Welcome back to the journal review. This is the February twenty twenty-six issue of the Journal of the American Academy of Dermatology, and we've got four pieces worth your time this month — an anatomic-site cohort study in acral melanoma, a brief report classifying oral photoprotective ingredients by strength of evidence, a case series on language barriers and melanoma stage at presentation, and a multicenter cohort study addressing one of the more anxiety-provoking findings you can get back on a squamous cell carcinoma pathology report: solitary large-caliber perineural invasion. Let's get into it.

First up is a retrospective cohort study out of Asan Medical Center in Seoul, looking at whether anatomic site — upper limb versus lower limb — predicts the pattern of first metastasis in early-stage acral lentiginous melanoma. The clinical gap here is real: we know acral lentiginous melanoma behaves differently from other cutaneous subtypes, and there's been some suggestion that tumor location itself carries independent prognostic weight, and that the mutational landscape differs by site. But nobody had specifically asked whether the hand versus the foot biases the route of first spread — nodal versus distant — which has obvious implications for what you image and how often.

Methodologically, this was a retrospective single-center review spanning nearly three decades, from 1997 to 2024, restricted to biopsy-confirmed acral lentiginous melanoma with a documented, photographically or operatively confirmed lesion location. Critically, they restricted the analytic cohort to early-stage disease — skin-confined at baseline, no nodal or systemic spread — specifically so they could capture the pattern of first progression during follow-up rather than just describing established metastatic disease. That's a sensible design choice for the question being asked: you can't ask what drives the first metastatic route if your cohort already has metastases at baseline. They further split cases into nail unit melanoma versus non-nail unit acral melanoma, which makes biological sense given how distinct the subungual compartment is anatomically and probably biologically. Out of 359 identified patients, 281 met criteria for the early-stage analysis — 83 upper limb, 198 lower limb.

The headline result: first lymph node metastasis was significantly more common in lower-limb disease, occurring in roughly a quarter of lower-limb cases versus about one in nine upper-limb cases. Flip side — first distant metastasis was significantly more frequent with upper-limb disease, occurring in about one in seven upper-limb patients versus one in twenty lower-limb patients, and the lung was the dominant distant site, again skewed toward the upper limb. This pattern sharpened when they stratified by stage: among stage two patients specifically, upper-limb disease had first distant metastasis rates around a third, versus under one in ten for lower-limb disease — a substantial, clinically meaningful gap. Despite these divergent metastatic routes, overall progression-free survival and overall survival did not differ significantly between upper and lower limb — it's the pattern of spread that differs, not the ultimate survival outcome, at least by these endpoints. On Cox regression, upper-limb location was an independent predictor of worse systemic metastasis-free survival, alongside ulceration, Breslow thickness, and lymphovascular invasion, with lymphovascular invasion carrying a strikingly large hazard ratio. For lymph node progression-free survival specifically, only Breslow thickness and bone involvement remained independent predictors — location dropped out for the nodal endpoint, which fits the overall narrative: it's specifically the route to systemic spread that's site-dependent, not systemic risk overall.

The authors' own limitation, stated plainly, is that this is a retrospective, single tertiary-referral-center study — so referral bias and center-specific surveillance protocols are real possibilities, even though they note follow-up imaging schedules didn't differ significantly by site in their cohort.

Practical takeaway: this is genuinely interesting and plausibly practice-relevant for surveillance planning, though it's not yet practice-changing given it's a single-center retrospective series awaiting external validation. But conceptually, it argues for asymmetric surveillance emphasis — leaning more heavily into nodal ultrasound surveillance for lower-limb and especially toenail melanoma, and being more vigilant about chest imaging for upper-limb, especially finger and fingernail, disease, particularly once you're dealing with stage two tumors. Worth filing away and watching for confirmatory cohorts.

Next, a brief report — not a full original study, more of an evidence-classification exercise — titled "Ingredients for Oral Photoprotection: An Evidence-Based Classification." This one's addressing a practical problem you've probably faced with patients: the oral supplement aisle is full of photoprotection claims with wildly inconsistent evidentiary backing, and there's been no standardized framework to sort signal from marketing.

The authors ran a PRISMA-guided systematic search across PubMed, Embase, and Cochrane through the end of twenty twenty-two, then applied a previously published weighted scoring algorithm that grades evidence by study type — in vitro studies worth minimal points, tissue-based studies more, animal studies more still, and then a tiered scale for human studies topping out with placebo-controlled trials and randomized controlled trials carrying the most weight, with caps placed on lower-tier evidence so that a mountain of weak in vitro data can't out-score a single good trial. Cumulative scores sorted ingredients into five bands from very weak to very strong.

From an initial pool of over fifteen hundred unique ingredients identified across nearly nineteen hundred articles, only forty-two had any clinical-level support at all, and just nine were backed by strong evidence. The standouts: Polypodium leucotomos extract, lutein, lycopene, beta-carotene, and eicosapentaenoic acid. Each maps to a different mechanism and a different target — Polypodium leucotomos extract showed the strongest evidence specifically for erythema reduction and DNA damage mitigation, working essentially immediately, within a couple of hours of dosing; lutein and lycopene carried very strong evidence for oxidative damage and photoaging endpoints at around three months of use; beta-carotene had moderate evidence for preventing UV-induced immunosuppression, with a caveat to avoid it in smokers given lung cancer risk at high doses; and eicosapentaenoic acid had very strong evidence specifically for suppressing UV-induced inflammatory mediators. The authors' central conceptual point is that no single ingredient covered every biological axis of solar damage — oxidative stress, DNA damage, inflammation, immunosuppression, pigmentation — so a combination formulation targeting complementary mechanisms is likely to outperform any single agent.

There's no results-versus-limitations arc here in the traditional sense since this isn't a clinical trial — it's a classification framework — but the authors are appropriately candid that overall clinical evidence for oral photoprotection remains limited and combination-specific trials are lacking. Worth flagging: this brief report was industry-funded, with several authors employed by or consulting for the sponsoring company, so read the ingredient rankings with that context in mind, even though the methodology itself looks reasonably rigorous.

Practical takeaway — this is useful counseling ammunition, not practice-changing therapy. If a patient asks about oral photoprotection, you now have a reasonably evidence-ranked shortlist — Polypodium leucotomos extract, the carotenoids, and eicosapentaenoic acid — to point them toward as an adjunct, while being clear it supplements rather than replaces topical sunscreen.

Third, a retrospective case series from UT Southwestern examining language barriers and advanced melanoma presentation in Hispanic patients, drawing on their institutional Surveillance, Epidemiology, and End Results registry from 2006 to 2022. The background problem is well established — Hispanic patients present with more advanced melanoma and have worse survival than non-Hispanic whites — but the specific social determinants driving that gap were poorly characterized, and that's the space this paper works in.

Of about forty-seven hundred registry patients, only two hundred eleven were Hispanic, and a hundred ninety-seven had staging data adequate for analysis — so right away, note this is a modestly powered single-institution sample, which the authors acknowledge limits statistical precision, especially for subgroup comparisons. They pulled demographics, language preference, interpreter use, point of entry into care, tumor characteristics, and stage, then ran univariate and multivariate logistic regression to identify independent predictors of late-stage disease at diagnosis. Logistic regression is the natural tool here since the outcome is binary — early versus late stage — and they wanted to isolate the effect of specific social variables like interpreter use while adjusting for confounders like insurance status and point of entry.

About a third of patients presented with late-stage disease, and five-year melanoma-specific survival was around two-thirds overall, notably worse in the late-stage subgroup. The point-of-entry data are the most clinically actionable finding: patients who presented first to dermatology had dramatically earlier-stage diagnoses, while those who first presented to the emergency department or primary care were far more likely to be diagnosed late — the emergency department point of entry carried an extraordinarily high odds ratio for late-stage disease, on the order of a fifty-fold increased risk in the multivariate model, which is a massive and clinically undeniable effect, not a statistical curiosity. Interpreter use was also an independent predictor of late-stage diagnosis, roughly tripling the odds even after adjustment, and this risk persisted even among privately insured patients, suggesting language access — not just insurance — is doing real work here. Interestingly, insurance status fell out of significance in the multivariate model once point of entry and interpreter use were accounted for. Chart review added qualitative texture: delays related to fear of missing work, and acral melanomas being misdiagnosed early on as infections or wounds — a pattern that should resonate given how often acral lentiginous melanoma masquerades as something benign.

Limitations are squarely acknowledged by the authors: single institution, retrospective, and a small, low-incidence population that limits statistical power — and they're careful to say association doesn't equal causation here.

Practical takeaway, and this one is genuinely actionable at the practice level even without further trials: language-concordant care and direct-to-dermatology access pathways matter enormously for this population. If your practice serves a significant Hispanic or Spanish-speaking population, this is a concrete argument for Spanish-language patient education materials, lower-barrier direct dermatology referral pathways, and heightened suspicion for acral lesions being dismissed as wounds or infections in primary care and emergency settings.

Last article: a multicenter cohort study addressing a question every Mohs surgeon has faced — you clear margins on a cutaneous squamous cell carcinoma, and the only adverse feature on pathology is large-caliber perineural invasion, no other risk factors. Do you push for adjuvant radiation, or not? Large-caliber perineural invasion — defined as nerve involvement at least zero-point-one millimeters in diameter, and or nerve below the dermis, and or a named nerve — is baked into both the Brigham and Women's Hospital and AJCC eighth edition staging systems as a risk factor, and NCCN guidelines suggest considering adjuvant radiation for perineural invasion after clear-margin surgery. But that guideline was built on data mostly reflecting perineural invasion in combination with other risk factors, and one small single-institution study of just sixteen solitary large-caliber perineural invasion tumors suggested these might behave more indolently on their own. This study set out to test that with real numbers, using a large multinational dataset spanning twelve centers.

Fourteen thousand two non-metastatic cutaneous squamous cell carcinomas, all treated with either Mohs surgery or wide local excision to clear margins, were stratified into three groups: no risk factors at all, large-caliber perineural invasion as the sole risk factor, and large-caliber perineural invasion plus at least one additional risk factor like size two centimeters or greater, deep invasion, or poor differentiation. They used Fine-Gray subdistribution hazard modeling to generate cumulative incidence curves for local recurrence and a composite major poor outcome endpoint — in-transit metastasis, nodal metastasis, distant metastasis, or disease-specific death. That competing-risk modeling choice makes sense here specifically because this is an older population where dying of something other than the squamous cell carcinoma is a real competing event that would otherwise bias a standard Kaplan-Meier estimate. Importantly, and the authors are upfront about this, they modeled unadjusted, because there simply weren't enough events in the solitary large-caliber perineural invasion group to support multivariable adjustment — which also means they could not assess whether adjuvant radiation actually changed outcomes in this subgroup.

The numbers: solitary large-caliber perineural invasion was rare, just forty-three tumors out of the entire fourteen-thousand-plus cohort, compared with two hundred sixty tumors that had large-caliber perineural invasion plus additional risk factors. Five-year cumulative local recurrence was about two percent for tumors with no risk factors, about five to six percent for solitary large-caliber perineural invasion, and climbed to seventeen percent when other risk factors were layered on. The major poor outcome composite followed the same gradient — under one percent with no risk factors, roughly six percent with solitary large-caliber perineural invasion, and twenty-four percent when combined with other risk factors, a statistically significant and clearly clinically meaningful jump between those latter two groups. In absolute terms, only three of the forty-three solitary large-caliber perineural invasion patients had any poor outcome at all — two local recurrences, one of whom also developed a nodal metastasis and died of disease, and one additional patient with an isolated nodal metastasis.

The authors are honest that the solitary large-caliber perineural invasion group is small and the study is underpowered to fully characterize local recurrence risk in that subgroup — this is a genuine statistical limitation, not just a formality. But the directionality is consistent and the gap between solitary large-caliber perineural invasion and large-caliber perineural invasion-plus-other-risk-factors is wide enough to be clinically believable even with modest numbers.

Practical takeaway, and I'd call this cautiously practice-informing rather than fully practice-changing given the small denominator: when large-caliber perineural invasion is truly the only adverse feature — clear margins, no size, depth, differentiation, or lymphovascular concerns riding along with it — the data now support that these tumors behave closer to routine low-risk squamous cell carcinomas than to the high-risk perineural invasion tumors adjuvant radiation guidelines were largely built around. That supports individualized, more conservative decision-making about adjuvant radiation in this specific solitary large-caliber perineural invasion scenario, ideally in tumor board discussion, rather than reflexively escalating every large-caliber perineural invasion result to radiation regardless of context. It does not settle the question definitively, and larger prospective data specifically powered for this subgroup would still be valuable.

That wraps this month's roundup — an acral melanoma paper arguing for site-tailored surveillance, an evidence-ranked look at oral photoprotective ingredients, a sobering case series on language access and melanoma stage, and reassuring though still preliminary data on solitary large-caliber perineural invasion in cutaneous squamous cell carcinoma. Thanks for listening, and we'll be back with the next issue.