Welcome back to this month's journal review — we're covering the February twenty twenty-six issue of the Journal of the American Academy of Dermatology, and I've pulled four brief reports and original studies that all sit squarely in your wheelhouse: melanoma risk stratification, Merkel cell carcinoma surveillance, and two papers directly comparing Mohs micrographic surgery with conventional excision for Merkel cell carcinoma. Let's get into it.

First up is a brief report, essentially a validation study, looking at whether a gene expression assay can identify high-risk T1a melanoma patients who fall outside the features we already use to trigger sentinel lymph node biopsy. The clinical problem here is one you know well — T1a melanoma is, by definition, a low-risk category, but a meaningful minority still harbor nodal disease, and our existing tools for flagging who needs a sentinel node biopsy are mediocre. The MIA nomogram is the standard comparator, and the authors note upfront that it has limited accuracy. The tool under study is the clinicopathologic gene expression profile, called CP-GEP, marketed as the Merlin assay, which combines Breslow thickness, patient age, and expression of eight target genes plus two housekeeping genes from the primary tumor, previously validated as a predictor of sentinel node positivity.

Methodologically, this was a retrospective, multi-institutional cohort — twelve international centers, a hundred fifty-seven T1a patients who actually underwent sentinel node biopsy between 2000 and 2023. The authors ran CP-GEP on archival tissue and kept the results blinded during testing, which is the right design choice here — you want a clean, retrospective diagnostic accuracy study using banked tissue and known outcomes, because prospectively enrolling enough T1a patients with node positivity to power this would take forever given how rare events are in this population. That's also why this reads like a proof-of-concept rather than a definitive trial — the authors are upfront that the number of node-positive cases was small.

The results are the interesting part. Overall node positivity in this cohort was about three percent. CP-GEP classified the large majority — a hundred thirty-seven of a hundred fifty-seven — as low risk, and only twenty as high risk. That would translate to roughly an eighty-seven percent reduction in biopsies performed, with a negative predictive value around ninety-eight and a half percent and specificity around seventy-two percent. And here's the clinically meaningful split: the post-test node-positivity rate was about one and a half percent in the CP-GEP low-risk group versus fifteen percent in the high-risk group. That fifteen percent figure clears the ten percent threshold where NCCN guidelines say biopsy should be offered. Compare that to MIA — even when the authors pushed the MIA threshold up to ten percent, the resulting high-risk group still had a positive rate under ten percent, meaning MIA never actually identified a subgroup that clears the guideline-relevant bar. Specificity for MIA at the five percent threshold was around twenty-four percent, versus CP-GEP's seventy-two percent — a substantial difference in how efficiently these tools triage patients.

The discussion places this in context with other genomic tools — this reproduces the prospective MERLIN-001 trial's zero versus twenty percent split between low- and high-risk groups, which is reassuring external validation. By contrast, the thirty-one-gene expression profile test, DecisionDx-Melanoma, has not identified a T1 subgroup exceeding ten percent positivity in other cohorts, and at least one independent series classified essentially all pT1a melanomas as low risk, limiting its discriminatory value in this specific population. The honest limitation here is the small absolute number of node-positive events feeding these percentages, so confirmation in larger datasets is still needed — and it's worth flagging that this study was industry-funded by the assay manufacturer, which doesn't invalidate the findings but is relevant context.

Practical takeaway: this is genuinely interesting and biologically plausible, and unlike MIA, CP-GEP actually produces a high-risk group that clears guideline thresholds for offering biopsy. Within the NCCN framework that already allows considering biopsy for T1a lesions over point-five millimeters with adverse features, a high-risk Merlin result looks like it could function as a strong additional adverse feature. I'd call this practice-informing rather than fully practice-changing yet — promising enough to discuss with your multidisciplinary tumor board, but the event numbers are still thin enough that I wouldn't consider it definitively validated.

Next, a prospective multicenter study on circulating tumor DNA, or ctDNA, in Merkel cell carcinoma surveillance — and this one has real implications for how you sequence follow-up imaging. Background: Merkel cell carcinoma recurs in about forty percent of patients, and ctDNA has emerged as a blood-based biomarker for detecting recurrence early, sensitive and specific enough that it's now already been folded into the twenty twenty-five NCCN surveillance guidelines. But the gap the authors identify is practical — nobody has laid out what to actually do when ctDNA turns positive before there's any clinically evident disease. How urgently should you image? How should the level of ctDNA itself inform that decision?

The design is a prospective, longitudinal, observational cohort across six tertiary Merkel cell centers, two hundred fifteen patients, six hundred sixty-nine total ctDNA tests, using a tumor-informed assay — meaning each patient's tumor undergoes whole-exome sequencing to identify their specific somatic mutations, and a personalized multiplex assay is then built to detect those same mutations in plasma. This tumor-informed approach is why the assay achieves such high specificity — it's not looking for generic circulating tumor markers but for a patient-specific mutational fingerprint. The prospective multicenter design makes sense for a rare cancer like this: you need pooled numbers across major referral centers to get enough surveillance events, and prospective collection with scheduled testing avoids the bias of only testing when clinicians already suspect recurrence.

The results are striking. Any positive ctDNA test during surveillance was associated with an eighteen-fold increase in the hazard of subsequent clinical recurrence, and this held up after adjusting for other risk factors. Among ctDNA-positive patients, seventy-seven percent went on to develop clinically evident disease within a year. The median lead time — meaning how far in advance ctDNA flagged the recurrence before imaging or exam caught it — was about two and a half months. And the level of ctDNA itself was informative: levels above ten molecules per milliliter tended to precede clinical recurrence within about three months, levels between one and ten molecules per milliliter within about six months, and levels below one molecule per milliliter took closer to nine months to manifest clinically. Meanwhile, patients who stayed ctDNA-negative throughout had a low cumulative recurrence risk — well under ten percent even out to a year.

The authors' own framing is refreshingly practical rather than purely mechanistic: a positive ctDNA test detects recurrence roughly three months earlier than imaging alone, negative ctDNA can be used to justify spacing out imaging, and positive ctDNA — especially at higher levels — should prompt closer, more urgent follow-up. The limitation they flag honestly is that this is real-world data with variable timing and frequency of both imaging and ctDNA testing across sites, though they note most patients did have both modalities tracked in parallel.

Practical takeaway: this is about as close to practice-changing as a biomarker paper gets for Merkel cell carcinoma, particularly because ctDNA surveillance is already guideline-endorsed — what this paper adds is a level-stratified framework for how aggressively to chase a positive result. A very high ctDNA level should trigger near-term imaging within weeks, not months; a borderline low-level positive still needs closer follow-up than a negative test, but perhaps not urgent same-week imaging. This is useful, actionable guidance for anyone co-managing Merkel cell patients with medical oncology.

Third, a retrospective chart review from a single tertiary academic center — Penn — looking at guideline compliance and timeliness of sentinel lymph node biopsy in Merkel cell carcinoma, comparing patients treated with Mohs micrographic surgery versus conventional excision. The background here is important context: NCCN recommends sentinel node biopsy for essentially all primary Merkel cell carcinoma given the thirty to forty percent risk of occult nodal disease, but prior national database studies had suggested that Mohs surgery is associated with lower rates of actually getting that biopsy done — a finding the authors worried might not generalize to a well-coordinated tertiary care setting.

This is a straightforward retrospective chart review, single institution, spanning 2006 through 2023, a hundred fifty-eight patients total — sixty treated with Mohs, ninety-eight with conventional excision. The rationale for a single-center design is really to stress-test the generalizability question directly: does the national database signal about Mohs being a barrier to biopsy actually hold up when you have a coordinated multidisciplinary team managing timing between the Mohs surgeon and the surgical oncologist doing the node biopsy?

The results say no, largely. Sentinel node biopsy compliance was high and statistically similar between groups — about eighty-three percent for Mohs versus ninety-three percent for conventional excision, a difference that did not reach statistical significance. Time from diagnostic biopsy to sentinel node biopsy was essentially identical between groups, around thirty-seven days either way. This is despite the Mohs cohort being heavily skewed toward head and neck tumors — eighty-eight percent of Mohs cases were head and neck versus only about twenty-one percent of the excision cohort — a location that's historically associated with lower compliance and more technical difficulty. Where the groups did diverge significantly was sentinel node biopsy failure — the biopsy was attempted but unsuccessful in thirty percent of the Mohs cohort versus about five and a half percent of the excision cohort, and essentially all of those failures in the Mohs group occurred in head and neck cases. That's a real, clinically meaningful signal, and it's about anatomic location and lymphatic mapping difficulty rather than about Mohs itself delaying or discouraging biopsy.

The authors' own interpretation is that their high compliance, in contrast to the national database literature, is likely attributable to a highly coordinated multidisciplinary team and tertiary care infrastructure — patient health literacy and insurance access may also play a role, though that's speculative on their part. They're appropriately honest about limitations: retrospective, single tertiary center, so this may not generalize to community practice settings where that same multidisciplinary coordination doesn't exist, and no long-term recurrence or survival data were captured.

Practical takeaway: reassuring rather than practice-changing — it pushes back on the narrative that choosing Mohs for Merkel cell carcinoma inherently compromises nodal staging, provided your practice has the same kind of tight coordination with surgical oncology that this center describes. The more actionable nugget is the head-and-neck sentinel node failure rate — worth an explicit conversation with your surgical oncology colleagues about mapping technique and timing when treating head and neck Merkel cell carcinoma with Mohs, since failure to localize the node was the dominant issue.

Which brings us naturally to the fourth paper, also from Penn, a single-center retrospective cohort directly comparing Mohs micrographic surgery to conventional excision for Merkel cell carcinoma — but this time looking at oncologic outcomes rather than biopsy logistics. Background: conventional excision carries a local recurrence rate historically cited between ten and forty percent, and the premise is that Mohs' comprehensive margin assessment should reduce positive margins and local recurrence, but prior comparative studies have been small case series, and larger database studies haven't captured recurrence or metastasis data at all.

The design is again a single-center retrospective cohort, 2006 through 2022, a hundred seventy-six patients — sixty-one treated with Mohs, a hundred fifteen with conventional excision — with treatment assignment based on provider or tumor board judgment rather than randomization. That's an important methodological point to flag: this is not a randomized comparison, so selection bias is baked in, and in fact it cut against the Mohs group here — the Mohs cohort had significantly larger preoperative tumor diameters, was overwhelmingly head and neck location, and received adjuvant radiation far less often, thirty-six percent versus sixty-four percent in the excision group. A retrospective design is really the only feasible option for a cancer this rare — you'd need decades to randomize enough patients — but it means any outcome difference has to be interpreted against a Mohs cohort that, if anything, looked higher-risk at baseline.

Despite that stacked deck, local recurrence was significantly lower with Mohs — one point six percent versus eleven percent overall, and three-year local recurrence risk by Kaplan-Meier was two percent for Mohs versus about fifteen percent for excision, a difference that was both statistically significant and clinically substantial. Rates of metastasis were similar between groups, and there were no significant differences in disease-free, disease-specific, or overall survival. The authors note their local recurrence rate with Mohs was even lower than a previously published nine percent, which they attribute to consistent use of cytokeratin-20 immunostaining — used in one hundred percent of their Mohs cases versus only about half of excision cases — allowing better intraoperative discrimination of tumor from surrounding tissue.

The authors' own conclusion is measured: Mohs with cytokeratin-20 may lower local recurrence with comparable survival, and the benefit may be greatest in localized, stage one or two disease, since metastases beyond initial staging in the Mohs cohort were concentrated in stage three patients. Limitations are exactly what you'd expect and the authors state them plainly — small sample size, modest average follow-up of about thirty-one months, retrospective design, and the inherent selection bias of treatment assignment by clinical judgment rather than randomization.

Practical takeaway: this is a meaningful, hypothesis-supporting addition to a still-thin evidence base, and it's notable that the local recurrence benefit held up despite the Mohs cohort being biased toward larger, more difficult, more radiation-undertreated tumors — if anything that strengthens the signal. It's not fully practice-changing on its own given the retrospective, non-randomized design and modest follow-up, but combined with the mechanistic rationale of complete margin assessment plus cytokeratin-20, it's a reasonable argument for favoring Mohs where feasible for Merkel cell carcinoma, especially in cosmetically sensitive or anatomically constrained head and neck sites, while continuing to rely on adjuvant radiation and multidisciplinary staging to manage systemic risk.

That wraps up this month's four papers. Taken together, there's a nice thematic throughline — better biomarkers and better surgical technique both pushing toward more individualized, more evidence-based management in melanoma and Merkel cell carcinoma, two diseases where our staging tools have historically been blunt instruments. Thanks for listening, and I'll see you next month.