Welcome to this month's journal review podcast, covering the Journal of the American Academy of Dermatology for July 2026. Today we're going through four articles relevant to Mohs surgery and dermatologic oncology, spanning a reconstructive surgical pearl, an artificial intelligence reader study in pigmented lesion diagnosis, a SEER database analysis of surgical utilization patterns in a rare sarcoma, and an ethics discussion on diagnostic discordance in melanocytic lesions. Let's get into it.

Our first article is a surgical pearl titled "Extended East-to-West Advancement Flaps for Large Defects of the Lower Face and Cervical Junction," by Ghafari-Saravi, Gramann, Roberts, and Nijhawan, from University of Texas Southwestern Medical Center and San Antonio Uniformed Services Health Education Consortium. This is a single-case technique report, not a comparative study, so there's no cohort or statistical outcome data to report — the value here is procedural.

The clinical challenge presented is reconstruction of a large, atypically shaped Mohs defect spanning the mandibular border and mobile cervical skin, a zone where subtle tension or distortion can alter the mandibular silhouette. The index case was an eighty-one-year-old man with paraplegia who developed a six by four point five centimeter defect from a sclerosing basal cell carcinoma involving the anterior lateral mandible and left submental triangle. The authors note that a combination repair would typically be considered for a defect of this size and shape, but in a medically complex, low self-care-capacity patient, that operative complexity was undesirable.

Their solution was a bilateral advancement flap design, essentially an extended east-to-west flap adapted from the original nasal reconstruction concept described by Geist and Maloney in 2012. The design uses a superior Burow's triangle with its apex at the superolateral defect margin extending just inferior to the oral commissure along the melomental fold, and an inferior Burow's triangle with its apex at the inferomedial defect margin extending along the anterior neck midline. The key modification — the "extended" component — is a horizontal incision placed along the submental crease to recruit additional tissue and allow tensionless closure. The patient achieved an acceptable cosmetic and functional result at follow-up.

From a practice standpoint, the takeaway is that this bilateral advancement approach can reduce flap footprint and operative complexity compared to combination repairs for atypical defects crossing the lower face-neck junction, which may be particularly useful in patients with limited capacity for multi-stage or high-maintenance reconstructions. The obvious limitation is that this is a single-patient technique description with no comparative outcomes, no recurrence data, and no quantification of cosmetic result beyond descriptive terms, so generalizability and reproducibility across defect sizes or patient anatomies remain unestablished.

Moving to our second article: "Improved Detection of Lentigo Maligna with AI-Assisted Dermoscopy: A Reader Study in Facial Pigmented Lesions," by Yilmaz, Erol Mart, Temelkuran, and Akay, from Imperial College London and Ankara University.

The clinical question here is whether a deep learning model can improve differentiation of lentigo maligna from its major benign mimickers — pigmented actinic keratosis and the combined solar lentigo–seborrheic keratosis category — on facial skin, and whether AI-generated malignancy scores improve resident diagnostic accuracy.

This was a retrospective study using seven hundred twenty-two unique lesions across eight hundred ninety-four dermoscopic images: one hundred ninety lentigo maligna lesions, two hundred thirty pigmented actinic keratoses, and three hundred two solar lentigo or seborrheic keratosis lesions. Twenty percent of lesions were held out as an independent test set — one hundred forty-four lesions, one hundred seventy-seven images — with the remainder used in five-fold stratified cross-validation, maintaining no patient overlap between training, validation, and test sets. The model architecture was Xception-based, evaluated for both binary, malignant versus benign, and three-class classification.

On the binary task, the model achieved a mean test accuracy of eighty-four point two percent, plus or minus two point five percent, sensitivity of ninety point eight percent — reported in the abstract, though the results section states ninety point zero percent, plus or minus eleven point five percent — and specificity of eighty-one point nine percent, plus or minus one point two percent. The mean confusion matrix across folds was forty-five true positives, twenty-three false positives, one hundred four true negatives, and five false negatives. For three-class classification, mean test accuracy was seventy-one point four percent, plus or minus zero point six percent, with a macro F1 score of seventy point eight percent and a weighted F1 score of seventy-one point three percent.

The reader study involved twenty-six dermatology residents with a mean of two point seven years of clinical experience, tested in a timed format with one minute per question, first unaided and then after disclosure of the AI malignancy score. Baseline resident accuracy was sixty-four point nine percent — one thousand six hundred eighty-six correct out of two thousand six hundred, ranging from fifty to eighty-four percent across residents — which improved to seventy-four percent, one thousand nine hundred twenty-five out of two thousand six hundred, with AI assistance. This improvement was statistically significant at p less than zero point zero zero zero one, with the largest gain specifically in lentigo maligna detection, an increase of sixteen point five percentage points.

Limitations acknowledged by the authors include the retrospective design, absence of multimodal clinical data such as patient history or lesion evolution, and a reader study limited to residents rather than a range of experience levels including board-certified dermatologists or dermatopathologists. For practice, this supports AI-assisted scoring as a potential adjunct particularly for trainees managing diagnostically ambiguous facial pigmented lesions, though it does not establish impact on biopsy rates, downstream margin-directed surgical planning, or performance in more experienced proceduralists such as fellowship-trained Mohs surgeons.

Our third article is a SEER-based brief report: "Predictors for Utilization of Mohs Surgery over Wide Local Excision in Pleomorphic Dermal Sarcoma," by Shelton, Taylor, Hendrickson, Xu, and Collins, from the University of Wisconsin-Madison, Advocate Illinois Masonic Medical Center, and the University of Oklahoma.

The clinical question was what sociodemographic and clinicopathologic factors predict utilization of Mohs micrographic surgery over wide local excision in localized pleomorphic dermal sarcoma, a rare, locally aggressive tumor formerly termed malignant fibrous histiocytoma. This matters because prior work — referenced as Moore, Schmults, and Ruiz, 2023 — has shown comparable survival and low recurrence with Mohs surgery relative to wide local excision in this entity, yet utilization predictors were previously unstudied.

This was a retrospective cohort study using the SEER-17 database, identifying patients with biopsy-confirmed localized pleomorphic dermal sarcoma treated with either wide local excision or Mohs surgery between 2000 and 2021. Regional or distant disease and unknown surgical approach were exclusion criteria. The final cohort included four hundred sixty-seven patients: three hundred four, or sixty-five point one percent, underwent Mohs surgery, and one hundred sixty-three, or thirty-four point nine percent, underwent wide local excision.

On univariable comparison, higher annual income of eighty thousand dollars or more was associated with Mohs utilization, seventy-two point seven percent versus fifty-five percent in the Mohs and wide local excision groups respectively, standardized mean difference zero point three eight. Head and neck tumor location showed a similar pattern, sixty-eight point three percent versus thirty-eight point five percent, standardized mean difference zero point four one. On multivariable logistic regression using backward elimination, both factors remained independently significant. Annual income of eighty thousand dollars or greater carried an adjusted odds ratio of two point two eight, ninety-five percent confidence interval one point five three to three point four one, for receiving Mohs surgery. Head and neck location carried an adjusted odds ratio of three point five nine, ninety-five percent confidence interval one point nine five to six point six zero. Variance inflation factors were under two for all variables, arguing against significant multicollinearity in the model.

The authors' interpretation is that these findings reflect disparities in access to Mohs surgery — potentially related to geographic availability of fellowship-trained Mohs surgeons, awareness, or perceived cost barriers for lower-income patients — while the head and neck association is consistent with the known tissue-sparing rationale for Mohs in cosmetically and functionally sensitive sites. Limitations explicitly stated include the retrospective design, reliance on SEER and census-tract-level rather than individual income data, small overall sample size, missing data on tumor size and differentiation, and the inability of SEER to capture referral patterns or institutional practice preferences that may also drive surgical selection. For practice, this data doesn't change treatment recommendations but does flag a utilization disparity worth institutional awareness, particularly for lower-income patients with non-head-and-neck localized pleomorphic dermal sarcoma who may be candidates for Mohs surgery but are not currently receiving it at comparable rates.

Our fourth and final article is an ethics journal club piece: "When Experts Disagree: Ethical Challenges in Clinical Management of Diagnostic Discordance in Melanocytic Lesions," by Adler, Feig, and Grant-Kels, from SUNY Downstate Health Sciences University and the University of Connecticut School of Medicine.

This is framed as an advice-column-style ethics discussion rather than a primary data study, prompted by a hypothetical case of a changing melanocytic lesion where three dermatopathologists rendered three discordant reads — severely dysplastic nevus, atypical intraepidermal melanocytic proliferation associated with a nevus, and invasive melanoma.

The piece cites data from Hosler and colleagues, referenced as a 2024 Journal of Cutaneous Pathology study, reporting that histologic discordance among dermatopathologists reviewing atypical melanocytic lesions resulted in equivocal diagnoses in twenty-three point eight percent of cases. The authors note that gene expression profiling is available as an adjunctive molecular tool but is explicitly not yet recommended as a standard of care, per a companion 2025 JAAD piece by Mehrmal and colleagues.

The ethical framework offered centers on several points. First, clinicians must decide whether to privilege one pathologist's read, often gravitating toward a trusted subspecialist, or to synthesize a middle-ground management approach — both strategies have precedent in the literature, citing Ronen and colleagues' 2021 Archives of Pathology and Laboratory Medicine study on discordance and its clinical impact. Second, over-treating based on the most severe read may minimize medico-legal liability but risks disproportionate morbidity from wider excision or intensified surveillance, while under-treating risks missing a malignant lesion with real mortality consequences. Third, on disclosure, the authors argue that informed consent obligates clinicians to disclose diagnostic disagreement to patients honestly but without inducing undue alarm, framing the discussion around the range of interpretations and the rationale for the chosen management path, with shared decision-making playing a central role given that no single "correct" answer may exist.

The authors point to systemic mitigations rather than individual-encounter fixes: consensus conferences, second-opinion pathways, subspecialty consultation, and structured reporting frameworks such as MPATH-Dx version 2.0, a four-class hierarchy intended to standardize diagnostic reporting and management recommendations for melanocytic lesions, referenced from Barnhill and colleagues' 2025 Clinics in Dermatology publication. They note explicitly that these mechanisms reduce but do not eliminate ambiguity, and that molecular adjuncts introduce their own uncertainties and costs rather than resolving the problem outright.

There is no original data, cohort, or effect size in this piece — it is a conceptual and ethics-focused discussion, so the appropriate takeaway for surgical practice is procedural rather than statistical: when facing discordant pathology reads prior to definitive surgical management, whether Mohs surgery or excision, transparent documentation of the discordance, consideration of consensus or subspecialty re-review, and explicit shared decision-making conversations with the patient about the uncertainty and its management implications are the responsible course, rather than silently defaulting to the most severe or most reassuring read.

That concludes this month's review. In summary: a practical bilateral advancement flap technique for atypical lower face and cervical defects in medically complex patients; promising but early-stage evidence that AI-assisted dermoscopy scoring meaningfully improves resident detection of lentigo maligna; SEER-based evidence of income- and anatomic-site-based disparities in Mohs utilization for pleomorphic dermal sarcoma; and an ethics framework for navigating and disclosing pathologist discordance in melanocytic lesion management. Thanks for listening, and we'll see you next month.