Welcome to this month's journal review, covering the September twenty twenty-six issue of the Journal of the American Academy of Dermatology. We've got four pieces on the docket today — two are "Highlights" commentaries synthesizing recent literature, one is a "Game Changer" retrospective narrative on phototherapy and skin cancer risk, and one is a clinician's perspective piece coining a new subspecialty term for nail oncology. Let's get into it.

First up is a Highlights from JAAD Reviews piece by Shari Lipner, framed around the diagnostic and management challenges of nail unit squamous cell carcinoma — a tumor she rightly points out gets misdiagnosed as verruca or chronic paronychia for years before anyone thinks to biopsy it. This is a synthesis commentary, not a primary study, and it pulls together two recent papers. The first is an expert panel clinical review by Curtis and colleagues in JAAD Reviews, laying out the evidence for HPV, human papillomavirus, association and genital-digital transmission, and reinforcing that Mohs micrographic surgery is standard of care here — recurrence in the range of about four to ten percent, compared to twenty-five to over fifty percent with conventional excision. That's not a subtle difference; it's the kind of number that should anchor your treatment counseling.

The second paper is a multinational retrospective series — the largest to date, two hundred sixty-nine tumors in two hundred sixty-one patients — and it's actually the same underlying study we'll circle back to in our fourth article, so I'll keep this pass focused on what Lipner's commentary emphasizes. Mean diagnostic delay was about three years, which is a sobering number for a squamous cell carcinoma. Tumors clustered on the thumb, index, and middle fingers, which the authors speculate reflects either genital-digital HPV inoculation patterns or simply the mechanics of cigarette holding — an interesting but unproven hypothesis worth flagging as such.

The clinically useful nugget here is the predictor data: subungual ooze independently predicted invasive disease, while pain and nail plate loss were associated with invasion on univariate analysis. Koilocytosis, on the other hand, tracked with in situ tumors, periungual location, and younger patients. This maps neatly onto a two-subtype model proposed elsewhere — a blue-basaloid, HPV-driven, periungual, in-situ phenotype versus a pink-keratinizing, non-HPV, subungual, invasive phenotype. If you're trying to triage a nail lesion clinically before you even get pathology back, that framework is genuinely useful.

Both source papers reinforce the genital-digital HPV transmission story. One cited study found that nearly a quarter of HPV-related nail cases had HPV disease elsewhere in the body, and in the larger retrospective series, roughly one in ten patients had a history of HPV disease at genital, cervical, perianal, or oropharyngeal sites. The practical message: ask your nail squamous cell carcinoma patients about anogenital HPV history, and consider referral for examination.

On limitations — and this is important since it's synthesized across two papers — HPV status in some patients was assigned by koilocytosis on histology rather than confirmed by PCR, polymerase chain reaction, which introduces real misclassification risk. Imaging criteria for bony invasion remain undefined, recurrence data are still thin, and there's essentially no data on patients with skin of color. Open questions include whether pre- and post-surgical PCR for high-risk HPV subtypes, or p16 immunohistochemistry on Mohs margins, could reduce recurrence — neither is ready for practice yet, but both are reasonable directions.

Practical takeaway: this is not practice-changing in the sense of introducing something brand new — Mohs as first-line for nail unit squamous cell carcinoma is already your practice — but it usefully sharpens your index of suspicion. Subungual ooze plus pain plus nail plate loss should raise your pretest probability of invasion, and a documented HPV history elsewhere in the body should prompt you to ask about genital HPV exposure. Interesting-but-not-yet-actionable: PCR-guided margin assessment and p16 immunostaining protocols.

Our second article is a JAAD Game Changer piece — these are short first-person narratives from senior dermatologists reflecting on how a particular study changed their practice, so it's not a full original research walkthrough; I'll follow its actual shape. Jane Grant-Kels writes about a retrospective cohort study looking at skin cancer incidence following ultraviolet phototherapy without psoralens — so broadband UVB, narrowband UVB, and combined UVA-UVB, explicitly excluding PUVA, psoralen plus ultraviolet A, which already has well-established risk data. The underlying study reviewed more than thirty-five hundred patients treated between nineteen seventy-seven and twenty eighteen, predominantly for psoriasis or atopic dermatitis.

The design here is a retrospective cohort compared against general population incidence rates — and although the text doesn't spell out the exact statistical methodology, this is the classic setup for what's often done as a standardized incidence ratio comparison: you count observed cancers in your exposed cohort and compare them to the number you'd expect based on population cancer registry rates, adjusted for age and follow-up time. I want to flag that inference as mine, not something the article states explicitly, but it's the standard tool for exactly this kind of question, and it makes sense here because you can't randomize decades of phototherapy exposure — you're stuck comparing an observed cohort to an external benchmark.

The result: no significant difference in melanoma, squamous cell carcinoma, or basal cell carcinoma incidence between the phototherapy cohort and the general population. They also looked for a cumulative dose-response relationship between UVB exposure and skin cancer and didn't find one. Grant-Kels' takeaway, in her own words, is that she no longer feels obligated to pursue enhanced skin cancer screening specifically because a patient received non-PUVA phototherapy — narrowband UVB, broadband UVB, or combined UVAB does not appear to carry the same carcinogenic signal that PUVA does.

Practical takeaway: this one is genuinely practice-relevant reassurance, particularly if you're the one being asked to co-manage phototherapy patients or field questions about cancer risk from psoriasis or atopic dermatitis patients your colleagues are treating. It does not change how you screen PUVA patients — that risk remains well established — but it should ease anxiety, yours and theirs, around non-psoralen UV regimens.

Third, another Highlights piece, this one from JAAD International, by Jonathan Kantor, covering two studies on noninvasive imaging for basal cell carcinoma diagnosis. The first is a head-to-head comparison of line-field confocal optical coherence tomography against conventional OCT in nearly two hundred clinically equivocal lesions, using histopathology as the reference standard — the right comparator, since you need ground truth to judge any imaging modality. The second is an external validation study of a deep learning model applied to reflectance confocal microscopy images for basal cell carcinoma diagnosis.

The line-field confocal OCT technology produced more high-confidence diagnoses and better sensitivity than conventional OCT when the question required subtype certainty, though it was actually less sensitive specifically for infiltrative basal cell carcinomas — worth remembering, since infiltrative subtype is exactly the one where you most want confidence before deciding on treatment approach.

The deep learning model is where I want to pause for a quick methodology point, because this is a great teaching example of internal versus external validation. Internally, the model performed almost perfectly, with an area under the curve approaching zero-point-nine-eight — area under the curve, for anyone who wants the refresher, is essentially a measure of how well a diagnostic test discriminates cases from non-cases across every possible threshold, where one-point-oh is perfect and point-five is coin-flip guessing. So point-nine-eight internally is outstanding. But when they tested the same model on external cohorts — different patients, different imaging setups, different populations than it was trained on — performance dropped to around point-eight-six, and in one cohort as low as point-seven-four. That gap between internal and external performance is the single most important thing to watch for in any diagnostic AI paper, because a model that only performs well on the data it was tuned on isn't yet a model you can trust in your own clinic. The authors deserve real credit for even doing external validation across two independent cohorts, since plenty of diagnostic AI papers skip that step entirely.

Limitations across both studies: the OCT comparison was single-assessor and single-center, so we don't know about interobserver reliability or generalizability across readers, and the deep learning analysis was retrospective and prevalence-dependent, meaning its apparent accuracy will shift depending on how common basal cell carcinoma is in whatever population you deploy it on.

Practical takeaway: interesting, not yet actionable. Neither of these technologies is ready to replace biopsy in your practice, but line-field confocal OCT's edge in high-confidence subtyping — tempered by its infiltrative-subtype blind spot — and the sobering external-validation drop for the AI model are both worth filing away as this field matures.

Our fourth and final piece is a Clinician's Perspective by Warren Heymann, proposing the term "onycho-oncology" as a unifying label for nail unit malignancy awareness, and walking through three recent papers to make the case. This is an editorial-style piece, so I'll follow it as a synthesis rather than a single study.

The first paper he discusses is the same large retrospective nail squamous cell carcinoma series we touched on earlier, but Heymann gives us more granular numbers worth having on hand. Of the two hundred sixty-nine tumors, fifty-five percent were in situ and forty-five percent were invasive. The tumors clustered on the right index and middle fingers and both thumbs in about six in ten cases, predominantly in men in their sixties. Koilocytosis was present in about a third of tumors overall, and significantly more common in the in situ group. Subungual ooze carried roughly a four-fold increase in the odds of invasive disease, and both pain and nail plate loss were significantly associated with invasion as well — so that predictor triad we discussed earlier now has real numbers behind it. Mohs micrographic surgery was the most common treatment, used in about four in ten cases, and overall recurrence was under ten percent, with no significant difference in recurrence across treatment modalities. The authors' summary is clean and clinically usable: periungual presentation with a warty, horizontally spreading appearance suggests HPV-associated in situ disease, often sexually transmitted, while an oozing, ulcerating subungual tumor suggests dermal invasion — and digit-sparing surgery remains first-line regardless of subtype.

The second paper Heymann covers is new territory relative to article one — a comparison of nail unit acral melanoma against non-nail acral melanoma, two hundred eighty-one patients total, fifty-eight with nail unit involvement and two hundred twenty-three without. Ulceration was significantly more common in the nail unit group — roughly a third versus under a fifth in the non-nail group. Most nail unit melanoma patients presented at tumor stage T2 or higher, meaning a Breslow depth of at least one millimeter. The authors raise two possible explanations, and importantly don't adjudicate between them: this could reflect diagnostic delay — the nail unit is notoriously easy to misdiagnose as trauma or infection — or it could reflect genuinely greater biological aggressiveness in nail unit melanoma. Since this is a commentary on someone else's retrospective comparison, we don't get a statistical methodology to teach here, but the clinical message stands regardless of mechanism: any pigmented nail unit lesion deserves a lower threshold for biopsy than an equivalent lesion elsewhere on acral skin.

The third piece is a Delphi consensus statement — an international expert consensus process, not a primary data study — on perioperative management in nail surgery, led by Sechi and colleagues. Consensus was reached across a long list of practical domains: antibiotic prophylaxis, bleeding risk management, anesthesia technique, antisepsis, tourniquet use, postoperative analgesia, dressing choice, wound healing expectations, follow-up strategy, and prevention of long-term complications like nail dystrophy and functional impairment. They've distilled this into an eight-step perioperative flowchart intended to standardize practice.

Heymann's closing argument is that all three papers point toward the same clinical vulnerability — nail unit tumors, benign and malignant alike, get mistaken for inflammatory or infectious mimics, and his proposed term "onycho-oncology" is meant as a cognitive nudge to keep malignancy on the differential.

Practical takeaway from this piece: the nail unit melanoma ulceration and staging data reinforces a low threshold for biopsy of pigmented nail lesions — useful awareness, though it doesn't change your existing surgical approach. The Delphi consensus flowchart is the most directly actionable item here if you're looking to standardize perioperative protocols in your own nail surgery practice — worth pulling the original paper if you do a meaningful volume of nail procedures.

That wraps our four articles this month — a nail squamous cell carcinoma diagnostic and outcomes picture that sharpens your clinical suspicion, welcome reassurance on non-psoralen phototherapy and skin cancer risk, a sobering lesson in internal-versus-external validation for diagnostic AI in basal cell carcinoma, and a call to keep nail unit malignancy squarely on your differential. Thanks for listening, and I'll see you next month.