Welcome back to the journal review. This is the August twenty twenty-six issue of the Journal of the American Academy of Dermatology, and we've got four pieces worth your time today — a dermoscopy brief report on cutaneous squamous cell carcinoma, a policy commentary on indoor tanning restrictions, a data-driven letter responding to a melanoma disparities paper, and a brief report characterizing Kaposi sarcoma subtypes. Let's get into it.

First up is a retrospective cross-sectional study out of an Italian skin cancer center looking at dermoscopy of invasive cutaneous squamous cell carcinoma, broken down by anatomic location and by National Comprehensive Cancer Network, or NCCN, risk category. The gap here is pretty clean: since the twenty twenty-two NCCN update added a "very-high-risk" tier for squamous cell carcinoma, nobody had really nailed down whether the dermoscopic appearance of these tumors tracks with that risk stratification, or whether dermoscopy varies meaningfully by body site in a way that could tip you off before the pathology comes back.

Methodologically, this is a chart-and-image review — they pulled histopathologically confirmed invasive squamous cell carcinomas with archived dermoscopic images from twenty eleven to twenty twenty-one, excluding nail and genital tumors, and had two dermatologists grade the dermoscopic features blinded to clinical data, risk group, and histology. That blinding matters a lot here — it's the only thing protecting you from circular reasoning, since if the graders knew the risk category going in, they could unconsciously see what they expected to see. For the analysis, they used simple univariate comparisons across anatomic sites and risk groups, then built a logistic regression model to find independent predictors of poor differentiation — but they restricted that regression to head and neck tumors only, because that's where the cohort distribution actually supported it. This is a nice teaching point on logistic regression itself: it's the right tool here because the outcome — poorly differentiated versus not — is binary, and logistic regression estimates, for each dermoscopic feature, the independent odds of that outcome while holding the other variables constant. It's doing the job a simple chi-square can't, which is disentangling overlapping features so you're not crediting ulceration for an effect that's really being driven by age. Worth flagging too: they explicitly chose not to correct for multiple comparisons, reasoning that a correction would inflate their type two error, meaning they'd rather risk a false positive than miss a true signal — a defensible but important caveat to keep in mind when you look at the individual p-values.

On to results. They analyzed seven hundred sixty-eight invasive squamous cell carcinomas, overwhelmingly on the head and neck — about eighty percent — and mostly in men. Poor differentiation, the histologic proxy for very-high risk, clustered heavily on the lower limbs and acral sites, hitting roughly seven in ten lower-limb tumors and about half of acral tumors, a statistically significant site effect. Dermoscopically, small-caliber vessels dominated on the upper limbs and acral sites, while diffuse whole-lesion vascularity was more of a trunk and lower-limb pattern, and a peripheral vessel distribution was the signature of acral and head-and-neck lesions. Dotted-glomerular vessels were common across the board, arborizing vessels leaned toward trunk and head and neck, and ulceration was seen in about six in ten tumors overall, again concentrated on head, neck, and limbs. Yellow crust — and this becomes important — was especially frequent on acral sites, head and neck, and lower limbs.

By NCCN category, about nine percent were low-risk, the large majority — nearly eighty percent — were high-risk, and about twelve percent hit that very-high-risk tier, with age climbing progressively as risk category increased. Low-risk tumors tended to show small-caliber, dotted-glomerular, or hairpin vessels and white scale; higher-risk tumors shifted toward linear irregular or arborizing vessels and ulceration. And in that head-and-neck-only regression, three features came out as independent predictors of poor differentiation: older age, with each additional year nudging the odds up slightly; ulceration, more than doubling the odds; and yellow crust, increasing the odds by about seventy percent. On the flip side, peripheral vessel distribution, dotted-glomerular vessels, and white circles were protective.

The authors' interesting interpretive twist in the discussion is on that yellow crust finding — they suggest it may actually represent purulent material or dried exudate from secondary infection rather than true keratinization, which would explain why it clusters on sites more prone to contamination, like the head and neck, lower limbs, and acral skin, rather than reflecting a distinct biologic feature of the tumor itself. The obvious limitations are the retrospective, single-center design, and the fact that the multivariable model only applies to head and neck tumors by necessity of sample distribution — so you can't extend those specific odds ratios to trunk or extremity lesions.

For your practice, this is useful pattern recognition, not a practice-changer. It won't replace histology, and the authors say so directly — dermoscopy here is framed as something that "may support earlier risk upgrading while histology is pending." So if you're staring at an ulcerated head-and-neck squamous cell carcinoma with yellow crust in an older patient before the permanent section report is back, this gives you some dermoscopic reinforcement to treat it with a higher index of suspicion for aggressive histology — worth knowing, worth validating in your own practice, but not something that should independently drive a treatment decision yet.

Next, a shorter piece — this is a commentary, a letter reacting to a recent U.S. Food and Drug Administration policy reversal. In March twenty twenty-six, the FDA withdrew its proposed rule that would have banned indoor tanning device use for anyone under eighteen nationally. The authors, who previously published their own policy analysis in this journal, are pushing back hard on that withdrawal using their own prior data plus newer mechanistic evidence.

Their own earlier analysis found that states with complete under-eighteen tanning bans had roughly forty-seven percent lower prevalence of indoor tanning among high school girls compared to states without such bans — a meaningful and statistically significant effect. In contrast, the weaker policy alternative many states use — parental consent requirements — showed no significant reduction in tanning behavior at all. They pair that policy data with mechanistic evidence from a separate study showing indoor tanning is associated with nearly a threefold increase in melanoma risk, and that it drives widespread DNA damage across melanocytes, including in body sites that don't normally see natural sun exposure — which they argue is direct biologic support for a causal, not just correlative, relationship between tanning bed use and melanoma.

There's no new methodology here to walk through — this is advocacy built on previously published data, and it reads that way. The message for you as a clinician is straightforward: this is squarely in your wheelhouse as counseling ammunition. If you're asked by patients, especially parents, or if you sit on any advisory or policy committee, you now have a clean, citable comparison — full bans work, consent laws don't — to advocate for state-level legislation given the federal vacuum. Not a clinical practice change, but a useful policy and counseling reference.

Third, a letter responding to a previously published Surveillance, Epidemiology, and End Results, or SEER, database analysis on melanoma survival disparities between Asian American and Pacific Islander patients — I'll call them AAPI — and non-Hispanic White patients. The original paper had flagged worse melanoma-specific mortality in AAPI patients and speculated about genetics, treatment delays, and language barriers as possible drivers. This response team decided to actually go test the treatment-delay hypothesis directly, since that piece of the puzzle had been under-explored.

They pulled histologically confirmed cutaneous melanoma cases from the SEER twelve database spanning nineteen ninety-two through twenty twenty-two, looking specifically at time to treatment, meaning the days between diagnosis and the first definitive cancer-directed therapy of any kind. Because SEER top-codes this variable at a ceiling value, they categorized time to treatment into bins and ran chi-square testing as the primary analysis, then did secondary analyses — Wilcoxon rank-sum testing, Kaplan-Meier curves, and a right-censored Cox proportional hazards model — restricted to patients with a calculable numeric time-to-treatment value.

Here's a nice stats teaching moment, and credit to the authors for flagging it themselves because it's genuinely counterintuitive. A Cox proportional hazards model normally estimates the instantaneous risk of some bad event — usually death — happening at any given moment, comparing that risk between groups while accounting for the fact that people are followed for different lengths of time. Here, the authors repurposed that same math, but the "event" isn't death — it's initiating treatment. So a hazard ratio below one, which would normally mean a protective effect and longer survival, instead means a lower moment-to-moment likelihood of starting treatment at any given point in time, which mathematically translates into a longer wait to treatment. They stratified the model by sex, age, and melanoma stage specifically because the underlying hazards weren't proportional across the whole population — a technical requirement of Cox models that this design step satisfies.

The cohort was large and lopsided, as SEER data usually is — about two hundred twenty-two thousand non-Hispanic White patients against roughly twenty-six hundred AAPI patients. The categorized time-to-treatment distribution differed significantly between groups. Medians were actually identical at zero days for both — most patients in both groups started treatment same-day — but the overall distribution shape differed significantly, and the Cox model, stratified as described, showed AAPI patients had a hazard ratio of point eight five, meaning a meaningfully lower likelihood of initiating treatment at any given time point, translating into longer real-world delays for a subset of that population.

The authors are appropriately cautious about causation. They lay out plausible contributors — linguistic discordance, unfamiliarity with the U.S. health care system, insurance gaps, cultural preference for traditional or complementary medicine, and reported experiences of discrimination in care — but they're upfront that none of these can be directly measured in SEER, so this is hypothesis-generating rather than a mechanistic answer. Limitations also include missing time-to-treatment data, the top-coding ceiling, and the ambiguity of what a "day zero" treatment entry actually captures in the same-day majority.

Practically, this doesn't change what you do at the bedside tomorrow, but it should sharpen your awareness that treatment-initiation delay is a real, measurable, and disparate phenomenon in AAPI melanoma patients, distinct from tumor biology — and it argues for systems-level fixes like interpreter access and patient navigation programs rather than anything you'd individually adjust in your surgical decision-making.

Last, a brief report characterizing Kaposi sarcoma, or K.S., across sociodemographic and clinical subtypes — a serial cross-sectional study from two tertiary academic centers in a large California urban area, spanning twenty ten to twenty twenty-two. The background problem is that with antiretroviral therapy driving down HIV-associated Kaposi sarcoma incidence, the overall epidemiology of the disease in the U.S. has likely shifted, and there's been strikingly little contemporary domestic data to characterize that shift.

They identified patients by ICD code, confirmed diagnoses by chart review, and required at least six months of follow-up. Patients were sorted into five clinically meaningful subgroups: HIV-associated Kaposi sarcoma, split further by CD4 count strata at diagnosis; classic Kaposi sarcoma, meaning HIV-negative patients over sixty-five; iatrogenic, for HIV-negative patients on immunosuppressive medications; endemic, for HIV-negative patients from sub-Saharan Africa; and a fifth category the authors specifically wanted to draw attention to — early-onset Kaposi sarcoma, meaning HIV-negative patients under sixty-five. Statistically, they used Wilcoxon rank-sum and chi-square tests for group comparisons, plus bivariable — meaning unadjusted, one-variable-at-a-time — logistic regression stratified by HIV status.

Worth pausing on that bivariable approach, especially against last article's multivariable model. Bivariable regression here is really just an odds ratio between two groups without adjusting for other covariates, and the authors leaned on it partly because some of their subgroups were tiny — the iatrogenic group had only nine patients, and had to be excluded from several comparisons entirely because a sample that small produces odds ratios with such wide, unstable confidence intervals that they're not clinically interpretable. That's a good general rule of thumb for reading small-cohort papers: when you see a subgroup denominator in the single digits, treat any point estimate attached to it with real skepticism regardless of how it's reported.

Of three hundred eighteen total patients, about seventy percent had HIV-associated Kaposi sarcoma, sixteen percent classic, three percent iatrogenic, and eleven percent fell into that early-onset category. Zooming into early-onset: the vast majority — over eighty percent — were men who have sex with men, and local therapy, rather than systemic treatment, was by far the dominant approach, also used in over eighty percent. Compared with classic Kaposi sarcoma, early-onset patients were more likely to receive local therapy, with an odds ratio of about two point nine — though notably the confidence interval's lower bound sat right at one, so treat that as a borderline, not a rock-solid, signal. Only about six percent of early-onset patients received systemic chemotherapy as first-line treatment, under eighteen percent received chemotherapy at all, and about four in ten had achieved disease resolution by last follow-up.

The authors frame this as reinforcing a growing recognition that HIV-negative men who have sex with men can develop an indolent form of Kaposi sarcoma that behaves very differently from the aggressive, HIV-driven disease most of us associate with the diagnosis — responding well to local measures alone without needing systemic therapy. They speculate this could reflect elevated human herpesvirus eight seroprevalence in this population, though that wasn't directly measured here, and they call for future multicenter registries incorporating viral and immunologic data. Limitations are the ones you'd expect from a two-center study in one urban area — generalizability is limited, and outcome measures like "resolution" were subjectively assessed by chart review rather than by any standardized instrument.

The practical takeaway is mostly diagnostic awareness rather than a management overhaul: if you're seeing an HIV-negative man under sixty-five who has sex with men presenting with what looks like Kaposi sarcoma, this data supports counseling toward an indolent expected course and a local-therapy-first approach, while keeping in mind that the evidence base behind that odds ratio is thinner than it looks at first glance.

That wraps up this episode. To summarize the throughline across today's four pieces: dermoscopic clues may help you anticipate high-risk squamous cell carcinoma histology while you wait on pathology, indoor tanning policy is moving in a direction the data doesn't support and gives you talking points to push back, treatment-initiation delay is a real and measurable disparity in AAPI melanoma patients worth systems-level attention, and early-onset Kaposi sarcoma in HIV-negative men who have sex with men is emerging as its own indolent clinical entity worth recognizing. Thanks for listening, and I'll see you next issue.