Welcome to this April 2026 review from the Journal of the American Academy of Dermatology. We've got four papers on the docket this month — a health economic analysis in lentigo maligna, a vulvar Paget disease cohort out of Shanghai, a large interobserver variability study in melanocytic pathology, and a SEER-based look at metastatic Merkel cell carcinoma. Let's get into it.

First up is a cost-effectiveness analysis comparing radiotherapy against topical imiquimod for lentigo maligna, and this is essentially a formal economic evaluation conducted alongside a randomized trial — the RADICAL trial — so think of it as a companion piece to a clinical outcomes paper rather than a standalone efficacy study.

Quick background on why this exists. Lentigo maligna is melanoma in situ on chronically sun-damaged skin, and its incidence is rising in aging populations in both the US and Australia — in the study referenced here it accounted for roughly one in six melanomas diagnosed in the States. For patients who aren't good surgical candidates — facial location, size, comorbidities, preference — radiotherapy and imiquimod are the two nonsurgical fallbacks. The RADICAL trial randomized 126 patients across Australia, New Zealand, and Brazil to one or the other, with treatment failure at twenty-seven months as the primary endpoint. The headline result there: treatment failure occurred in about one in five radiotherapy patients versus one in ten imiquimod patients, a difference that did not reach statistical significance — and importantly, the trial was disrupted by the pandemic and ended up underpowered, so this is a "no proven difference" rather than a "proven equivalence" result. Given that null finding, the authors reasoned: if we can't show one treatment is clinically superior, cost becomes a legitimate, clinically relevant axis for decision-making — which is exactly the gap this second paper fills.

Methodologically, this is a cost analysis performed prospectively alongside the trial, pulling actual resource-use data — visit counts, treatment courses, adverse event prescriptions — directly from the RADICAL dataset, then costing those out using Australian Medicare benefit schedule rates and pharmacy pricing, converted to US dollars. They looked at it from two angles: a health system perspective, capturing direct medical costs only, and a societal perspective, which layers on indirect costs — lost productivity and travel time to appointments, calculated using the human capital approach with actual driving times to treatment centers. That societal lens is a smart methodological choice here specifically because radiotherapy is such a visit-heavy therapy; if you only counted the health system's bill, you'd miss the burden falling on patients themselves.

The results are stark and honestly not surprising in direction, though the magnitude is worth knowing. Direct health system costs came out to about twenty-seven hundred US dollars per patient for radiotherapy versus about seventy dollars for imiquimod — roughly a thirty-seven-fold difference. That's driven almost entirely by visit burden: radiotherapy patients averaged around twenty-seven visits over the treatment and follow-up period compared with about six for imiquimod. Adverse event management costs were trivial and essentially identical between arms, so that's not where the gap comes from. Once you add in productivity loss and travel time, the societal cost gap widens further — around forty-four hundred dollars for radiotherapy versus about seven hundred dollars for imiquimod.

On limitations, the authors are upfront that this is built entirely on Australian costing inputs — Medicare benefit schedule fees, Australian pharmacy pricing, Australian average wages for the travel-time proxy — so the absolute dollar figures won't transport cleanly to other health systems, though the relative magnitude of the gap likely will. They also note they couldn't capture monetary impact on retired patients, who made up a substantial chunk of this older cohort, since retired patients were assigned zero income for productivity-loss purposes — probably underestimating the true societal cost of radiotherapy's visit burden in a population that's disproportionately retired.

Practically, here's the takeaway for you as a Mohs surgeon or dermatologic oncologist counseling a patient who isn't a great surgical candidate for LM: this doesn't tell you imiquimod is more effective — the underlying trial couldn't establish that. What it tells you is that when efficacy is a toss-up, the cost and burden difference is enormous and lopsided toward imiquimod, particularly in visit frequency, which matters enormously for an older, less mobile population. This is genuinely useful for shared decision-making conversations — you can honestly tell a patient that if efficacy is comparable, imiquimod will cost them a fraction of the clinic visits and out-of-pocket burden. It's not practice-changing in the sense of altering your efficacy assumptions, but it is a legitimate new data point to bring into that conversation.

Next, a brief report out of three centers in Shanghai looking at clinicopathologic features and prognostic factors in surgically treated extramammary Paget disease of the vulva — a retrospective cohort analysis. This is a smaller, focused paper, so we'll walk through it in that shape rather than a full five-beat structure.

The clinical problem: vulvar Paget disease is notorious for high positive margin rates, quoted here at forty to seventy-five percent, and recurrence rates in the thirty-to-fifty-five percent range, largely because of complex vulvar anatomy and the disease's multifocal growth pattern. There's been genuine disagreement in the literature about whether margin status actually predicts recurrence, and that's the specific question this cohort tries to clarify.

The authors pulled pathology database cases from three institutions spanning 2007 to 2024, identifying just over a hundred vulvar Paget disease cases, of which about two-thirds underwent definitive surgical resection — the rest were biopsy-only, largely elderly patients who declined excision after a risk-benefit discussion. After excluding incomplete records and biopsy-only cases, they landed on fifty-eight surgically treated patients for analysis — wide local excision in the majority, with radical vulvectomy, reconstruction, and lymphadenectomy in smaller subsets. This is a retrospective design essentially by necessity — vulvar Paget disease is rare enough that a prospective trial isn't realistic, so pooling pathology archives across three centers over nearly two decades is the pragmatic way to get an analyzable sample.

On results: nearly half the cohort had positive margins — about forty-seven percent. Two cases turned out to be secondary Paget disease from underlying urothelial carcinoma, and one from cervical adenosquamous carcinoma, a reminder that vulvar Paget disease always warrants a search for an underlying visceral malignancy. Among primary cases, invasive disease was significantly more common in patients with advanced stage. With a median follow-up right around four years, the recurrence rate was about twenty-nine percent and mortality about fourteen percent. On multivariate Cox analysis, two factors emerged as independent predictors of progression: radical vulvectomy carried roughly a four-fold higher progression risk compared with wide local excision, and positive margins carried a similarly elevated four-fold risk compared with negative margins — both statistically significant and clinically meaningful. Positive margins also trended toward worse overall survival, though that association didn't reach significance. Notably, adjunctive therapies — photodynamic therapy, chemotherapy, imiquimod — did not improve prognosis, which the authors reasonably attribute to confounding by indication: these treatments were disproportionately used in patients who already had positive margins or advanced-stage disease, not a true treatment failure signal.

The authors don't dwell heavily on limitations in the text provided, but the retrospective, multi-decade, multi-institutional design carries the usual caveats around treatment heterogeneity and selection bias — patients who got radical vulvectomy versus wide local excision weren't randomly assigned, so that hazard ratio likely reflects a sicker, more extensive-disease population being selected for more radical surgery, not necessarily surgery itself driving progression.

For your practice, the actionable point is squarely about margins: this adds another data point to the literature supporting margin status as a genuine independent driver of recurrence and progression in vulvar Paget disease, which reinforces the case for meticulous margin assessment — and for those of us doing Mohs or staged excisions for extramammary Paget disease elsewhere on the body, it's a reminder that the field broadly continues to validate margin control as the lever that most directly affects outcomes in this entity.

Third, and this one is squarely relevant to your daily diagnostic life: a retrospective cohort study on interobserver variability in the histopathologic diagnosis of melanocytic neoplasms, out of Washington University in St. Louis.

The background here is the familiar epidemiologic paradox — melanoma incidence has climbed steadily for decades without a matching rise in mortality — and the authors lay out three possible explanations: genuine increased incidence of curable early lesions, overdiagnosis of low-malignant-potential entities that look histologically like melanoma, or diagnostic drift, where shifting criteria push more lesions toward a malignant label over time. Multiple prior studies have shown pathologists disagree with each other and even with themselves on melanocytic lesions, but this paper wanted to quantify that disagreement in real-world consultation practice and, critically, identify what's driving it.

Methodologically, this is a retrospective review of nearly fifteen hundred melanocytic lesions that had a diagnosis rendered at an outside institution and were then sent for consultation review at the reference institution's dermatopathology service between 2016 and 2021. The authors built an eight-tier ordinal classification — benign nevus, then mild, moderate, severe dysplastic nevus, then a genuinely clever innovation: they split the ambiguous "melanocytic proliferation of uncertain potential" category into two subclasses based on whether the outside pathologist did or didn't recommend definitive excision — followed by melanoma in situ and invasive melanoma at the top. That ordinal structure lets them capture not just whether the two institutions disagreed, but which direction the disagreement went — upgraded toward malignancy or downgraded toward benign — which is far more clinically informative than a flat concordance rate. They then layered on machine learning models — logistic regression, random forest, gradient boosting, and others — to identify which case-level features actually predict discordance, which is a reasonable way to handle a many-variable problem where you want to know relative feature importance rather than just isolated univariate associations.

The topline number: overall discordance between the outside and reference institution was thirty-three percent — one in three cases got a materially different diagnosis on second look. But that average badly undersells where the real problem lives. Dysplastic nevi, even though a minority of the total cohort, showed uniformly high discordance, peaking at ninety percent for moderately atypical dysplastic nevi. And the indeterminate-risk category — melanocytic proliferation of uncertain potential — was the biggest story: discordance ran about eighty to ninety percent depending on subclass, and when the outside pathologist had explicitly recommended excision, over seventy percent of those cases were downgraded on review, and most of those downgrades landed all the way at completely benign or mildly dysplastic. That's not a semantic disagreement — that's a case where the first read recommended surgery and the second read said it wasn't warranted. In sharp contrast, invasive melanoma diagnoses were quite stable, with discordance under eleven percent — so the variability problem is concentrated almost entirely at the ambiguous middle of the spectrum, not at the extremes.

On what drives that discordance: intent-to-consult and inclusion of a differential diagnosis in the outside report were both strongly associated with disagreement — makes sense, since a pathologist who flags uncertainty by requesting consultation or listing a differential is essentially self-selecting the ambiguous cases. More actionable clinically: lack of dermatopathology board certification and non-academic practice setting at the outside institution were both associated with higher discordance, while board certification and academic setting tracked with concordance. The random forest model performed best at predicting discordance, correctly classifying about eighty-three percent of cases, with predictors including the original diagnostic class, use of melanocytic lineage immunohistochemistry, anatomic site outside the head and neck, and absence of board certification. About one in five cases overall were flagged as involving a genuine management change — shifting between a "watch it" category and one with an explicit excision or treatment recommendation — and again, the indeterminate MPUP-with-excision-recommendation category drove the bulk of that, with over half of those cases downgraded all the way to benign.

Limitations are squarely acknowledged by the authors: single-institution retrospective design, which means there's no ground truth or outcomes correlation here — they can tell you two pathologists disagreed, but not which one was right, since there's no long-term follow-up data tied to this cohort. There's also inherent referral bias, since consultation cases are not a random sample of melanocytic lesions — by definition they're the ones someone thought worth a second opinion, which likely inflates the discordance rate relative to unselected practice.

Here's the practical takeaway, and I'd call parts of this genuinely practice-relevant rather than just interesting. This isn't news that melanocytic pathology has gray zones — you already know that. What is useful is the specificity: the indeterminate, uncertain-malignant-potential category is where discordance is highest and where it most often changes management, and a strikingly large share of "excise this" outside reads get walked back to benign on expert second review. That's a strong empirical argument for a lower threshold to seek dermatopathology consultation — particularly board-certified dermatopathology consultation — specifically for lesions sitting in that ambiguous MPUP zone before committing a patient to definitive excision, since board certification and academic setting were the two features most consistently tied to concordance. It's not telling you to second-guess every invasive melanoma diagnosis that crosses your desk — those were quite stable — but for the borderline, "atypical melanocytic proliferation, recommend excision" report, this data supports pausing for a second read before you take that patient to the OR or the Mohs suite.

Last, a brief report using the Surveillance, Epidemiology, and End Results program — SEER — database, examining survival trends, prognostic factors, and metastatic patterns in metastatic Merkel cell carcinoma.

The gap being addressed: we don't have great population-level data on what drives site-specific metastatic spread in Merkel cell carcinoma at diagnosis, and prognostic factors specifically within the metastatic subgroup are understudied. The authors pulled SEER-17 data from 2010 through 2022 using the relevant histology and topography codes, excluding patients without documented metastatic status, then ran multivariate logistic regression to look at what predicts metastasis to specific sites, and multivariate Cox regression to look at what predicts disease-specific survival within each metastatic site group. They also split the cohort at 2017 — the year avelumab was approved for metastatic Merkel cell carcinoma — to see whether survival improved in the immunotherapy era, which is a sensible natural experiment given SEER can't tell you who actually received which drug.

Out of about seventy-seven hundred total Merkel cell carcinoma patients meeting inclusion criteria, the metastatic subsets were small — around 164 with liver metastases, 145 bone, 84 lung, and just 22 brain — so treat these numbers with appropriate caution given the sample sizes, especially for brain metastases, which the authors didn't even have enough events to analyze survival predictors for. Head and neck primary tumors carried significantly increased odds of liver metastases compared with extremity primaries, with a similar trend toward bone. Tumor size over 20 millimeters was, unsurprisingly, a strong and significant predictor of metastasis to liver, bone, and lung alike.

On the survival-trends side, the results were site-specific and genuinely interesting: two-year disease-specific survival in patients with bone metastases nearly tripled between the earlier and later era — from about fifteen percent up to forty-three percent, a statistically significant and clinically substantial gain, plausibly tracking with immunotherapy availability. Liver metastases showed a numerical improvement, roughly thirteen to twenty-two percent, but that did not reach statistical significance. Lung metastases showed no meaningful improvement at all — survival was poor and essentially flat across both eras. So the immunotherapy-era survival gain being celebrated in Merkel cell carcinoma more broadly does not appear to be distributing evenly across metastatic sites. On prognostic factors within the metastatic groups, older age over 75 was associated with significantly worse survival in the bone metastasis group, and nonmetropolitan residence was associated with significantly worse survival in both the liver and lung metastasis groups — the lung group also showing worse survival with lower household income — which the authors flag as a possible signal of disparities in access to specialty or immunotherapy care rather than a biological effect.

Limitations here are the ones inherent to any SEER brief report: no data on whether patients actually received immunotherapy, no molecular or biomarker detail on the metastatic tumors, and the very small subgroup sizes, particularly for brain metastases, limit how far you can generalize these hazard ratios.

Practically, I'd frame this as interesting and directionally useful rather than something that changes your Mohs or dermatologic oncology practice directly, since it's downstream of your usual scope — but it's worth knowing for patient conversations and referral urgency. The big-picture message worth carrying forward: not all metastatic Merkel cell carcinoma is prognostically equal in the modern era — bone metastases have seen a real, substantial survival improvement, while lung metastases remain a persistently poor-prognosis group even with immunotherapy access, and that's useful context when you're the one delivering a new metastatic diagnosis or coordinating with medical oncology. The rural-urban survival gap is also a reasonable prompt to be proactive about facilitating access to tertiary or immunotherapy-capable centers for patients in nonmetropolitan settings.

That wraps our four articles for April. To sum up the throughlines: cost matters and is now quantified for lentigo maligna's nonsurgical options; margin status keeps proving itself as the dominant modifiable driver of outcomes in Paget disease of the vulva; melanocytic diagnostic uncertainty is real, measurable, and concentrated in the indeterminate zone where a second dermatopathology opinion has genuine value; and metastatic Merkel cell carcinoma's prognosis in the immunotherapy era depends heavily on which organ is involved. Thanks for listening, and we'll see you next month.