Introduction

Oral lichen planus (OLP) is a chronic immune-mediated disorder of the oral mucosa that presents with reticular, plaque-like, atrophic, or erosive lesions. Diagnosis requires clinicopathological correlation because its clinical and histopathological features may overlap with those of oral lichenoid lesions and other interface mucosal disorders [1,2]. OLP lesions characteristically contain T lymphocytes beneath and within the epithelium, accompanied by basal keratinocyte degeneration and apoptosis [1,3]. Although topical or systemic immunomodulatory treatment can suppress symptomatic inflammation, lesions may persist or recur over prolonged periods [2]. The mechanisms responsible for active epithelial injury may therefore differ, at least partly, from those that allow inflammation to remain localized or re-emerge after clinical improvement.

Conventional immunopathogenic models emphasize antigen presentation, T-cell recruitment and polarization, interferon (IFN)-γ signaling, and CD8⁺ T-cell-mediated epithelial cytotoxicity [3-5]. These mechanisms account for major features of active OLP, including interface inflammation, basement membrane disruption, and basal keratinocyte injury [3,4]. They do not fully explain, however, why inflammatory activity remains restricted to previously affected mucosal sites, why clinically improved lesions frequently recur, or why patients with similar histological findings can follow different clinical courses.

Single-cell RNA sequencing (scRNA-seq), immune-repertoire sequencing, multiplex imaging, and spatial transcriptomics have revealed substantial cellular and spatial heterogeneity within the OLP microenvironment. OLP lesions contain tissue-associated and clonally expanded CD8⁺ T-cell populations [6,7], inflammatory fibroblast subsets with immune-regulatory and lymphoid-organizing features [8,9], and B-cell and plasma-cell responses associated with tertiary lymphoid structures (TLSs) [10,11]. These observations suggest that chronic OLP may depend not only on repeated recruitment of circulating effector cells but also on local cellular and structural conditions that retain, reactivate, or regenerate pathogenic immune states.

Previous reviews have broadly addressed OLP immunopathogenesis, molecular pathways, clinical heterogeneity, and therapeutic implications [3,12-15]. The present review instead focuses on a narrower unresolved question: whether local persistence and recurrence are explained primarily by long-lived tissue-resident memory T (T_RM) cells, by repeated generation of cytotoxic CD8⁺ T-cell states, or by a multicellular niche comprising stromal cells, antigen-presenting cells, and TLSs. Qing et al. [7] linked a T_RM-associated transcriptional and spatial phenotype to erosive OLP and recurrent erosion. Jiang et al. [16], using a broader atlas of cutaneous and mucosal lichen planus, found no prominent canonical T_RM signature and instead identified CXCL13⁺ cytotoxic CD8⁺ T-cells associated with an inferred classical type 2 dendritic cells subset A (cDC2A)–interleukin (IL)-15 interaction. These findings are not necessarily mutually exclusive and may represent alternative or overlapping tissue-associated pathogenic states. Accordingly, this review compares cellular persistence with niche persistence while distinguishing cross-sectional associations from evidence that particular cells or tissue structures remain through clinical improvement and contribute to recurrence. The proposed framework is summarized in Fig. 1, and the principal supporting studies and their evidence boundaries are presented in Table 1.

From Epithelial Cytotoxicity to Local Immune Persistence

OLP is classically described as a T-cell-mediated interface mucositis in which cytotoxic activity is concentrated at the epithelial–connective tissue junction. CD8⁺ T-cells accumulate within the epithelium and along the basement membrane, where they are associated with basement membrane disruption and basal keratinocyte injury [4]. Th1-associated cytokines, particularly IFN-γ, reinforce inflammatory signaling at this interface [5]. Dendritic cells are also recruited to OLP lesions and may contribute to local antigen presentation and T-cell activation [17]. This model accounts for the effector phase of an active lesion but does not explain which immune components remain—or are re-established—after clinical improvement.

Immune-repertoire analyses extend this framework beyond the repeated recruitment of circulating effector cells. ScRNA-seq combined with T-cell receptor (TCR) and B-cell receptor (BCR) repertoire sequencing identified activated tissue-associated CD8⁺ T-cell populations and substantial TCR and BCR clonal expansion within OLP lesions [6]. Clonal enrichment is consistent with local selection or expansion of antigen-experienced lymphocytes, although the relevant antigens remain unknown [6]. Moreover, detection of expanded clonotypes in an active lesion does not demonstrate that the same clones survive treatment or clinical remission.

Systemic immune heterogeneity does not fully account for site-restricted chronicity. A blood-based study of 91 participants identified differences in circulating lymphocyte features, stimulated cytokine secretion, and inhibitory-receptor expression between reticular and erosive OLP [18]. Because the analysis was cross-sectional, it could not determine which immune components remain at previously affected mucosal sites or predict recurrence at the same location [18]. The remaining question is therefore which tissue-associated cells and structures persist, or are locally reconstructed, between clinically apparent episodes.

Tissue-resident Memory T-cells and Persistent CD8⁺ T-cell States

T_RM cells are antigen-experienced lymphocytes retained in non-lymphoid tissues with limited recirculation. Their phenotype varies according to tissue and inflammatory context: CD69 is widely used as a retention-associated marker, whereas CD103, CD49a, and CXCR6 are expressed less consistently [19]. Because CD69 can also indicate recent activation, tissue residence cannot be assigned from a single marker in an actively inflamed lesion [19]. Anatomical localization, transcriptional state, and persistence over time must therefore be considered together when interpreting T_RM-like cells in OLP.

Li et al. [6] identified a T_RM-associated CD8⁺ T-cell population by single-cell profiling of OLP tissue and paired peripheral blood. Multiplex immunofluorescence confirmed CD8⁺CD69⁺ cells within lesions, and repertoire sequencing demonstrated clonally expanded T- and B-cell populations [6]. These results indicate a spatially restricted and clonally structured immune response but do not show that the same cells remain in the mucosa through treatment, improvement, and recurrence.

Qing et al. [7] related the T_RM-associated phenotype more directly to clinical presentation. ScRNA-seq and spatial transcriptomics localized the annotated CD8⁺ T_RM population mainly to the lamina propria adjacent to the epithelial basement membrane, with greater abundance in erosive than in non-erosive OLP [7]. These cells expressed IFN-γ, tumor necrosis factor-α, IL-17, granzymes, and perforin, and cells isolated from erosive lesions produced more inflammatory cytokines than those from non-erosive lesions [7]. In a separate cohort followed for one year, several T_RM-associated and effector genes were expressed at higher levels in patients with recurrent erosion than in those with persistent non-erosion [7]. Thus, the study links a T_RM-associated molecular and spatial phenotype to erosive OLP and recurrent erosion, without demonstrating survival of the same resident clones over time.

The OLP-specific findings of Li et al. [6] and Qing et al. [7] do not establish a uniform resident-memory compartment in all lesions. CD69 alone cannot distinguish tissue residence from recent activation [19], and OLP contains multiple CD8⁺ T-cell states. Chen et al. [20] identified an expanded terminally exhausted CD8⁺ population expressing PDCD1, LAG3, and TIGIT, together with reduced average expression of several effector molecules. Although based on a small cross-sectional cohort, this finding indicates that active OLP contains heterogeneous cytotoxic, activated, and dysfunctional T-cell states rather than one continuously active effector population [20].

Jiang et al. [16] subsequently reported a different T-cell landscape that limits a simple T_RM-dominant interpretation. Their integrated single-cell and spatial study examined several forms of lichen planus and included six oral mucosal biopsies [16]. Classical T_RM-associated genes, including ITGA1 (CD49a), CD69, and CXCR6, were not prominent across the identified T-cell populations [16]. Instead, CXCL13⁺CD8⁺ T-cells were enriched in cutaneous and mucosal lesions and expressed IFN-responsive and cytotoxic programs characterized by IFNG, GZMB, PRF1, and TNFRSF9 [16]. Spatial analyses linked inflammatory T-cell signals with epithelial responses, and cDC2As were identified as a major source of IL-15 and a potential contributor to the CXCL13⁺ cytotoxic state [16].

The findings of Qing et al. [7] and Jiang et al. [16] are not directly contradictory. Qing et al. [7] examined OLP specifically, compared erosive and non-erosive disease, and evaluated recurrent erosion during follow-up. Jiang et al. [16] constructed a cross-subtype atlas of cutaneous, follicular, and mucosal lichen planus; their study was not designed to determine whether oral T-cell clonotypes survive remission or predict recurrence. The studies also differed in cohort composition, tissue processing, analytical strategy, and criteria for T_RM annotation. Qing et al. [7] therefore support a T_RM-like pathogenic state in erosive OLP, whereas Jiang et al. [16] demonstrate that mucosal lichen planus can also contain CXCL13⁺ cytotoxic CD8⁺ T-cell states without a prominent canonical T_RM signature.

The unresolved issue is whether these populations are separate lineages, successive differentiation states, or overlapping responses to chronic antigenic stimulation. Some OLP lesions may retain long-lived T_RM-like cells, whereas others may rely more heavily on a local environment that repeatedly generates or sustains cytotoxic CD8⁺ T-cell states. Serial same-site sampling with TCR clonotype tracking and spatial phenotyping before treatment, after clinical improvement, and at recurrence will be required to distinguish durable tissue residence from recurrent local differentiation.

Tertiary Lymphoid Structures as Organizers of Local Immunity

Whereas T_RM-like cells provide a possible cellular basis for tissue-restricted recall responses, TLSs provide an anatomical setting for sustained local adaptive immunity. TLSs develop in chronically inflamed non-lymphoid tissues and range from loosely organized lymphocyte aggregates to compartmentalized structures containing T- and B-cell zones, antigen-presenting cells, specialized stromal cells, and high endothelial venule-like vessels [21,22]. Depending on their degree of organization, TLSs can support local antigen presentation, lymphocyte recruitment and activation, and B-cell differentiation [21,22].

Wang et al. [10] provided direct histological evidence of TLS formation in OLP. TLSs were defined by organized CD20⁺ B-cell and CD3⁺ T-cell aggregates associated with peripheral node addressin (PNAd)-positive vessels and were detected in 44 of 51 OLP specimens [10]. Higher TLS scores were accompanied by greater numbers of CD20⁺ B cells, CD208⁺ dendritic cells, and PNAd-positive vessels [10]. These findings establish that OLP lesions frequently contain organized lymphoid microenvironments, although they do not determine whether TLSs sustain disease or develop as a consequence of prolonged inflammation.

Earlier histological studies documented CD27⁺ and CD38⁺ lymphocytes, mixed T-cell/B-cell infiltrates, CD20⁺ B cells, and CD138⁺ plasma cells in OLP lesions [23-25]. Epstein–Barr virus (EBV)-positive CD138⁺ plasma cells were also enriched in selected erosive lesions, and increased circulating CXCR5⁺CD4⁺ T follicular helper-like cells have been reported in OLP [26,27]. These observations demonstrate B-lineage and humoral immune heterogeneity but do not establish that the corresponding responses occur within functional mucosal TLSs.

Given the presence of EBV-positive plasma cells in erosive lesions [26], latent viral reactivation could potentially contribute to a persistent inflammatory circuit. Chronic viral antigens may provide the continuous stimulation necessary for CCL19⁺ fibroblast activation and subsequent TLS organization, although this mechanistic link requires direct experimental validation in OLP.

Transcriptomic studies have related humoral immune features more directly to clinical course. Qing et al. [28] identified B-cell-related molecular modules that distinguished stable OLP from recalcitrant erosive OLP. This finding implicates humoral immune dysregulation in unfavorable disease courses but does not identify TLSs as the source of the B-cell signal [28]. Yang et al. [11] subsequently classified OLP transcriptomic cohorts using a TLS-related gene signature. A humoral subtype enriched for B-cell pathways and the plasma-cell-associated markers IRF4 and TNFRSF17 was associated with erosive and recurrent OLP [11]. Tissue analysis showed that CD20⁺ B cells were concentrated within TLS-like structures, whereas CD38⁺ plasma-cell infiltration was associated with erosive disease [11]. Because the validation cohort included both OLP and oral lichenoid lesions and the analyses were observational, TLS-associated humoral activity has not been established as a prospective predictor or cause of recurrence.

The stromal compartment may help organize these lymphoid structures. Lai et al. [9] combined scRNA-seq with multiplex immunofluorescence and identified CCL19⁺ fibroblasts associated with T-cell chemotaxis and inducible TLS formation in OLP. Lymphotoxin-related signaling was implicated in CCL19⁺ fibroblast activation, and inducible TLS-positive lesions were associated with atrophic-erosive disease and higher clinical disease scores [9]. The retrospective design, however, did not determine whether CCL19⁺ stromal niches or inducible TLSs remain after treatment and clinical improvement.

TLSs may therefore provide a local platform for coordinated T-cell, B-cell, dendritic-cell, and plasma-cell interactions in OLP. They should not yet be described as an established immune-memory system. No study has shown that OLP-associated TLSs retain antigen-specific clonotypes during clinical remission, initiate subsequent recurrence, or directly generate or maintain pathogenic T_RM cells. The principal studies supporting this interpretation and their evidence boundaries are summarized in Table 1.

Stromal and Dendritic-cell Networks Connect Persistent Immunity to Epithelial Injury

Fibroblasts and dendritic cells may connect local immune organization with epithelial injury by influencing T-cell recruitment, activation, and effector function. Most available studies, however, have examined established OLP lesions and demonstrate support of active inflammation more directly than persistence through treatment and remission.

Zhang et al. [29] reported increased fibroblast activation protein-α and CCL5 expression in OLP tissues and lesion-derived fibroblasts. In vitro, OLP-derived fibroblasts promoted CD4⁺ T-cell proliferation and migration while reducing T-cell apoptosis [29]. Fibroblasts from active lesions can therefore support T-cell accumulation, although their persistence after clinical improvement was not assessed.

Cheng et al. [8] further resolved fibroblast heterogeneity using scRNA-seq, primary-cell analyses, and multiplex tissue imaging. A disease-associated SFRP2⁺WNT5A⁺ subset expressed antigen-processing and presentation-related genes, including human leukocyte antigen class I molecules and ERAP2, together with CXCL12 and CXCL14 [8]. These fibroblasts were located near CD8⁺ T-cells in the superficial inflammatory compartment, and computational analyses implicated them in communication with both CD8⁺ T-cells and epithelial cells [8]. The results support stromal amplification of T-cell–epithelial inflammation but do not demonstrate functional antigen presentation in vivo, selective support of T_RM cells, or persistence of the fibroblast state after remission.

Jiang et al. [16] identified a complementary dendritic-cell network in cutaneous and mucosal lichen planus. cDC2As were increased in both tissue compartments and represented the principal myeloid-cell source of IL-15 [16]. Ligand–receptor and upstream-regulator analyses implicated this signal in supporting CXCL13⁺CD8⁺ T-cells expressing IFNG, GZMB, PRF1, and TNFRSF9 [16]. The inferred cDC2A–IL-15 interaction was increased in mucosal lesions relative to healthy mucosa, although spatial proximity between cDC2A and CXCL13⁺CD8⁺ T-cells was more pronounced in cutaneous disease [16]. Spatial analyses also linked inflammatory T-cell signals to epithelial responses [16].

TNFRSF9 encodes the costimulatory receptor 4-1BB. In an IFN-γ-primed keratinocyte–peripheral blood mononuclear cell co-culture system, 4-1BB blockade reduced keratinocyte death, providing functional evidence that the 4-1BB pathway can contribute to T-cell-mediated epithelial cytotoxicity [16]. The experiment did not use OLP-derived cells, and cDC2A-derived IL-15 was not directly manipulated in oral mucosal tissue. The study therefore supports individual components of the proposed network without establishing a complete causal pathway from dendritic-cell signaling to recurrent epithelial injury in OLP.

Together, these studies indicate that fibroblast and dendritic-cell networks can create local conditions that support pathogenic T-cell activity in established lesions. Whether the same stromal and myeloid states remain after clinical improvement or are reconstructed before recurrence is unknown. Accordingly, Fig. 1 presents these networks as candidate contributors to local immune persistence rather than as an established causal sequence.

An Integrated Model of Local Immune Persistence

The available evidence can be organized into two non-exclusive hypotheses: cellular persistence and niche persistence. The cellular-persistence hypothesis proposes that antigen-experienced pathogenic lymphocytes remain in previously affected mucosa and rapidly resume effector activity. The niche-persistence hypothesis instead proposes that stromal, myeloid, and lymphoid components preserve the capacity to retain, reactivate, or repeatedly generate pathogenic T-cell states even when individual effector cells are not permanently maintained. Both possibilities are incorporated into the conceptual framework shown in Fig. 1.

Clonally expanded tissue-associated CD8⁺ T-cell populations in OLP [6] and the association of a T_RM-like phenotype with erosive disease and recurrent erosion [7] are compatible with cellular persistence. Neither study, however, tracked the same cells or clonotypes through treatment and clinical improvement. The coexistence of activated and exhausted CD8⁺ T-cell states in OLP [20], together with CXCL13⁺ cytotoxic CD8⁺ T-cells lacking a prominent canonical T_RM signature in mucosal lichen planus [16], further indicates that chronic cytotoxic activity need not depend on a single stable resident-memory phenotype.

Evidence for niche persistence is indirect but involves several tissue compartments. FAP-α⁺ and SFRP2⁺WNT5A⁺ fibroblasts promote T-cell accumulation or T-cell–epithelial inflammatory interactions in active OLP lesions [8,29]. CCL19⁺ fibroblasts are associated with lymphocyte recruitment and inducible TLS organization [9], whereas an inferred cDC2A–IL-15 interaction has been implicated in supporting CXCL13⁺ cytotoxic CD8⁺ T-cell states in mucosal lichen planus [16]. OLP-associated TLSs contain organized T-cells, B cells, dendritic cells, plasma cells, and PNAd-positive vessels [10]. Inducible TLS formation and TLS-associated humoral activation have been linked to atrophic-erosive disease, greater clinical activity, and recurrent phenotypes [9,11]. These cross-sectional and retrospective observations do not show that the same stromal, myeloid, or lymphoid architecture remains after clinical improvement. A direct functional relationship between TLSs and pathogenic T_RM cells has also not been demonstrated.

This framework therefore separates the immune architecture of an active lesion from components that remain after clinical activity subsides. Demonstration of local immune persistence will require paired sampling of the same mucosal sites before treatment, after clinical improvement, and, where clinically and ethically feasible, at recurrence. TCR and BCR clonotype tracking combined with spatial transcriptomics or multiplex imaging could determine whether recurrence reflects survival of pathogenic lymphocytes, persistence of an immune-organizing niche, or reconstruction of both.

Clinical Implications and Future Directions

No current evidence supports routine immune profiling or immune-subtype-based treatment selection in OLP. T_RM-associated features have been linked to erosive OLP and recurrent erosion [7], whereas TLS-associated humoral activation has been associated with erosive and recurrent phenotypes [11]. CXCL13⁺ cytotoxic CD8⁺ T-cell states have also been identified in mucosal lichen planus without a prominent canonical T_RM signature [16]. These findings raise the possibility that lesions with similar clinicopathological appearances contain different immune architectures, but none of these patterns has been prospectively validated as a biomarker of recurrence or treatment response [7,11,16].

Available treatment-response studies remain exploratory. Among patients receiving systemic corticosteroids, B-cell-positive OLP and oral lichenoid lesions showed greater reductions in reticulation, erythema, and ulceration scores, and B-cell infiltration was associated with a more favorable response [30]. Because the study included both OLP and oral lichenoid lesions and did not assign treatment prospectively according to B-cell status, B-cell infiltration cannot currently guide corticosteroid selection [30]. In a phase II open-label study of 12 patients with predominantly cutaneous lichen planus, 83.3% achieved a clinical response to baricitinib at week 16, accompanied by suppression of IFN-associated activity and reduction of CXCL13⁺CD8⁺ T-cells [31]. Only two participants had concomitant mucosal involvement; the study therefore does not establish efficacy or prevention of recurrence in OLP [31].

These pathways are mechanistic candidates, not validated treatment targets. CCL19⁺ fibroblasts and inducible TLSs are associated with atrophic-erosive disease and higher disease scores [9], while TLS-associated B-cell and plasma-cell activation is linked to unfavorable clinical phenotypes [11]. The inferred cDC2A–IL-15–CXCL13⁺CD8⁺ T-cell network and 4-1BB-associated epithelial cytotoxicity provide additional candidates, but much of the functional evidence derives from broader lichen planus cohorts or non-OLP experimental systems [16]. Oral mucosal validation is required before these pathways can support OLP-specific interventional studies.

Future studies should distinguish three outcomes: suppression of active inflammation, persistence of tissue-associated immune architecture after clinical improvement, and prediction of subsequent recurrence. Prospective cohorts should distinguish OLP from oral lichenoid lesions and document clinical phenotype, anatomical site, treatment exposure, and time to recurrence. Where repeated same-site biopsy is clinically and ethically acceptable, paired repertoire analysis could test whether identical T- or B-cell clonotypes remain after treatment, while spatial profiling could assess whether fibroblast subsets, dendritic-cell networks, or TLS architecture persist despite clinical improvement. Minimally invasive blood-, saliva-, or cytology-based measurements may reduce the need for repeated biopsy, but any proposed surrogate must first be validated against the corresponding tissue state.

Conclusion

Current evidence expands the conventional epithelial-interface injury model of OLP to include tissue-associated CD8⁺ T-cell states and local immune organization. OLP-specific studies link T_RM-like CD8⁺ cells to erosive disease and recurrent erosion, whereas broader mucosal lichen planus data identify CXCL13⁺ cytotoxic CD8⁺ T-cell states without a prominent canonical T_RM signature [6,7,16]. Together with CCL19⁺ fibroblasts and TLS-associated humoral responses [9-11], these findings support two non-exclusive explanations for chronicity: survival of pathogenic lymphocytes and persistence or reconstruction of a niche that sustains or regenerates cytotoxic T-cell states. Neither persistence of identical clonotypes, direct T_RM–TLS coupling, nor a causal role of these structures in recurrence has been demonstrated. Longitudinal same-site studies integrating clonotype tracking with spatial profiling are required to identify which components remain after clinical improvement and predict recurrence.