Welcome to this month's journal review podcast, covering key dermatologic oncology literature from the Journal of the American Academy of Dermatology, July twenty twenty-six issue. Today we're covering four articles spanning noninvasive imaging for basal cell carcinoma, an update on rare cutaneous sarcomas, a real-world study on immune checkpoint inhibitors and skin cancer risk, and a comprehensive review of liquid biopsy in cutaneous oncology. Let's get into it.

Our first piece is a Highlights from JAAD International summary by Jonathan Kantor, titled "Noninvasive imaging for basal cell carcinoma." This piece synthesizes two separate studies published in JAAD International. The clinical question driving both papers is how well emerging noninvasive imaging modalities can diagnose basal cell carcinoma, and in particular basal cell carcinoma subtype, without a biopsy.

The first study, by Wolswijk and colleagues, was a head-to-head comparison of line-field confocal optical coherence tomography, or LC-OCT, against conventional optical coherence tomography, in one hundred ninety-seven clinically equivocal lesions, using histopathology as the reference standard. The key finding was that LC-OCT produced more high-confidence diagnoses and greater sensitivity than conventional optical coherence tomography, particularly when subtype certainty was required. However, LC-OCT identified infiltrative basal cell carcinomas with lower sensitivity, a limitation with direct relevance to margin-control surgery planning.

The second study, by O'Hagan and colleagues, externally validated a deep learning model applied to in vivo reflectance confocal microscopy images, again for basal cell carcinoma diagnosis. Internally, the model performed very well, with an area under the curve approaching zero point nine eight. But on external validation across independent cohorts, performance dropped to an area under the curve of zero point eight six, and in one subset, as low as zero point seven four. This is a substantial performance gap between internal and external validation, underscoring the classic problem of diagnostic artificial intelligence models overfitting to their training population.

Methodological strengths across both studies included use of a histopathologic reference standard, and, for the deep learning analysis, external validation across two independent cohorts, a step that is often skipped in diagnostic artificial intelligence literature. Limitations included the single-assessor, single-center design of the imaging comparison study, and the retrospective, prevalence-dependent nature of the computational analysis, meaning the reported accuracy metrics may shift substantially depending on the basal cell carcinoma prevalence in a given clinical population. The practical takeaway for Mohs surgeons is that these noninvasive imaging tools show promise for triage and subtype characterization, but the sensitivity drop for infiltrative subtypes and the external validation performance drop caution against relying on them as a replacement for histopathologic margin assessment at this stage.

Moving to our second article, a clinical review titled "Atypical fibroxanthoma, pleomorphic dermal sarcoma, and undifferentiated sarcoma: Updates on diagnosis, management, and surveillance," by Jason Park and colleagues from Columbia University and collaborating institutions. This is a comprehensive review addressing the clinical question of how to distinguish and manage this spectrum of rare cutaneous mesenchymal malignancies, which has been complicated by decades of shifting nomenclature.

There is no single new study here, but the review compiles substantial epidemiologic and prognostic data relevant to surgical decision-making. Atypical fibroxanthoma and pleomorphic dermal sarcoma have estimated incidences of eight and one point five per one million patients annually, respectively. Atypical fibroxanthoma and pleomorphic dermal sarcoma share ultraviolet-signature mutations and lie on a biological spectrum, while undifferentiated sarcoma lacks these ultraviolet-associated mutations, shows a complex karyotype, arises on sun-protected skin, and remains a diagnosis of exclusion once lineage-defined sarcomas are ruled out.

On outcomes, a large study of one thousand one hundred eighteen patients found five-year local recurrence rates of ten percent for atypical fibroxanthoma versus seventeen percent for pleomorphic dermal sarcoma, with metastasis risks of zero point eight percent versus sixteen percent, respectively. Pleomorphic dermal sarcoma metastasized within three years in more than ninety percent of cases, most commonly to the lungs, accounting for fifty percent of metastatic sites. Regional lymph node spread occurred in thirty-eight percent of metastatic atypical fibroxanthoma cases versus eight percent of metastatic pleomorphic dermal sarcoma cases. A separate cohort of one thousand nine hundred eleven pleomorphic dermal sarcoma cases identified tumor size greater than four centimeters as the most pertinent prognostic factor for metastasis. Comparing undifferentiated sarcoma to pleomorphic dermal sarcoma, one study found undifferentiated sarcoma carried higher five-year local recurrence, twenty-four percent versus twenty percent, higher overall metastasis, nineteen percent versus seven percent, higher distant metastasis, twelve percent versus two percent, and lower five-year survival, fifty percent versus sixty-seven percent. Interestingly, lymph node spread was found in only four percent of metastatic undifferentiated sarcoma compared to twenty percent of metastatic pleomorphic dermal sarcoma.

On ultrasonography, a small case series of thirteen tumors found that twenty-three percent of lesions initially diagnosed as atypical fibroxanthoma were reclassified as pleomorphic dermal sarcoma after ultrasound identified subcutaneous involvement, later confirmed histopathologically. This has direct implications for preoperative planning and margin strategy.

For surgical practice, the review's capsule summary is direct: Mohs micrographic surgery and other margin-control surgical techniques are preferred over wide local excision for both atypical fibroxanthoma and pleomorphic dermal sarcoma. Sentinel lymph node biopsy is not routinely recommended for atypical fibroxanthoma given metastatic potential comparable to low-risk keratinocyte carcinomas, though the melanoma-derived ten percent regional metastasis threshold is suggested as a possible cutoff for considering the procedure in higher-risk pleomorphic dermal sarcoma or undifferentiated sarcoma cases, albeit without established guidelines. As a review article, the main limitation is the absence of prospective, standardized diagnostic and management guidelines given the rarity of these tumors, and the persistent classification ambiguity between pleomorphic dermal sarcoma and undifferentiated sarcoma in sun-protected sites.

Our third article is a brief report titled "Future Nonmelanoma Skin Cancer Development Following Immune Checkpoint Inhibitor Therapy: A TriNetX Real-World Study," by Eli Kasheri, Colin Kincaid, Xiying Fan, and Janellen Smith from the University of California Irvine. The clinical question is whether immune checkpoint inhibitor exposure affects subsequent nonmelanoma skin cancer incidence, given prior conflicting reports of a possible chemopreventive effect.

This was a retrospective real-world cohort study using the TriNetX federated database, examining adults with melanoma, lung, bladder, or kidney cancer treated with or without immune checkpoint inhibitors. Patients with immunosuppression, cemiplimab exposure, or death within the outcome window were excluded. Cohorts were propensity score matched and stratified by prior nonmelanoma skin cancer history, with outcomes assessed from three months to three years after immune checkpoint inhibitor initiation or cancer diagnosis.

In the primary analysis of patients without prior nonmelanoma skin cancer, twenty-one thousand eight hundred twenty patients were matched per cohort. Immune checkpoint inhibitor exposure was not associated with overall nonmelanoma skin cancer risk, with a hazard ratio of one point zero five and a p-value of point three nine eight. However, cutaneous squamous cell carcinoma risk was higher, with a hazard ratio of one point three zero, ninety-five percent confidence interval one point one two to one point five one, and a p-value of point zero zero four. Basal cell carcinoma risk was lower, with a hazard ratio of zero point eight three, ninety-five percent confidence interval zero point seven two to zero point nine six, p equals point zero two zero. Total procedure burden did not differ significantly, at zero point zero three nine versus zero point zero three five procedures per patient. In a screening-enriched subanalysis, the cutaneous squamous cell carcinoma signal was directionally similar, hazard ratio one point three four, but lost statistical significance after correction, at p equals point zero five two, and basal cell carcinoma showed no significant association.

In patients with prior nonmelanoma skin cancer, two thousand three hundred twenty-seven patients were matched per cohort. Immune checkpoint inhibitor exposure was associated with lower basal cell carcinoma risk, hazard ratio zero point eight three, p equals point zero two eight, and lower total malignant procedure burden, zero point two two four versus zero point two eight four procedures per patient, p equals point zero two eight. Overall nonmelanoma skin cancer and cutaneous squamous cell carcinoma were not significantly different in this group. In the screening-enriched version of this subgroup, the lower basal cell carcinoma risk persisted, hazard ratio zero point seven five, ninety-five percent confidence interval zero point six three to zero point nine zero, p equals point zero zero eight, again with lower total procedure burden.

The authors' takeaway is a subtype-specific pattern: higher cutaneous squamous cell carcinoma risk and lower basal cell carcinoma risk in patients without prior nonmelanoma skin cancer, and lower basal cell carcinoma risk with reduced procedure burden in patients with prior nonmelanoma skin cancer. They explicitly note, however, that these effect sizes are modest and may not represent clinically meaningful differences warranting a change in clinical decision-making based on immune checkpoint inhibitor exposure alone. Proposed mechanisms include immune-mediated unmasking of subclinical squamous cell carcinomas, consistent with prior reports of eruptive lesions after programmed cell death protein one blockade, and enhanced immune clearance of less immunogenic basal cell carcinomas. Limitations explicitly acknowledged include residual confounding, coding-based outcome ascertainment, and exclusion of cemiplimab, meaning prospective validation is needed before these associations inform surveillance practice.

Our fourth and final article is a clinical review titled "Liquid Biopsy in Cutaneous Oncology: Current Practice and Future Directions," by Sabah Choudhury, Adewunmi Adelaja, and colleagues from Harvard-affiliated institutions. This review addresses the clinical question of how circulating tumor biomarker technologies, collectively termed liquid biopsy, are being applied across cutaneous malignancies, and what dermatologists and Mohs surgeons need to know as these tools move toward standard of care.

The review is not built around a single study, but rather synthesizes the current technological landscape. It covers circulating free DNA and RNA, noting that circulating tumor DNA, a fragmented, malignancy-associated subtype of cell-free DNA, has been proposed for diagnosis, prognosis, monitoring tumor burden, detecting treatment resistance, and identifying minimal residual disease prior to clinical or radiographic relapse. Circulating tumor cells are highlighted as already Food and Drug Administration approved for monitoring metastatic breast, prostate, and colorectal cancers, with emerging interest in earlier-stage disease screening and prognosis. Exosomes, small extracellular vesicles carrying tumor-specific genetic material and proteins, are noted for their stability across bodily fluids, making them attractive for early detection and treatment-response monitoring. For Merkel cell carcinoma specifically, the review discusses antibody titers against Merkel cell polyomavirus, including the anti-Merkel panel assay, as tools already used to risk-stratify patients for recurrence, with other anti-polyomavirus oncoprotein antibodies proposed to estimate tumor burden and detect recurrence.

On assay technology, the review distinguishes several platforms with clear tradeoffs. Droplet digital polymerase chain reaction offers low cost, rapid turnaround, and excellent precision for serially monitoring known driver mutations at very low variant allele fractions, but is limited to preselected hotspot variants and cannot detect novel resistance mechanisms or comprehensively profile tumors; BEAMing, a related digital polymerase chain reaction method, offers higher sensitivity at the cost of a more complex and expensive workflow. Broad multiplex targeted next-generation sequencing panels can detect single nucleotide variants, insertions, deletions, select copy number alterations, and gene fusions, and are useful when tissue is insufficient, but their sensitivity drops in low-burden or minimal residual disease settings, and clonal hematopoiesis of indeterminate potential can introduce false positive variants unless bioinformatic filtering or paired leukocyte sequencing is used. Tumor-informed minimal residual disease assays, which sequence the patient's tumor first and then track a personalized panel of somatic variants in plasma, achieve improved sensitivity and specificity for minimal residual disease detection and are less susceptible to clonal hematopoiesis interference, though they require adequate tumor tissue and have longer turnaround times, and cannot detect new untracked alterations arising during therapy. Finally, ultra-low-pass whole genome sequencing estimates tumor fraction via genome-wide copy number alterations at shallow sequencing depth, offering a cost-efficient option that does not require prior tumor sequencing, though its sensitivity for very low tumor fractions lags behind deep mutation-tracking methods; newer approaches incorporating fragmentomics and methylation profiling aim to improve signal detection when mutational burden is low.

For surgical oncology practice, the clinical takeaway is that liquid biopsy technologies are increasingly poised to complement, though not yet replace, tissue-based diagnosis and surveillance for aggressive cutaneous malignancies, including melanoma, cutaneous squamous cell carcinoma, and Merkel cell carcinoma. As this is a narrative review rather than a primary data study, its main limitation is the absence of comparative sample sizes or effect sizes for cutaneous-specific validation; the authors note that adoption into skin cancer clinical practice is still in early stages, and that dermatologists should build familiarity with these assay classes now given the trajectory toward incorporation into standard-of-care pathways.

That concludes this month's review. To summarize: noninvasive imaging modalities for basal cell carcinoma show diagnostic promise but reveal real gaps in infiltrative subtype sensitivity and external validation performance; the sarcoma review reinforces margin-control surgery as the preferred approach for atypical fibroxanthoma and pleomorphic dermal sarcoma while highlighting persistent nosologic challenges with undifferentiated sarcoma; the TriNetX study offers reassurance against a clinically meaningful chemopreventive or carcinogenic effect of checkpoint inhibitors on nonmelanoma skin cancer overall, despite statistically significant but modest subtype-specific shifts; and the liquid biopsy review lays out a rapidly maturing technological toolkit that dermatologic oncology will need to engage with as it moves toward the clinic. Thank you for listening, and we'll see you next month.