Welcome back to the journal review. This month we're working through the November twenty twenty-six issue of the Journal of the American Academy of Dermatology, and I've pulled four pieces that are genuinely useful for anyone doing cutaneous oncology and Mohs surgery — two original research studies, a dermoscopy clinical pearl, and a practical review on perioperative management of oncology patients on systemic therapy. Let's get into it. First up is a retrospective cohort study using the National Cancer Database, asking whether adjuvant radiation therapy actually improves overall survival in low-risk, node-negative Merkel cell carcinoma. This is a question that's been nagging at the field for a while. We know Merkel cell carcinoma is aggressive, with a real propensity for local recurrence and early nodal and distant spread, and adjuvant radiation has become something of a default in many practices because it improves locoregional control. But the National Comprehensive Cancer Network guidelines already carve out a low-risk subgroup — tumors one centimeter or smaller, not on the head or neck, without lymphovascular invasion, and without immunosuppression — where radiation can reasonably be omitted. The problem the authors identify is that prior survival data supporting or refuting radiation in this specific subgroup has been muddied by including patients with unknown nodal status and by not actually stratifying by these NCCN risk features. So this group set out to isolate the question cleanly. They queried the National Cancer Database from 2010 to 2020 for stage one and two, pathologically node-negative Merkel cell carcinoma treated surgically, then split patients into low-risk and high-risk buckets based on those same NCCN features, and within each bucket compared those who got adjuvant radiation to those who didn't. Here's where the methodology gets genuinely instructive. Because this is registry data and not a randomized trial, the patients who received radiation and those who didn't are not comparable at baseline — sicker or higher-risk patients might be steered toward radiation, or alternatively, frailer patients might be steered away from it, and either direction would bias a naive comparison. To handle this, the authors used propensity score matching before building their Kaplan-Meier survival curves. In plain terms, propensity score matching estimates, for each patient, their probability of having received radiation based on all their observed characteristics — age, comorbidity, tumor features, facility type, and so on — and then pairs treated and untreated patients who have very similar predicted probabilities. The effect is to recreate, as best the observed data allows, something that looks like a randomized comparison, stripping out the confounding that comes from channeling certain patients toward or away from treatment. They layered on adjusted Cox proportional hazards models as well, which instead of comparing survival at one fixed time point, compare the instantaneous risk of death between groups at every moment during follow-up, which is the right tool when patients are followed for different lengths of time and radiation's effect might not be constant throughout that follow-up. And on top of that, they ran landmark analyses at three, six, and twelve months — this is a specific fix for something called immortal time bias, where patients who received radiation necessarily had to survive long enough to receive it, which can artificially inflate the apparent benefit of treatment if you don't account for it. Restarting the survival clock at a fixed landmark after treatment decisions were made removes that distortion. The final cohort was a little over two thousand patients, with about one in five classified low-risk and the remaining four in five high-risk — which itself tells you something about how this disease typically presents. Among high-risk patients, there was a notable selection pattern: those who got radiation were actually less likely to have head and neck primaries, despite head and neck location being one of the defining high-risk features, likely reflecting real-world reluctance to irradiate that anatomically complex, morbidity-prone region. The key result: in low-risk patients, radiation was not associated with improved overall survival. Five-year survival was about eighty-five percent with radiation versus seventy-nine percent without, numerically favoring radiation but not statistically significant. In high-risk patients, however, radiation was associated with a real, statistically significant survival benefit — five-year survival around seventy-two percent versus sixty-four percent without it, a meaningful absolute gap. The landmark analyses told the same story, so this isn't an artifact of immortal time bias. The multivariable Cox model also flagged immunosuppression as an enormous independent risk factor in the high-risk group, nearly tripling mortality risk, along with positive margins and tumor size over one centimeter. The authors' take is that this validates the NCCN's risk stratification — these criteria are actually doing real work in identifying who benefits from radiation and who can likely be spared it. They reasonably note that the absence of benefit in low-risk patients probably reflects a floor effect: baseline cancer-specific mortality is already low in this group, so even if radiation reduces local recurrence, that doesn't translate into a detectable survival difference. The honest limitations here are substantial and worth naming: this is retrospective registry data without disease-specific survival, progression-free survival, distant metastasis-free survival, or locoregional recurrence data — all of which would have been more sensitive than overall survival for detecting a true local-control benefit from radiation. So this study can tell you radiation doesn't move the needle on survival in low-risk disease, but it cannot tell you radiation doesn't reduce local recurrences, which still matters clinically even without a survival signal. Practically, this is a genuinely useful, if not entirely practice-changing, piece of evidence. It won't overturn NCCN guidelines — it validates them. For the surgeon having that adjuvant radiation conversation with a genuinely low-risk, node-negative patient, this gives you solid data to support omitting radiation without an overall survival tradeoff, while reinforcing that your high-risk patients, especially the immunosuppressed and those with positive margins, are the ones where radiation referral should be pursued more assertively. Next, a smaller but conceptually interesting piece: a genomic analysis asking what mutations correlate with perineural invasion in high-risk cutaneous squamous cell carcinoma. The clinical gap here is straightforward — perineural invasion is one of our most important adverse prognostic features, baked into staging and treatment algorithms, but it's a histologic diagnosis made after the fact, usually on final pathology or even only after Mohs sectioning reveals it. There's no preoperative way to flag a tumor as likely to have it. This group mined a publicly available MD Anderson cutaneous squamous cell carcinoma cohort on cBioPortal — patients selected for aggressive features like metastasis, perineural invasion itself, large size, or subcutaneous invasion — all of whom had targeted next-generation sequencing with a five-hundred-plus gene panel. Methodologically, this is a straightforward but small retrospective genomic correlation study. With only thirty-nine patients total, the authors restricted their analysis to genes mutated in at least five patients, which is a sensible guardrail against chasing noise in ultra-sparse data, and then ran logistic regression to test each gene's association with perineural invasion, adjusting for tumor size and depth in multivariable models. It's worth pausing on why logistic regression is the right tool here: when your outcome is binary — invasion present or absent — rather than a continuous or time-to-event variable, logistic regression models the odds of that outcome as a function of predictors, producing an odds ratio for each gene that tells you how much the odds of perineural invasion change when that mutation is present versus absent. It's the natural counterpart to the Cox model we just discussed, suited for a yes-or-no outcome rather than a time-to-event one. The results: just about half the cohort had perineural invasion. Four genes stood out on univariate analysis as significantly associated with it — DDR2, FLT3, NOTCH2, and NOTCH1 — with odds ratios ranging from roughly five and a half up to nearly fourteen for DDR2, though that confidence interval was wide given the small numbers. Strikingly, every single patient with perineural invasion carried a mutation in at least one of those four genes, compared to under half of patients without invasion — a clean separation, even if the absolute numbers are small. One gene, MAP3K14, actually ran the other direction, associated with the absence of perineural invasion. When the authors filtered down to mutations annotated as functionally oncogenic, NOTCH2 held up as significantly associated with invasion, but NOTCH1 lost statistical significance — the authors are upfront that this likely reflects underpowering rather than a true null effect. The honest framing here is that this is hypothesis-generating, not practice-changing. Thirty-nine patients is a small dataset for a multi-gene analysis, there's no outcomes data linking these mutations to recurrence or death, and the clinicopathologic variables available were limited. But conceptually, it's an interesting signal that perineural invasion may not be a purely stochastic histologic event but could trace back to a distinct genomic phenotype involving Notch signaling and other invasion-related pathways. For now, this belongs in the "interesting, keep an eye on it" category rather than anything that should change how you counsel patients or order sequencing today — though if validated in larger cohorts, you could imagine a future where genomic profiling flags high perineural-invasion-risk tumors before the knife ever touches them. Moving to something very different in format: a clinical pearl on parallel-polarized dermoscopy for acral parallel pattern analysis. No statistics here, just a practical imaging tip. The challenge the authors lay out is one every dermoscopist doing acral lesion work knows intimately — distinguishing the parallel furrow pattern from the parallel ridge pattern, a heavily weighted component of the BRAAFF scoring system for acral melanocytic lesions. That distinction hinges entirely on where the pigment sits relative to the acrosyringium, the eccrine duct opening at the center of the ridge. Furrow pattern means pigment in the sulci with visible white acrosyringial dots along the ridges; ridge pattern means the pigment has effaced those dots. The trouble is that standard cross-polarized dermoscopy, the default mode on most hybrid dermatoscopes, works by suppressing superficial reflectance — polarized light randomizes after very shallow penetration, which is great for seeing vascular and deeper structures but tends to wash out keratin-based surface detail, making those acrosyringial landmarks faint or invisible on acral skin exactly when you need them most. Their solution requires no new hardware, just using a mode most hybrid dermatoscopes already have: switching to parallel-polarized imaging, which preserves that superficial reflectance and renders the acrosyringia as bright, conspicuous dots along the cristae. Their recommendation is simple — capture both views of the same acral field, toggle between them, and use the parallel-polarized image specifically to re-establish the anatomic grid of acrosyringial openings before committing to a furrow-versus-ridge call. They're careful to note cross-polarized dermoscopy isn't obsolete — it's still your tool for dotted vessels, rosettes, and shiny white structures — the two modes are complementary, not competing. For anyone doing serial surveillance or tricky acral lesion triage, this is a free, immediately actionable addition to your exam technique. Last, and most dense practically, is a review article out of Memorial Sloan Kettering on perioperative management of dermatologic surgery patients who are on active systemic cancer therapy. This isn't an original study with its own dataset — it's a structured, expert-based synthesis built from multidisciplinary meetings between four Mohs surgeons and two medical oncologists, cross-referenced against FDA package inserts and a targeted literature search. The authors are upfront that this is a real and growing gap: as targeted and immunotherapies extend survival, more patients are showing up for skin cancer surgery while actively on these agents, and there's been no good consolidated guidance on what to hold, what to continue, and for how long. They work through drug class by drug class. For VEGF inhibitors — bevacizumab, pazopanib, sorafenib, sunitinib — the concern is impaired angiogenesis translating into dehiscence and bleeding risk, and the practical guidance is to hold based on half-life: bevacizumab three to four weeks preoperatively given its long clearance, versus seven to ten days for the shorter-half-life agents like sunitinib and sorafenib, resuming only once the wound has adequately healed. For MEK inhibitors like trametinib, cobimetinib, and binimetinib, the concern shifts to major hemorrhagic events and wound healing complications, with a recommendation to hold roughly forty-eight hours before surgery and resume around a week out. BRAF inhibitors are a nice contrast — rather than a bleeding or healing risk, the main perioperative issue is the paradoxical MAPK activation that drives squamoproliferative lesions and keratoacanthomas, so the emphasis is on baseline skin exams, surveillance every couple of months while on therapy, and the reassuring note that these agents may actually enhance wound healing, meaning continuation through surgery is generally considered safe in conjunction with oncology. When BRAF and MEK inhibitors are combined, which is now the standard approach in BRAF-mutant melanoma, the practical compromise is to hold the MEK component — given its bleeding and healing signal — forty-eight hours preoperatively while the BRAF inhibitor can often continue, resuming per the individual drug timelines afterward. For venetoclax, a BCL-2 inhibitor, the dominant perioperative concern isn't bleeding or wound healing but neutropenia and infection risk, with febrile neutropenia reported in a meaningful minority of patients, so the recommendation is simply a preoperative complete blood count check and vigilant postoperative infection monitoring rather than drug discontinuation. Tyrosine kinase inhibitors like cabozantinib, axitinib, and lenvatinib again center on bleeding and wound healing, with hold times scaled to half-life — roughly three weeks for cabozantinib given its long half-life, as little as twenty-four hours for axitinib, and about a week for lenvatinib. Bruton tyrosine kinase inhibitors, ibrutinib and acalabrutinib, carry real reported bleeding risk after dermatologic surgery, and the guidance leans toward checking a preoperative blood count and considering a short three-to-seven day hold around surgery in discussion with oncology. What's worth flagging honestly is the evidence grading itself, since the authors report it transparently: most of these recommendations sit at level four evidence — expert opinion and institutional experience — with only the VEGF inhibitor and tyrosine kinase inhibitor sections reaching the slightly more robust level two-A. There is essentially no dermatologic-surgery-specific prospective data behind most of this. That doesn't make it less useful — quite the opposite, it's filling a genuine vacuum — but it means this is best read as a thoughtful, codified expert consensus rather than evidence-based protocol in the strict sense. Practically, this is one of those review articles that earns a permanent spot as a reference document rather than something you absorb once and forget. The actionable core for a practicing Mohs surgeon is to know, drug class by drug class, which agents threaten bleeding and wound healing versus which threaten infection via cytopenias, to anchor your hold times to half-life logic rather than a blanket rule, and to treat every one of these decisions as a conversation with the treating oncologist rather than a unilateral call — because for many of these patients, interrupting systemic therapy carries its own oncologic cost that has to be weighed against your surgical risk. That wraps our four articles for this November issue — a survival analysis that sharpens who truly needs adjuvant radiation in Merkel cell carcinoma, an early genomic signal for perineural invasion risk in squamous cell carcinoma worth watching but not yet acting on, a simple dermoscopy mode switch that costs nothing and clarifies acral pattern reads, and a practical perioperative roadmap for the growing population of surgical patients on active systemic cancer therapy. Thanks for listening, and I'll see you next month.