Welcome back to the journal review. This is your walkthrough of the August twenty twenty-six issue of the Journal of the American Academy of Dermatology, and today we're covering four brief reports that all have real relevance to a Mohs and dermatologic oncology practice — a dermoscopy study on lentigo maligna, a minimally invasive combination treatment for squamous cell carcinoma in elderly patients, a large real-world look at niacinamide for skin cancer chemoprevention, and a cohort study on baseline imaging in Merkel cell carcinoma. Let's get into it.

First up is a retrospective diagnostic study out of Memorial Sloan Kettering and collaborating centers, looking at how to differentiate small-diameter lentigo maligna and lentigo maligna melanoma — I'll just call it LM slash LMM going forward — from solar lentigines that are difficult enough to warrant biopsy in the first place. The clinical problem here is one you know well: on the head and neck, early LM can look deceptively like a banal solar lentigo, and getting that call wrong in either direction has real costs — treating melanoma as benign leads to inadequate destructive therapy, while treating a benign lentigo as melanoma means an unnecessary biopsy in a cosmetically sensitive area.

Methodologically, this is a retrospective image review pulled from two tertiary dermatology centers, restricted specifically to small lesions — one centimeter or less — because those are the ones that actually create diagnostic uncertainty; the easy, large, obviously atypical LMs weren't the point here. They excluded recurrent LM/LMM and atypical intraepidermal melanocytic proliferations, presumably to keep the comparison clean between two well-defined diagnostic categories rather than muddying it with ambiguous in-between lesions — that's my inference, not something the authors state outright, but it's a sensible way to isolate a genuine discrimination problem. Two blinded dermatologists scored dermoscopic features independently, with a third expert as tiebreaker, which is the right design choice for any study whose whole premise is about visual pattern recognition — you want to know the findings aren't just one person's gestalt.

On the statistical side, they used LASSO regression — that stands for least absolute shrinkage and selection operator — to whittle down a large list of candidate dermoscopic features to the ones that actually carry independent information. Think of LASSO as a variable-selection tool that penalizes complexity: it will shrink the effect of weak or redundant predictors right down to zero, effectively dropping them from the model, so you're left with a lean set of features that aren't just correlated with the outcome by chance or redundant with each other. That's particularly useful here because dermoscopic features tend to travel together — a lesion with more colors also tends to have more structures — so you need something that can cut through that collinearity. The features that survived LASSO selection were then fed into a multivariable logistic regression, adjusted for age and sex, to generate odds ratios for each independently associated feature.

So what did they find. Out of a hundred twenty-three lesions — forty-one LM slash LMM, the vast majority in situ, and eighty-two challenging solar lentigines — the melanocytic lesions were disorganized far more often, roughly eighty percent versus forty percent for the lentigines, and they carried more colors and more distinct dermoscopic structures on average. All of that was highly statistically significant. On multivariable analysis, four features stood out as independently predictive of LM slash LMM: angulated lines, granularity, gray color, and the absence of scalloped borders, with odds ratios in the three-and-a-half to five-and-a-half range — that's a meaningful effect size, not just statistical noise. The final model's discrimination was excellent, with an area under the curve of zero point eight eight, which for a dermoscopy-based model on small, deliberately ambiguous lesions is quite strong.

The interesting wrinkle is hyperpigmented follicular openings, which turned out to be common in both groups — present in over half of the solar lentigines — so on their own they're not diagnostic. But when hyperpigmented follicular openings covered more than half the lesion's surface, or showed a mix of thick and thin pigmentation, or appeared alongside multiple other LM-associated structures, the balance tipped meaningfully toward melanoma. The authors also flag follicular size variability as a newly described clue, which they speculate reflects heterogeneous follicular involvement by melanoma cells — a plausible but unproven mechanistic explanation. The obvious limitation here is sample size — a modest number of lesions from two centers, and it's a retrospective image-based design without outcome or follow-up data, so this is about diagnostic pattern recognition, not proof of what happens if you leave equivocal lesions alone.

Practically, this doesn't change your surgical algorithm, but it sharpens your dermoscopic threshold for biopsy versus reassurance on small facial pigmented lesions. If you're staring at a lentigo-like macule with gray color, angulated lines, granularity, non-scalloped borders, or extensive and heterogeneous follicular pigmentation, this data supports lowering your threshold to biopsy rather than assuming it's just a sun spot. That's a genuinely useful addition to your dermoscopic checklist, even if it's not practice-changing in the sense of altering management once melanoma is already confirmed.

Next, a brief report from a group in Guiyang, China, describing electrocoagulation combined with five-aminolevulinic-acid photodynamic therapy — I'll call it ALA-PDT — for primary cutaneous squamous cell carcinoma in elderly patients who are not good surgical candidates. This is best thought of as a small case series demonstrating feasibility of a combination technique, not a comparative trial, so there's no control group or formal statistical modeling here — just descriptive outcomes.

The clinical rationale is one you'll recognize immediately: some elderly or frail patients with cutaneous squamous cell carcinoma simply aren't going to tolerate Mohs or conventional excision, and ALA-PDT alone has a track record in in-situ disease but a much shakier one in invasive tumors, with the authors citing a prior recurrence rate as high as thirty percent for invasive squamous cell carcinoma treated with photodynamic therapy alone. The idea here is to debulk the tumor first with electrocoagulation — cauterizing down to visible carbonized collagen, typically within about twenty minutes — and then follow with ALA-PDT as a consolidative step aimed at cleaning up residual tumor cells the electrocoagulation might have missed.

Fourteen patients were treated, six with invasive disease and eight in situ, median age eighty-three, essentially all deemed unsuitable for Mohs due to location, frailty, or refusal. The protocol was electrocoagulation followed by weekly ALA-PDT sessions, three to five total. Nearly eighty percent of lesions were on the head and face. All patients achieved clinical tumor clearance. Over a median follow-up of about thirteen months, there was exactly one recurrence — an in-situ case — which was successfully re-treated with the same combination and did not recur again. That gives an overall recurrence rate of about seven percent, which the authors note is in the same ballpark as standard wide excision, around six percent, and only modestly higher than Mohs, which typically runs one to four percent. Notably, none of the six invasive cases recurred at all, which stands in contrast to that thirty percent recurrence figure quoted for photodynamic therapy used alone — suggesting the electrocoagulation debulking step may be doing real work. Cosmetic and functional outcomes were described as good, with only minor scarring or pigment change, and no severe complications.

The honest caveats here are exactly what you'd expect from a series this size: fourteen patients, a single center, no comparator arm, and follow-up that, while reasonable, isn't long enough to fully characterize recurrence risk in invasive disease over years rather than months. The authors are appropriately modest, calling this a potential alternative rather than a validated standard.

Where does this land practically. It's not practice-changing for patients who are good surgical candidates — Mohs remains your gold standard. But for the genuinely inoperable or treatment-averse elderly patient with invasive squamous cell carcinoma, this gives you another data point supporting a non-surgical combination approach beyond photodynamic therapy alone, worth knowing about as an option to discuss or refer for, even though the evidence base is still thin.

Third, a real-world pharmacoepidemiology study — a retrospective cohort analysis using the TriNetX United States collaborative network — asking whether oral niacinamide is associated with reduced risk of basal cell carcinoma, squamous cell carcinoma, and, notably, melanoma. The background here is familiar: niacinamide's chemopreventive benefit for nonmelanoma skin cancer is reasonably well established, but melanoma data has been sparse, despite a shared biological rationale around ultraviolet-induced DNA damage and immunosuppression.

The design choice worth dwelling on is the comparator group. Rather than using an unexposed general population, which risks all sorts of confounding by indication, the authors chose patients diagnosed with xerosis who did not receive niacinamide — a group that's clinically similar in terms of coming in for dermatologic care and follow-up, but with no inherent link to skin cancer risk. That's a smart active-comparator design, because it controls for the fact that people who see a dermatologist regularly, for any reason, are going to have different surveillance and detection patterns than the general population. They then did one-to-one propensity score matching across demographics, comorbidities, immunosuppression status, other medication exposures, and health care utilization, yielding two matched cohorts of just under half a million patients each.

Let's spend a moment on propensity score matching, since it's the analytic backbone of this paper and it comes up constantly in retrospective database studies. The problem it solves is confounding by indication — patients who get prescribed niacinamide are systematically different from patients who don't, in ways that could independently affect cancer risk, and you can't randomize people after the fact. So the propensity score collapses all of those baseline differences into a single number representing each patient's predicted probability of receiving niacinamide, based on their measured characteristics. You then pair up patients from the treated and untreated groups who have very similar propensity scores, which — if it works well — leaves you with two groups that look statistically alike on everything you measured, mimicking some of the balance you'd get from randomization. The tool the authors used to confirm this worked is the standardized mean difference, and here every covariate came in under zero point one, which is the conventional threshold for good balance — so the matching did its job on the variables they measured, though of course it can't balance what wasn't captured.

On top of that, they used both simple risk ratios and Kaplan-Meier-based hazard ratios to characterize the outcomes over a ten-year window. Kaplan-Meier methodology and its associated hazard ratio account for the fact that patients are followed for different lengths of time and some are censored before the full ten years elapses — instead of just comparing crude proportions, it compares the instantaneous rate at which the event occurs in each group across the entire follow-up period, which is the appropriate approach when you have time-to-event data with variable follow-up, as opposed to a fixed endpoint.

The results were striking. Over roughly four years of average follow-up, basal cell carcinoma occurred in about one percent of the niacinamide group versus three and a half percent of controls — roughly a two-thirds reduction in relative risk. Squamous cell carcinoma showed a similar pattern, under one percent versus two and a half percent. And melanoma — the novel finding here — occurred in about zero point four percent of the niacinamide group versus zero point eight percent of controls, roughly a fifty percent relative risk reduction, and this was statistically significant. The hazard ratios tracked closely with the risk ratios across all three outcomes.

In discussion, the authors note this melanoma signal is broadly consistent with in vitro and animal work showing niacinamide promotes apoptosis in melanoma cells and boosts interferon gamma production, though they're careful to say direct human trial evidence for melanoma specifically remains limited. They also contextualize their findings against a prior veteran cohort study that showed a similar protective association but was ninety-eight percent male, versus this cohort's forty-one percent male composition — meaningfully broadening generalizability. The limitations are the ones inherent to any claims-based retrospective design: reliance on diagnostic coding rather than confirmed histology in every case, no data on niacinamide dose or duration, likely underreporting of both exposure and outcomes, and the fundamental point that association is not causation — this design cannot prove niacinamide prevented these cancers, only that it's strongly associated with lower incidence in a real-world population.

Practically, this is a reassuring, hypothesis-strengthening study, not a practice-changing one on its own — you're likely already recommending niacinamide for high-risk keratinocyte carcinoma patients, and this adds a large, well-matched real-world data point supporting that habit, plus a genuinely novel and provocative melanoma signal that's worth watching but not yet something to promise patients will reduce their melanoma risk. Frame it as encouraging chemoprevention data, not proven melanoma prophylaxis.

Last, a retrospective cohort study from Washington University on the role of baseline imaging in Merkel cell carcinoma. The background is one every dermatologic oncologist should have memorized: Merkel cell carcinoma has a disproportionately high rate of clinically occult nodal and distant metastases at diagnosis, cited here from prior single-institution data at around thirteen percent, and while National Comprehensive Cancer Network guidelines recommend baseline imaging, real-world uptake is inconsistent and not universally covered by insurers. The gap this paper addresses is straightforward: what patient and tumor characteristics predict upstaging on baseline imaging, and how often does it actually happen.

The design is a retrospective cohort pulled from an institutional repository of biopsy-proven Merkel cell carcinoma patients treated between 2017 and 2023, specifically excluding patients with unknown primary tumors — a sensible exclusion, since those patients are almost by definition diagnosed because imaging already found something, which would artificially inflate the upstaging rate and bias the analysis toward the very outcome being studied. Upstaging was defined operationally as identification of regional or distant metastatic disease on imaging that wasn't apparent on physical exam, with cases counted only if they went on to receive treatment for metastatic disease — a pragmatic definition given that not every imaging-detected abnormality got histopathologic confirmation.

Statistically, this is a simpler paper — the main tool is the chi-squared test, used to compare the proportion of patients upstaged across different subgroups, like tumor size categories, age brackets, sex, and immunosuppression status. A chi-squared test is the basic workhorse for comparing categorical proportions between groups — it asks whether the pattern of counts you observed across categories deviates from what you'd expect if there were no real association, and it's the right tool here because both the exposure, like tumor size category, and the outcome, upstaged or not, are categorical rather than continuous.

Of a hundred eighty-four patients, the large majority — eighty-eight percent — actually received baseline imaging, and of those, just over twenty-two percent were upstaged. That's notably higher than the thirteen percent benchmark from prior literature, which strengthens rather than weakens the case for universal imaging. Among upstaged patients, the large majority — nearly eighty percent — had occult regional nodal disease, while the remainder had occult distant metastases. Primary tumor size over two centimeters was significantly associated with upstaging, at thirty-five percent, compared to twelve percent for tumors two centimeters or smaller, and this difference was statistically robust. Notably, age eighty and above, male sex, tumor location, and immunosuppression status did not reach significance as predictors.

The critical point the authors hammer home in discussion is that even that lower-risk group — small primary tumors under two centimeters — still had a meaningful upstaging rate of about twelve percent, split between occult nodal and occult distant disease. In other words, there's no safe subgroup where you can reasonably skip imaging based on tumor size alone. Nearly forty-two percent of upstaged patients went on to receive systemic immunotherapy, which underscores that this isn't an academic staging exercise — it's directly changing treatment selection for a meaningful fraction of patients. Limitations are appropriately disclosed: a fairly small sample that limits subgroup comparisons, like formally comparing PET-CT against CT alone, lack of universal histopathologic confirmation of imaging-positive findings, and no capture of the false-positive rate, which matters clinically since false positives drive unnecessary biopsies and patient anxiety.

This one is genuinely practice-relevant, even practice-affirming in a strong way. If you're involved in Merkel cell carcinoma staging decisions, or advocating with insurers who push back on imaging for smaller primary tumors, this data gives you a concrete number to cite: even small, node-negative-on-exam Merkel cell carcinomas carry a double-digit occult upstaging rate, and tumor size, while significantly associated with higher risk, does not identify a subgroup low-risk enough to forgo imaging. That's about as close to an actionable, guideline-reinforcing takeaway as a brief report gets.

That wraps up this month's four articles — a sharpened dermoscopic checklist for small facial lentigo maligna, a promising but early combination technique for inoperable squamous cell carcinoma, a large real-world signal for niacinamide extending possibly into melanoma prevention, and a strong reaffirmation that baseline imaging belongs in every Merkel cell carcinoma workup regardless of primary tumor size. Thanks for listening, and we'll see you next month.