional sense — this is an opinion and ethics piece — but the practical takeaway for you is straightforward: this is a reasonable framework to start applying now. Consider incorporating something like the SUNTRAC score, informally or formally, into how you triage transplant follow-up intervals, and treat annual-forever TBSE as a default that can be renegotiated with lower-risk patients rather than a fixed rule. It's not practice-changing in the sense of new data, but it's a useful nudge toward more deliberate, resource-conscious scheduling.

Next, a full clinical review on dermatofibrosarcoma protuberans, or DFSP — a nice comprehensive update covering epidemiology, diagnosis, and management, authored by a multi-institutional group including surgical oncology and pathology co-authors, which shows in the depth of the multidisciplinary discussion.

On epidemiology, DFSP remains rare — incidence around four and a half per million in the U.S., translating to roughly fifteen hundred new cases a year, making it one of the more common dermal sarcomas even though that's a low bar. Typical age of onset is thirty to fifty, no sex predilection, rare in children, and — importantly for your differential threshold — Black patients are disproportionately affected, roughly twofold compared to lighter-skinned patients. Clinically it tends to show up as an indurated plaque, often on the trunk, evolving slowly over years, which sets up the diagnostic trap the authors keep returning to: cutaneous exam routinely underestimates the true subcutaneous extent of disease, which is part of why recurrence rates have historically been high with less rigorous margin techniques.

On histopathology, this is your classic storiform spindle cell proliferation with a honeycomb pattern of fat infiltration, and diffuse CD34 positivity is essentially required for confirmation — negative factor thirteen-A rules out dermatofibroma, negative SOX10 and S100 rule out neurogenic tumors and melanoma. The review runs through the variant zoo — myxoid, pigmented or Bednar tumor, atrophic, giant cell fibroblastoma, and fibrosarcomatous DFSP — and flags that only the fibrosarcomatous variant, with its higher mitotic index and reduced CD34 expression, carries a worse prognosis; the others don't change the outcome story.

Molecularly, over ninety percent of cases carry the t(17;22) translocation producing the COL1A1-PDGFB fusion gene, which drives autocrine signaling through Ras-MAPK and PI3K-AKT-mTOR pathways — and this is directly actionable, because confirming that fusion is recommended before starting imatinib, given that a meaningful minority of fusion-negative cases instead involve PDGFD rearrangements that don't respond the same way.

On treatment, the review is unambiguous that Mohs micrographic surgery, or complete margin assessment more broadly, is first-line, and they lean on some numbers worth knowing. In one retrospective series, only about half of tumors had clear margins after the first Mohs stage, climbing to over ninety percent by the second stage — reinforcing that a single conservative excision plan is inadequate for this tumor. A meta-analysis of roughly seven hundred patients found Mohs associated with substantially lower recurrence than wide local excision — the odds ratio was point three one, meaning roughly a two-thirds reduction in the odds of recurrence, and that was a statistically robust finding, not just a trend. Worth pausing on that odds ratio interpretation for a second: an odds ratio of point three one means the odds of recurrence in the Mohs group were about a third of the odds in the wide excision group — a meaningful, clinically relevant effect, not a marginal one. Mohs also preserved meaningfully more tissue than wide excision margins of two or three centimeters in a prospective comparison. Interestingly, conventional frozen-section Mohs and paraffin-embedded Mohs showed similar recurrence rates in a head-to-head retrospective comparison — about three percent versus seven percent — which the authors interpret as reassuring equivalence, though that difference didn't reach statistical significance, so treat it as hypothesis-supporting rather than proof of true equivalence.

For radiation, the review positions it strictly as a fallback — for unresectable tumors or positive margins where re-resection isn't feasible — with reported local control in the seventy-five to one hundred percent range across a large systematic review, but a clear signal that radiation after positive or close margins carries meaningfully higher recurrence than after negative margins. And for systemic therapy, imatinib is FDA-approved for unresectable, recurrent, or metastatic disease, with the strong caveat that fusion confirmation should precede starting it, since response depends on the presence of that COL1A1-PDGFB rearrangement.

There's no results-and-limitations arc here because it's a synthesis review, not a primary study, but the practical takeaway is clean and mostly reinforces what you likely already practice: Mohs remains your first-line tool for DFSP everywhere on the body, preoperative biopsy should confirm the diagnosis before you plan surgery, imaging is reserved for suspected subfascial extension or larger head-and-neck tumors rather than routine use, and a high index of suspicion is warranted in skin-of-color patients given the documented disparity in recognition and delay. The genuinely newer material here is on the molecular and therapeutic-target side — EGFR, PD-L1, and PRAME are flagged as emerging targets from single-cell sequencing work — but that's research-stage, not yet actionable in clinic.

Now the main course — an original study, and the only piece this month with a real statistical analysis worth teaching. This is a repeated cross-sectional ecological study looking at geographic variation in melanoma mortality decline following the introduction of checkpoint inhibitors and BRAF-targeted therapy.

The background is straightforward: population-level melanoma mortality began falling sharply around 2013, coinciding with approval of ipilimumab, pembrolizumab, and dabrafenib-trametinib, after decades of dismal stage four survival. But whether that mortality benefit was distributed evenly across the country, or concentrated in certain kinds of communities, hadn't been examined at the population level — prior work showed uneven immunotherapy uptake by age, insurance, and practice setting, but not uneven mortality outcomes.

Methodologically, the authors used county-level data from the Surveillance, Epidemiology, and End Results program, comparing age-adjusted melanoma mortality in a baseline window of 2008 to 2013 against a follow-up window of 2019 to 2023, deliberately skipping the 2014-to-2018 transition years to isolate a clean before-and-after comparison around the therapy rollout. This ecological design — meaning they're analyzing county-level aggregates rather than individual patients — makes sense here because you can't easily get individual-level treatment and outcome data linked at national scale, but county aggregates from SEER, Medicare claims, the American Hospital Association survey, and physician registries are available and let you approximate health system exposure. They restricted to counties with at least sixty thousand residents specifically to avoid a data suppression artifact — SEER suppresses death counts under eleven, which would otherwise make small counties look artificially volatile.

The core analytic technique is ordinary least squares regression — plain multivariable linear regression — used first in unadjusted, single-predictor form to see how much each county characteristic alone explains of the variation in mortality decline, and then in a fully adjusted model with everything entered together. This is really the simplest tool for this job: you have one continuous outcome, percent change in mortality, and you want to know which county features are independently associated with it once you account for the others. The unadjusted R-squared values reported are small on their own — often just one or two percent of variance explained by any single factor — which the authors are transparent about, and that's an important teaching point: statistical significance here does not mean a factor explains most of the variation. It just means the association is unlikely to be due to chance; the magnitude of explained variance tells you a different, and often more clinically important, story.

Onto results. Nationally, mortality was flat from 2000 to 2011 — essentially unchanged — then dropped about twenty-seven percent from 2011 to 2023, concentrated after 2013, tracking the drug approvals. Across the seven hundred fifty-one counties analyzed, the median decline was about twenty-four percent, but with wide spread — the interquartile range ran from under eight percent up to over thirty-five percent — and notably, almost one in five counties saw no improvement at all, or actually got worse. That spread is the headline finding: the national number papers over enormous local heterogeneity.

The single strongest predictor, by a good margin, was county population size. Counties above the median population size saw a twenty-five percent mortality decline versus about fifteen percent in smaller counties — an absolute gap of roughly ten percentage points, and that difference was highly statistically significant. In the regression, each doubling of county population was associated with roughly a five-percentage-point greater mortality decline even after adjusting for urbanicity, patient demographics, insurance mix, income, education, and hospital and specialist infrastructure — meaning population size was capturing something about treatment uptake or system capacity that wasn't fully explained by any of the other measured variables. Urbanicity mattered on its own but dropped out once population size was in the model, suggesting it was really a proxy for the same underlying phenomenon. A few patient-level factors held up in the adjusted model too — counties with higher uninsured rates and higher percent male population actually showed larger declines, which is a bit counterintuitive and the authors don't fully unpack why, though it may reflect baseline differences in mortality rates that leave more room for improvement. And to quantify the stakes: the authors estimated that if every county had matched the mortality reductions seen in the largest counties, upward of twenty-eight hundred additional deaths could have been prevented nationally over the study period.

The authors are appropriately honest about limitations, and this is worth internalizing as a general lesson in reading ecological studies: because the unit of analysis is the county, not the individual patient, you cannot say which specific patients did or didn't get novel therapy, or draw a causal line from "lives in a big county" to "received better care" — it's association at the aggregate level, vulnerable to what's called the ecological fallacy, where patterns seen in grouped data don't necessarily hold for individuals within those groups. There's also inherent residual confounding — anything correlated with population size that wasn't measured could be doing the real work.

Practically, this isn't a paper that changes what you do chairside tomorrow, but it's an important health-services-research data point for anyone thinking about equity in melanoma care delivery. It's a strong argument, if you're involved in tumor board, referral network design, or advocacy, for actively pushing telemedicine, outreach dermato-oncology clinics, or regional referral partnerships into smaller and more rural counties, since the data suggest that geography and system capacity — not just biology — are shaping who benefits from the immunotherapy era. File this under "interesting and important for systems-level thinking," not "changes your Mohs practice on Monday."

Last, a letter to the editor — a rapid chemistry-based rebuttal to a previously published wound care technique. The original article had proposed mixing povidone-iodine with erythromycin ointment and petrolatum gauze as a combined dressing for extensive skin barrier defects. This letter's authors raise a compatibility red flag: povidone-iodine is an aqueous formulation with free iodine as its active agent, while erythromycin ointment is lipid-based — so on a purely physical level, mixing the two causes immediate phase separation, which the original paper's own photos apparently showed.

But the more important finding here is chemical, not just physical. The letter authors ran nuclear magnetic resonance spectroscopy — a technique that tracks molecular structure by how atomic nuclei respond in a magnetic field — on the mixture over time, and found that free iodine released from povidone-iodine oxidatively degrades the macrolide ring structure of erythromycin A. Practically, they saw the characteristic erythromycin spectral signal fade progressively at ten minutes, twelve hours, twenty-four hours, and forty-eight hours after mixing, with new degradation peaks appearing in its place — happening well within the typical twenty-four to forty-eight hour window between dressing changes, meaning this isn't a theoretical concern, it's happening in real time during actual clinical use of the dressing.

The clinical implication is that mixing these two agents likely destroys much of the erythromycin's antimicrobial activity before it ever does its job, and the authors also raise a secondary concern about a potentially exothermic reaction causing local thermal injury, though they present that as a concern to investigate rather than something they've directly measured. There's no dedicated methods-and-limitations arc here because this is a letter presenting a single confirmatory experiment, not a clinical trial, and the authors are appropriately modest — they call for standardized animal-model studies comparing sequential versus combined application before drawing firm clinical conclusions.

The takeaway is simple and immediately practical: don't premix povidone-iodine and erythromycin ointment in a single dressing. If you're using both, in Mohs reconstructive or general wound care contexts, apply them sequentially rather than combined — the underlying rationale for combining them, presumably convenience or fewer dressing components, doesn't hold up once you know the active antibiotic component is being chemically degraded by the iodine on contact.

That wraps the August issue. To summarize the through-line: individualize transplant surveillance rather than defaulting to universal intervals, keep leaning on Mohs as your DFSP workhorse while watching the molecular-target pipeline evolve, recognize that novel melanoma therapy benefits are geographically uneven in ways worth advocating against, and keep your antiseptic and antibiotic dressings separate rather than mixed. Thanks for listening, and we'll see you next month.