Welcome back to the journal review. This month we're working through the March twenty twenty-six issue of the Journal of the American Academy of Dermatology — four pieces, ranging from a nice little 3D-printing workaround for confocal imaging, to a phase two trial of an oncolytic virus in low-risk squamous cell carcinoma, to two epidemiologic brief reports touching melanoma outcomes. Let's get into it.

First up is a JAAD Online technology pearl out of Fujian Medical University: a novel three-D-printed soft resin adapter to improve reflectance confocal microscopy imaging of elevated lesions. This is a pure technique piece — no cohort, no outcomes data, just an engineering solution to a recognized imaging problem, so I'll walk it through as the authors frame it: the challenge and then their fix.

The challenge is one every RCM user will recognize immediately. Reflectance confocal microscopy lenses are built and calibrated for flat skin. The moment you're imaging a raised or dome-shaped lesion, the probe's rigid geometry doesn't conform to the lesion's slope, so you get focus drift — the image plane keeps sliding off the tissue of interest as pressure and contact angle shift with even small hand movements. Practically, that means repeated re-scanning of the same lesion, which eats into your imaging time and your diagnostic confidence.

Their solution is a tapered lens sleeve, custom-designed in Solidworks and printed on a light-curing resin printer using an elastic, rubber-like fifty-A durometer resin — so it's soft, not rigid. The sleeve diameter is matched to the RCM lens itself, and the base of the adapter has a wavy, accordion-like fold built into it specifically so it can accommodate lesions of different heights without needing a different adapter for each one. Because the base is soft and flexible, it drapes and conforms to the surrounding skin surface, which lets the tip of the lens sleeve sit down at or below lesion level rather than perching unstably on top of a curved surface. Functionally, that gives the operator a fixed, stable platform — the adapter is doing the mechanical work of maintaining consistent contact pressure and angle, rather than relying on the operator's hand to compensate in real time.

There's no formal validation cohort here, no measured improvement in image quality metrics or scan time reduction — this is presented as a proof-of-concept fabrication note with supplementary video showing the before-and-after contact quality. So the honest takeaway is: this is an interesting, low-cost, reproducible fix for anyone doing high volume RCM on papules, nodules, or exophytic lesions — not something with outcomes data behind it yet, but the mechanical rationale is sound and the barrier to trying it yourself, given open access to CAD software and resin printers, is low. If your practice does confocal imaging of raised lesions and you're fighting focus drift, this is worth a look purely as a workflow fix, not as a claim of improved diagnostic accuracy.

Second article, and the meatiest one this month: a single-arm phase two study of intralesional talimogene laherparepvec — T-VEC — for lower-risk invasive cutaneous squamous cell carcinoma, out of the University of Arizona Cancer Center.

The clinical gap here is a real one for those of us treating field-cancerized patients. A meaningful minority of patients — the authors cite roughly a quarter to a third — develop numerous cutaneous squamous cell carcinomas over time, and repeated surgical procedures in that population are costly, cumulative, and in some anatomic sites like the lower leg, carry real wound-healing risk. Surgery also does nothing preventive — it treats the tumor in front of you and leaves the field behind. Intralesional chemotherapeutics like five-fluorouracil or methotrexate exist but are barely addressed in guidelines and don't engage the immune system. T-VEC is an FDA-approved herpes-simplex-virus-1 based oncolytic virus already used in unresectable melanoma; injected intratumorally it replicates preferentially in tumor cells, lyses them, and triggers local and potentially systemic antitumor immune priming. It had never been studied as monotherapy in lower-risk invasive squamous cell carcinoma, which is the tumor type most of us are actually managing day to day — so that's the gap this trial targets, including whether local immune priming might reduce subsequent primary tumor development, not just treat the injected lesion.

On methods: this was a single-site, single-arm phase two design, non-randomized. The authors don't frame it this way explicitly, but a single-arm design here makes sense for a first-in-disease-type study — before you invest in a randomized comparison against surgery or topical chemotherapy, you need proof that an oncolytic virus can even produce meaningful responses in this population, and with expected surgical cure rates near universal, a placebo or observation arm would raise real equipoise concerns. They used a Simon two-stage optimal design, which is a classic phase two oncology tool for stopping early if a drug isn't hitting a prespecified efficacof threshold — here testing whether overall response rate was at least eighty percent against a null of fifty percent or less. Eleven patients enrolled with twenty-four target lesions, well below their original target of twenty, because COVID-era recruitment truncated the second stage — worth flagging up front since it shrinks statistical power. Patients had zero-point-five to five-centimeter, well or moderately differentiated invasive squamous cell carcinomas, less than two millimeters deep, without perineural or vascular invasion — so genuinely lower-risk tumors — and lesions on the face, neck, hands, feet, nail units, and ankles were excluded, presumably both for cosmetic and functional risk-mitigation reasons and likely to avoid confounding by sites with different healing or immune microenvironments. Dosing followed the standard melanoma T-VEC protocol — a low starting titer, then a higher maintenance titer for subsequent injections — every two weeks for up to four injections, with structured follow-up out to two years.

The results are frankly striking for a phase two dataset. Overall response rate was one hundred percent — every one of the eleven patients and all twenty-four treated tumors responded to some degree. Complete response occurred in about nine in ten patients, involving twenty-three of the twenty-four tumors, with the sole non-complete responder still showing a sixty percent size reduction and long-term stability, felt clinically to represent residual scar rather than active tumor. Median time to response was around five weeks, and median duration of response was about seven months, with no recurrences observed during follow-up. A durable response — defined as lasting at least six months — was seen in roughly five out of six tumors. And critically for the immunoprevention angle, at two years post-treatment, patients had significantly fewer new primary invasive squamous cell carcinomas compared to their own pretreatment history — suggesting a possible field effect beyond just the injected lesion, though this was a secondary, exploratory observation, not the primary powered endpoint.

On limitations, the authors are upfront: this is small, non-randomized, single-site, with meaningful lesion-site exclusions that limit generalizability — you can't yet extrapolate this to facial or acral squamous cell carcinomas, which is where a lot of our own field-cancerized patients actually accumulate tumors. The trial was also underpowered relative to its original design because of pandemic-related accrual disruption, so while the response signal is large enough to be compelling, the confidence interval around that one hundred percent response rate is wide — the lower bound was only around seventy-six percent, still comfortably above the fifty percent null, but a reminder not to over-read a small numerator.

Practically, I'd frame this as genuinely exciting but not yet practice-changing. It is not something to start offering off-label tomorrow for your surgery-fatigued patients — this is one single-arm study from one center with eleven patients. But the combination of a very high complete response rate in lesions patients often want to avoid cutting, tolerability data that read as manageable, and a hint of reduced subsequent tumor burden, makes this the kind of paper that should make you watch for a randomized follow-up. For patients with multiple lower-risk truncal or extremity squamous cell carcinomas who are genuinely averse to repeated surgery, this is a paper worth knowing about and discussing as an emerging option in trials, not yet as standard care.

Third, a brief report using the SEER database — Surveillance, Epidemiology, and End Results — looking at melanoma survival disparities in Asian American and Pacific Islander patients compared with non-Hispanic White patients.

The background here builds on prior work already showing that AAPI patients, despite lower overall melanoma incidence, have worse melanoma-specific survival than non-Hispanic White patients. What wasn't established is whether that survival gap simply reflects AAPI patients presenting at more advanced stages — which would point squarely at an access-to-diagnosis problem — or whether the disparity persists even after you account for stage, which would implicate something beyond just late detection.

Methodologically, this is a retrospective population-based cohort analysis using SEER's Detailed Asian and Pacific Islander Groups Database spanning three decades, nineteen ninety-two through twenty twenty-two. A SEER-based design is really the only feasible way to study this question — melanoma in AAPI patients is uncommon enough that no single institution or even multi-site academic collaborative would accumulate adequate numbers, and SEER's population-based structure lets you capture stage at diagnosis and cause-specific mortality across a huge, generalizable sample. They used Cox proportional hazards models to compare melanoma-specific mortality between groups, adjusting for stage, with non-Hispanic White patients as the reference.

The numbers: just over three thousand AAPI melanoma cases against three hundred seventy-five thousand non-Hispanic White cases — reflecting the known incidence disparity — and AAPI patients had significantly worse melanoma-specific mortality overall. They were also significantly more likely to present with regional or distant disease and less likely to present with in situ or localized disease, a highly significant stage-distribution difference. Here's the key finding though: even after adjusting for stage at diagnosis, AAPI patients still had about a one-third higher risk of melanoma-specific death compared with non-Hispanic White patients — a statistically significant and clinically meaningful residual disparity. And when they broke it down stage by stage, the excess mortality risk was significant within localized disease, regional disease, and distant disease individually — roughly a one-and-a-half-fold increased risk for localized melanoma specifically, which is the group where you'd expect outcomes to be closest to equivalent if stage alone explained everything.

The authors' interpretation is that later-stage presentation only accounts for part of this gap — the persistence of excess mortality even in early, localized melanoma points toward other contributing factors: potential biologic or genetic differences in tumor behavior, delays in accessing treatment after diagnosis, language barriers, insurance access, and differing health-seeking behavior or cultural relationships with the medical system. As with any SEER brief report, the limitations are the ones you'd expect — this is registry data without granular clinical detail on time-to-treatment, comorbidities, treatment received, or socioeconomic status, so the "why" behind the residual disparity remains genuinely unresolved; the authors are appropriately calling for further investigation rather than claiming a mechanism.

Practically, this isn't a paper that changes your Mohs or excision technique, but it should sharpen clinical suspicion and counseling. Even in a lower-incidence population, we should not be reassured by a seemingly early-stage AAPI melanoma into complacency about follow-up intensity or margin adequacy, and it's a reminder to actively address access barriers — get these patients into your clinic faster, don't let diagnostic delay compound an already unfavorable biology or systems-level disadvantage. It's a call to action on equity and vigilance rather than a change to your surgical algorithm.

Fourth and last, another brief report — this one on the utility of the thirty-one gene expression profile test, or thirty-one GEP, specifically in AJCC pT1a melanomas, from Baylor College of Medicine.

The setup is a direct response to a discrepancy in the literature: one prior single-institution study found that one hundred percent of pT1a melanomas tested with the thirty-one GEP came back Class 1A — meaning essentially no prognostic information was being added by testing this group, since Class 1A is the lowest-risk category and consistent with what you'd already expect from stage alone. Another study found only about eighty-five percent were Class 1A, implying a meaningful minority of ostensibly low-risk thin melanomas were being reclassified as higher molecular risk. Given that gap, the authors turned to the SEER-DecisionDx linked database to get a much larger, real-world answer.

Methodologically, this is a retrospective database analysis, again the pragmatic choice given how granularly you'd need to sample pT1a melanomas with paired GEP results to answer this at any single center. They identified pT1a melanomas by histology and site codes, excluded anyone missing Breslow depth or ulceration status, and then split the cohort into low-risk and high-risk pT1a groups — high-risk defined by conventional adverse features: age under or equal to forty-two, head and neck location, Breslow depth over point-five millimeters, or mitotic rate of two or more per square millimeter. They note explicitly that they couldn't incorporate lymphovascular invasion into that risk stratification because SEER doesn't capture it, which they flag as a real limitation since some ostensibly low-risk cases could actually harbor that adverse feature unrecognized.

Results: across thirty-five hundred pT1a patients, ninety-five percent of the low-risk subgroup came back Class 1A on the thirty-one GEP, and just over ninety percent of the high-risk subgroup did as well — so the overwhelming majority of both groups tested as reassuringly low molecular risk, consistent with what stage and conventional histology already tell you. Within the low-risk group, none of the small handful reclassified as Class 1B or higher actually died of melanoma — zero events — so there was no ability to even assess whether that reclassification carried prognostic weight. In the high-risk pT1a group, however, patients who scored Class 1B or higher did have a substantially and significantly worse melanoma-specific survival compared to their Class 1A counterparts — a roughly five-to-six fold increase in hazard, a large and clinically meaningful effect. Within that high-risk group, older age was independently associated with worse survival, while mitotic rate, absolute Breslow depth, and head-and-neck location were not independently significant once age was accounted for.

The authors' own interpretation is fairly pointed: testing low-risk pT1a melanomas with the thirty-one GEP appears to be low yield — you get almost uniformly reassuring results, and even the rare non-reassuring result didn't translate into an observed survival difference in this sample. The signal that actually matters seems concentrated in the high-risk pT1a subgroup, where a non-Class-1A result does track with meaningfully worse survival. They do append real caveats: this cohort skews toward the high-risk pT1a group, lack of lymphovascular invasion data could be misclassifying some high-risk tumors as low-risk, transected specimens may include tumors that were actually understaged pT2a or higher, and the retrospective registry design can't support firm management recommendations.

Practically, I'd call this an incrementally useful, but not fully practice-changing, refinement of an already-live debate. It won't settle guideline committees on its own, but it does offer a reasonably data-driven argument for being selective about which pT1a melanomas you send for thirty-one GEP testing — reserving it for those with adverse features like younger age, deeper Breslow depth within the pT1a range, elevated mitotic rate, or head-and-neck location, rather than reflexively testing every thin melanoma. For the genuinely low-risk pT1a patient sitting in front of you, this data suggests the test is unlikely to change your surveillance intensity, and you can have that conversation about cost and utility with more confidence.

That wraps the March issue. To summarize the actionable versus the interesting: the T-VEC data is the one to watch most closely for a future randomized trial, potentially meaningful for surgery-averse squamous cell carcinoma patients down the line; the AAPI melanoma survival disparity data should sharpen your vigilance and access advocacy today; the thirty-one GEP pT1a analysis offers a reasonable, if not definitive, argument for more selective testing; and the confocal adapter is a nice piece of shop-class problem solving worth trying if focus drift on raised lesions is a recurring headache in your clinic. Thanks for listening, and I'll see you next month.