Every clinician who has stood at the bedside of a non-communicating patient with confluent oral erosions has felt the same instinctive pull: empirical antifungals first, biopsy later. The reasoning is humane — relieve suffering, treat the most common cause, defer invasive workup until the situation is calmer. The reasoning is also frequently wrong, and the cost of being wrong is measured in months of misdirected therapy and irreversible structural damage to the oral mucosa. The discipline that prevents this error is morphologic, not historical: the lesion itself, examined carefully, tells you whether you are looking at a primary infection, a fibrin-floored ulcer, or — most consequentially — the floor of a ruptured intraepithelial blister.

This last category, autoimmune blistering disease of the mucosa, is where the bedside reading matters most. The patient cannot give you a clean blister history because the blisters do not survive long enough to be reported. The intraepithelial split that defines the family is fragile by construction: the blister roof, only a few cell layers thick, ruptures within hours of formation. What the patient brings to the emergency department is not a blister but its aftermath — a ragged erosion, often coated by a sloughed-off membrane that the inexperienced eye reads as exudate or fungal pseudomembrane.

Reading that membrane correctly is the single act that anchors the differential.

When the history is unreliable, the lesion still speaks

Translation gaps are the most familiar case. A patient and a clinician operate in different languages; a family member or a phone interpreter bridges the gap, fluently or not. Clinical vocabulary is precisely where the bridge buckles: “painful sores” survives translation; the temporal sequence of “small thin-walled blisters that broke in hours” often does not. What the family conveys is the present-tense description of erosions, not the lost prodrome.

Translation is the most common scenario but not the only one. Cognitive impairment in dementia, post-stroke aphasia, intubation, sedation, severe depression, and paediatric pre-verbal age all produce the same epistemic situation: the patient cannot reliably narrate the disease, and the clinician must extract the diagnosis from what is on the surface and what the laboratory will later confirm. In each of these settings, the temptation toward empirical broad-coverage therapy — antifungals, antivirals, sometimes both — is strongest, and the diagnostic cost of giving in to that temptation is highest.

This is also where the morphology earns its place. A grayish-white membrane on an erosion is not a single entity; it is a differential. Wiped away, candida pseudomembrane reveals erythematous, intact mucosa. Peeled away, the pseudomembrane of an intraepithelial blistering disease reveals a bright-red, friable, bleeding surface — the floor of a vesicle that has lost its roof. The two findings are bedside-distinguishable in seconds, by anyone trained to look for the difference.

Why the distinction is worth a steroid course

The cost of confusing the two is not symmetric. Treating an autoimmune blistering disease as candidiasis delays effective therapy by weeks to months. Treating candidiasis as autoimmune blistering disease — which essentially never happens, because the morphology forces a biopsy — costs a single histology slide. Asymmetric cost favors the diagnostic gesture every time.

The pseudomembrane is not a coating

The mechanism behind the morphology deserves more attention than it typically receives in trainee teaching. A true pseudomembrane in candidiasis is a superficial fungal-fibrin-keratinocyte coagulum sitting on top of intact epithelium; the epithelium underneath, while inflamed, is structurally continuous. Diphtheritic pseudomembrane, similarly, is a deposited layer.

The grayish-white sheet on an erosion of pemphigus, by contrast, is the epithelium — specifically, the suprabasal portion of it, separated from the basal layer by an antibody-mediated split between desmoglein-bound keratinocytes. The roof of the blister loses adhesion to its floor, fluid accumulates briefly between them, the roof rolls back or sloughs forward, and what remains adherent to the lesion edge is the rolled-up, partially detached upper epithelium. Acantholysis — loss of cell-to-cell adhesion within the epithelium — is the histologic correlate of what the eye sees clinically as “membrane that, when peeled, exposes a raw bleeding floor.”

The cleavage plane is diagnostic. Subepithelial split (mucous membrane pemphigoid, bullous pemphigoid, linear IgA, dermatitis herpetiformis) leaves the full thickness of the epithelium as the blister roof — the roof is therefore relatively durable, the blisters survive long enough to be observed, and the resulting erosion has a non-membranous, often fibrin-floored appearance. Suprabasal intraepithelial split (pemphigus vulgaris and its variants, paraneoplastic pemphigus) leaves only the upper epithelial layers as roof, which are too thin to survive — hence no blister history, only erosion plus rolled membrane.

This is not academic. The clinical translation is direct: membrane peels away cleanly to a friable bleeding base argues for an intraepithelial process before any laboratory has been drawn.

Multi-mucosal pattern as the second anchor

Once the morphology has narrowed the field to intraepithelial autoimmune blistering disease, the next reading is anatomic: which mucosae are involved, and which conspicuously are not. The map is informative because the autoantigen distribution is informative. Desmoglein 3 is enriched in oral, genital, and conjunctival mucosa. Desmoglein 1 dominates skin. The clinical extension of an antibody profile follows where its target is expressed.

This is why the patient with simultaneous oral and perianal or genital involvement, but unremarkable skin and unremarkable conjunctivae, fits a recognizable pattern. The pattern is not specific to a single diagnosis on its own — mucous membrane pemphigoid, paraneoplastic pemphigus, and pemphigus vulgaris all produce multi-mucosal disease — but it sharply excludes isolated aphthosis (single-site, recurrent, self-limited), localized infection (regional, often unilateral), erythema multiforme (target lesions on extremities, mucosae are involved but skin dominates), and Stevens-Johnson syndrome (febrile, systemic, palmoplantar lesions).

The map reading also signals which subtype within the autoimmune blistering family is in play. Conjunctival predominance with scarring favors mucous membrane pemphigoid. Severe polymorphic skin disease plus oral involvement, plus palmoplantar erosions, raises paraneoplastic pemphigus and prompts the search for an underlying lymphoproliferative malignancy. Predominantly oral disease with secondary genital or oesophageal involvement, no scarring of the conjunctivae, and no palmoplantar component, fits classical mucosal-dominant pemphigus vulgaris.

The negative findings carry equal weight

What is absent from the examination earns its place in the report. Absent palmoplantar lesions argue against erythema multiforme major and Stevens-Johnson. Absent conjunctival involvement argues against mucous membrane pemphigoid as the dominant entity. Absent truncal rash argues against typical paraneoplastic pemphigus. The pertinent negatives are not space-fillers; in autoimmune blistering disease, the absence of involvement of certain anatomic sites is part of the clinical fingerprint that distinguishes one entity from its neighbors.

What B-cell biology has to do with the lesion

Behind the bedside picture is a B-cell story that has been only partially told in clinical teaching. Pemphigus vulgaris, the prototype intraepithelial blistering disease, is driven by IgG autoantibodies — predominantly IgG4 — directed against desmoglein 3 (and, in mucocutaneous disease, desmoglein 1). The antibodies are not passive markers; they are pathogenic. When desmoglein function is blocked at the cell-cell junction, keratinocyte adhesion fails, and the suprabasal split that produces the entire clinical picture follows.

The cells that produce these antibodies are not, contrary to traditional teaching, a homogeneous “B-cell pool.” They are a stratified population. Naïve B-cells, having never encountered antigen, contribute to the broad protective repertoire and respond to vaccination and new infections. Short-lived plasmablasts produce a transient burst of antibody during acute response. Long-lived plasma cells in bone marrow niches sustain the durable antibody titer that protects against pathogens previously encountered. And memory B-cells — the central population for autoimmunity — circulate quietly between germinal centers, retain the molecular memory of their cognate antigen, and re-expand on antigen re-exposure into both plasmablasts and the plasma-cell niche.

In autoimmune disease, the offending memory B-cell population is long-lived, antigen-specific, and resistant to elimination by the routine turnover of the broader B-cell compartment. Broad-spectrum B-cell-depleting therapy — historically successful in pemphigus and other antibody-mediated autoimmune diseases — depletes naïve B-cells, plasmablasts, and a portion of memory B-cells alike, and accepts the cost of broad humoral suppression for the benefit of attriting the pathogenic memory pool. The cost is real: vaccine response is blunted, hypogammaglobulinemia accumulates with repeated dosing, and infection risk rises in the months following depletion.

The therapeutic question that follows is whether the autoreactive memory subset can be targeted selectively, sparing the naïve repertoire, the plasmablast compartment, and the protective memory pools. Surface markers that segregate cleanly between long-lived autoreactive memory B-cells and the rest of the compartment are the technical prerequisite. Conformational variants of CD20 with restricted expression on memory B-cell subsets are one of several candidate axes; selective markers on the memory population, identified through deep B-cell phenotyping, are another. The shared logic is the same: precision over breadth.

The translational implication, set aside from any specific clinical program, is that the design space for next-generation B-cell-directed therapy in autoimmune blistering disease is now defined less by total B-cell ablation and more by which subset of the B-cell compartment is targeted. The morphology that sent the patient to biopsy, in other words, also points toward the immunology that defines the next decade of treatment.

From morphology to mechanism

Returning to the bedside, the diagnostic sequence is short and inexpensive. Read the membrane. If it peels to a friable bleeding base, the lesion is intraepithelial, the differential is autoimmune blistering disease unless proven otherwise, and the next step is biopsy of the erosion edge with perilesional intact tissue — for histology and direct immunofluorescence — before empirical systemic immunosuppression. Histology will show the cleavage plane and the acantholytic basal layer; direct immunofluorescence will show the intercellular IgG fishnet pattern that confirms the diagnosis and distinguishes the pemphigus subtype.

The therapeutic decisions that follow — initiation, regimen choice, and the increasingly central question of how broadly the B-cell compartment is depleted — belong to the dermatology service rather than to the emergency department or the referring clinic. But the diagnostic anchor that brings the patient to that decision begins at the bedside, and depends on a single morphologic reading: the membrane is not a coating, and the surface beneath it is not intact mucosa. It is the floor of a blister whose roof has been undone by an antibody whose origin can, in principle, be traced to a small and identifiable population of long-lived B-cells.

The therapeutic frontier is the precision with which that population is now being targeted.