Welcome back to this month's journal review. This is the July 2026 issue of JAMA Dermatology, and we've got four pieces worth your time — a dermoscopy study on melanonychia in skin of color, a two-case report on a nasty EGFR-MET combination toxicity, an editor's commentary putting the brakes on nicotinamide chemoprevention enthusiasm, and a claims-based look at who's actually destroying actinic keratoses in the Medicare population and how that's shifting. Let's get into it.

First up is an original cross-sectional study — an international, multicenter effort characterizing dermoscopic features of longitudinal melanonychia specifically in patients with skin of color. You know the clinical problem here better than most: a pigmented nail band that could be anything from benign melanocytic activation or ethnic melanonychia all the way to subungual melanoma, and the dermoscopic criteria we've all learned — asymmetry, blurred borders, Hutchinson sign — were largely derived in lighter-skinned populations. Patients with skin of color have a higher background rate of both subungual melanoma and benign ethnic melanonychia, which stacks the diagnostic uncertainty even higher, so the gap this group is addressing is straightforward: do the standard criteria even hold up in Fitzpatrick four-to-six skin, or do we need a different rulebook.

Methodologically, this was a retrospective collection of dermoscopic photographs pooled through the International Dermoscopy Society's imaging task force and the Council for Nail Disorders — eighty images from eleven institutions. Malignant lesions required histopathologic confirmation; benign lesions were confirmed either by biopsy or by at least two years of documented clinical stability. Three masked expert reviewers independently scored a battery of dermoscopic and clinical features as present or absent, and their scores were averaged into a consensus score. This design makes sense for a rare-outcome, image-based question — you're never going to prospectively accrue a meaningful number of biopsy-proven subungual melanomas in skin of color at any single center, so pooling retrospective images across institutions and continents, with masked multi-rater scoring to control for the inconsistent dermoscopy training the authors themselves flag as a limitation, is really the only feasible approach.

The results: of the eighty lesions, twenty-one were malignant melanocytic — seventeen melanoma in situ and four invasive — and fifty-nine were benign. Two features came out statistically significant: multiple colors within the band, and irregular band thickness, both more common in the malignant group. Black-brown coloration specifically was seen in roughly six in ten melanomas versus one in four benign lesions, a meaningful separation. But — and this is the clinically important part — the classic criteria we lean on in lighter skin, asymmetry, blurred borders, Hutchinson sign, loss of parallelism, did not discriminate in this cohort. Pigmentation width also didn't separate benign from malignant, which pushes back against the commonly cited forty-percent-of-nail-plate cutoff. In an exploratory sub-analysis, the authors also found that a brown background color tracked more with melanocytic hyperplasia than with malignancy per se, which helps explain some of the conflicting prior literature on background color as a cancer predictor.

The authors are appropriately humble in the discussion — fewer dermoscopic criteria may be meaningful in skin of color than the standard algorithms assume, and multiple colors, irregular band thickness, and black-brown pigmentation should raise concern and prompt biopsy, particularly for a solitary digit. Limitations are real: this is a small sample even though subungual melanoma is genuinely rare, image quality was inconsistent across contributing sites, interobserver agreement was only fair for most features by Fleiss kappa, and using two-year stability as a benign surrogate risks verification bias — some of those "benign" lesions might simply not have been followed long enough to declare themselves.

Practical takeaway for you: this is not practice-changing in the sense of a validated new algorithm, but it is a legitimate signal to recalibrate your threshold in patients with skin of color — don't be reassured by the absence of asymmetry or blurred borders the way you might in a lighter-skinned patient, and do weight multicolor, irregular thickness, and black-brown pigment more heavily, especially on a single digit. Biopsy threshold should stay low regardless; this data just refines which dermoscopic red flags are actually earning their keep in this population.

Next, a research letter — really a two-case report — on ecthymalike ulcerations in lung cancer patients on combined amivantamab and osimertinib. Quick framing: osimertinib is a familiar EGFR inhibitor with the expected xerosis and papulopustular acneiform toxicity profile, and amivantamab is a bispecific EGFR-MET antibody increasingly added on for resistance. The combined cutaneous toxicity profile hasn't been characterized, and that's the gap this letter fills.

Case one: a woman in her sixties on osimertinib who had amivantamab added for progression, developed xerosis and asteatotic dermatitis within a month, which progressed despite doxycycline and topical steroids into punched-out, crusted ulcerations that spread from her fingertips to her scalp, back, and thighs, flaring with each amivantamab infusion. Biopsy showed a neutrophil-rich fibrinonecrotic crust without a significant dermal or perifollicular infiltrate, and — critically — tissue culture grew Pseudomonas aeruginosa, despite negative surface swabs and negative viral PCR. Stopping both drugs and treating with ciprofloxacin and topical gentamicin cleared it, and she was able to resume both oncologic agents without recurrence.

Case two mirrors this almost exactly: a man in his seventies, osimertinib restarted with amivantamab added at recurrence, prophylactic doxycycline and topical steroid on board, who nonetheless developed asteatosis progressing to painful ulcerations. Same biopsy pattern, same negative special stains, same story — negative swab, positive tissue culture for Pseudomonas. Stopping therapy and a week of ciprofloxacin plus gentamicin ointment resolved things, and he resumed osimertinib.

The teaching point here is really about pattern recognition and a diagnostic pearl. This eruption doesn't map cleanly onto either drug's known toxicity alone — it borrows the ulceration and scalp and intertriginous predilection of amivantamab toxicity and the acneiform, xerotic background of osimertinib toxicity, but with a pseudomonal infection sitting underneath that a surface swab will miss every time. The authors propose a two-hit mechanism: EGFR inhibition disrupts the barrier first, producing the xerosis and fissuring, and then steroid and tetracycline-driven dysbiosis, compounded by impaired MET-dependent wound repair from the amivantamab, opens the door for pseudomonal colonization. The actionable point for us: if you're seeing refractory ecthymalike ulcers in a patient on this combination that isn't responding to tetracyclines and topical steroids, get a tissue culture, not just a swab, before you write this off as a drug-class toxicity — targeted antipseudomonal therapy can clear it and let the patient stay on their cancer treatment without a prolonged interruption.

Third is an editor's note — a commentary, not a primary study — responding to a companion cost-effectiveness paper on oral nicotinamide for keratinocyte carcinoma prevention. Worth walking through because it's a good exercise in reading economic modeling skeptically. The companion paper, by Perez and colleagues, modeled nicotinamide implementation within the Veterans Health Administration and found it dominant — meaning cost-saving and effective — with roughly six hundred keratinocyte carcinomas prevented annually, net savings approaching three hundred sixty-five thousand dollars, and an incremental cost-effectiveness ratio well within accepted thresholds even under non-VHA cost assumptions.

The editorialists' point is that this model is only as good as the clinical evidence feeding it, and that evidence is shakier than it looks. The original ONTRAC trial showed a roughly 23% reduction in new nonmelanoma skin cancers at twelve months, but in a small, high-risk Australian cohort. The follow-up ONTRANS trial, in solid organ transplant recipients — a very different immunosuppressed population with different KC biology — missed its primary endpoint entirely, though it still showed a 24% reduction in invasive squamous cell carcinoma specifically. A separate observational VHA study of more than thirty-three thousand veterans found a 14% overall reduction in KC hazard, rising to 54% in patients who started nicotinamide only after their first KC diagnosis — a large sample with propensity-score adjustment, which lends some statistical weight, but it's still observational, so it can't rule out confounding by healthier behaviors among nicotinamide users.

The commentary's conclusion is blunt and, frankly, well taken: the economics are favorable and the biology is plausible, but we do not yet have an adequately powered randomized trial in a representative, non-immunosuppressed population confirming real-world efficacy. This is squarely in the "interesting but not yet actionable" category — a good cost model built on hypothesis-reinforcing rather than hypothesis-confirming data. For your practice, nicotinamide remains a reasonable discussion point for motivated, high-risk patients, particularly those with a KC history, but it should not yet be framed to patients as a proven, guideline-level chemoprevention strategy the way we'd frame, say, field therapy for AKs.

Last, an original cohort study using Medicare claims to look at actinic keratosis destruction patterns from 2016 through 2023. The background gap is workforce-related: with advanced practice clinicians increasingly common in dermatology, the authors wanted to know not just aggregate AK destruction volume, but how that volume is distributed across dermatology physicians, nondermatology physicians, and APCs, and how it's changing per individual clinician.

Methodologically, this is a claims-based cohort study using the publicly available CMS Medicare Physician and Other Practitioners dataset, capturing Part B fee-for-service billing. They used the standard CPT code scheme — 17000 for the first lesion, 17003 billed per lesion for lesions two through fourteen, and 17004 as a bundled code for fifteen or more, multiplied by fifteen to estimate lesion count — which is the same approach prior AK-burden papers in this space have used, making this reasonably comparable to earlier snapshots. Claims data is really the only tool available here: it's the only source that captures billing behavior across the entire national Medicare population and across clinician types, though the authors are upfront that this inherently undercounts anyone not billing destruction — patients on topical or photodynamic field therapy are invisible to this analysis. Trends were tested with Kendall tau-B correlation, appropriate for monotonic trend detection over this eight-year span, and clinician type was assigned using a previously validated classification approach.

The results: nearly thirty-five thousand clinicians billed for AK destruction over the study period — about 41% dermatology physicians, 30% nondermatology physicians, and 28% APCs practicing dermatology. Total destroyed lesions came in around 286 million over the eight years, climbing from about 36 million in 2016 to roughly 37 million in 2023, with allowable payments exceeding 401 million dollars in that final year alone. The more interesting story is in the trend breakdown: total AK destruction volume and per-clinician volume both significantly declined among dermatology physicians and among nondermatology physicians over this period, while both metrics significantly increased among APCs. Overall, AKs treated per thousand Medicare beneficiaries rose significantly, driven predominantly by the APC segment.

The authors' discussion frames this as consistent with a broader dermatology workforce shift — more APCs entering the field and taking on a growing share of procedural AK management — though they also raise the possibility of a beneficiary demographic shift or increased emphasis on skin cancer prevention driving overall volume up. Limitations are appropriately modest here: Medicare-only means this doesn't generalize below age sixty-five, though AKs are of course more common in this older population anyway; the CPT-based lesion counting is conservative and likely undercounts true AK burden since it captures destruction only, not topical or photodynamic field treatments; and this is pure claims data with no clinical outcome or appropriateness information attached.

Practically, this isn't a paper that changes how you treat an individual patient's AKs, but it is a genuinely useful workforce and health-policy signal — the center of gravity for procedural AK management in the Medicare population is visibly shifting toward APCs, both in absolute terms and in per-clinician volume, while dermatology physician per-clinician destruction volume is trending down. That's worth knowing for practice structuring, supervision planning, and understanding where the next round of AK-related quality and training conversations in our field are likely to come from.

That wraps this episode. To summarize the throughline: refine your dermoscopic threshold for melanonychia in skin of color toward multicolor, irregular thickness, and black-brown pigment rather than the classic ABCDEF criteria; keep tissue culture on your radar for refractory ulcerations in patients on combined amivantamab and osimertinib; treat nicotinamide chemoprevention as economically attractive but not yet evidence-proven for routine recommendation; and recognize the ongoing shift of AK destruction volume toward advanced practice clinicians as a workforce trend worth tracking. Thanks for listening, and I'll see you next month.