Welcome to this April twenty twenty-six review of the Journal of the American Academy of Dermatology. We've got four brief reports worth your attention this month — a collision tumor pathology study, a SEER-based comparison of surgical approaches for acral lentiginous melanoma, a TriNetX pharmacovigilance study on PD-1 inhibitors and squamous cell carcinoma risk, and a surveillance imaging study in very high-risk cutaneous squamous cell carcinoma. Let's get into it.

First up is a retrospective cohort study out of a national pathology group looking at collision tumors involving atypical fibroxanthoma and pleomorphic dermal sarcoma. You know the biology here — both are dermal-based spindle cell neoplasms sitting on a UV-mutation background that overlaps substantially with undifferentiated squamous cell carcinoma, sharing things like TP53, NOTCH1 and 2, CDKN2A, and TERT promoter alterations. The literature on actual collision tumors — atypical fibroxanthoma or pleomorphic dermal sarcoma physically colliding with another skin cancer in the same specimen — was essentially two case reports. So the authors set out to actually quantify how often this happens.

Methodologically, this is a big pathology-database study: they screened over one million accessioned cases from January twenty sixteen through June twenty twenty-three and pulled every top-line diagnosis of atypical fibroxanthoma or pleomorphic dermal sarcoma — six hundred thirty-eight and ninety-one cases respectively, for a combined cohort of seven hunder twenty-nine tumors. Where both a biopsy and an excision were available, a collision found on either counted, and the final atypical-fibroxanthoma-versus-pleomorphic-dermal-sarcoma call was made off the excision. Every collision case got a second dermatopathologist review plus immunohistochemistry to exclude mimics — which matters a lot here, because the differential for these spindle cell tumors is basically a rogues' gallery of spindled squamous cell carcinoma, leiomyosarcoma, and desmoplastic melanoma, so confirmatory staining is the only way to trust these numbers.

The cohort was classic for this tumor type — mean age about seventy-seven, the vast majority male, and the vast majority head and neck. Of the seven hundred twenty-nine tumors, five and a half percent had a genuine collision with another skin cancer. Squamous cell carcinoma was by far the dominant partner, present in about five percent of all cases, and within that, squamous cell carcinoma in situ alone accounted for the majority — about four percent of the whole cohort, versus less than one percent for invasive squamous cell carcinoma. Basal cell carcinoma collisions were rare, and there was a single melanoma in situ collision and one triple collision of atypical fibroxanthoma with both squamous cell carcinoma in situ and superficial basal cell carcinoma. One case is worth flagging specifically for your practice: a lesion originally read as atypical fibroxanthoma on the initial biopsy was upgraded to pleomorphic dermal sarcoma only after review of the Mohs debulk — the authors point out that a thin shave biopsy can simply miss the deeper, more aggressive dermal tumor entirely.

The authors' takeaway is that shared UV mutational signatures likely explain why squamous cell carcinoma in situ, especially, turns up riding along with these dermal sarcomas so often — roughly three-quarters of all collisions were with squamous cell carcinoma in situ specifically. There isn't much here in the way of limitations discussion beyond the inherent constraints of a retrospective pathology-database design and the fact that this doesn't tell you anything about mechanism, only frequency — the authors explicitly say the mutational relationship needs dedicated molecular follow-up. For your practice, the actionable point is a diagnostic vigilance one, not a management-changing one: when a superficial or exophytic lesion's biopsy shows squamous cell carcinoma in situ or superficial basal cell carcinoma, keep in mind there's a real, not trivial, chance of a deeper dermal-based sarcoma being missed underneath, and your final Mohs or excision pathology should be read with that possibility in mind rather than assuming the shave biopsy told the whole story.

Next, a national cohort study using the SEER database comparing Mohs micrographic surgery against wide local excision for acral lentiginous melanoma, looking specifically at disease-specific survival. This is squarely in your wheelhouse — acral lentiginous melanoma is UV-independent, disproportionately affects patients of color, has a well-known tendency toward subclinical extension, and the optimal surgical modality has genuinely been unsettled.

The authors queried SEER for biopsy-proven invasive acral lentiginous melanoma treated with either wide local excision or Mohs from two thousand to twenty twenty-one, explicitly excluding in-situ disease, and ran chi-square, Mann-Whitney, Kaplan-Meier with log-rank, and stepwise Cox regression. SEER is the obvious tool here because you need national-level numbers to get any meaningful Mohs sample size in a rare melanoma subtype — but keep that trade-off in mind, because SEER buys you scale at the cost of granularity: no recurrence data, no complication data, no information on Mohs technique or margin-assessment method, and no immunotherapy details.

Of sixteen ninety-nine total patients, the split was heavily lopsided — about ninety-six percent wide local excision, only about four percent Mohs, seventy-three patients. And critically, these two groups were not clinically equivalent at baseline: wide-excision patients had roughly double the rate of regional-stage disease, more than double the distant-stage disease, twice the ulceration rate, and roughly double the median Breslow depth compared to the Mohs group. In other words, Mohs was disproportionately used on thinner, more favorable tumors — an important selection effect to hold onto as you read the survival numbers. Despite that imbalance, unadjusted five- and ten-year disease-specific survival looked numerically better with Mohs — eighty-four and sixty-eight percent versus seventy-two and fifty-six percent for wide excision — but this difference was not statistically significant. And after the stepwise multivariable Cox model adjusted for age, sex, race, income, rurality, stage, Breslow depth, and ulceration, there was no significant difference in mortality between the two, with an adjusted hazard ratio around zero-point-eight favoring Mohs but a confidence interval that comfortably straddled one.

The authors frame this as reassurance that Mohs achieves comparable oncologic outcomes to wide excision while offering the tissue-sparing and margin-control advantages that matter enormously on palms, soles, and nail units. They cite two smaller external studies — one showing no survival difference at all between the approaches, and interestingly another smaller multivariable analysis that actually found an eighty-five percent increase in disease-specific mortality with wide excision relative to Mohs. So the literature, taken together, leans toward Mohs being at least non-inferior and possibly favorable, but this SEER analysis itself, with only seventy-three Mohs patients and a wide, non-significant confidence interval, doesn't nail that down definitively. The honest limitations are the retrospective design, the missing recurrence and complication data, and no ability to distinguish Mohs subtypes or capture systemic therapy exposure.

Practically: this is useful reassurance, not a practice-changer in the sense of overturning anything — most of you were likely already leaning toward Mohs or a staged/complete-margin-assessment approach for acral lentiginous melanoma given the subclinical extension issue and the anatomic constraints of these sites. What this paper gives you is a national-level data point you can cite to patients and colleagues that Mohs is not sacrificing survival to gain margin control and tissue conservation in this setting — interesting and supportive, but the underlying evidence base still needs a properly powered, ideally propensity-matched, prospective look before you'd call this definitive.

Third, a retrospective cohort study using the TriNetX network examining whether PD-1 inhibitor therapy for metastatic melanoma raises the risk of subsequent cutaneous squamous cell carcinoma. This question exists because of a handful of case reports describing squamous cell carcinomas emerging eight to fourteen months after starting pembrolizumab or nivolumab — but as the authors point out, this signal is muddied by the inconvenient fact that both of these drugs are also approved to treat squamous cell carcinoma directly, and there's actually some literature suggesting PD-1 blockade might have a preventive effect on keratinocyte carcinogenesis. No population-level study had addressed this before.

They queried TriNetX Global for metastatic melanoma patients getting their first exposure to either nivolumab or pembrolizumab against unexposed controls, from twenty fifteen through twenty twenty-five, and deliberately excluded anyone exposed to other checkpoint inhibitor classes — CTLA-4, PD-L1, LAG-3, or other PD-1 agents — to isolate the effect of these two specific drugs rather than immunotherapy broadly. One-to-one propensity matching covered the usual demographics plus, importantly, UV skin damage history, immunosuppression, HIV, transplant status, nicotine, and alcohol use — all plausible confounders for squamous cell carcinoma risk. This is the right design for this question: you can't run a randomized trial to ask whether a cancer drug secondarily causes a different cancer, and case reports simply can't give you a denominator, so a large matched retrospective claims-based cohort is really the only feasible way to get a population-level risk estimate here.

The results ran counter to the case-report signal. Nivolumab, in a matched cohort of about thirty-seven hundred per arm, was associated with a reduced risk of squamous cell carcinoma at every follow-up interval — one, three, five, and ten years — with the largest reduction in patients over sixty-five, roughly a seventy percent relative risk reduction. Pembrolizumab, in a larger matched cohort of about six thousand per arm, showed no difference at one year, but a significant risk reduction emerged from three years onward, again most pronounced in the over-sixty-five group, with roughly a forty percent relative risk reduction by five and ten years.

The authors' interpretation is that this aligns with enhanced effector T-cell function and immune surveillance suppressing UV-driven keratinocyte neoplasia, rather than a treatment effect creating new squamous cell carcinomas — and the earlier signal with nivolumab may simply reflect its more frequent dosing schedule driving quicker cumulative immune activation. But they're appropriately conservative in the framing: given residual confounding, competing mortality risk in a metastatic melanoma population, coding inaccuracies inherent to claims data, a predominantly White cohort limiting generalizability, and probable undercapture of squamous cell carcinomas diagnosed in outpatient dermatology settings outside the parent hospital systems, they explicitly say this is better described as an absence of increased risk rather than a proven protective effect.

For your practice, the takeaway is reassurance rather than a new intervention: patients on nivolumab or pembrolizumab for metastatic melanoma do not appear to need heightened suspicion or intensified skin-cancer surveillance specifically because of the checkpoint inhibitor itself — standard melanoma survivorship skin surveillance remains appropriate. The case reports of post-PD-1 squamous cell carcinoma are probably better explained by baseline risk, incidental timing, or hyperprogression of a pre-existing indolent lesion rather than a causal carcinogenic effect of the drug. This is interesting and clinically reassuring, but the authors themselves call it hypothesis-generating — retrospective claims data, however large, isn't the final word.

Last, a retrospective cohort study out of a tertiary cancer center looking at whether surveillance imaging detects subclinical recurrence in very high-risk cutaneous squamous cell carcinoma. The backdrop here is that adjuvant radiation cuts locoregional recurrence roughly in half in clear-margin, high-T-stage tumors, and a prior study had shown subclinical disease detection on imaging in about a fifth of high-risk cases generally — but nobody had specifically looked at a cohort where every single patient received radiation as part of initial management, which is exactly the population where a positive imaging finding is most likely to change what you do next.

The authors reviewed a prospectively collected database of ninety very high-risk squamous cell carcinoma patients treated with radiation oncology and followed with surveillance imaging — imaging modality, interval, and duration weren't standardized but were left to treating-physician discretion, which the authors note actually reflects current NCCN-guided real-world practice rather than being a purely artificial limitation.

Over two years of follow-up, about a quarter of patients — twenty-three of ninety — developed a local recurrence, nodal metastasis, or distant metastasis. Across those ninety patients, three hundred imaging studies were done in total, most for surveillance rather than initial staging — a mix dominated by CT and MRI, with a substantial chunk of PET scans, and only a handful of ultrasounds and x-rays. Now here's the key operational finding: of the twenty-three recurrences, about seventy percent were picked up clinically — palpable tumor, palpable nodal disease, or perineural symptoms — while the remaining thirty percent, seven cases, were caught purely on surveillance imaging while the patient was clinically silent. PET scans did most of that subclinical detection work, picking up nodal metastases, a distant metastasis, and a simultaneous local-plus-nodal recurrence, with CT and MRI each catching one additional local recurrence. And the number that should really stick with you: in every single case of recurrence or metastasis identified in this study — clinical or radiographic — detection changed management, whether that meant additional surgery, radiation, systemic therapy, or a combination.

The authors are candid about the limitations — single-institution, retrospective, no standardized imaging protocol, and CT being the workhorse modality even though they note ultrasound has documented utility for locoregional staging in experienced hands and could plausibly play a larger surveillance role. They frame the contribution here as characterizing a population not previously studied in this way — very high-risk squamous cell carcinoma patients who all received radiation therapy up front — and argue that finding subclinical disease in thirty percent of this group's recurrences supports a real role for structured surveillance imaging, while also suggesting that limiting the surveillance window to the first few years, when most cutaneous squamous cell carcinoma recurrences actually occur, could help balance the benefit against imaging burden and cost.

Practically, this is a genuinely useful data point for the very-high-risk, radiation-treated subgroup you're likely already flagging for close follow-up: it reinforces that a meaningful minority of recurrences in this population will be clinically silent, that PET in particular is doing a lot of that subclinical detection work, and that finding it early changes what you do for the patient one hundred percent of the time. What it doesn't yet give you is a validated protocol — modality, interval, and total surveillance duration remain physician-judgment calls, and there's no survival endpoint here to confirm that earlier detection actually improves outcomes rather than just changing management sooner. So call this practice-informing for how you counsel and structure follow-up in your very high-risk patients, not yet a guideline-level mandate.

That wraps our four papers this month — a pathology-vigilance note on collision tumors with atypical fibroxanthoma and pleomorphic dermal sarcoma, reassurance on Mohs for acral lentiginous melanoma, reassurance on PD-1 inhibitors and squamous cell carcinoma risk, and a real-world case for imaging surveillance in your highest-risk squamous cell carcinoma patients on radiation. Thanks for listening, and we'll see you next month.