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 we've got four pieces on deck — a major review on a family of aggressive dermal spindle cell tumors you're almost certainly biopsying and excising already, a multicenter case series on a nasty checkpoint inhibitor toxicity you may not be thinking about until it's in front of you, and two short letters-to-the-editor that sharpen some methodology around acral melanoma. Let's get into it. First up is a review article — not a primary data paper, but a state-of-the-art synthesis — on atypical fibroxanthoma, pleomorphic dermal sarcoma, and undifferentiated sarcoma, pleomorphic subtype. If you do any volume of Mohs on bald scalps and ears in elderly patients, this is squarely your territory, and the nomenclature history alone is worth knowing cold because it explains why the literature on this topic is such a mess. The authors start by untangling what is actually a single biological spectrum. Atypical fibroxanthoma, or AFX, and pleomorphic dermal sarcoma, or PDS, share the same ultraviolet-signature mutations — the classic C-to-T transitions — and the same driver genes, including TP53, FAT1, NOTCH1 and 2, CDKN2A, PIK3CA, and TERT promoter alterations. A methylation-profiling study of over two hundred of these tumors and their mimics found AFX and PDS are essentially molecularly indistinguishable from each other. The distinction between them is purely histologic: PDS is diagnosed the moment you see subcutaneous invasion, necrosis, perineural invasion, or lymphovascular invasion — any one of those upgrades the diagnosis. Undifferentiated sarcoma, by contrast — what used to be called malignant fibrous histiocytoma, then undifferentiated pleomorphic sarcoma — is a completely different animal molecularly. No ultraviolet signature, a complex karyotype instead, and critically, it's a diagnosis of exclusion that only gets made after ruling out every lineage-defined sarcoma that can mimic it. The authors make a clean recommendation here: stop using MFH, UPS, and subcutaneous UPS as separate labels and consolidate everything under the current World Health Organization term, undifferentiated sarcoma, with "pleomorphic" as an optional modifier. On workup, there's no established imaging or sentinel node guideline, but a few data points are worth banking. Ultrasound can be clinically useful — in one small series of thirteen tumors, nearly a quarter of lesions called AFX clinically were upgraded to PDS once subcutaneous involvement was seen on ultrasound and confirmed histologically. That's a meaningful clinical pearl: if you're on the fence about depth, a quick ultrasound before you commit to a Mohs plan may change your margin strategy. On prognosis, the numbers are worth knowing because they genuinely diverge between these three entities. A large series of over eleven hundred patients found five-year local recurrence around ten percent for AFX versus seventeen percent for PDS, with metastasis risk under one percent for AFX but sixteen percent for PDS — and when PDS does metastasize, it's usually within three years, most often to the lung. A separate large PDS series found tumor size over four centimeters was the single most important predictor of metastatic risk. And when you stack undifferentiated sarcoma against PDS head-to-head, US comes out worse across the board — five-year local recurrence around twenty-four versus twenty percent, overall metastasis about nineteen versus seven percent, distant metastasis twelve versus two percent, and five-year survival only fifty percent compared with sixty-seven percent for PDS. Immunosuppressed patients and solid organ transplant recipients fare particularly poorly across this whole spectrum. On treatment, the surgical data is about as clean as you'll get for a rare tumor. A review of nearly nine hundred AFX patients found Mohs micrographic surgery recurrence around two percent versus almost nine percent for wide local excision — and notably, no immunocompromised patient treated with Mohs recurred or metastasized, whereas forty percent of immunocompromised patients treated with wide excision recurred and ten percent metastasized. That's a clinically meaningful, practice-relevant gap, not a statistical curiosity. For those situations where Mohs isn't available, a systematic review using probabilistic modeling on just over a hundred AFX cases found that a two-centimeter peripheral margin was needed to clear ninety-five percent of tumors, leading to graduated recommendations — one centimeter for sub-centimeter tumors, up to two centimeters for anything larger than two centimeters. The authors are explicit that margin data in the older literature is confounded by inconsistent reporting, which is exactly why this modeling exercise mattered. For sentinel node biopsy, there's no consensus, and the authors are refreshingly honest that the mechanism of metastasis here — lymphatic versus hematogenous versus direct extension — isn't even settled, so borrowing the melanoma ten-percent threshold is really just an extrapolation. In their own practice, they skip sentinel node biopsy and routine imaging entirely for AFX, treating its metastatic risk as comparable to low-risk keratinocyte carcinoma, but they do get cross-sectional imaging of the regional nodal basin and chest for PDS, specifically because they've seen real metastatic PDS, especially in transplant recipients. The practical takeaway here is fairly clean. Complete margin control — Mohs or other complete circumferential peripheral and deep margin assessment techniques — should be your default for AFX and PDS; that's about as close to practice-changing as a rare-tumor review gets, given the recurrence gap. The nomenclature consolidation is useful for how you read pathology reports and talk to your pathologist, but it's housekeeping rather than something that changes what you do at the bedside. And the imaging and sentinel node guidance remains genuinely unsettled — reasonable to individualize based on tumor depth and immune status, but don't expect a guideline to bail you out here anytime soon. Next is a brief report — a retrospective multicenter case series with a comparative cohort analysis — looking at eosinophilic fasciitis triggered by immune checkpoint inhibitors. This is a toxicity most of us know about in passing but probably underappreciate in its distinct clinical behavior. The background here is straightforward: checkpoint inhibitors cause all sorts of cutaneous immune-related adverse events, but eosinophilic fasciitis specifically has been reported only rarely and poorly characterized. This French multicenter group identified fourteen patients with checkpoint-inhibitor-associated eosinophilic fasciitis, all meeting standard Pinal-Fernandez diagnostic criteria, and compared them against a historical cohort of seventy-eight patients with ordinary, non-checkpoint-inhibitor eosinophilic fasciitis from a single center. They also folded in a literature review of published cases for a sensitivity analysis. Methodologically, this is a nice teaching moment. With only fourteen cases in the exposure group, a historical comparator cohort is really the only feasible design — you're not going to randomize a rare drug toxicity — and the authors lean on nonparametric statistics throughout, which is the right call here. Notice that continuous variables like eosinophil count and age are reported as medians with interquartile ranges rather than means with standard deviations — that's a signal the data are skewed rather than normally distributed, which is typical for lab values like eosinophil counts that have a long right tail. When you're comparing two independent groups on a skewed or small-sample continuous variable, the standard tool is the Mann-Whitney U test, sometimes called the Wilcoxon rank-sum test, which doesn't assume a bell curve — instead it ranks all the observations from both groups together and asks whether one group's ranks are systematically higher or lower than the other's. For the categorical comparisons, like proportion with edema or proportion achieving remission, they're using chi-square or Fisher's exact test, with Fisher's being the appropriate choice for sparse cells, which you'd expect with a fourteen-patient arm. Now the findings. Ten of the fourteen patients had metastatic melanoma, and every one had received an anti-PD-1 agent, with about a fifth also exposed to anti-CTLA-4. The onset pattern is the standout clinical pearl: median time to diagnosis was almost a year of checkpoint inhibitor exposure, and about a month after the most recent infusion — so this is a late, delayed toxicity, not something you'd catch in the first few cycles. Compared with ordinary eosinophilic fasciitis, these patients were older — mid-sixties versus mid-fifties, a statistically significant and clinically sensible difference — and they presented with a more inflammatory phenotype across the board: edema in roughly nine in ten versus about six in ten, tendon retraction in half versus about one in five, muscle involvement in over ninety percent versus about sixty percent, and blood eosinophil counts more than double the comparator group, all statistically significant. Imaging findings were essentially universal in the checkpoint-inhibitor group — MRI was abnormal in all of them compared with about half of the non-checkpoint-inhibitor cohort, with fascial enhancement seen in the large majority. Despite this more aggressive presentation, treatment was intensified — more intravenous steroid pulses, and notably, intravenous immunoglobulin was used in about a fifth of the checkpoint-inhibitor patients and essentially never in the comparator group — yet outcomes were still worse. Complete remission without any residual sequelae occurred in only about one in seven checkpoint-inhibitor patients, compared with nearly half of the comparator cohort, and that gap was statistically significant. When the literature-review cases were folded in for a sensitivity analysis, the pattern held. At last follow-up, cancer had progressed in half the checkpoint-inhibitor cohort and four patients had died, underscoring that this is happening in a sick oncology population, not in isolation. Interestingly, three patients were rechallenged with checkpoint inhibition, with or without concurrent steroids, and none flared. The authors are appropriately cautious about limitations: this is retrospective, there's no quality-of-life data, the cohort is heavily skewed toward melanoma patients, and there's no matched oncologic control group to tell us whether eosinophilic fasciitis itself affects cancer survival versus simply co-occurring in a population with aggressive disease. For your practice, the actionable point is recognition and timing. If a patient on a checkpoint inhibitor — especially anti-PD-1 for melanoma — develops progressive limb edema, new peau d'orange change, or unexplained eosinophilia months into therapy or even after it's stopped, eosinophilic fasciitis needs to be on the differential, and an MRI of the affected fascia is a reasonable next step given how often it was abnormal here. The finding that rechallenge didn't provoke flares in this small subset is interesting but not yet something to hang a confident oncology-dermatology discussion on, given the tiny numbers. Moving to our two letters. The first is a comment responding to a recent review of acral lentiginous melanoma, and it's a nice piece of surgical pedantry that's actually clinically important. The original review advised cutting perpendicular to the dermatoglyphs — the ridges and furrows on palms and soles — for incisional biopsies, citing a source that was really describing how pathology sections should be cut, not how the surgical incision should be oriented. The letter writers point out this is a transposition error with real consequences. Surgically, a fusiform incisional biopsy should run parallel to the dermatoglyphs, because dermatoglyphs run roughly transverse to flexural creases, and transversely-oriented fusiform closures across flexures are what you want to minimize contracture and web formation during healing — that's basic reconstructive principle. Separately, on the pathology bench, breadloafing that specimen perpendicular to its long axis — which is standard gross sectioning technique — will then naturally produce histologic sections cut perpendicular to the dermatoglyphs, which is exactly the orientation that lets the pathologist properly assess rete ridge and suprapapillary melanocyte distribution without a false impression of confluence in a benign nevus. So the elegant point is that incising parallel and sectioning perpendicular aren't competing recommendations — doing the surgery correctly automatically sets up the pathology correctly, as long as you don't confuse which step the word "perpendicular" applies to. Practically: if you biopsy acral melanocytic lesions, orient your fusiform incision parallel to the dermatoglyphs, and for digit pads where dermatoglyphs loop or whorl rather than run in straight lines, consider communicating directly with your dermatopathologist or bisecting the specimen yourself at the bedside in the orientation you think best captures the lesion. The second letter is a methodologic critique of a retrospective cohort study of two hundred eighty-one early-stage acral melanoma cases that looked at how anatomic site — upper versus lower limb — affects prognosis and metastatic pattern. The letter writers raise several fair points without presenting any new data of their own. They note the original study didn't account for the fact that upper and lower limb lymph node basins differ fundamentally in size and lymphatic flow, which could itself explain differences in nodal metastasis rates independent of tumor biology. They flag that using "first metastasis" as an endpoint is vulnerable to detection bias from variable surveillance intensity and imaging sensitivity across sites. More substantively, they point out the original study didn't incorporate sentinel lymph node biopsy results at all, which means apparent differences in nodal involvement could reflect differences in diagnostic practice rather than true tumor behavior. They also raise a statistical power concern — the strongest site-based associations were seen in stage two disease, a biologically heterogeneous group unevenly distributed across sites, and without stage-stratified subgroup analysis across stages two-A, two-B, and two-C, the findings in that upper-limb subgroup may simply reflect small numbers. Finally, they note that lumping palm, finger, and nail unit melanomas together under "upper limb" glosses over real embryologic and lymphatic drainage differences between those sites. None of this invalidates the original study, but it's a useful reminder that anatomic-site analyses in rare melanoma subtypes need SLNB-stratified data and finer-grained site categorization before changing surveillance protocols based on limb alone. That wraps our four articles for this episode. The clearest actionable message this month is the AFX-PDS-undifferentiated sarcoma review's reaffirmation that complete margin control surgery is the right default for these tumors, paired with a genuinely useful new toxicity pattern to keep on your differential — late-onset, highly inflammatory eosinophilic fasciitis in patients on checkpoint inhibitors. The two letters are worth filing away less as practice-changers and more as sharpening tools for how you read the next acral melanoma paper that crosses your desk. Thanks for listening, and we'll see you next month.