Welcome back to the journal review. This is the August 2026 issue of the Journal of the American Academy of Dermatology, and we've got four pieces to get through today — two original research reports, a disparities-focused review roundup, and a clinician's perspective piece that itself surveys five other SCC papers in this same issue. Let's get into it.

First up is a retrospective cohort study out of UC San Diego looking at full-thickness skin grafts for scalp reconstruction after Mohs, specifically asking whether donor site choice affects complication rates. You already know the clinical tension here — the scalp's inelastic, local tissue is limited, and grafts are a reliable workhorse, but there's not much data guiding which donor site to actually pick beyond habit and convenience.

The design is a single-center chart review, three hundred sixty-three patients over about fifteen years, all full-thickness grafts after Mohs for scalp cancer. This is really the only feasible design for this question — you're not going to randomize donor site in a retrospective sense, and complication rates for grafts are low enough that you need a large accumulated volume, which naturally means retrospective review of a high-volume single center. The tradeoff, as always, is that donor site wasn't chosen at random — it was chosen based on tissue quality, availability, and patient preference — so any association you find is confounded by the very reasons a given site was picked in the first place.

Donor sites broke down roughly into thirds among abdomen, adjacent tissue, and supraclavicular skin, with smaller numbers from postauricular, upper arm, forearm, and preauricular sites. Complications overall were reassuringly rare — the most common was partial graft necrosis at about three percent, with hematoma, infection, bleeding, and total necrosis all less common than that. No hospitalizations, no deaths.

On the statistics: the authors first ran chi-square and Fisher's exact tests comparing complication rates across donor sites, suture type, comorbidities, and graft size categories — and none of those simple pairwise comparisons reached significance. But they also ran a logistic regression, which is the right next step when you want to isolate the independent effect of one variable — here, donor site — while implicitly accounting for the fact that groups differ in other ways, like graft size. Logistic regression models the odds of a binary outcome, complication versus no complication, as a function of one or more predictors simultaneously, which lets you ask "does abdominal donor site predict complications above and beyond what's explained by graft size alone" rather than just eyeballing a raw rate difference. That regression found abdominal grafts carried roughly three-fold higher odds of a recipient-site complication compared with adjacent tissue grafts, and this was statistically significant. A separate linear regression also linked abdominal graft size itself to complications, though that same relationship didn't hold up under Fisher's exact testing, and the R-squared was low, around 0.05, meaning graft size alone explains only a small slice of the variance in complications.

The authors' interpretation is straightforward and plausible — abdominal grafts were simply bigger on average, about seventeen square centimeters versus five for adjacent tissue and ten for supraclavicular, and abdominal skin differs in thickness and elasticity, which could plausibly impair early adherence and revascularization. Cosmetically, blinded review of healed abdominal versus supraclavicular grafts found no difference in appearance. Donor-site complications themselves were rare and clustered at the abdomen too, with four hematomas and one infection.

Limitations are the ones you'd expect from this design: retrospective, single-center, non-randomized donor site selection, and a real risk of confounding by indication, since surgeons likely reached for abdominal skin preferentially for larger defects to begin with, which is exactly the scenario that would produce this signal without donor site itself being causal.

Practical takeaway: this isn't practice-changing in the sense of telling you to abandon abdominal donor sites — overall complication rates are low across the board, and the primary chi-square analyses were negative. But it's a useful signal to file away: if you're using a large abdominal graft, particularly one pushing into that larger size range, it's reasonable to counsel patients on a slightly higher chance of partial necrosis and to schedule closer early postoperative follow-up, regardless of which donor site you ultimately chose.

Next, a retrospective diagnostic accuracy study from Compass Dermatopathology in San Diego, examining the 23-gene expression profile test — marketed as MyPath Melanoma — and how well it performs against PRAME immunohistochemistry and final clinicopathologic diagnosis in melanocytic lesions. The gap they're addressing is real-world performance versus the original validation cohorts, which tend to inflate test performance relative to everyday practice.

Methods: two hundred seven consecutive lesions that had been sent for 23-gene expression profiling between 2014 and 2017 were pulled and independently, blindly restained and graded for PRAME using standard quartile scoring, plus an extra "weak or blush" category for sparse faint staining that's treated as negative in practice but tracked separately here to see if it behaves differently. This is a straightforward diagnostic accuracy design — you need a reference standard, which here is the final clinicopathologic diagnosis, and then you calculate sensitivity, specificity, positive and negative predictive value for the index test, in this case the gene expression profile. They also calculated a kappa statistic to assess agreement between the two ancillary tests, gene expression profile and PRAME. Kappa is the right tool here rather than simple percent agreement, because plain agreement is inflated by how often two tests will match just by chance alone — if both tests overwhelmingly call things benign, they'll "agree" often even if they're not really tracking each other. Kappa corrects for that chance agreement, so a kappa in the moderate range tells you the two tests are meaningfully aligned beyond what coincidence alone would produce.

The cohort had only ten melanomas among a hundred ninety-seven nevi — a real skew toward benign lesions that matters for interpreting everything downstream. The gene expression profile classified the majority as benign, a smaller group as malignant, and the rest indeterminate. Overall performance came out to sensitivity of sixty percent, specificity around ninety percent, positive predictive value of only twenty-one percent, and negative predictive value of ninety-eight percent. That negative predictive value looks great, but the authors are upfront that it's an artifact of the low melanoma prevalence in this referral population, not evidence the test is excellent at ruling out melanoma. The more clinically alarming number is the sensitivity — the test missed four of ten melanomas, correctly catching only six.

The concordance analysis with PRAME is the most interesting mechanistic finding. Overall agreement between the two tests was high, around ninety percent, with a kappa in the moderate-to-good range, suggesting the gene expression profile assay track record often just mirrors PRAME positivity rather than adding independent information. And when they looked at the missed melanomas, three of the four gene-expression-profile false negatives were also PRAME-negative — meaning the two ancillary tests were failing together on the same lesions, not offering independent checks on each other. Interestingly, "weak or blush" PRAME staining, which is treated as negative in routine practice, still showed up disproportionately in lesions the gene expression profile called malignant, including in a couple of correctly identified melanomas, hinting that this faint staining category might carry more diagnostic signal than it currently gets credit for.

Limitations are explicitly acknowledged by the authors: only ten melanomas total, a heavy skew toward benign referrals which artificially inflates NPV, and this is retrospective single-institution data from tests run several years ago. They did confirm no missed diagnoses on nine to twelve years of follow-up, which is reassuring but doesn't offset the core sensitivity concern in this dataset.

Practical takeaway: this is genuinely useful, cautionary information rather than a call to abandon the test — the message is that a benign gene-expression-profile result should not be treated as independently reassuring if PRAME is also negative and there remains real clinicopathologic concern, because the two tests appear to be failing in tandem on the same difficult lesions rather than providing orthogonal information. Continue to lean on morphology, clinical context, and consultation for genuinely ambiguous melanocytic lesions rather than anchoring on a benign ancillary test result alone.

Third, a Highlights piece from Dr. Shari Lipner synthesizing three linked papers, one JAAD Reviews systematic review and two JAAD original studies, on geographic and racial disparities in melanoma outcomes. This is a synthesis commentary, so there's no independent methods or results to walk through — just the throughline across the three source papers.

The systematic review found that patients living farther from dermatologic care presented with thicker melanomas, that rural areas have measurably reduced access to dermatologists and specialty cancer care, and that poverty compounded this further, independently correlating with greater Breslow depth and diagnostic delay. Interestingly, stage at diagnosis often didn't differ much between rural and urban patients, even though rural mortality was higher — which tells you the disparity isn't purely about late detection, it's likely also about downstream access to treatment once diagnosed.

The SEER-based study adds nuance by race and ethnicity: non-Hispanic White patients actually had lower melanoma incidence in rural versus urban settings, but non-Hispanic Asian and Pacific Islander patients showed significantly higher rural incidence, and non-Hispanic American Indian and Alaska Native patients had high incidence overall. So the geography-risk relationship isn't uniform — it flips direction depending on race and ethnicity, which argues against a one-size-fits-all rural outreach strategy.

The third study looked at county-level melanoma mortality trends since the introduction of checkpoint inhibitors and BRAF-targeted therapy around 2013. Mortality dropped substantially nationally, but nearly one in five counties saw little improvement or actually got worse despite these advances being available. County population size was one of the strongest predictors of who benefited, pointing squarely at health system infrastructure and specialist access as the bottleneck, not drug availability itself. Immunotherapy uptake is also documented elsewhere as lower among uninsured and Medicaid patients, and higher at academic centers, reinforcing that this is as much an access and insurance problem as a geographic one.

There's no separate limitations section since this is a synthesis piece, but the practical takeaway for you is straightforward: don't assume equitable access to modern melanoma therapy just because it exists — for patients from smaller counties, rural areas, or without academic center access, proactively facilitate referral pathways and consider that delayed presentation and treatment access, not just tumor biology, may be driving worse outcomes you're seeing in these populations.

Finally, a Clinician's Perspective piece from Dr. Warren Heymann surveying five separate squamous cell carcinoma papers in this same issue, framed around the theme of a rising global SCC burden. Since this is a commentary rounding up other articles, I'll walk through his highlights briefly rather than treating each as a full standalone study.

First, a large propensity-matched TriNetX database analysis of about a million patients total, comparing oral niacinamide users against patients with xerosis who weren't on niacinamide, found a significant reduction in ten-year incidence of basal cell carcinoma, SCC, and melanoma with niacinamide exposure — SCC incidence was well under one percent in the niacinamide group versus over two and a half percent in controls, a meaningful relative risk reduction. This supports what many of us already do empirically for chemoprevention in high-risk patients.

Second, a dermoscopy study characterizing SCC vessel and scale patterns by risk stratification — low-risk tumors tended to show small dotted or glomerular vessels with white scale, while higher-risk tumors showed irregular linear or arborizing vessels and ulceration, a significant and clinically intuitive pattern that could inform triage before biopsy.

Third, a small case series of fourteen elderly patients ineligible for conventional surgery, treated with electrocoagulation combined with photodynamic therapy, using visual carbonization differences intraoperatively as a rough guide to tumor margins. All fourteen achieved clearance, with only one recurrence in an in-situ case over a modest follow-up — promising as a minimally invasive alternative for poor surgical candidates, but this is a small case series without comparator, so treat it as hypothesis-generating rather than practice-changing.

Fourth, and most relevant to your daily practice: a case series of Mohs-facilitated excision, a multidisciplinary model where Mohs clears tumor up to critical anatomic structures and then surgical colleagues excise the remainder under general anesthesia. Across fifty-nine high-risk keratinocyte carcinomas, periosteal or bony invasion was the most common reason full Mohs clearance wasn't anticipated to be feasible. Final microscopic margin clearance was achieved in about eighty-five percent of cases, with a local recurrence rate of just over three percent and a metastasis rate under seven percent over roughly two years of follow-up — quite reasonable numbers for a high-risk population that couldn't be fully managed with Mohs alone, and a nice validation of collaborative staged approaches for tumors approaching bone or other critical structures.

Fifth, a retrospective cohort comparing integrated versus non-integrated multidisciplinary care models for advanced cutaneous SCC found that integrated care shaved about seventeen days off time-to-treatment-initiation and required fewer clinic visits, though there was no detectable difference in disease-specific mortality — but the authors note the study was underpowered to detect a mortality difference, so that null result shouldn't be over-interpreted as proof integrated care doesn't affect survival.

Heymann closes with a personal note as a two-time cancer patient himself, underscoring that even without a proven mortality benefit yet, integrated multidisciplinary care meaningfully reduces logistical and emotional burden for patients — a point worth remembering as you think about how your own practice coordinates with medical oncology, radiation, and head and neck surgery colleagues.

That wraps our four pieces for this August issue — a cautionary tale on abdominal donor grafts, a sobering look at gene expression profiling's real-world limits alongside PRAME, a call to look beyond simple rural-urban framing in melanoma disparities, and a five-paper snapshot of where SCC management is heading. Thanks for listening, and I'll see you next issue.